From - Sun Jan 06 19:26:49 2008 X-Account-Key: account2 X-UIDL: <477D4F88.2070504@shaw.ca> X-Mozilla-Status: 0005 X-Mozilla-Status2: 00000000 X-Mozilla-Keys: $label2 Return-Path: Received: from hrndva-mxlb.mail.rr.com ([10.128.255.7]) by hrndva-imta13.mail.rr.com with ESMTP id <20080103225319.JQDG29039.hrndva-imta13.mail.rr.com@hrndva-mxlb.mail.rr.com> for ; Thu, 3 Jan 2008 22:53:19 +0000 X-IronPort: hrndva-mx04.mail.rr.com 296171809 X-RR-Connecting-IP: 128.143.137.19 Received: from ares.cs.virginia.edu ([128.143.137.19]) by hrndva-mxlb.mail.rr.com with ESMTP; 03 Jan 2008 22:53:19 +0000 Received: from pd2mo3so.prod.shaw.ca (idcmail-mo1so.shaw.ca [24.71.223.10]) by ares.cs.Virginia.EDU (8.13.8/8.13.8/UVACS-2007111101) with ESMTP id m03MrFLX009915 for ; Thu, 3 Jan 2008 14:53:15 -0800 (PST) Received: from pd2mr3so.prod.shaw.ca (pd2mr3so-qfe3.prod.shaw.ca [10.0.141.108]) by l-daemon (Sun ONE Messaging Server 6.0 HotFix 1.01 (built Mar 15 2004)) with ESMTP id <0JU300BCB8SCMF30@l-daemon> for mccalpin@cs.virginia.edu; Thu, 03 Jan 2008 14:53:00 -0700 (MST) Received: from pn2ml4so.prod.shaw.ca ([10.0.121.148]) by pd2mr3so.prod.shaw.ca (Sun Java System Messaging Server 6.2-7.05 (built Sep 5 2006)) with ESMTP id <0JU3003HK6UWIU90@pd2mr3so.prod.shaw.ca> for mccalpin@cs.virginia.edu; Thu, 03 Jan 2008 14:11:22 -0700 (MST) Received: from [192.168.1.108] ([68.147.43.233]) by l-daemon (Sun ONE Messaging Server 6.0 HotFix 1.01 (built Mar 15 2004)) with ESMTP id <0JU300F686UX8T60@l-daemon> for mccalpin@cs.virginia.edu; Thu, 03 Jan 2008 14:11:21 -0700 (MST) Date: Thu, 03 Jan 2008 14:11:36 -0700 From: Doug Fortune Subject: stream results, AMD Athlon X2 4600+ dual core To: "John D. McCalpin" Message-id: <477D4F88.2070504@shaw.ca> MIME-version: 1.0 Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 7bit User-Agent: Thunderbird 1.5.0.14pre (X11/20071023) re: http://www.cs.virginia.edu/stream/#PeeCeeResults Prof McCalpin: I don't know if you are still collecting results, but here they are for my AMD Athlon X2 4600+ dual core Linux box. I have installed the fastest memory allowed by this motherboard. Also, soon to arrive is my new HP Proliant DualCPU Quadcore DL360 G5 server, I notice you don't have any results for the Generation 5 box. Let me know if you are interested, and I'll forward that bench to you when I can. cheers Doug Fortune Pentam Aerospace Calgary gcc -O3 stream.c -o stream ./stream ------------------------------------------------------------- STREAM version $Revision: 5.8 $ . . . ------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 3728.3001 0.0088 0.0086 0.0093 Scale: 3693.0179 0.0091 0.0087 0.0096 Add: 3993.6767 0.0122 0.0120 0.0127 Triad: 4052.3428 0.0122 0.0118 0.0128 ------------------------------------------------------------- Solution Validates ------------------------------------------------------------- doug@warp5:~$ cat /proc/cpuinfo processor : 0 vendor_id : AuthenticAMD cpu family : 15 model : 75 model name : AMD Athlon(tm) 64 X2 Dual Core Processor 4600+ stepping : 2 cpu MHz : 2400.000 cache size : 512 KB physical id : 0 siblings : 2 core id : 0 cpu cores : 2 fdiv_bug : no hlt_bug : no f00f_bug : no coma_bug : no fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt lm 3dnowext 3dnow pni cx16 lahf_lm cmp_legacy svm cr8legacy ts fid vid ttp tm stc bogomips : 5000.18 processor : 1 vendor_id : AuthenticAMD cpu family : 15 model : 75 model name : AMD Athlon(tm) 64 X2 Dual Core Processor 4600+ stepping : 2 cpu MHz : 2400.000 cache size : 512 KB physical id : 0 siblings : 2 core id : 1 cpu cores : 2 fdiv_bug : no hlt_bug : no f00f_bug : no coma_bug : no fpu : yes fpu_exception : yes cpuid level : 1 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt lm 3dnowext 3dnow pni cx16 lahf_lm cmp_legacy svm cr8legacy ts fid vid ttp tm stc bogomips : 5000.18 From - Wed Jan 30 10:41:39 2008 X-Account-Key: account2 X-UIDL: <479FD027.7030005@azulsystems.com> X-Mozilla-Status: 0015 X-Mozilla-Status2: 10000000 X-Mozilla-Keys: $label2 Return-Path: Received: from hrndva-mxlb.mail.rr.com ([10.128.255.4]) by hrndva-imta02.mail.rr.com with ESMTP id <20080130011740.YKYP2031.hrndva-imta02.mail.rr.com@hrndva-mxlb.mail.rr.com> for ; Wed, 30 Jan 2008 01:17:40 +0000 X-IronPort: hrndva-mx01.mail.rr.com 378973193 X-RR-Connecting-IP: 209.217.50.103 Received: from smtp-andy.catalog.com (HELO andy.catalog.com) ([209.217.50.103]) by hrndva-mxlb.mail.rr.com with SMTP; 30 Jan 2008 01:17:40 +0000 Received: (qmail 11679 invoked by uid 10000); 30 Jan 2008 01:17:38 -0000 Delivered-To: mccalpin.com-john@mccalpin.com Received: (qmail 11673 invoked by uid 0); 30 Jan 2008 01:17:38 -0000 Received: from unknown (HELO spamfilter.catalog.com) (209.217.50.108) by smtp3.catalog.com with SMTP; 30 Jan 2008 01:17:38 -0000 X-ASG-Debug-ID: 1201655853-29ec029d0000-TpEisi X-Barracuda-URL: http://209.217.50.108:8000/cgi-bin/mark.cgi Received: from apollo.azulsystems.com (localhost [127.0.0.1]) by spamfilter.catalog.com (Spam Firewall) with ESMTP id 0A170CE0FF for ; Tue, 29 Jan 2008 19:17:33 -0600 (CST) Received: from apollo.azulsystems.com (mail.azulsystems.com [208.66.102.26]) by spamfilter.catalog.com with ESMTP id GHf8uBCmdYEPCU1v for ; Tue, 29 Jan 2008 19:17:33 -0600 (CST) Received: from apollo.azulsystems.com (localhost [127.0.0.1]) by apollo.azulsystems.com (Postfix) with ESMTP id 2BAAB16265 for ; Tue, 29 Jan 2008 17:22:19 -0800 (PST) Received: from XCHMTV1.azulsystems.com (gw-10-10-110.azulsystems.com [10.10.110.1]) by apollo.azulsystems.com (Postfix) with ESMTP id 1E0A516247 for ; Tue, 29 Jan 2008 17:22:19 -0800 (PST) Received: from [10.10.11.215] ([10.10.11.215]) by XCHMTV1.azulsystems.com over TLS secured channel with Microsoft SMTPSVC(6.0.3790.3959); Tue, 29 Jan 2008 17:17:27 -0800 Message-ID: <479FD027.7030005@azulsystems.com> Date: Tue, 29 Jan 2008 17:17:27 -0800 From: Cliff Click Organization: Azul Systems User-Agent: Thunderbird 2.0.0.9 (Windows/20071031) MIME-Version: 1.0 To: john@mccalpin.com Cc: David Lebsack X-ASG-Orig-Subj: Re: Java-based STREAM result for Azul? Subject: Re: Java-based STREAM result for Azul? References: <47602BC4.1090206@acm.org> <4798CEB0.7060908@austin.rr.com> <479928CD.1030804@azulsystems.com> <479A077D.7020308@austin.rr.com> In-Reply-To: <479A077D.7020308@austin.rr.com> Content-Type: multipart/mixed; boundary="------------010300060000060706000206" X-OriginalArrivalTime: 30 Jan 2008 01:17:27.0340 (UTC) FILETIME=[E0794EC0:01C862DD] X-Virus-Scanned: ClamAV using ClamSMTP X-Barracuda-Connect: mail.azulsystems.com[208.66.102.26] X-Barracuda-Start-Time: 1201655855 X-Barracuda-Virus-Scanned: by Catalog.com Spam Firewall at catalog.com X-Barracuda-Spam-Score: 0.20 X-Barracuda-Spam-Status: No, SCORE=0.20 using per-user scores of TAG_LEVEL=4.0 QUARANTINE_LEVEL=1000.0 KILL_LEVEL=5.0 tests=BSF_SC7_SA015e, HTML_MESSAGE X-Barracuda-Spam-Report: Code version 3.1, rules version 3.1.40830 Rule breakdown below pts rule name description ---- ---------------------- -------------------------------------------------- 0.00 HTML_MESSAGE BODY: HTML included in message 0.20 BSF_SC7_SA015e Custom Rule SA015e This is a multi-part message in MIME format. --------------010300060000060706000206 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Ok, I did a Java version "The Right Way" - parallelize only around the individual loops, just like the OpenMP pragmas suggest, and a full barrier between loops.  In order to get decent runtimes for each loop (and not get swamped by thread start/stop - since I'm use a real "os puts these threads to sleep" barrier) I raised N to 10billion elements.  This makes an array larger than Java can express (limit to 2billion elements), so I made the actual timing loops double-nested: a total of 10billion elements get touched, but I end up repeating over the same elements several times.

The invariant is: N == N1 *  a.length

In the spirit of Java (and not Fortran MP) I stripe across the loops expressly.
Inner loops look like:

        for( int i=0; i<N1; i++ )     // repeat inner loop, to make runtime longer
          for(int j=min; j<max; j++)  // striped per CPU, still much larger than all caches
            a[j] = b[j]+scalar*c[j];  // original loop body

with a little Java boilerplate around it to control the parallelism.
I also tossed out the first 3 runs as warm-up, to give the JIT a chance to profile & compile optimized.
Other than these changes, I made as clean a translation into Java from C as I could; the printing and formatting remains all the same.  So does the verify, etc.  Source code included.

I hope this result is acceptable to you; I would very much like to see an Azul result on the STREAM webpage.
Let me know what you think!

Thanks,
Cliff


[swx2021:bugs] java -Xmx25g -Xms25g -Xmn22g -XX:+UseParallelGC stream
-------------------------------------------------------------
STREAM version $Revision: 5.8 $
-------------------------------------------------------------
This system uses 8 bytes per DOUBLE PRECISION word.
-------------------------------------------------------------
Array size = 10000000000, Offset = 125000
Total memory required = 228881.8 MB.
Each test is run 20 times, but only
the *best* time for each is used.
-------------------------------------------------------------
Number of Threads requested = 730
-------------------------------------------------------------
Printing one line per active thread....
-------------------------------------------------------------
Your clock granularity/precision appears to be 2 microseconds.
Each test below will take on the order of 2373563 microseconds.
   (= 1186781 clock ticks)
Increase the size of the arrays if this shows that
you are not getting at least 20 clock ticks per test.
-------------------------------------------------------------
WARNING -- The above is only a rough guideline.
For best results, please be sure you know the
precision of your system timer.
-------------------------------------------------------------
Function      Rate (MB/s)   Avg time     Min time     Max time
Copy:       83968.3776       1.7372       1.9055       2.0023
Scale:      83223.5639       1.7872       1.9225       2.0820
Add:        90757.0839       2.3871       2.6444       2.7425
Triad:      91394.1568       2.3662       2.6260       2.6663
-
-------------------------------------------------------------
Solution Validates
-------------------------------------------------------------



John D. McCalpin, Ph.D. wrote:
This is why my timers are external to the parallel sections -- I count the wall time from before the first thread starts until after the last thread stops.   Bad scheduling can make performance worse, but it cannot make performance "better".   Of course the difficulty of describing what you want the computer to do in these cases is one reason why creating more threads than processors is uncommon in homogeneous throughput applications.

Cliff Click wrote:
Looking over this benchmark more (the Java version & the C version I port'd from - NOT necessarily anything from your website), there's an obvious flaw in the multi-threaded measurements: the many threads are not required to run simultaneously.  They are all released from the barrier at the same time of course, but the OS is free to not bother to give them any cycles.  I can observe this unfair behavior easily on my 2-way linux box.

The effect is that if I launch 1000 threads on my 2-way box, I'll get essentially 500 pairs-of-threads running, where each pair gets the entire bandwidth of the machine.  Sometimes during the outer-loop best-of-10 tries the OS will switch threads and some poor loser thread will get a really sucky time for that try.  But unless the OS switches him out on all 10 attempts it's likely he'll get at least 1 run in at the full bandwidth of the machine.  The best score per thread is reported & summed; the "sucky loser" score that realistically represents a lack of machine resources is tossed out.

Hence the final score will be 500 pairs, each reporting 3G/sec or 1500 G/sec on my el-cheapo linux blade.  Blah.  If I switch to "avgtime" instead of "mintime", and report bandwidth as (total memory moved) / (total time across all threads)  I get a much more reasonable result that waffles around between 3G/sec to 4G/sec total depending on the phase of the moon.

This next version reports a respectable 86G/sec on Azul's currently-shipping hardware.
Sample output and source code included.
I'm open to suggestions on improving the measurement.

Thanks,
Cliff


[swx2021:bugs] java -Xmx50g -Xms50g -XX:+UseParallelGC STREAM 750
-------------------------------------------------------------
This system uses 8 bytes per DOUBLE PRECISION word.
Array size = 2000000, Offset = 0
Total memory required = 32043.45703125 MB.
Each test is run 10 times, but only the *best* time for each is used.
=== Warmup 0
PID 0 begins setup (0 secs since last report)
PID 0 begins to init array (0 secs since last report)
PID 0 waits on lock (0 secs since last report)
PID 0 starts main loop (0 secs since last report)
PID: 0 Copy: 1075.8308521769286MB/sec  Scale: 587.273524049533MB/sec  Add: 933.9771012072665MB/sec  Triad: 603.0731362242764MB/sec 
PID 0 exits (2 secs since last report)
=== Warmup 1
PID 0 begins setup (0 secs since last report)
PID 0 begins to init array (0 secs since last report)
.....

PID: 662 PID 621 exits (17 secs since last report)
Copy: 98.767475684564MB/sec  Scale: 98.42488154787503MB/sec  Add: 117.56838098470504MB/sec  Triad: 115.49884828210581MB/sec 
PID: 35 PID 662 exits (17 secs since last report)
Copy: 110.98078010891447MB/sec  Scale: 107.66635593588056MB/sec  Add: 122.13051560896352MB/sec  Triad: 126.85864547720726MB/sec 
PID 35 exits (17 secs since last report)
PID: 171 Copy: 114.38891792050323MB/sec  Scale: 106.18831294828703MB/sec  Add: 129.38213050929102MB/sec  Triad: 130.12129703049794MB/sec 
PID 171 exits (17 secs since last report)
PID: 389 Copy: 104.5781427749813MB/sec  Scale: 109.57665020255511MB/sec  Add: 122.10444364468034MB/sec  Triad: 122.04530606238013MB/sec 
PID 389 exits (17 secs since last report)
=== Caught the last thread.
Average cpu bandwidth:  Copy: 108MB/sec/cpu Scale: 108MB/sec/cpu Add: 123MB/sec/cpu Triad: 123MB/sec/cpu
Total system bandwidth: Copy: 76287MB/sec  Scale: 75932MB/sec  Add: 86282MB/sec  Triad: 86448MB/sec 




John D. McCalpin, Ph.D. wrote:
Sorry for the delay -- lost almost two weeks with seasonal allergies and then lost the last week packing and unpacking my office and lab (AMD is in the midst of moving all its Austin folks to a new consolidated campus).

The code below looks good to me, but I don't know very much about Java.   It worked right away on my dual-core Windows box with 1 or 2 threads.

The biggest problem from a philosophical point of view is that all the data is locally allocated, rather than globally shared.   Of course this is also true of the MPI version of STREAM, but it is obvious in that case that there is no other choice, while a multithreaded Java implementation might be expected to operated on a globally shared memory pool (like the Fortran and C versions).

Is it difficult/impossible/slow to build a Java implementation that uses multiple threads to update non-overlapping portions of a single globally accessible address range?

john

Cliff Click wrote:
Can we use a multi-threaded Java port of STREAM to measure bandwidth, and still call it 'STREAM'?
Would you "bless" the code and add the results to your stream results table?

A very plain-vanilla Java port of stream.c running on a 768-way Azul box achieves 70Gb/sec of total system bandwidth - which puts us in the top 20 results in a 14-U sized box.  We'd like to tout that, if we can get your blessing.

Thanks,
Cliff 


-- 
John D. McCalpin, Ph.D.                                              "Dr. Bandwidth"
john@mccalpin.com                                          http://www.streambench.org/
http://home.austin.rr.com/mccalpin/                 http://www.cs.virginia.edu/stream/
--------------010300060000060706000206 Content-Type: text/plain; name="stream.java" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="stream.java" /*-----------------------------------------------------------------------*/ /* Program: Stream */ /* Revision: $Id: stream.c,v 5.8 2007/02/19 23:57:39 mccalpin Exp mccalpin $ */ /* Original code developed by John D. McCalpin */ /* Programmers: John D. McCalpin */ /* Joe R. Zagar */ /* */ /* This program measures memory transfer rates in MB/s for simple */ /* computational kernels coded in C. */ /*-----------------------------------------------------------------------*/ /* Copyright 1991-2005: John D. McCalpin */ /*-----------------------------------------------------------------------*/ /* License: */ /* 1. You are free to use this program and/or to redistribute */ /* this program. */ /* 2. You are free to modify this program for your own use, */ /* including commercial use, subject to the publication */ /* restrictions in item 3. */ /* 3. You are free to publish results obtained from running this */ /* program, or from works that you derive from this program, */ /* with the following limitations: */ /* 3a. In order to be referred to as "STREAM benchmark results", */ /* published results must be in conformance to the STREAM */ /* Run Rules, (briefly reviewed below) published at */ /* http://www.cs.virginia.edu/stream/ref.html */ /* and incorporated herein by reference. */ /* As the copyright holder, John McCalpin retains the */ /* right to determine conformity with the Run Rules. */ /* 3b. Results based on modified source code or on runs not in */ /* accordance with the STREAM Run Rules must be clearly */ /* labelled whenever they are published. Examples of */ /* proper labelling include: */ /* "tuned STREAM benchmark results" */ /* "based on a variant of the STREAM benchmark code" */ /* Other comparable, clear and reasonable labelling is */ /* acceptable. */ /* 3c. Submission of results to the STREAM benchmark web site */ /* is encouraged, but not required. */ /* 4. Use of this program or creation of derived works based on this */ /* program constitutes acceptance of these licensing restrictions. */ /* 5. Absolutely no warranty is expressed or implied. */ /*-----------------------------------------------------------------------*/ import java.util.concurrent.*; /* INSTRUCTIONS: * * 1) Stream requires a good bit of memory to run. Adjust the * value of 'N' (below) to give a 'timing calibration' of * at least 20 clock-ticks. This will provide rate estimates * that should be good to about 5% precision. */ public abstract class stream { static final long N = 10000000000L; // Because Java is limited to 31bit array indices, I loop over the same // array data enough times to reach 'N' total array elements. The actual // array is sized below, and must exceed any cache by a substantial amount. static final int N0 = 10000000; static final int N1 = (int)(N/N0); static final int NTIMES = 20; static final int OFFSET = 125000; /* * 3) Compile the code with full optimization. Many compilers * generate unreasonably bad code before the optimizer tightens * things up. If the results are unreasonably good, on the * other hand, the optimizer might be too smart for me! * * Try compiling with: * cc -O stream_omp.c -o stream_omp * * This is known to work on Cray, SGI, IBM, and Sun machines. * * * 4) Mail the results to mccalpin@cs.virginia.edu * Be sure to include: * a) computer hardware model number and software revision * b) the compiler flags * c) all of the output from the test case. * Thanks! * */ static final String HLINE = "-------------------------------------------------------------\n"; static double a[] = new double[N0+OFFSET]; static double b[] = new double[N0+OFFSET]; static double c[] = new double[N0+OFFSET]; static double avgtime[] = {0.0, 0.0, 0.0, 0.0}; static double maxtime[] = {0.0, 0.0, 0.0, 0.0}; static double mintime[] = {Double.MAX_VALUE,Double.MAX_VALUE,Double.MAX_VALUE,Double.MAX_VALUE}; static final String label[] = {"Copy: ", "Scale: ", "Add: ", "Triad: "}; static final double BytesPerWord = 8.0; static final double bytes[] = { 2 * BytesPerWord * N, 2 * BytesPerWord * N, 3 * BytesPerWord * N, 3 * BytesPerWord * N }; // Worker threads static Thread workers[]; static volatile stream _work; static CyclicBarrier _barrier_start; static CyclicBarrier _barrier_stop; abstract boolean work( final int min, final int max ); void do_par( ) throws InterruptedException, BrokenBarrierException { _work = this; // Inform workers of what to do _barrier_start.await(); // Start workers _barrier_stop .await(); // Main continues when work is done } static public void main( String args[] ) throws InterruptedException, BrokenBarrierException { /* --- SETUP --- determine precision and check timing --- */ System.out.printf(HLINE); System.out.printf("STREAM version $Revision: 5.8 $\n"); System.out.printf(HLINE); System.out.printf("This system uses %d bytes per DOUBLE PRECISION word.\n", (int)BytesPerWord); System.out.printf(HLINE); System.out.printf("Array size = %d, Offset = %d\n" , N, OFFSET); System.out.printf("Total memory required = %.1f MB.\n", (3.0 * BytesPerWord) * ( (double) N / 1048576.0)); System.out.printf("Each test is run %d times, but only\n", NTIMES); System.out.printf("the *best* time for each is used.\n"); System.out.printf(HLINE); final int num_threads = args.length > 0 ? Integer.parseInt(args[0]) : Runtime.getRuntime().availableProcessors(); System.out.printf ("Number of Threads requested = %d\n",num_threads); // Classic cyclic-barriers. Number of threads is +1, so the // workers will await for the main thread before proceeding. _barrier_start = new CyclicBarrier(num_threads+1); _barrier_stop = new CyclicBarrier(num_threads+1); // Array of worker threads workers = new Thread[num_threads]; final int NSLICE = N0/num_threads; for( int i=0; i epsilon) { System.out.printf ("Failed Validation on array a[]\n"); System.out.printf (" Expected : %f \n",aj); System.out.printf (" Observed : %f \n",asum); } else if (Math.abs(bj-bsum)/bsum > epsilon) { System.out.printf ("Failed Validation on array b[]\n"); System.out.printf (" Expected : %f \n",bj); System.out.printf (" Observed : %f \n",bsum); } else if (Math.abs(cj-csum)/csum > epsilon) { System.out.printf ("Failed Validation on array c[]\n"); System.out.printf (" Expected : %f \n",cj); System.out.printf (" Observed : %f \n",csum); } else { System.out.printf ("Solution Validates\n"); } } } --------------010300060000060706000206 Content-Type: text/x-vcard; charset=utf-8; name="cliffc.vcf" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="cliffc.vcf" begin:vcard fn:Dr. Cliff Click n:Click;Cliff org:Azul Systems adr;dom:;;1600 Plymouth St;Mountain View;CA;94043 email;internet:cliffc@acm.org title:Distinguished Engineer tel;work:650-230-6512 tel;home:408-323-8283 tel;cell:408-307-0321 x-mozilla-html:TRUE url:http://www.azulsystems.com version:2.1 end:vcard --------------010300060000060706000206-- From - Mon Apr 07 21:45:36 2008 X-Account-Key: account2 X-UIDL: X-Mozilla-Status: 0005 X-Mozilla-Status2: 00000000 X-Mozilla-Keys: $label2 Return-Path: Received: from hrndva-mxlb.mail.rr.com ([10.128.255.4]) by hrndva-imta06.mail.rr.com with ESMTP id <20080408003506.KZVG16536.hrndva-imta06.mail.rr.com@hrndva-mxlb.mail.rr.com> for ; Tue, 8 Apr 2008 00:35:06 +0000 X-IronPort: hrndva-mx01.mail.rr.com 585437697 X-RR-Connecting-IP: 128.143.137.19 Received: from ares.cs.virginia.edu ([128.143.137.19]) by hrndva-mxlb.mail.rr.com with ESMTP; 08 Apr 2008 00:35:05 +0000 Received: from over.ny.us.ibm.com (over.ny.us.ibm.com [32.97.182.150]) by ares.cs.Virginia.EDU (8.13.8/8.13.8/UVACS-2007111101) with ESMTP id m380Z03d005900 for ; Mon, 7 Apr 2008 17:35:00 -0700 (PDT) Received: from e31.co.us.ibm.com (e31.co.us.ibm.com [32.97.110.149]) by pokfb.esmtp.ibm.com (8.12.11.20060308/8.12.11) with ESMTP id m37NxSGQ017661 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK) for ; Mon, 7 Apr 2008 19:59:29 -0400 Received: from d03relay04.boulder.ibm.com (d03relay04.boulder.ibm.com [9.17.195.106]) by e31.co.us.ibm.com (8.13.8/8.13.8) with ESMTP id m37NxMZU015108 for ; Mon, 7 Apr 2008 19:59:22 -0400 Received: from d03av02.boulder.ibm.com (d03av02.boulder.ibm.com [9.17.195.168]) by d03relay04.boulder.ibm.com (8.13.8/8.13.8/NCO v8.7) with ESMTP id m37NxMv5127800 for ; Mon, 7 Apr 2008 17:59:22 -0600 Received: from d03av02.boulder.ibm.com (loopback [127.0.0.1]) by d03av02.boulder.ibm.com (8.12.11.20060308/8.13.3) with ESMTP id m37NxMjq032674 for ; Mon, 7 Apr 2008 17:59:22 -0600 Received: from d03nm116.boulder.ibm.com (d03nm116.boulder.ibm.com [9.17.195.142]) by d03av02.boulder.ibm.com (8.12.11.20060308/8.12.11) with ESMTP id m37NxMj5032671 for ; Mon, 7 Apr 2008 17:59:22 -0600 To: mccalpin@cs.virginia.edu Cc: jacobt@us.ibm.com MIME-Version: 1.0 Subject: STREAM results for IBM Power 595 (5.0 GHz, POWER6) X-Mailer: Lotus Notes Release 6.0.2CF1 June 9, 2003 Message-ID: From: Ly Vu Date: Mon, 7 Apr 2008 18:59:21 -0500 X-MIMETrack: Serialize by Router on D03NM116/03/M/IBM(Release 8.0|August 02, 2007) at 04/07/2008 17:59:21, Serialize complete at 04/07/2008 17:59:21 Content-Type: multipart/alternative; boundary="=_alternative 0083C6E086257424_=" This is a multipart message in MIME format. --=_alternative 0083C6E086257424_= Content-Type: text/plain; charset="US-ASCII" These are standard STREAM results on an IBM Power 595 with sixty-four 5.0 GHz cores. This is a POWER6 SMP machine. Large pages were used in all cases. Function Rate (MB/s) Avg time Min time Max time Copy: 679207.1738 .0509 .0506 .0515 Scale: 624707.8380 .0554 .0550 .0557 Add: 777334.8234 .0664 .0663 .0666 Triad: 805804.6461 .0640 .0640 .0640 Here is the full output file: -------------------------------------------------- Requesting LARGE pages Running TRIAD kernel Setting up for 64 chips Setting up for 64 threads Number of segments per array = 64 Reading cpu binding list for page placement DATA binding list : 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 3 3 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 Beginning allocation of data segments.... GETSHRSEG: requesting large pages GETSHRSEG ENTRY: shmgetflag -2147481216 Starting Initialization Done With Initialization a(1) 1.00000000000000000 a(M) 1.00000000000000000 b(1) 1.00000000000000000 b(M) 1.00000000000000000 c(1) 1.00000000000000000 c(M) 1.00000000000000000 Moff = 75775 Incremental Offset = 706 504403158265495552 504403175445370376 504403192625245208 ---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- Array size = 2147407873 The total memory requirement is 49150 MB You are running each test 5 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------------- Your clock granularity appears to be less than one microsecond Your clock granularity/precision appears to be 1 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 679207.1738 .0509 .0506 .0515 Scale: 624707.8380 .0554 .0550 .0557 Add: 777334.8234 .0664 .0663 .0666 Triad: 805804.6461 .0640 .0640 .0640 ---------------------------------------------------- Solution Validates! ---------------------------------------------------- --=_alternative 0083C6E086257424_= Content-Type: text/html; charset="US-ASCII"
These are standard STREAM results on an IBM Power 595
with sixty-four 5.0 GHz cores. This is a POWER6 SMP machine.
Large pages were used in all cases.

Function     Rate (MB/s)  Avg time   Min time  Max time
Copy:           679207.1738             .0509             .0506             .0515
Scale:          624707.8380             .0554             .0550             .0557
Add:            777334.8234             .0664             .0663             .0666
Triad:          805804.6461             .0640             .0640             .0640

Here is the full output file:
--------------------------------------------------
 Requesting LARGE pages
 Running TRIAD kernel
 Setting up for  64  chips
 Setting up for  64  threads
 Number of segments per array =  64
 Reading cpu binding list for page placement
 DATA binding list :  0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 3
3 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
 Beginning allocation of data segments....
GETSHRSEG: requesting large pages
GETSHRSEG ENTRY: shmgetflag -2147481216

 Starting Initialization
 Done With Initialization
 a(1) 1.00000000000000000
 a(M) 1.00000000000000000
 b(1) 1.00000000000000000
 b(M) 1.00000000000000000
 c(1) 1.00000000000000000
 c(M) 1.00000000000000000

 Moff =  75775
 Incremental Offset =  706
 504403158265495552 504403175445370376 504403192625245208
----------------------------------------------
 Double precision appears to have 16 digits of accuracy
 Assuming 8 bytes per DOUBLE PRECISION word
----------------------------------------------
 Array size = 2147407873
 The total memory requirement is   49150 MB
 You are running each test   5 times
 --
 The *best* time for each test is used
 *EXCLUDING* the first and last iterations
 ----------------------------------------------------
 Your clock granularity appears to be less than one microsecond
 Your clock granularity/precision appears to be          1 microseconds
 ----------------------------------------------------
Function     Rate (MB/s)  Avg time   Min time  Max time
Copy:           679207.1738             .0509             .0506             .0515
Scale:          624707.8380             .0554             .0550             .0557
Add:            777334.8234             .0664             .0663             .0666
Triad:          805804.6461             .0640             .0640             .0640
 ----------------------------------------------------
 Solution Validates!
 ---------------------------------------------------- --=_alternative 0083C6E086257424_=-- From - Mon Apr 07 21:45:36 2008 X-Account-Key: account2 X-UIDL: X-Mozilla-Status: 0005 X-Mozilla-Status2: 00000000 X-Mozilla-Keys: $label2 Return-Path: Received: from hrndva-mxlb.mail.rr.com ([10.128.255.13]) by hrndva-imta07.mail.rr.com with ESMTP id <20080408000156.TNSL27158.hrndva-imta07.mail.rr.com@hrndva-mxlb.mail.rr.com> for ; Tue, 8 Apr 2008 00:01:56 +0000 X-IronPort: hrndva-mx10.mail.rr.com 446907233 X-RR-Connecting-IP: 128.143.137.19 Received: from ares.cs.virginia.edu ([128.143.137.19]) by hrndva-mxlb.mail.rr.com with ESMTP; 08 Apr 2008 00:01:56 +0000 Received: from e34.co.us.ibm.com (e34.co.us.ibm.com [32.97.110.152]) by ares.cs.Virginia.EDU (8.13.8/8.13.8/UVACS-2007111101) with ESMTP id m3801oOE006875 for ; Mon, 7 Apr 2008 17:01:51 -0700 (PDT) Received: from d03relay02.boulder.ibm.com (d03relay02.boulder.ibm.com [9.17.195.227]) by e34.co.us.ibm.com (8.13.8/8.13.8) with ESMTP id m37NxqMp008150 for ; Mon, 7 Apr 2008 19:59:52 -0400 Received: from d03av02.boulder.ibm.com (d03av02.boulder.ibm.com [9.17.195.168]) by d03relay02.boulder.ibm.com (8.13.8/8.13.8/NCO v8.7) with ESMTP id m3801jNg220946 for ; Mon, 7 Apr 2008 18:01:45 -0600 Received: from d03av02.boulder.ibm.com (loopback [127.0.0.1]) by d03av02.boulder.ibm.com (8.12.11.20060308/8.13.3) with ESMTP id m3801iEQ007531 for ; Mon, 7 Apr 2008 18:01:44 -0600 Received: from d03nm116.boulder.ibm.com (d03nm116.boulder.ibm.com [9.17.195.142]) by d03av02.boulder.ibm.com (8.12.11.20060308/8.12.11) with ESMTP id m3801ijE007523 for ; Mon, 7 Apr 2008 18:01:44 -0600 To: mccalpin@cs.virginia.edu Cc: Jacob Thomas MIME-Version: 1.0 Subject: STREAM results for IBM Power 575 (4.7GHz, POWER6) X-Mailer: Lotus Notes Release 6.0.2CF1 June 9, 2003 Message-ID: From: Ly Vu Date: Mon, 7 Apr 2008 19:01:44 -0500 X-MIMETrack: Serialize by Router on D03NM116/03/M/IBM(Release 8.0|August 02, 2007) at 04/07/2008 18:01:44, Serialize complete at 04/07/2008 18:01:44 Content-Type: multipart/alternative; boundary="=_alternative 000028A986257425_=" This is a multipart message in MIME format. --=_alternative 000028A986257425_= Content-Type: text/plain; charset="US-ASCII" These are standard STREAM results on an IBM Power 575 with thirty-two 4.7 GHz cores. This is a POWER6 SMP machine. Large pages were used in all cases. Function Rate (MB/s) Avg time Min time Max time Copy: 142708.1898 .1204 .1204 .1205 Scale: 142612.4358 .1205 .1205 .1205 Add: 159010.6786 .1624 .1621 .1628 Triad: 162844.0220 .1585 .1582 .1588 Here is the full output file: -------------------------------------------------- Environment variable MEMSUITE set to DETAILS Requesting LARGE pages Running TRIAD kernel Setting up for 32 chips Setting up for 32 threads Number of segments per array = 32 Reading cpu binding list for page placement DATA binding list : 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Beginning allocation of data segments.... GETSHRSEG: requesting large pages GETSHRSEG ENTRY: shmgetflag -2147481216 Moff = 75776 Incremental Offset = 678 504403158265495552 504403166855435560 504403175445375576 ---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- Array size = 1073666048 The total memory requirement is 24574 MB You are running each test 5 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------------- Your clock granularity appears to be less than one microsecond Your clock granularity/precision appears to be 1 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 142708.1898 .1204 .1204 .1205 Scale: 142612.4358 .1205 .1205 .1205 Add: 159010.6786 .1624 .1621 .1628 Triad: 162844.0220 .1585 .1582 .1588 ---------------------------------------------------- Solution Validates! ---------------------------------------------------- --=_alternative 000028A986257425_= Content-Type: text/html; charset="US-ASCII"
These are standard STREAM results on an IBM Power 575
with thirty-two 4.7 GHz cores. This is a POWER6 SMP machine.
Large pages were used in all cases.

Function     Rate (MB/s)  Avg time   Min time  Max time
Copy:           142708.1898             .1204             .1204             .1205
Scale:          142612.4358             .1205             .1205             .1205
Add:            159010.6786             .1624             .1621             .1628
Triad:          162844.0220             .1585             .1582             .1588

Here is the full output file:
--------------------------------------------------
 Environment variable MEMSUITE set to DETAILS
 Requesting LARGE pages
 Running TRIAD kernel
 Setting up for  32  chips
 Setting up for  32  threads
 Number of segments per array =  32
 Reading cpu binding list for page placement
 DATA binding list :  0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
 Beginning allocation of data segments....
GETSHRSEG: requesting large pages
GETSHRSEG ENTRY: shmgetflag -2147481216

 Moff =  75776
 Incremental Offset =  678
 504403158265495552 504403166855435560 504403175445375576
----------------------------------------------
 Double precision appears to have 16 digits of accuracy
 Assuming 8 bytes per DOUBLE PRECISION word
----------------------------------------------
 Array size = 1073666048
 The total memory requirement is   24574 MB
 You are running each test   5 times
 --
 The *best* time for each test is used
 *EXCLUDING* the first and last iterations
 ----------------------------------------------------
 Your clock granularity appears to be less than one microsecond
 Your clock granularity/precision appears to be          1 microseconds
 ----------------------------------------------------
Function     Rate (MB/s)  Avg time   Min time  Max time
Copy:           142708.1898             .1204             .1204             .1205
Scale:          142612.4358             .1205             .1205             .1205
Add:            159010.6786             .1624             .1621             .1628
Triad:          162844.0220             .1585             .1582             .1588
 ----------------------------------------------------
 Solution Validates!
 ---------------------------------------------------- --=_alternative 000028A986257425_=-- From - Sat May 03 10:06:03 2008 X-Account-Key: account2 X-UIDL: <481A5688.6000205@yottayotta.com> X-Mozilla-Status: 0005 X-Mozilla-Status2: 00000000 X-Mozilla-Keys: $label2 Return-Path: Received: from hrndva-mxlb.mail.rr.com ([10.128.255.5]) by hrndva-imta07.mail.rr.com with ESMTP id <20080501234726.HWPI27158.hrndva-imta07.mail.rr.com@hrndva-mxlb.mail.rr.com> for ; Thu, 1 May 2008 23:47:26 +0000 X-IronPort: hrndva-mx02.mail.rr.com 632622235 X-RR-Connecting-IP: 209.217.50.103 Received: from smtp-andy.catalog.com (HELO andy.catalog.com) ([209.217.50.103]) by hrndva-mxlb.mail.rr.com with SMTP; 01 May 2008 23:47:25 +0000 Received: (qmail 17311 invoked by uid 10000); 1 May 2008 23:47:25 -0000 Delivered-To: mccalpin.com-john@mccalpin.com Received: (qmail 17300 invoked by uid 0); 1 May 2008 23:47:25 -0000 Received: from unknown (HELO spamfilter.catalog.com) (209.217.52.101) by smtp3.catalog.com with SMTP; 1 May 2008 23:47:25 -0000 X-ASG-Debug-ID: 1209685631-783a01600000-TpEisi X-Barracuda-URL: http://209.217.52.101:8000/cgi-bin/mark.cgi Received: from mx1.yottayotta.com (localhost [127.0.0.1]) by spamfilter.catalog.com (Spam Firewall) with ESMTP id DDCBCD2FF4AA for ; Thu, 1 May 2008 18:47:11 -0500 (CDT) Received: from mx1.yottayotta.com ([198.161.246.32]) by spamfilter.catalog.com with ESMTP id EbLGkIlUWyJckbEm for ; Thu, 01 May 2008 18:47:11 -0500 (CDT) Received: from [192.168.1.2] (helo=edm-exchange.yottayotta.com) by mx1.yottayotta.com with esmtp (Exim 4.63) (envelope-from ) id 1JriZV-0008Ub-JZ for john@mccalpin.com; Thu, 01 May 2008 17:52:17 -0600 Received: from [10.0.3.198] ([10.0.3.198]) by edm-exchange.yottayotta.com with Microsoft SMTPSVC(5.0.2195.6713); Thu, 1 May 2008 17:47:20 -0600 Message-ID: <481A5688.6000205@yottayotta.com> Date: Thu, 01 May 2008 17:47:20 -0600 From: Chris Youb User-Agent: Mozilla Thunderbird 1.0 (X11/20041206) X-Accept-Language: en-us, en MIME-Version: 1.0 To: john@mccalpin.com X-ASG-Orig-Subj: STREAM: Result set - Dell PowerEdge 1950 III - 2 x Intel Xeon Quad Core E5440 Subject: STREAM: Result set - Dell PowerEdge 1950 III - 2 x Intel Xeon Quad Core E5440 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 01 May 2008 23:47:20.0896 (UTC) FILETIME=[B2662C00:01C8ABE5] X-Barracuda-Connect: UNKNOWN[198.161.246.32] X-Barracuda-Start-Time: 1209685631 X-Barracuda-Bayes: INNOCENT GLOBAL 0.8401 1.0000 2.6087 X-Barracuda-Virus-Scanned: by Catalog.com Spam Firewall at catalog.com X-Barracuda-Spam-Score: 2.71 X-Barracuda-Spam-Status: No, SCORE=2.71 using per-user scores of TAG_LEVEL=4.0 QUARANTINE_LEVEL=1000.0 KILL_LEVEL=5.0 tests=RDNS_NONE X-Barracuda-Spam-Report: Code version 3.1, rules version 3.1.49403 Rule breakdown below pts rule name description ---- ---------------------- -------------------------------------------------- 0.10 RDNS_NONE Delivered to trusted network by a host with no rDNS Here are my results from running STREAM on the configuration listed below. For the most part the results are tightly grouped with little variation. I noticed that I achieved better results as I increased the number of OMP_NUM_THREADS, until they peaked at 8. Incidently this is the number of cores in my system (2 x Quad core). However, on my dual CPU AMD 2212HE based system the results are not very consistent at all. When using identical software and configuration to that listed below, my results can range from 3500MB/s to 7700MB/s. Do you have any advice or suggestions about what could be wrong? -------------------------------------------------------------- Hardware: - Dell PowerEdge 1950 III - Two (2) Intel Xeon Quad Core E5440 - Memory stick confirguration (from factory): 2G x 0 x 2G x 0 x 2G x 0 x 2G x 0 = 8G total memory -------------------------------------------------------------- Operating System: - Ubuntu 8.04 AMD64 Desktop Compiled as follows: gcc -O3 -fopenmp -D_OPENMP stream.c -o stream Run as follows: export OMP_NUM_THREADS=8 ./stream -------------------------------------------------------------- Results: [cyoub@cyoub.edmonton.yottayotta.com]$ cat results ------------------------------------------------------------- STREAM version $Revision: 5.8 $ ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 2000000, Offset = 0 Total memory required = 45.8 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Number of Threads requested = 8 ------------------------------------------------------------- Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 3380 microseconds. (= 3380 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 6341.4944 0.0051 0.0050 0.0051 Scale: 5992.3979 0.0054 0.0053 0.0054 Add: 5572.2832 0.0086 0.0086 0.0086 Triad: 5771.3161 0.0083 0.0083 0.0083 ------------------------------------------------------------- Solution Validates ------------------------------------------------------------- ------------------------------------------------------------- STREAM version $Revision: 5.8 $ ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 2000000, Offset = 0 Total memory required = 45.8 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Number of Threads requested = 8 ------------------------------------------------------------- Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 3644 microseconds. (= 3644 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 6282.1310 0.0051 0.0051 0.0051 Scale: 5982.2485 0.0054 0.0053 0.0056 Add: 5561.3544 0.0087 0.0086 0.0087 Triad: 5761.5715 0.0083 0.0083 0.0084 ------------------------------------------------------------- Solution Validates ------------------------------------------------------------- ------------------------------------------------------------- STREAM version $Revision: 5.8 $ ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 2000000, Offset = 0 Total memory required = 45.8 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Number of Threads requested = 8 ------------------------------------------------------------- Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 5143 microseconds. (= 5143 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 6268.3415 0.0054 0.0051 0.0071 Scale: 5839.3617 0.0056 0.0055 0.0060 Add: 5910.5922 0.0084 0.0081 0.0085 Triad: 5944.9754 0.0082 0.0081 0.0083 ------------------------------------------------------------- Solution Validates ------------------------------------------------------------- ------------------------------------------------------------- STREAM version $Revision: 5.8 $ ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 2000000, Offset = 0 Total memory required = 45.8 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Number of Threads requested = 8 ------------------------------------------------------------- Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 2902 microseconds. (= 2902 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 6566.7463 0.0051 0.0049 0.0053 Scale: 6043.3936 0.0054 0.0053 0.0056 Add: 5772.8055 0.0086 0.0083 0.0087 Triad: 5993.2898 0.0082 0.0080 0.0084 ------------------------------------------------------------- Solution Validates ------------------------------------------------------------- ------------------------------------------------------------- STREAM version $Revision: 5.8 $ ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 2000000, Offset = 0 Total memory required = 45.8 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Number of Threads requested = 8 ------------------------------------------------------------- Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 5073 microseconds. (= 5073 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 6210.0462 0.0058 0.0052 0.0074 Scale: 5911.6335 0.0061 0.0054 0.0078 Add: 5853.0276 0.0090 0.0082 0.0104 Triad: 5756.6292 0.0087 0.0083 0.0106 ------------------------------------------------------------- Solution Validates ------------------------------------------------------------- From - Sat May 17 10:10:24 2008 X-Account-Key: account2 X-UIDL: <482D0230.6D4B@Yahoo.com> X-Mozilla-Status: 0005 X-Mozilla-Status2: 10000000 X-Mozilla-Keys: $label2 Return-Path: Received: from hrndva-mxlb.mail.rr.com ([10.128.255.21]) by hrndva-imta01.mail.rr.com with ESMTP id <20080516044037.SDZR1943.hrndva-imta01.mail.rr.com@hrndva-mxlb.mail.rr.com> for ; Fri, 16 May 2008 04:40:37 +0000 X-IronPort: hrndva-mx18.mail.rr.com 1597182083 X-RR-Connecting-IP: 209.217.50.103 Received: from smtp-andy.catalog.com (HELO andy.catalog.com) ([209.217.50.103]) by hrndva-mxlb.mail.rr.com with SMTP; 16 May 2008 04:40:37 +0000 Received: (qmail 14949 invoked by uid 10000); 16 May 2008 04:40:36 -0000 Delivered-To: mccalpin.com-john@mccalpin.com Received: (qmail 14928 invoked by uid 0); 16 May 2008 04:40:36 -0000 Received: from unknown (HELO spamfilter.catalog.com) (209.217.36.23) by smtp3.catalog.com with SMTP; 16 May 2008 04:40:36 -0000 X-ASG-Debug-ID: 1210912801-450503470000-TpEisi X-Barracuda-URL: http://209.217.36.23:8000/cgi-bin/mark.cgi Received: from smtp4.telebecinternet.com (localhost [127.0.0.1]) by spamfilter.catalog.com (Spam Firewall) with ESMTP id 37B2D232460 for ; Thu, 15 May 2008 23:40:01 -0500 (CDT) Received: from smtp4.telebecinternet.com (smtp4.bellnordiq.ca [142.217.217.73]) by spamfilter.catalog.com with ESMTP id V1FUrqKvCBnzfiFf for ; Thu, 15 May 2008 23:40:01 -0500 (CDT) Received: from Saratoga (NTL208H101-89-192.nt.net [208.101.89.192]) by smtp4.telebecinternet.com (8.14.2/8.13.1) with SMTP id m4G4eKie005168; Fri, 16 May 2008 00:40:20 -0400 Message-ID: <482D0230.6D4B@Yahoo.com> Date: Thu, 15 May 2008 23:40:32 -0400 From: Michael Lancop Reply-To: MichaelandShoSho@Yahoo.com X-Mailer: Mozilla 3.04 (Win95; U) MIME-Version: 1.0 To: john@mccalpin.com CC: MichaelandShoSho@Yahoo.com X-ASG-Orig-Subj: Stream for Windows Subject: Stream for Windows Content-Type: multipart/mixed; boundary="------------597E6A4842B8" X-Scanned-By: MIMEDefang 2.63 on 192.168.150.24 X-Barracuda-Connect: smtp4.bellnordiq.ca[142.217.217.73] X-Barracuda-Start-Time: 1210912802 X-Barracuda-Virus-Scanned: by Catalog.com Spam Firewall at catalog.com X-Barracuda-Spam-Score: 0.20 X-Barracuda-Spam-Status: No, SCORE=0.20 using per-user scores of TAG_LEVEL=4.0 QUARANTINE_LEVEL=1000.0 KILL_LEVEL=5.0 tests=BSF_SC0_MJ615 X-Barracuda-Spam-Report: Code version 3.1, rules version 3.1.45799 Rule breakdown below pts rule name description ---- ---------------------- -------------------------------------------------- 0.20 BSF_SC0_MJ615 Custom Rule MJ615 This is a multi-part message in MIME format. --------------597E6A4842B8 Content-Type: application/x-zip-compressed; name="Stream.zip" Content-Transfer-Encoding: base64 Content-Disposition: inline; filename="Stream.zip" UEsDBBQAAAAIAO8MsDhEbLyOHVMAAADQAAAKAAAAU3RyZWFtLmV4ZexZb3Qb1ZV/kkb2JJEi JdjFkASURaGBJBwXJ1unjoNcIsdp41aOsRyV2IElMcIn/J+h2XPsRN6RWI8HQWjZQlqWhjXs 9mw5e+i2UEPTRK6CHZuwmJCmTpGCgbSMkNk14NoKcTL7u2/k4GzP7n7aD3uO3/HMu+++e++7 97777rsj135nH7MxxgQ8hsFYDzObj/3vbQjP/Ktfnc9+MeeNpT2WzW8svSV814Oe+x64984H br/bc8ft99xzr+T5q52eB+R7PHfd49nw7XrP3ffu2HmD0znXm5dx7idNPz08UJabftb9tjL3 M/TvZ8tzh3i/NhdFnz7pyR3kNDfmHkM/AjzNb7nrjjDxTesU8DO22WJjC69aVTONG2EuyzxL AWP3kdIpK+F+V4KXG0+ERj4Tdpt+oDbds0DayufQbPztNmkv9hc73nbHGFvO/g9ahLHE/zB9 g7Rzt4R++E5mKnQfu1QxNA9jt93wwI7bpdsZe3UOM22/6IMvmg9/N5hkTPxX+GufieT9pXQJ Nttm22ybbbNtts222TbbZttsm22z7f9Fa9BGlTO5jtGIhbH6hmBjmD5zW5n+tULGwuL3fUwf xIdz+Mz3ACUKpnEvA2oVwy8Q9p8vYg8QloUj0atY+BGaeoymbvug2gdsmBB7iGJu2EfwA4CV I77wYhrsxKAs8QqbU8kOl4vXWT/7aXf5mdWWfvs6vKGm75ULYiU7RLjDPrw+e767POFjmMeb sbGb8vMYHfbhxRX67HndyRcR4xu8noBR7Ij42Lu7vB5lVExVjmCQcg7jnV7sDQ8/Di3etTOm rRKBUo44Wl3Gb7U1AgYB46S2ZBuA+Gavu854u4eWSTMlIVaSQqlEemuu7T/it3hLthhvtzLj t8QgeUs0kGtt3sXaBm/Rik8W5VTJW6q2eVdDhVKogG756bNfqPA0qbANKjQGwxGCA4DDmwmq JqiIoHX2aW/faCfTvKQmTCM94y94D8AZ/Ru8pXPRw2h3wHibO9X0DPfs87CiBFZo0BdauqFl ETRcvOLkol71Fm+5usG7jgyFeuV1eHlPn003vWi8XZZIR0gAmc284/DdECn+VClc0HYmbz0E mUJvgdDNENq3aBACfepm7wbI8oGMBHu2kFCWFwoBF4VGShNcKBaCUEiDBRBUAj+S0MUQXrTi zUUnIbAG2m6GsBoylIQGSGjgv9MUorEQRGOBttMznUCiN+RF/25RH/RdCdHLaX+AL+UR85pY A3wpVvj7p+BbV+yZC4aBvS4WeYCY/pe8Ww/T7yraybKBtDXVUb5yiS29n787SvfBLuvpU++O TBypkkUTkbWmren9fKpjK5HlCSySaCJMAoKUXlH5TUA562l7XukTvy3/KHwCbPpKG0WBcDiA QdlE+mFaTHt0G94p51HopP2AmLWHV2O+Ln2Ld6X2KF/uGSLckn74NgzghNUpZ4IMgb2r61Iv E4nm2hJAlCIy68Iv0lK9VizVLyhHlkOBDoM42x41Y/EfMQUp5C7vloB+OZsO0jjxHHFvL0s0 N92q/aEpaeYbZXR12s46fKw+2NhDsVpzKpGPJwTIcG96m7ckPZzG9uf3Djssv5Ny5gBjd7d5 yzHnNscYuU3K/KgkbR8jaMb71OClAXZq8NSAGQ+8p2BLgXN5OPoEF+HFs1ofQuZBUPgQIOuQ GXdhTj9IuG3eDbDUVx8IBwj3E8Jt9pbCAyvrtoTLCfcE4TZ4NwNfE6gz6SjHgoacvHlLIOwh 3C6TdxfwYWRDsHPSBkL7HVqtCOYp8uTNwMDMQGqHt0Y5UpIqJrvMWLGeGkkhtaSW6HlUlVwa ThHTfDBB8wTBVnN1X/2W8HM0/pTR1ize3sz3hftzh9fz53I9M+WuC0d5/mfcOspwRVCQy/8H xuWvpiMD6/gaj/7XNUqwxvI/X2P5pWvsIN4tjHvGQ6cebuFrrDfXXQd8Odbla1x/yRrhzYT7 EscJedy+Ga0+eGgEDm4M6hacHC1nphgk66D+qo2cpQ3rEj9TJVoS/BSwdcEfnWqfMuQp/Wu0 /hg2ZnLoGnTiNX7HvLGg/hLQIdXvCJ2626H1NQ8koaO4zDAmDRIt6j2GYYRCyXoEjliWaJhI WholW6t760RSkKwOcSIpohMmkm501omkB51lImmV7E7GROb65etB7a1DzMHYYfoN+pqTv2Y4 VtLVHZ/Tz9RSEXoRvVhxUp43rz9bOO+1rC1yzHLNm83SvIo3ZAekWFiWhFnZ9qaJfvetXLLA pjUK0tK2VsvWi5rk9bKic0NddJ7JJClDGmqvv8r1GDzE9VjScRbymbSw4t+kORWDsjNyzJEV I8dE0kLQXu94j2oKyQEy0MkF8/qy1nm/aSZ5ArtVApE7qV8GF+pvwsTuNiQm/RMkhXfeTyZ7 vnUd3fLHnvSxD5b5WPfRJ2k4hPemawk6AUhdRtAwoK8A150CQLvWaoRFZJiwVOZjO56lWiew Q11GvTLqRt5Rd+pGcREllONdJ9TiI5Qmj3a8Z4BAdSYw6n1PtA2qzoME6u7uHvQojDIl2OtY QhZbF+s/pH1Xdk4x/efFDHeBG9K6ha9D//uLYEmELLlyIQzbtgCvBjdjz54ArnsYr4Cx5hg6 Y81RopInDAM1h+o/E9C/46azP6qJ2hy1diRQp0/gTkH4aFBYHtE3Y1qp7AQXAkTvvBILS15B 7wEQJuWN4hcR4KD/guiOaaLvAwjlqUqgbDKI/CYaD7n1PYsZiyVC0nc7LtD5kIWyRHZexwX2 d4DvQGQY9icgrGWy2uGwyFuUdbFHMGT1jXET+hX9H0H21OvRRVyOWihfFq8uarWq1aJasFWt LlGrhaxTu0ltLlH3iOp6tVlo6fiWw7oAsrff2pzEBZFU2rwCC0p/AZ0cE9VOt3RdUN8IvSIt zn9XWkqYci1TWkQWaqbz5ZjwgaIwoC8DBVDY8Y7RGsQjtlZrd5RNdF29MfqOKo+r8qja7nBF x1EKarVjZcezIjpNHio7vrZh1PVICvjIOU/bZZFzu/fOdx1KdkXG4RplxJW1liVch6J9RQij 3pyg5ApV/5Drb38IBsMb+2gpiGrHDLXhmNowrDaMqA1n1IYptSGV2Q0KCFJ0j1Sh6DZplaKL 0tU1NdJCjFyxEBWltVNiJmACQmajCVgy60mZ2qlIZo2JsWZWcvWul6+J+90B/U8Fpnv9I674 wjz35DCGGYHMA/rQ63EhEt9QNZTNqA1jmY/tJvtVnP3lafYxV/xtqh+hf2YQQFxQiRNsmMv0 cKZN0g2Rc2HJEzm3S1oYOfddV+wA4WvPi5knOe+UO/OoCXgyMZq66S/loshey2p5/saNoIvA 75ldmFAbhjTnvqdwVBtSrpeTExuqLBHpirh/NGDIIhwV0PeCqutqnICN4Jk50cKrXEdmmd10 ae5O12OLqajI7XTFf0CHObfV9ViWSrLobsLrN7uiP6aKoka6D88uRXe4Yhmkq5ZJ/1SpKAtk gCaPKRcM2X4IiXkv98zTdpPCLYa0o2ptzhW9gJwWG5DnaM6DOBBA9ftT9Cmi9WonvxU77oqe wKhFKcfCg4B8vuyH/f5jRBFJeuATX/zXPecNo2JQbc+54m9SYe4fg/9Z5i12qTbmuqSGVKGX QpFAPEasAX3/ldx61T8eS7Tf2O8fJvHZ6xT9O9KNij5Heq3GFdWBy4gwUB/GbRDq5Kz9/nEi NeWo/lzmQ/iGG/pzwTzoZVqNEBuQVk74z4vS5a7D7C0dNmKlvJmZl8ibO1NkOWYz3Rg+20Oe 8OcCGbrYZUfk3J1yIazibvG7DfkYyiyVgDGtJqfW9mi+KdV/kDYyb1Hc34NimbK4Jk8pR4o6 /hhhdIUp7WNMnp8nMooprYdI+sLYxAz0MY7mbl2Fa6b2vCX+61fIwPZco/4pJebMlIVi4C5X lJys6HZX9BkKBlc0Sp1URwcweh48NcD9jNPgRA5YaHdOfJm8u25F71u5zC8u7n6Pufv5bZfE SDmTbL7sJyt6M49xtjGR7+0JfDGzCf9UBL6nf6pKTV3+E33+j0qFun7/GXiPdfqz2Zo8BfFI FTDBml3D92WRjXN7QtKcFq12tEUVs3b0qpjf9cw1Vn7UfGCey1c6H5EcnPVjzISyVcQtLcXb x0PJoR8lvOtwYn92mX6Qw79MZD+9SCDq/8QZR/QfU192nGakRbHjey9vE2KJBxdMDivHUe5M /p48rX2eFVD6fE7ryjblQ0Zb1j4nv/tZAdfgpNY3r18GdAah9kvEieYfM2rHYom9BTQjVWj+ 8f1bAo1B1Z9QG47oB7BjhnxEG1aSpYacQE12FTDKubW0zh6bremE5s8ZtbnOkuxLELgCA38O 61sRpdIVWm0usrZUhqayE0Q+jPsspZhS2ofxfTZOMdk0NZH0SddO9FukAmzFqmwRFTMErsjO ofJmHkAPtodHsDyyQj6zQh6fSDrkhfkkFGz16Jdb+EcMYeL+j4A0ZFzTN5pYKpkuR8E0g6NU /wPjc0p7iklVSvs420O2a025+G6jRbPdDDlHAzfrr36J1xKhPTfGa0frwMvx+rDJrfV/G7YN ZYvzgg153JBz+qvmLDZLkGZqeX+eq91h3ny482KXnTUMlBKoYweS4cU+tn4RDhCe5/Gk8JzB cwXwL+DBBarVOoxNwu752oIuv9hpNSyul1EFkGL6XZdTDaLkIM4t2xXRaEoarCmJu5a+nknT PUvpN48g1dJ4ivQXDF7SKOsNqZAT7P1gezOvOB1a8pvBRnBcy38dkbxu16GCxmpwLQ7ocp7t JmJLfhNs71MZjhtfUJigMaSU/EBEhVSV0+qmvphsodlnGSqwYGtRLNEsF/ZQiZotqOhrL9T6 ummmVQjoL5I13eNfWWKD9+WlrUJQXUNz+pOXTIitV+qPozILlQ30JJAGtGKaUYUFSsKj9AqV PuSjto/Kjr8yRJO/0T7tzdmVP16tCbGRUlSTwhj8JVGA2pQ5htLr+XrH2VIUVG0nmpP6ICre SOUoVVGSPTOEdZP8t7cwFcn6rz7jXwzYk3ry/hoLT9quqIc+4raKsXdc0WKKv3N22aHcJjJU URknR1hc0QkKZudTEK3WotZxqBxGvlfOiq7YUT7bQym4+CDetreCFYPtxfHVm1a8FV/9Ug1Q VGj1Ob4hT8Lda44AUYl8n5Ad/XYaKHTe76vcNwPVQaimysdnoCjfZr9R+f0ZqBihvloZn4H6 G0J5Kx+bgXqYUAsqv0eouSaqCyiDK9IqGm+HVA6GmrMimV4HDD7FuslG5aiRnWPIDqN4MUq4 W5uS+HgUNGcC5GuLabeCPfSWyFrYSrCScFT0PSisdct/ijs2NccdccJW9BJKQgHIf7EYLUnb 96EKTDfl0vZODoyl/WMpOZeqzaXuHks36Wm/nloywj9in3nIjv0YQE2LU/cLUBvFXkzEElJx +KkO/nORxxzbW5kxlMzsR8zzShznIKA//iVe3oSkldVK5YEf+phljwsTFLcWShihLA3NgqpL qFJGKMJOdDmfA2lXQef86q75nQVdvV3zq6MD8me8WBa6ivmsGB2QHF8blMSuGosPs+NdIjrp ej7hjJzdievtrJ+ut4+1ozWR8lLp83K3bKnpWuCrih7vdMsTXFqapZaYn/DPPCDA1KTppro6 pV1ERndLi+IoEvRlV/IfMUIhpGYVgTil1ggDSX5lffWLKbcqDhB3pPIA/6QISl/WmtzPPo1B 2YA2hlD/cW9d2YBSvsr1RK+t1lGnUy0DL5Rj1+vq0vTbqd62iLK9WxsLag2O+M2WQFlCWQ+G RNkAF2Absg3W6e/ycqoIM1hbFvrYKl/snT1FcEZjfLelUzR9qPVv7xTLBsIJbFW5k5SihY/a Xq/T/+UKCGgvwuraUcnWZ/Fr1Y4qctM6rVn45l/bJ9/ps6yqUs7taFugJVt37G+evMAsWnL/ 5AXt90DP5ei5l6BZyAJlmhGr0w5o0HZ53cFG6fpnn6Njuw0pcg15oyIpb4Qf1q8y1XlD6+1j pZHb8GlfSr8wT7vi/iv5L1Re8kW1YEtedIXttYCevuKiB87LhX1+tirebgEZ0qy71dKoP4Ic EuorLA1tzFptOTinirAPAXuAXBPqLNQeEjaGYhPty/PqSsLk6WzB5OmK021W7XR9o74JxPWt pY36gkV8re3aa81N/HcU/mVGW1ZIeg5Ale6nzcS0CXunOp/jW46NDtSbW619EtRXgIwHA0QR 5VpO5nq6t9P5E2L2CTVKbqqbkG1LK5LtSxDJc7sKqqMJ+ayn+gLSp6NRr+J3oyOLyrTIxIjm bbmdHI+Lx91nJ2ssBE4/+p0oEyPLuJUgJTLHF2QOevSaCzNIHDxXtLt3ojhvd/ullXi3yJdh zf9s723AoirTxvEzMwc4wsCMBoqJOSaWhhmKFjSagzCIH4OjIzNiAZEMwYRAwzl+tIBDBzYO x0nbbb/ebdt1zWpr37Z3P8ytVhlpAc01P9qy9DUq1w4dM0yWDx05v/t+zoBoub/9///v9b+u 97p48Jnn637u5/O+n/t+Po7IkoCN3zpI1jYokYu/FvsZxMpjr4v6KoiAxvxWXF71jQp7K7IH vVOKjgFNw+7U9TmlcWQyGwsKW3e+8CzhJlrlaGnsDjz8wO3aI9DCUhqDFAb3qw1u9VAeilQ9 Tu0TjCDtkehQhMujKUxb2ANt4sbi9pUSN8tHmNYMVIuhPhQu4rHCwm4AaZqMSxi3YGc3WTNB eY8+Dz6n//Ggudt7ixiu1ou/QJvnI0j1VwWlJqxU/I7Q3l1eXukMjEjEiIkkovHUgxwOASOQ aqh7FYXqZoUZqlodPYy3jZa/Aqki5VTaZAnLdXFh0noTUMB83H1JUZrD729lxy1YyGqWOMn+ Up6fDdfIkp/+C7bXLp2IxC29PPMlYT7uzcCaP0b6xQVYfcWFiCF1CpuW65J+BDlFKc/qW7SQ vT1Xeh4ymVvzhHACXoHgLuVeaXsc7ge10Xm6q6lTuIP5CqmE9OIk3B6cjH7htsKC9rCzUFOU ONXdxVXX1373FIra1w+wpTTAOB1KXDK0fiduKwnzz+KUH0yNZ7ViP/JYvzVodwDR2XBRB4Hb FvTTTXZpcAw2in8n3nwJashIh7+CGg7Bi1YA+r7d6SL79BACysoWyG5UYcEu3I56kDBk0RYv WmMdTn4qiMZG3FRpp1RNj4EZyQj38e00SJm+1Nu5zGa7xuK7fDs7s+E4O92w53jvk4F7Juto Vq9oGnvZ8GZj0/gsC/H/nfjlIyRA++6gshBBlkXOIDFhJLnZaDHs+biXH4kkrDm+6bYsKMXE joFSxvguj+EONxzndNkyQ1CVKKCvxlM+1Mg7oZ7NRoQNQzjU10g8Cj8jKy/mxiradiuD7Uo5 5Ut9iKMtGXIfVmkWqArc9BQlrTYelsbNmtv9dg1HiwdknVDL4A7T8bTceMOzB4Tc+GOyeGI5 6IYxGViRtvCHshTNcu5SsxHquYyvjae4SKzKIqgKez+ukexsAJw4oonhav+ofWMY0YE6iFU0 lq61ICIMdZii6ZqAQYXV8XfCcMQCF0NhfNWqncG7J+tgAbk7E6aM05VyKqU3RYF1I0X5u9ht vsSCiIUQRGObocRZAErsN18xNHwJMWoaKqpd25EV9GsMjSjYwRzSq/kNDQcgXJJ2m9jGxlgs JWnx3IDqfJPUJi5cAOiO9TscFofD7nQ4hMzEecqHomSW2bm50mMRKslkJhrZaQ5HLu6MJwKU 8iFILzGE39ahBi88BPoEWKcSl4rVa5UdfL+WW2G+gmLdUI3ZBalTxHY20pIaz13Bn96kdoso 5Up0RIicOTqlV47KBaVySwLhay4lbh7mbsNtvUJYknCL9HymeugNPSIdDCdZG3uBZCZJP/sS SEaYj4J8e1gnECcwFcr/fYQ0d1ff6iuhKT5GacsKi9yJMHwbE4Zp8mfqsWvcGiirBLjlYph1 f0WqyUzMbGwxNPwZ/cn+Unovbtqbu/0Js2Hh7E6bj2XUrt5/FpCp9UlTN+vYVF4tX4jwN5Hi A9XxvoeheEZps4RFiga+hakPkMLPwqxWMdW8JYeLxJcC86YuS9ShLnJvs43qDWjY2b0BhouC WtpVOfQ2sV18D5WXT4eUFz/dIUYJDMA3ZdNLrHwrbR6seTPllNrYBScV/7yTnK3xVFsWXc+F eXrXypFia3bfyUktfMs/AfEKMlC8orATXMPF6KeT3fUHcEDDfTm0RTbw/TogjhyakWn49WXw V3Q1P1dIxZW4bGT7MFL8O5kq8/egaqJK0yX1qbZ1HIgECxpBdNUFmrNvbYpptkxqytKQzeC9 s3DjfcNU3AQrBWpx2NuJ+yzqEnGUWqFb+b9shjgtN5b/Szl4wrwMalJ4VOZvfB+YZukOIywS pZ1EiF9LMhkajhJm4fcjQOov8Je9pVmzoKiGWfDEVnqWqUmTnjqR+0rF4BlvL+2ox/MxE5mG eHSgpshL/Y3FsPYTPm8P8fnUX2CU34+/N8GKSXYVhV06PAUXTHMr+4BnFvD92yAEVMYmW8QO WCVifGlmjm6aKJO833giHXbplyo5wBiwk6H32XggdPYWXBikn6I0UxsMz8NRxFUgpcVDpWUm MqW4vQ8iJAxsMemEXR0wDdh7pR+iRNOv23UYglx0aT9uQ84w4a5SJA5yvFQ+AycxZ1DmYw4l 7mEfnuC2otJwWBUJc1EiKX12LvQwSD4PWCkqP3RuJ65IpPvScWOb2ossizXguRqodEZBuxcf Y7SiRInk5hksfQ4QFCvzIZYKHfNAigl0ZbFKzw8q7GwQHBLZGf55G0X9H4QVIFtk6/maP9AU N65Ug1p15h8YyYKTY80fGLm1eHsY4oCOwt2HlJZiMRrD9VtoLIgziqv04mp92iqGCyN1a3Ws eus8KI5yZMgj2hhhOSNk0MJj+rwHD9DU4cPOjLnJazOSk4lGyWDTg2K+HpbTPgtppKER1WGy b2YUYJG29gsZQbFaL36PwRYU+fUb+ZrtNMVanbn+x4zKmu10fgFgiGphUzYvACiH9BsiYdKQ 6nRIZ0iAgWwejZi5nZGKYIqL9HYhQw95GcgASeK87fJO8upHTiAtkeMQuUdxSM+r+fNJ9NCg 5MKoMOJMMT3BLqbH26UtISBo4U48b8LZFA0jfhe/8I8glmq4qaVByC7he5ud1OzJuryQpIqs QMLVJk89u3kBByEXFK+9FCi2ZiMbBXwtfSFqud6LTvFi4JzO/GNMMjQmAOhQYYYGiSJTLDpU oKHhQxyoy1iqoeEihaT80DCZ4ZWNZDKFuVtC8UcbLCgHRiEhqVEuu/8v6ErPQ/0seXz/go0z 1RS7JEIUyF4kxEtmj07XX3oYMLikObPUbcLGh6GbS18nWGujsHiMcNkl33Cg9LcktRIilD00 tEmNtkv5EFNqNGmG4BoI3PIkxDyrUL7Db2X8JIMHBPJ7JTyjyAOPp0fluECfLDgFB1uV+Sug r8npM1IKrnU1iXTnptvJOWAC6bI8bjwSN2skcVWz1LgzQJ47kZ4Jx0X+HGee/ywg27TIKUlT 1DVpXkEr0vfdSJAwnoy/ysCf65Z1HmOBeT79toXaNEVsC0i0XVp/p4r2NjkaVladR1PAtyGs E0YixBM2Y5XJxoKZrHReHVRB7OA7pwQGwkS9H5ec5swAzXda1GUAEI1xSs9guwpandI0tTVs GN+hFLT2ZkVbWAOeaf4Zuq3vZF7BpBZyvOggOlTKKRco7YD9U51vQMvdV/IbRsPOEbMY8YGk /kal7g6XHfO+T7pcb27nYgD8ZR8bk3hVyNLLEXyO3sS/q0AGHkQAIbxQbAUG4dGEpjJQNchc p0BsJBuKdbaduGUo0q9DL6WK2freAM8mw1AYNZzBLs3Gzc5+JY+dnCHfirEUd0tvQMvG2KVb QkmcNqo/y0yQ1fyBZBRbQfpvR7Z0+PBhwiMDOMRsbLMmvaGFtahB7iJhYKLmTWVwsE53lO9W Ug7wAfo3lEZTz0piehBg5zacZRN/A5OQYm/5DfI+Viv/zZ+uINakQKs/fXDYd3XYFwz5UEUA VaCF0+a38gsfghkCvLREtOlL6gcUalOaR9PErM1v9dsY3s1Qq3C8cTcLdBC9R2MvNWlxU68c Z2ncAnU5DieycXR72G+Btsm9jg4lH1t5zcAgZibGR7Vw8aF7a8aU430XxZN4IgzuCTld7EDt FFcJjDvWdezUsS+PnYw6yPX0XRK7+4oT4wEiMzG275zumDfGDHCbGG+EGWA3anJSjottBQ8e MFLD5SFoqDiCkBQHsx5+D6Qcl+2kOLyapsfYG4rrgOJin4EfY6gsPZQVA2UxG5kkqPukNYmx 4DLg6vs+7jvJf0xh4WRhFDk9TtQC48tPGhouQWe8bDE04kHnyxp2PM69l42GhlM4xDlM4qAQ A7MT4DGBd+upxK/12pLf6DUQy92unumzTOg8P8ohbQ+C/gBUGeGU5sBMz8ODfUbjYotDhODP 1ggxYk780kZFyKHrjC67QwrMoNTDSL2cwl9R2PFiq/jujUKjHLa3E1a/XgttYqM9Wg/lkO69 i+QDzM02pomRYz0aYGOFiLJiGGVarb4Qxh+ISzbArCKCoszoTVB/pOoHyc0x6BPG8DsO1vNO l8Axm67nKCkKxG9aS3gXsq1bAY85jHCu2UBquEPYn9eocCvkKci7bg0xqnEhRvXm7UMAi5we ZiRvEoEvcIn8QuSoFDvJKVlVnsZOlEWQWTzA+6czahuBLPNbnfL5lBb5n06XB4TvghAjyBzB CO7pteh5NtouDcxEPQvoX7OEH4ytmUBglHmvS9WwDpBAHp/4OjVE/28jfrJnAE2DIVIv3mzS q3vBhjfodHm12OEU52MwMMAY9hzrzXRqfAUgCLoHqSZbsMl61aOVx/jcV6kmaxCkML8t0qNB CWwV2RsAPvYW1umdBKTrg62GN6yR06x6sh+wKkTkwOK4yVABvj/dsO0VpNL+dYbtO9HTYur6 Gc7IfGY/1syjMX9SUDs9jdB9XRSIJBZyE0qO3KluXq+d1gbl0KDkucU4jBJbA/1Q7Y7ezGUa H2vwuSMpj7bJyjTlj1mL1R5DNeUzYhssHn5b0KPDSjN2lzIfKyYtIxPNBB38mrmNCzfssQbl aPi9athjCwY6magAOQhDOnV6TD/L6xvwTBG7kYf8rG8gD6OTDvGHlBOf9R1vjH2dS8jiB3W1 Y/3063wqxcVk8S00RNR8A72m3lfBHNAZHhMyHERHkMUCMv+CP7g8U/IAMunQiU67cp90VV1P cjYnQBqeBLg8Gog+F4rmv6Ybe2u/GLpwBuOrB8z+zYrHRC4Y+Qs1UExrwc9InTvUvWMrU+DR CLZ+P90EFJT0zTrxyInPGo3s1AxyaBSLfY1EHuPCYA4GQazMA9zQSoT16PKmGkVrv8UcwK0Q JqnDv+DJlJZLr4rW4HBXIbd3enRJHevEZAFv6tLLuH8WkJ1JJqXl1VdfLSRLN27cUlx0OvbS gNpLrUOnJ3q1OdCCWuZB/ojyM/HddmsQV6m+AcHaA6nQLuRa4qET55I+7vtAjBA7ph3BU/dD J6Qoq16IwM2td0G7PdGp2ILQWu60WNvj0T4o9nmYvMA5rfmTGgO0QAwktYm0UaDHyhE4RRM1 1HK+i97SSZqjHuVAPFTG8EZ2pPhR/af4+BmPybLDBVuPmK01vEGJtcb6z5WtIETKyiJM7Q90 GuuvaFrwKkWQAyGsJ6XXLoUrhIcaGvGqAcTZJQlkwjyPdm1XIBTjt0p26Q2IVTYxBGIKkZaN MAE0GeISWuxIWsmYP8IrL5CIwnuevMB8cuuipFaxGzOT+I2oEVp7nHbptRGo3r8KwvwSvUWF eZYgNqX0dtUAiHlKzQwS7SQZ9D5bn4+EN0BY3BQLU053n0fTtVohyRDjc/dtRQjdJWkpViMP VL8peUkr9eJW44lzoMz0fTTpIN7ko3xRF6O4Hl6xcNGw1ou5/cLjtDAW9D3FxIXhaeMc9bj5 daD54X3yJ9Wd79bSjpGx5aFYdZKktNjBKBx0hl5tVNldeEFAba56nJx48KbAC0YA/2gI+DCZ prLH/wL54IBIJBtXSGAyqAKTeUhg+tey0jSQlW5VZaWYa7LSFkWO9m8ZlBn/lquyzr8FhSN9 CP/E5rHpDcfZ4qYJS0JFXJDDhAkwi0k5oiZUkjh2uKxPAX5eQy97Z98BUtI4dElR7cIEpMbC 1hLV07ZESw0HClubQi4KZblGB2iFuFiKy/Quf52mLTy58ZTYXjdFyGdSTjWPhULCDfuOp4O8 nryuyWhRbAzXBdl8JqrxVM04X9r8muhsX2oaF9ZGJcv/HKR8D8zhwpTFtDzOJTX/U1EsdhfI 35prRyjqoDiBG7vU0ViJ0rCf68krZWbjvUnPWGQDwd27S3ZL6++mqF3Js/FIJ96wLywSfLtm wE+hYX8YQu+aNRs33ebBrx7sXkwrUJcgMRcZSYm4OHy/gjt9HdORfgNd9LT3gCmIi2lxnOGN j/bhNJ2uKAZojsAF2VvMn4CK8mdqAW577kfKlqemnDK3cuPMJ0C+tzLCWlqILxHWMvJy/581 ACbkB8VDga64QGdUVAcARMEC8lmUkE0LxsYjsPoZdQfFjwPylKiP/PpxAuavb8fPZ4jvyp8A DUS9VyJkM4QODl9viKAIc7DxOLskpaUZxswlDvCXae/dfR+BSsIySceaIpYs5S6LHbBi6QLo GME5xH+mC3Rp2fBLr0LWcIBZBjBkyJH7DhcDneQioqPNCMxWDIjHdG3mwY2Mud/w5GYYsL52 MhFjARdg5AcY78xLu5XEH8VbYfFtxyvQFPBLr57v1Ok6lMTG5Eyg10T/DEjeAe5vIbjDAj/l 6AE77VgzVKY5S9O0RNOcpQW0TUu0fJuObyfIDw8j9y+9BkugNHyblm/XItSLw1A7VJgsgFiC KdtG5jeAE4Qye8CeB3s2E18GWKj3wR4GK2Y+fVbI3H4WXAlcCdzz4J4HtxtcWJyf7gG3B9x+ cPvBDYIb9NN+ZA66bl3/UE8o2OBksAvA2jEGFtvCg607sHYj/WrLQzFQx+t6A7piCNbPzgn6 a9KCof5PvNb/0Vcv7Q5Cua9nYbl9H0EvP5xF2js0HgPqeCTheOy1kvFA4B0d4JfAGiHQnKWb dqxpiS4HEK9EtHtGoMV6hdKvjdEFLf816f1tIyB3jIADoGtjdUHHf43VNTy5Dnjr9biLwSkH y4KtAdsAdhvYZ8G+lEXGJQH6OQHceHDjwY0FF7S3p43gglr1tB5cUKCeZsBlwKXBpa8fF7V/ WuC3A+xJsOezRoxLlg5qPWIcSPj6MRgZd924ebTSK8dh0VMv+wPfbOxl1w3rEeOd0i+iyEUt J3uvXXqbbBTJ96MOgfsxRG52Sp/EhbZ32bHqZa2gNG28eupsGLHXUa5uz6iSCLAylMyaNOqZ xfyfAqMDvcbQ4NHgDjMeuDjRd3yvBMtmWjLr5VuSFy88+XPcEkPxwKnE1QBCclSTiAK3JaUl b3/LPVDxWkkjEISXXn1FmPwOFG74/tfIF2svUIaGt1BAsX2NS4Gh4TUIGPakK4Y9x80HDE/9 EKGayPGTRb5wDeWlV/3zVgNjDJyjX/E/3w818pmB4bGzmldp1GM/jWHPZXP7plgxN9j8g04A YHYgGou5feM/09N9aRT3vmIN8q0MXxukvQHofVD1Z6vnUfZdWFmpDSu25zzk3f8rCO/enbcL 675bzgUEWxTK0JgN8w+ES7WGjGUhaorei05pG644BPgaUjlcmIz+lBaCDYRVOTptcip42Rjy OEX6PYUPD2W8OTQk8dL8dISAXh3kjO1hqSFcyvQG0tVXr4/0kcggZ9hZTlatoRSC6e4XgIDZ 23mJZifwUjSrz2Zxb2svTQPA3nHkl2bwdyz6XR4LyF2hzn4lVGnSL7uwFaRxROAmG6axeG/b et5pV8g4j5gK43HStOxDFaypIew8SFkWc5u3p9n6TVsjhkwgztzncPlzLxn2RBr2tJgDmyaD 7P82AU3fawIjfgSED4rAu7AoXXpl8eLmxUpDC3e08EHQvBrPX4YxmI+1UYu2+wlaJ6qPeBOH wFxFGEx1Dic71UzSzoQQTPDbMKUU2RcicOrJ3UMpLX4/Flkiju09oGHHq0NvbH42DAtp2tEC COSE3gNa1qgmmTAJcMuf+36FiZQlPR06YL+8NNQroLp6J4LaCnJvCFeHL2BqaiKYpvD9Rd5b +P4nNsUMoevwdZnkziFkgGof0DC/EM/sKW6c56o0C0rLaw+TQgf711b9/7fr8csrvns9bllO +P/eFep6vADdoxAXBGta8W+ux4j8/74eI9R3r8eh/L8EZxvYBrA1YFmw5WCLwT604n9qPSY9 oWCDW8AeBtu54v+P9diSM7wed9tuuh5n28h4IPCOteDfDPZXtn+xHiPkv7ceE5z/j9bjEO4e KD8IloagHmws2ASwiWDvy/mfWo/V/rHD71qwpWBrcv6n1mPquntY6sUkY0jJcPDuforcJGue 4t+kAfn+2i3JPa0exi6dOk82mQyhbbE9reSWJJ5ThDXfliUf9V25G3Jx4bwtViuH+a4kceEQ D1GiFTRzw7ZthENoDA088SQatj+Ba6sR91ISfVeS2THtViPewZfnNocv2MxGL1hLovBOmTxF sBnTbEZuAgBy40g6vWAtxzQXaLKwDFhoUo73ccb9yCjwtrpaecOeS3V6D+2UblMrHxXaIasi 1ztZxrDvXb4rWZ471NYoUMAuySNAS0S6alo7gC4j72Z+cDushmIH33Wf2Wasvlfk+qG1jJnr 8cZy+qE6mA9txE6g5TGG33FG3UWB6xetQViim+O7ivGs0Bar2IKgc+F1ASPqHagrwQoOIyDb egM0PmzcyiX0BjTcA/ynWnZMfS1+ic+3aUzjQW5WmrV/45291liNuqEPNaJYKK9fkccCZt+n 2t6PJh3s+0iH720AeRiULWgBXssyorUfnzVBdfrlKNGKYpVWSMV7GzeoVP7FyoOtIe3eX5xI qTqVQzwJ2hPoWkzfAaKK39IcuTj1I/ZYw0F2lRrFfRV10iW2Bj43OqNOipH7iUYeEPtS3oX5 PcBfVFA1fy+ljT9A13+KujmXMJ2o6CfQ0XBM/TlsKfdOQeGDeFy1OJh6kr0XFtavU0+zdzac Zb8ISEaIioOo0xAVDlHvyz8uKFRrvHgQ8oFzVXWC6BD9cAVohA5n81i8Vw9SVVnz45qGXnal 2IfnHc0RWalHCcqo5nD0RsKi/M+CBwuhBhjRzcn+OqW5SAuFzWgOB46e2sm905yugxzxzSUK H4AI9hP5v1JaIFPzga4TuLW0RIFAq9iOv2S9csBq9fkd4ke4YvWltHzzGgzSu+IgLGOXXmzO UlIO3r9E2USz9NKlfcfEwINIskPjsWohXhXwQyu8sfVd6L8bfxo1JL76fFKH+E6jRvxUZEQL bW8ttSBlXXcS2Niyawcss5yGnP/z05uILDZ9Gzoeza4GlJGjnyWyVg384lObta07Mdmv9+3c Qdwmc6Aaj4DoJLu+/jJ+5NAbwbfEytLQjqRRtDHqEUu60SUO8m3BpC36wNcG8b2ygzQIIP5t mkykT1u3GLMUprdgCxoaz+HRb+wcf8IcIbdHdAaF3H4xtwdEEiFfz9UFLtPL+MuLNuo8i9aJ y2nzcoZbwV82Vd+GsrZPPHRC9q/Q0n0npj60N3NwDDcDajB1ojzVv6yTAAwnP7f3HQgDRDhC 5NGiDW8JyDp0HYz4IK2z9QjraKBKqJQIntWMAPHHris/v5//HKrVY2jE+y5JQNf5/YErtGeR wOmXF5jPbNSKZ3S2fvEY4MCM5g8g5gPzx+x63NN7nDY/znAW/oqpOoHspYjvnrjQ98FUFmq+ Qkdzd2DN5shT/Ms7STJJ/KNab0wnNX+cJqgYcSkt5Kh+WlzKCDkMdDs0J5/BhzEcY/7Y0FAP OfFR5kPHaL/+mFAAORgBwHNoARDk0OYchnu4GfoHOrTJZhgMNIG/erqv1kBxMcP9C6WOmXit tzP3Zo6Jl2cSIOO1bkaoPBrCb5Lwl/4GUvExRmQ/4UJmchCfLTGGxj4KbxnMxmuthobTWMNo H4TE+aUwSwOfGnTZ+rdw96mUslCUY5XygZpOsoRwj7EzJBjKaTHyLzRhYSpeCA9mZ6jebKNv SwbFjeEfpgdJDL8ZdMgyB9BGlV75gEQpVcbrFUky68U4MudpX6BTK3YQQJDlV4w9v8qfHbvK Ln35Fb4BYkARVaIR3hyGIBtpfgMTS8iFkBQy9oVYyV2Yem23PHZEGQ4nlOLyL3gSgWuD/uwJ /KfdQCKBAXopP2CqjeGPKCcuIIMXuB7ZyAc6caHpPSHhK1xr/05SzknzJ0I+Uz0RSEQ0T7P1 TxuMeo+L4ltjzflBSPB+AS63RfzYfBLB4q4DCwMw+Zz5JJcHQNVj+e8xMH8gDiDlrzDfHUP5 okkagb8A8DFSKtLuycZTkMjGOqSPtbg8LjcKGjHDyKcqXERKS5cBIoXJpJYZRtHIX1aE3CAb K27bgWMm1uzIHKrIfaIaVZU5Lbd/GtdDjmiyM6M+ELkeboL4TAMkwjIKiXyAFgenpUV9yHaB ViweSlFuYDA94grAe4SLDhXjMUEhhY0Ha8KOfbpE7hMBYM0ALUYmAXlcxKOIQTqHH1xUB2Re YO4zNOCrH3EZbV7GcEW8YqpNUgngPZzoIOGmAksR8s9Pe6gpUwA7aOBSgfJE2/mpUfIc/8pr REFg/dvI9AQuNHHQANm4qBDwcppQrLiMEZ+geVjkhS00lCo+wQhbGEPDs4R+e8Sai0C/F4Un aGEZI6xCCGEVE6pdcfOKcJofBPqNqp05GPDVRlFNEMNF7ScVPnLi6zHmYS/UZ0xNU6Z8zwjA aH9OJwEAyO8hW/S/TWoL0SEuNOY+ESSJg2yUECmsGBtEntkvct26Y/4cDUbNDcL4iDGNB/1b NMID3K3myTjZuXGAy0wolIsIrTf4WIXxL6YJIagkEI2rjdOVopgDXLI/U7ADvYHqjYThGiLK bUCU5na2iFeXJrKQhXKjt/SddPxuElCRmsE/r15d88ztQpaRvctDl1ImZCiwYLmUuKZQip6L VTYaXSMpP6VFnsDnMIrwgFBHK2GkiCwjzyhi29AJMS5wLvFxo5jOpJxqbIH51JkB86nVs6is BWfgOn9DMpmCwPYsjGCh+RZmGXTPFQ8d+MKg26Inl/5KkcOF6oJXF4Giu3CDyP8T/MKw+fKm Bf4lRj67X+Gbzhtg0vubguC0W4Ld4EAXNgW/VhQhm273SRCBkWFEQMjuBuGzHa+ir9SH9aMQ oCnE7fV0xr9ML2QzQjrNFz+fSQlvP49j3JyV0dwxGGhoAdbTlJPB6cZspgmPhZXpy77jU+0M +VoHKuXdPweNDG/1chH7OsEvTy41mW4UMsTuxl5Dgx5p525s2rWex6tVyt/xyN/Q8HekR/ot fJFHOsPhUv6Oi8HvIH4v4jZ3c7fxC9EHzGYXuvzCsySUsAtdOU4I34keIYsW5qNHBE+y/yks x/80fklIyIr1Z8ULOXqhzihkMSlf/RmvrKUc568aDb84sGzasWVWwQg9g5d8r1KgCtV87sAb B3/4gpydwajrWs391ZN821HyBmGOESLal9DdEKhvx0rLp8U2+U4yjh4cSVy96gexruyNG5Pq vWj1jlUoA0w8NUvafOptC7BbaMAu9IVQmCAsxgAegRHDcb3Opq/b6mzlo8mAqJMSRC7SCy6+ zqiobIOwzn1G6Py3qO5FVP0irDmXJraSZVWXZSQ9b1crAWrAFN6gKNELyIRHtYA2t2/UCXV6 YLeKjWGj62WE579mGk/V/p6vDVLiEXEjzc7it8Qr3B0ejT/dtA4Zvn4x9ij5gkrN16rLRayS KDwlN5NKspF4Hcmw3Q5Rw3uDb2KrneIhc7pxY5g5PdbLiGPMh9j75AsvG7g5YvebuL1Uf06B iW7+xmvCG2dCDPkiVkcJ6BHdSQEhJulAQaCf9mcyzNA1AhrF/qRsY2BA78/swLtGesOesRqj bypVn4od0m6hcVpwE5QwbDq/EH9NHOMxSi8MDu1ck+7FsXGJTly91A8hiHWM/+l46F+xPamN l+ID/bpAp153wDwAlF99v2gAvm5+t+bWtEJ6Y5wDBIY/a0I72RyHIlEhzbeHMBDSJfm8Hep6 kAXSmdFfFw/Lu7kfZMnqFDHC3FozEZHFOuy4C+4ZRncH4Kq3Bsn4tDPm2n7vcRXLvWZSbTau KwP6WiCBmVOEOqZrhiZEaWI/v9WogPq4VU+B1L7I3K/e85QlsVCPc0i8JLYm9fEX4gODusA5 vW6TkTcr3FiYAcZavEmKG8+pCvs1NAkYXKfK3exOqHiI0bYZGvEaLEx5rt8prTkLUlNuECt+ XHy3LtkPdJsb9DX1ECKz04I9+CYODNDm2PZ0Mjikd/gAAyoz94GwEBvhfwq7zdxRrefTFDYM EuVvzB2TWqYFhCV48bcpixEKmZm6mVm0f0U4oz4vEGLkedK8K3i5gJ0JgmFTHuNfM5EhVCeM JaUmnYDV1bDnxEx7vL+e3PzXhnTjEOvHM2AnCNwuMc3/M2QM5m9AYhPbUcKa6qcnNE0wt1ZH aMbOXAXKP19EU/50RtAA8+jCZ2O6Pt8DFKsV2/zzjOZL1RnN8Q2nuGSgnbWaAp+Z4uiMLLnP /E11jtka5MJAUJf1Mzl9Wq7e8CS5/gMltcphhj2XdAQDx3j/4n+cMfdV1/np8WYrU72xORwU VwtgLNCsRYzTMyxynz/BaAbZbUzSN8JERKpW7HFGmNNkhBGU7zUfqL5lphUL8s4TL8mvGfa0 6Lrll/wJ49V8ByBf+sgGNY33ZzGPHTQEOumkwHAfkfdWFrKac4tL4xvxbv8OXBTwYcmj4nx8 nFuqh2iX8gFoowjJ2kuDDSSiVEJ3VybK4x/sysYFHv0wVuOVoyhhsjE7s9WIMAfGhB7EE+tQ 4iyqeA2UL/er+r7eBSLLE6B0f4RfoOtrw90J1P65iECXlquUp4Jmj2fC7HTQ4meq6ZxMNui5 VfznOjZaBYDUe5Zzl/A+E6jkrerOJjsOsxoa8Al1aK/zgpq1HBKWc5cfLFBPkcm9hKXs79Sr CeFDVxPG822hqwlnGo6zsxp62dtCVxP06tUErk2YIMfXf4an3+CLAR9FfHTKcYIzpWUpG5nS IkQgfg70Uh3XOFTHnXa1pwx4A1k5infqht4TX7sOXL1Q7BCPBgbC+M+mhG4D92YeozXsdLt0 0UguQMf9NCSeKHHP+iyEGgDruF2vzyKnQfCLn+UD2SV0ckKjuMo4zNGd5NWdgd+CkkBAvKg+ 4PcnkOuMgXM6UdecmUy/rDE0uPBMRUmrNQq1/UJtN1EYMO5lE6v3aF2rpJUqnxN1urZeC+0z NARwHPFmI17mF2oZw/e/oPAC+PkruLMWTLLpcWtuJnhxx22MbyCSi1Cs3W1UtKWJsYgd/h8h ZNKh+gGySB3GW+ckpt9v7fHgR2Iw5LKLOmVesnrxGxqdIVp7NMCN22uigNUlWfVmq9HbnAIq IMQZGkWoQlotwxZ6wtamWRluxbXuIb3V5cNBJb0oWBm53W/rca1SOCM+TL1WCKDugRYB85Rf ugb8HB5mPmfA1x8LRRijZNrC4tUvX+qthoa/wTKlloWXz4WFWJg8PsnaLcfy0zFAXTdQwwuu Q50ETld1NUyDbpwGMEAL/EMDZOjNnANzwWmX/hFDkcuU+CGBIUzySiQ7sp+cANPkOaTyk/wV 5YaZMsYunY3BepvECF9xMn3VD5UHHnazqrWqn8Mia4j0zFF8LYp3OlpAfGWj+aV6pj09nshe YyFA+9NjIQGitDgj0hlkT4JGFbNc/HsKXjC3kNvr8v0v8+w8p7TqGH4QoF8qQ5F6o1F6LSa0 htaGAbw8TsxKwFGwS8s/BSmwJAF02Xa+RE+psr9HKU1uvMmXz+LVL5/tZIHwyBWYB/zWs3aP psBZOgPyOJW4l9UhpkFqnIxpztJETADW9YJPff+ES4RHu3YX4sBT8ETRmgCYwnbiLUzyiNmo vmO2q3m6UAwhB5Mo8KXlx3MMOSwVrPEOhwosWk3k7VgnY7HjS+i4X5E51WluZdcL+UH/ApBe +BZdYlD6uQwSglsSewTutMvhlF49TeRgOQr5huiW8M0jDyRlvsTOVLhOp4rfoxnGCijxGbjd SSbBy6FJ4HcfFm3dxQJ55TO0kkJHWm7WkQlqRwJLWLgZNccsl8ujedBBepGcMTpB1n5d7UtG mIww8u0ul0PtTUj7bag3b9fg00tM1xLajK0zqn0oPa/BmwaCNXbn5lBPT1a4hKGH4qE+VhF1 nRnq4hMEiWm4i00ulwouWhNHdLFJ7QzxpJB/1vyJoeE5JMraoJ+epHbzvC/Vbn5HsJ7G7pRv GepfRaglMWL+2bTa02wJjOBptQTs5BBeaLQNGkuS1FpiV4jdAnfS3MfO7bVG06wlrTbB0IAH 8GYuYettatNuzAI91IivUXGEOm4YoXYrfu2J8tP3qnX2dal1Pi3kd5I6jxuuc8opUmUo/RP2 jzCXubPfKod9DurjcnG0yyWHq5VxOtRXCtiqBaT+uOPyOJ6MtHXhe0E8YUAm1Xjc6V+msNH4 bShg3ReXNvZylyBQwIUl4e0s8QD5Eon6AKXFr59rPuSlzUeq8ZuL+FKhQIgMsQNc+J3Kfcq9 0rvkDiOqOBNA+AQSVZKlN4bjdEoWg9HKJhrif3lDPFDeJgbin1LjC1Ulw+kSk8XNtHhE1ws6 6LxFga4pURfFbEb8UKiiA19OOfZp1CFQwplhaNCqs2nxG3EgcG7Ksa6oI6DciUzgC/Af+zLq vULBzghMwdCGulXvEPONKcfNB5wwMDkWmE+3CA5acDAbV7kEq1G85K/t3r3bsVuqglo5JI+C t1q77Q5JS8J4jon7lO4eCs+XDPuA/rtVgE/JPdEExQqytZHbk3K8MM3BcDNwbzAQkIxCBiPq YGm4YAx8aoy6iJd1rcFulEeqaUEnn9iHqmxjNcMZHVIPoIJUlFbyZLkElsiCEiEjcmi78yHy jQqj3/qCRyNYe1Ja1gnWk0KuhCu4tRtoAXiSYH1fsB4VrGdh+ROs54VaY7PBN2Bi4WcMGwnL OBvmG4jmdEvkLzx0QfOUJXx/lOGpTTiREhtfrbRQvitzavS+K2lbIzy6rskQf/9k/LgGF+EJ 68rShj4dmMTelp3NxpDPk72vDb3J9mjbrSexNWvll3j3SeTC8vOATsjtrgkDjHWvqagMDW8D tO9KEjvHd+VudhYeWq72XckwPC1oMd5aN853Zb3h6e+RkNuw3QseT3hXB9DTSk9UV4uGIrUc AzgNTzthZIbw5oEfsBka/4QS04Guyaiw5XanxZEr/mMNe1phJV73Hh5MejTr5JiRZ7B42Z+9 DTSyidUwXXpA6vHJyYq1pyle1inWfugp+Y1QVzwKBHYARx0UeEAv5J7movFAFSYAgbt0fYFd R25eYAIpcOzIAsk3xRDNn6GLDA2RpE13Gxo04On6OxSrYodShwpounkBhgZ8AwGrpcu/chDH wmZUxyIzNI4sjGMxGUcD7nMxslvI7cTGhGEVeiGLYds3ZLjSDNu7UGiJxL6PwRkyxjNm7cou /OAdZsiUA3ytiWLvAsz+lQogh8LY2OxsQ+Mfkf/bJMUTsbZLAHB069GNXLmW7480NPwIALoq 8c54rbGricIv1nUi94TVcuTxOW2Xnvvbdx2f4yl0Avm65B4/m360ftAeS1FbSRP27VsFfoE7 eyOm1TfFRPIdXSlrxQF8jlDbTQmphoaPcIrHg1wa3Bjnq30rrEY3aH0LSA/GLhsmiKwT8etu Qcrw/d0AmQ0MFidEEJO+8Vtb7EC2Che0Sx6UsK1nUw4C70uzSbXavpM6a3+arZPT6azxabbT nC7JmrAQW1CX3G7FDzmSrwfhImkE1ZmvfZ9iH045tRevCpO95ZTj7dbzuD7LBQtfBwkEP3yH n4em5HaUJbBwKYIBLLktYv4BkWsTrYdQKvlTyvGUFnwQKL865ANEeKFBjuNrj1KsYUQ0qYaN ibKeLBQeDxfytCVCeqRoPQ8caiwwKPYxC9UK9tZyC5UJ9pdgx2+wUCvA/hgsB/Yy2LNgX99A PugGYgq0CPhyglPs2EeeNJAHD9PaShoPutqsZw+3Wf8B9hzYL8BKYLvAfglWBnse7FdgL1xp s369KLTeikfZ8LYM7d0yDb8msVZf0nicu61R4W4FxYKLK+GnUZjQlqHTtGXQyV34SLXE/AH3 BG6v+EradRrK3M+pb/0Mfe0QaeEiSqtALJKzoFpsXD1qkPigFyC42aWlmGKALNNIxOTShyHC nwHzawpK3lBOGHkj01UBBZWuHUrUhhLD8+RPDW8cEN+DFaIrvqzFaKOowAWmhHd3Ux6NX59T IuRefuygTajt9f9Ro//qbXzVGBgEpe0bw/62vpN2XIOku8KJBF5SX3tZWeSFBeJslppwUYdC Wq81XlMipLBjxdpYyKXrb1S2Ml0X8bVFbSyvxNXpPHGFhje4y/UXBhfhZ0Pdl4EAE/DGCMGS 3w+sx5aQx11s7K2d0Hem/hz2QV00SayAxJw87pJ/pQZa1nhQsCbU2QVbrJ/r9tcyu3cj0G7p 1n51PZXGEg+jVu/FvlDsFfA0W6+ItgSf+wo0JZYGTpig2GKbNNweEOCbs5XsbN/AfP9yuibZ HPAa8P5/eBuVnC1fxK+hWUoU3SB1V+ssXVOGzrCvxVccT1ND74QAHrS8ZPwWSw8Af3TdHNAk +9aH4Rfq5BMjX5GF3pAVGbbhdjrf/4Rh+4vokUxdzxN6ZOqvoBxdO2OIrYCe7ZAGO667H7OW 79SGnpA99l1PyGLwCZn6gEx9+EZekJGHb0Hy8C30hix07ZC8I9P2kLc0wKZ+AyL1dz4kQ+UU 30WpOwiu6twRmwcLhjcPIshWgIa9D3ce5w49iJ+S0pLUwRqXssxS7ha7p0dm7J5/yjq7p1eJ +yNKmRG8PllJabmZ9jlyayThhmfSvZkBWuOnA6xOpFq/KzupdspxIRqT9hrfBn2IYU18C7N4 4ea/4LVifuAW72R+IHFTvJoPxe1Wkf57LIDuwojW+gEULrzj6gcO4TJCoHAib4xU/fjx9VY8 gAFZNta8EOtp4PGrfmK72PftDRbcJ9CwZS5puhJ6DL2AVzSgDitabrxHK8WQ+76iJIWRrfm/ 5rEJLmkAN71BvdxLRHcuUtXmRQlUua9cUhHSnS4PNXmq8St2TC55q5Unx928U4eOPVAX59nJ LzPsRGUjI8mgiJfUl+j7rqS05Anh+DXrLDok8iYOvbIpEZdHprS4UBYMCYhnS8SsSHHgTeS4 Kf3TDkw7VP8F8t2ShcpWf6xGyGUMPH5VAV9KbzXw/0m8l6/uNzz1a9W7f9Em/AL3/SWNLW+S O1bxIPw2OkGETWlJy6K5qLRwLqKrEvekWkrMHdxtmAwy8fi0DJobm6bjotWqCuFdj+GWMkrS UK9gu1WPz8JFa7euha/VU3VLdW3+5Qy5aaUXbD2CVSKfkeoxvEEZ3hhj+F1AtAX5QHCVXZkj pV0a0jLCgfDxc0vWfi2/oUer2CSug7cGtQp+8FWPiLc+V29ltC3I6oBp103vtX7h4+L91tN2 6fFLKIIzRAQnqVs/U520UCwJ1SYZ9luZvpMaKyOe7LWe1oBqYz1PEBQAguV53Jd87XmKpX22 0xpgy6cp3yaDaD0Ns0hDLaR8Goqbz9eeUbhZvPsMxdd+pnCTefdn0Lm15xSFi4IOK2m3nqN8 ckyJYj0nM4r1MxRDzwBXwmGqnlQi2s5EtZcIMaKtU8jRirazQk54iZAVKU8q6ftkkoJPretx S89X3075lK38HZRQx1y7OqF3vIUkBVKI/HCaTc+ur0Ul6a2T+ywU3m3Tk2gjh3fqSgQGIdKB ZiC+dU2gVq/jOyPMAfZOP3De+gU85fPP4716f23P7t27FVuvn+vBDQzJe5U81jsI2Tn1BFSZ XgWkMGpGzagZNaNm1IyaUXMzs7qJop4A4fxpsL8GexjsGbDnwV4GO/4pipoBdgFYO9hisE+A fRbsa2DfAXsabC9YBnDFg70DbCrYhyH/I+BWga0Duw3sT8G+APZ1sG+D7QB7HOwZsBfABsHq BYpKAJsMdjHYtWA9YGvAbgP7HNjfCmr9/V/96Vd9ba8/kND+OPx7e5Eay1jw19cwO10Ndy+6 f1vyZ09vsT4wsu2+hgdC6RR+l0T5yUQL9SpYUPXDUfM0UrpwKlxrpK1Waxia+cmUHaFNM1Lt ayMY6r7kZJc9AsImhpT6cOnDYKDBVZvBMLghFEExDIP/cRJDYZCaQVVQHFUOfzNJwTMquPJy 1dtSp7FjfXyKsiR9abXdW7neXV1d6c1yF7Gc1233uqvdFSwALLeuzrGuSJmLmdxJycnJ4BYW LlmxcnH6isJsa7q90GFdYc1YY83EeJvDmbF6zch4zOflKtiyDW6T2+ut9JooKjqSotasWOlw qDEYpBxLc5aMCFKZK23pS3Ouxay+N3luanTk3SauouiRcreJrTSVVZSxZUXlZU+4TaXuoio1 H8Ldh3AVlazJXVHJPVpqqq4qWu82lUDZ5ZWbykZkLGLLKiuu5bv3pvmq2eJ/kW8+5quCbjNt LPOyXFG5qYSrWI8wpvVF5eUEEOHm3RR/YWWFe3MZe08Ri46JxSYO45+Tdn27K6vcgLiyoroS wsXujWXrR8CG+si9ucq9nnUXk54Z2a8Ic98NMBu4craMLfW6i4qhh9Y/diP8zfsllKm4iC0C 8OjIokcqKr0boAOqvJWPeos2mFi3d0NZxfX9lZx2U3zuCujByooNMPcAHGFTbwpb5H2UQ8Dq of5NnouwJeWVUFzFo6aqyrIKlmSFmGJ3sVq+rWy9t7K6soQ1OcuqcagykpJMq0MTdEXZI94i 7xaEi8Q5F4q2YndMjYy0q426H6YwNXv2bGrBUCsrigCKq3isonJTBVD9Eje7oqiaTYcpsNFt r6ziqkicGnaVVRRXbqJsQG5Fj7oXV25Op7hqtzdl7uzi8nLqOoN84lCJhToCdih8V6mFmg0W /WfA/QzsnGmrctJzKHCX5mRRqptJXAeJ/1XjNZz7cHPER32n+d/On8O1qzi3d4vd7S3BWVix 3p1RCUPo9lJPaR1udg2ZrHZvWaW3jN1CDeCYZHBeL8whNQk46LcwZHndj3PuivVbhjmhOk6H SO7KDQBUvKKswp1OsRqIcbq91TDXKaqWsgIlh/gq9Z/a3IpSgCx3F1s3r3dXIT1klZVj1agH MB8CphcXexGYugNjbJXFXLk7m2QCTv28dk2IlNxDSKm+ES0Yiky8lhcKcOfAzITcv6egHW7r NepysF6gkep06g83SXFRx7WusmJ3RmmRd02lDTnE4i2smwpH/N8Gp5jvjndR1Abse7UdZDhg niGsgy1WIykqDsNY2zVbqtyUXU0tAk5atbSipDKdGnMDbmeRtwyZIVDOiD63AiX9QpMN/C7T Xc16K7dQPyehDBhaqDi1X+tUuTM2mPolSSNe6h7tarY8t2ITUCb1idYFs4P0HfUzApNeDjyR 2k/62o4Vov5MqDkDluc5WP5Kqw39bw3hVzNQvya5V7vV4AHNCuBCIQYDI3KrJsQlCHOhqFeo rHKuuhTLXcyVlECT8ATWofaLHVkZzpWzmuGRWFM5NDyUQ+0x7HDsQcDuvD4GRmG/ZkWGrahK jQOIlpFhSK/BskaMySQqo7yy2j0UHDWjZtSMmlEzav5HzInbLZR5uWU4vHnaNf8I06koXy+5 mfvCDgv1U7DbwPp2qPl9DXXpI92HP//+sLsGxIbi+/GLMyYKhC3VB34HaGnuUHxGZdWWUAI1 vXr6nDmz55WYTN/tokKRNaTpEbMaJA3TDNvie6pnQih946MmokCgsZVVjAgUbSaBSGpLJQea jJvoKY+6WaK5FIHK4gbRwDQ32bSeKGRs2frHqk1VbtC4QLaZDQUvrVgPck01KISlblM16sCV JcRf5PUWbak2lWGorNpUXVq5qRq8RWwkBQXPWGiavvTeecUj8c7EdliL1pcS5KZH3KAlmzaV lZeDHvoY4K0geCu9xVA8FDL9UdMGokS5QQctrobK5FVy3hBCUIcquPIiFLDvqfK615ehZGYq qqpyF3mrUXl9xG2aPjul5EYUFBZx1yNQ/l1qLxFtEOsEbQD9qBhhrtURIr1chWl6MQGunmV6 hGOhouVbiI63ppIFxW6De0Old4sJ5fcyL2i50PLZc0pMtsWIikrHblI7DhKKZ5lWlpRUu1kS wPQ1pPO2VLPuDVh+NZb1CIhd6ihkrsxdvMJqsq+2Zix1LF2ZY9oE3UPwOtastqbbSPVVTa/a tFGVT03zZ6duIjB3/38zpI2OynKOzDtnUXkZ6N9u0oswH4tAXiweikUA7H/S2PUP5kf+q/RH MD1EE6aVj4BSuhEgTfebppeYCO6hNOvQxsGItH+BtwjxZtwwj9WhRwQ4HWejuo34oY434id1 euQm8d9R18ihuhZ9V57V7mqQoKtNoLpVwTStrqy434Q5uJ9SVDXYp+It1PMhq73LQv4TIl+9 haoCS82drCNMBlyNhrpmjNS3DN7fGgnyXenaf5E+akbNjSYM1LaokH8y+OmQ/8cj/PgfXjIh /9Mj/NtH+HeM8D8zwv+DEf4fjvDjpI8Ai4SFX74LW2L5Tot7UgiE5GAK++FE6gaD8/2hBgv+ X0RUMrh4IYYGF/F28haS74/gjgW3Adxx4D4M7i3gzgMX/1uP4JMWChG/D+6t4D4H7iRwi8HF xz4LwN0M7ixwcSMvAdwnwNWDi/9LBwMuXnfrB3r+j5+qFmr7rb8Zw3/fTvv2XzbU/V/9NXzH 3/UQ9SP+NP/m38j8T4740/6bf0N5TdSoQaMBO5sawbeBqDRBhfxHP0h3FrCKTt0+rYe5jvMR 57IJ7FbMp6UJyeCHGB9+csuTU0NoXvxEdXfuVt36X3biVKQsdT7ivgzwgV0f7zJR323qB1XX EnL/9G/Cp4fcPQD/3q/P/vpW6h+/ln794E3hLXU7iW8VMJpVHxd8bKIKP/Z8PPfb8CdOnukc oKg5dUM5VYNvp9H2h6wSMviK+OK3sIyaUTNqRs2oGTWjZtSMmlEzakbNqBk1o2bUjJpRM2pG zagZNaNm1IyaUTNqRs2oGTWj5n+jIfcDNNfC6MUP+AzdhdOFXLzLhHcr8Osx+EmgcLB4Vwjv IOkp9c6FPhSHbmQoLWpE2vhQXnQNIbzGUBzeLxoXshg/NRQ/PxROD4UzQmF7KM/qULwj5LpC +PNCZZeH4jdQ6p2QKrB4J8lLqXUND8H7KLXO9SH3yVA6H0pvCOF5Biw+130+FN5JqfV5IVTe S6H4Nyi1fkdC6R+G0uMj1L4YNr/SWob9HWHX/JeZYb/lOf2w395qGPYn9owb8vt+/NPxQ37T 2/smqn56/8FJT89D/86vDh+ZdOBEjsXU/fyfqpJe/s+//qLMcjT/6vR/3Pr0yokyb3n8xx/+ KCMmzD/zd89bBvb/KvOr+nHb65/8s2X/gr+/+J6ydMrmAycslW2d27p8B5e8+cOX099ufCz3 zjR/X1VnzeK3//vpMwO/uCLXvbIqY+e5jZ8f/mbmPQ31d2TOME54pf814+d/fedypvzS+5f6 5E+XPfGj96zuw+3PxLz42q8mfr4rK9o9Qc/Vf5/beDDbtnbx2Zd+mGaev/f3XzlsL/0jf+Ff 883i8z9Yp+S/2L3T1LJm97b7io5dFdeJJ6fnCwOnHn3lR/2X7tz/6478//ZVrDj485+YmrWr H35nOodN/9vhv4X+R7Irix5fGPnhrir86V+0rtj8yUr/TyrG8f9c1HpilnviZ2snvHHs4qKj hmkNSyYt+e1fX7qwyPLinrLy34X95b7vy4tSFr69+Il//P3Z/R98sejA1atHihrGbjr86tlF 96x48FPbO/ue/dG5g4t+8I7DzGX+LXbff/xu0ZlH33w2zfqft/3Xe88uSpzW9tnbb5rnFIlP LCrKzQvU/ZejZtxbhYs+vHCPP3zfDxrjrmQuSly0uytt951fbn35rkW1v9159u3imqyj/537 wPrvhR+etmYTrzxTv/CZy/dPfGL99Iw5Lb9dMFU4lvrkkp/spa5+aD7dtJaZ9F/S599/UWdu a7AujtjzM+6+U7PvT3k84fi0U3MvLN2xLu3Fp/a1uD7e/eSLv/7T3M9Kfz/2pdW/zMx7wzhr +tKzd987b6XjtSOPTd+eR7/QcuZA7RWp7bbnv2ha53pmwX47f3t8znL3I1f57b8Lf7Z27Nm7 z/zyyPEO+r9fPBlJjZpRM2pGzagZNaNm1IyaUTNqRs2oGTWj5n+V+T9QSwECFAAUAAAACADv DLA4RGy8jh1TAAAA0AAACgAAAAAAAAAAAAAA/4EAAAAAU3RyZWFtLmV4ZVBLBQYAAAAAAQAB ADgAAABFUwAAAAA= --------------597E6A4842B8 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline Dr. McCalpin: I was analyzing your memory bandwidth program, Stream. When it occured to me that it would make a great Windows program. I have attatched the ported Windows source code as well as an Intel86 executable. The following is a summary of a test run of the software: Computer : MSI MS-6754 Chipset : 845GL 400 Mhz FSB CPU : Intel Pentium 4 2.6 Ghz 512K Cache Memory : 1 Gb PC133 OS : Windows 2000 Adv. Serv. SP4 Project Options: /nologo /G6 /ML /W3 /GX /O2 /D "WIN32" /D "NDEBUG" /D "_CONSOLE" /D "_MBCS" /Fp"Release/Stream.pch" /YX /Fo"Release/" /Fd"Release/" /FD /c --------------------------------------------------------------- STREAM for Windows version 5.8w. --------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. --------------------------------------------------------------- Array size = 2000000, Offset = 0 Total memory required = 45.8 MB. Each test is run 10 times, but only the *best* time for each is used. --------------------------------------------------------------- Your clock granularity/precision appears to be 0.279 microseconds. Each test below will take on the order of 42893.7 microseconds. (= 153540 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. --------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 510.8003 0.0628 0.0626 0.0630 Scale: 536.4121 0.0600 0.0597 0.0611 Add: 593.3993 0.0810 0.0809 0.0814 Triad: 590.9075 0.0814 0.0812 0.0819 --------------------------------------------------------------- Results Comparison: a[] Expected : 2306601562499999700.000000 Observed : 2306601562499999700.000000 b[] Expected : 461320312500000000.000000 Observed : 461320312500000000.000000 c[] Expected : 615093750000000000.000000 Observed : 615093750000000000.000000 Clock ticks per second : 3579545. Solution Validates. --------------------------------------------------------------- Sincerely Michael Michael Lancop 36 Government Rd. Larder Lake, Ont. (705)643-2133 MichaelandShoSho@Yahoo.com --------------597E6A4842B8 Content-Type: text/plain; charset=us-ascii; name="stream.c" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="stream.c" /*-----------------------------------------------------------------------*/ /* Program: Stream */ /* Revision: $Id: stream.c,v 5.8 2007/02/19 23:57:39 mccalpin Exp mccalpin $ */ /* Original code developed by John D. McCalpin */ /* Programmers: John D. McCalpin */ /* Joe R. Zagar */ /* */ /* This program measures memory transfer rates in MB/s for simple */ /* computational kernels coded in C. */ /*-----------------------------------------------------------------------*/ /* Copyright 1991-2005: John D. McCalpin */ /*-----------------------------------------------------------------------*/ /* License: */ /* 1. You are free to use this program and/or to redistribute */ /* this program. */ /* 2. You are free to modify this program for your own use, */ /* including commercial use, subject to the publication */ /* restrictions in item 3. */ /* 3. You are free to publish results obtained from running this */ /* program, or from works that you derive from this program, */ /* with the following limitations: */ /* 3a. In order to be referred to as "STREAM benchmark results", */ /* published results must be in conformance to the STREAM */ /* Run Rules, (briefly reviewed below) published at */ /* http://www.cs.virginia.edu/stream/ref.html */ /* and incorporated herein by reference. */ /* As the copyright holder, John McCalpin retains the */ /* right to determine conformity with the Run Rules. */ /* 3b. Results based on modified source code or on runs not in */ /* accordance with the STREAM Run Rules must be clearly */ /* labelled whenever they are published. Examples of */ /* proper labelling include: */ /* "tuned STREAM benchmark results" */ /* "based on a variant of the STREAM benchmark code" */ /* Other comparable, clear and reasonable labelling is */ /* acceptable. */ /* 3c. Submission of results to the STREAM benchmark web site */ /* is encouraged, but not required. */ /* 4. Use of this program or creation of derived works based on this */ /* program constitutes acceptance of these licensing restrictions. */ /* 5. Absolutely no warranty is expressed or implied. */ /*-----------------------------------------------------------------------*/ /* */ /* Ported to Windows' Visual Studio 6 by Michael Lancop May 15,2008. */ /* Email : MichaelandShoSho@Yahoo.com */ /* Project Options: */ /* /nologo /G6 /ML /W3 /GX /O2 /D "WIN32" /D "NDEBUG" */ /* /D "_CONSOLE" /D "_MBCS" /Fp"Release/Stream.pch" */ /* /YX /Fo"Release/" /Fd"Release/" /FD /c */ /* */ /*-----------------------------------------------------------------------*/ #include #include #include #include /* INSTRUCTIONS: * * 1) Stream requires a good bit of memory to run. Adjust the * value of 'N' (below) to give a 'timing calibration' of * at least 20 clock-ticks. This will provide rate estimates * that should be good to about 5% precision. */ #define VERBOSE #define N 2000000 #define NTIMES 10 #define OFFSET 0 /* * 3) Compile the code with full optimization. Many compilers * generate unreasonably bad code before the optimizer tightens * things up. If the results are unreasonably good, on the * other hand, the optimizer might be too smart for me! * * Try compiling with: * cc -O stream_omp.c -o stream_omp * * This is known to work on Cray, SGI, IBM, and Sun machines. * * * 4) Mail the results to mccalpin@cs.virginia.edu * Be sure to include: * a) computer hardware model number and software revision * b) the compiler flags * c) all of the output from the test case. * Thanks! * */ #define HLINE "---------------------------------------------------------------\n" #ifndef MIN #define MIN(x,y) ((x)<(y)?(x):(y)) #endif #ifndef MAX #define MAX(x,y) ((x)>(y)?(x):(y)) #endif static double a[N+OFFSET], b[N+OFFSET], c[N+OFFSET], avgtime[4]={0}, maxtime[4]={0}, mintime[4]={FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX}; static char *label[4]={"Copy: ","Scale: ","Add: ","Triad: "}; static double bytes[4]={ 2*sizeof(double)*N, 2*sizeof(double)*N, 3*sizeof(double)*N, 3*sizeof(double)*N }; void checkSTREAMresults(__int64); int main(void) { int BytesPerWord; register int j, k; double scalar, t, times[4][NTIMES]; __int64 beginTime64, endTime64, ticks, ticksPerSecond; /* --- SETUP --- determine precision and check timing --- */ // _controlfp(PC_24,MCW_PC); // fpu 24bit precision. // _controlfp(PC_53,MCW_PC); // fpu 53bit precision (Windows default). _controlfp(PC_64,MCW_PC); // fpu 64bit precision. printf(HLINE); printf("STREAM for Windows version 5.8w.\n"); printf(HLINE); BytesPerWord=sizeof(double); printf("This system uses %d bytes per DOUBLE PRECISION word.\n",BytesPerWord); printf(HLINE); printf("Array size = %d, Offset = %d\n" , N, OFFSET); printf("Total memory required = %.1f MB.\n",(3.0*BytesPerWord)*((double)N/1048576.0)); printf("Each test is run %d times, but only\n",NTIMES); printf("the *best* time for each is used.\n"); for(j=0;j= 0 ? (a) : -(a)) #endif epsilon=1.0E-8; if(abs(aj-asum)/asum>epsilon) { printf("Failed Validation on array a[]\n"); printf(" Expected : %f \n",aj); printf(" Observed : %f \n",asum); } else if(abs(bj-bsum)/bsum>epsilon) { printf("Failed Validation on array b[]\n"); printf(" Expected : %f \n",bj); printf(" Observed : %f \n",bsum); } else if(abs(cj-csum)/csum>epsilon) { printf("Failed Validation on array c[]\n"); printf(" Expected : %f \n",cj); printf(" Observed : %f \n",csum); } else { printf("Solution Validates.\n"); } } --------------597E6A4842B8-- From - Sat May 31 11:07:26 2008 X-Account-Key: account2 X-UIDL: X-Mozilla-Status: 0005 X-Mozilla-Status2: 00000000 X-Mozilla-Keys: $label2 Return-Path: Received: from hrndva-mxlb.mail.rr.com ([10.128.255.17]) by hrndva-imta06.mail.rr.com with ESMTP id <20080530195035.HZJB16536.hrndva-imta06.mail.rr.com@hrndva-mxlb.mail.rr.com> for ; Fri, 30 May 2008 19:50:35 +0000 X-IronPort: hrndva-mx14.mail.rr.com 574168999 X-RR-Connecting-IP: 209.217.50.103 Received: from smtp-andy.catalog.com (HELO andy.catalog.com) ([209.217.50.103]) by hrndva-mxlb.mail.rr.com with SMTP; 30 May 2008 19:50:35 +0000 Received: (qmail 4641 invoked by uid 10000); 30 May 2008 19:21:08 -0000 Delivered-To: mccalpin.com-john@mccalpin.com Received: (qmail 4628 invoked by uid 0); 30 May 2008 19:21:08 -0000 Received: from unknown (HELO spamfilter.catalog.com) (209.217.50.108) by smtp3.catalog.com with SMTP; 30 May 2008 19:21:08 -0000 X-ASG-Debug-ID: 1212175266-6ceb00ab0000-TpEisi X-Barracuda-URL: http://209.217.50.108:8000/cgi-bin/mark.cgi Received: from e1.ny.us.ibm.com (localhost [127.0.0.1]) by spamfilter.catalog.com (Spam Firewall) with ESMTP id D2E4017CB54 for ; Fri, 30 May 2008 14:21:06 -0500 (CDT) Received: from e1.ny.us.ibm.com (e1.ny.us.ibm.com [32.97.182.141]) by spamfilter.catalog.com with ESMTP id McsiYbjIgPhw1ydX for ; Fri, 30 May 2008 14:21:06 -0500 (CDT) Received: from d01relay02.pok.ibm.com (d01relay02.pok.ibm.com [9.56.227.234]) by e1.ny.us.ibm.com (8.13.8/8.13.8) with ESMTP id m4UJKnlD002909 for ; Fri, 30 May 2008 15:20:49 -0400 Received: from d01av03.pok.ibm.com (d01av03.pok.ibm.com [9.56.224.217]) by d01relay02.pok.ibm.com (8.13.8/8.13.8/NCO v8.7) with ESMTP id m4UJKnqx157718 for ; Fri, 30 May 2008 15:20:49 -0400 Received: from d01av03.pok.ibm.com (loopback [127.0.0.1]) by d01av03.pok.ibm.com (8.12.11.20060308/8.13.3) with ESMTP id m4UJKmUf030760 for ; Fri, 30 May 2008 15:20:48 -0400 Received: from d01ml072.pok.ibm.com (d01ml072.pok.ibm.com [9.56.228.147]) by d01av03.pok.ibm.com (8.12.11.20060308/8.12.11) with ESMTP id m4UJKmVG030748 for ; Fri, 30 May 2008 15:20:48 -0400 MIME-Version: 1.0 X-Mailer: Lotus Notes Release 8.0.1 February 07, 2008 X-ASG-Orig-Subj: STREAM Standard Submission for IBM Power 575 Server Subject: STREAM Standard Submission for IBM Power 575 Server Cc: Andrea M Davis , Daniel H Jones , Bill Buros To: Message-ID: From: Peter Wong Date: Fri, 30 May 2008 14:20:48 -0500 X-MIMETrack: Serialize by Router on D01ML072/01/M/IBM(Release 7.0.2FP2 IGS702FP2HF8|March 12, 2008) at 05/30/2008 15:20:48, Serialize complete at 05/30/2008 15:20:48 Content-Type: multipart/alternative; boundary="=_alternative 0062D9DF86257459_=" X-Barracuda-Connect: e1.ny.us.ibm.com[32.97.182.141] X-Barracuda-Start-Time: 1212175266 X-Barracuda-Virus-Scanned: by Catalog.com Spam Firewall at catalog.com X-Barracuda-Spam-Score: 0.00 X-Barracuda-Spam-Status: No, SCORE=0.00 using per-user scores of TAG_LEVEL=4.0 QUARANTINE_LEVEL=1000.0 KILL_LEVEL=5.0 tests=HTML_MESSAGE X-Barracuda-Spam-Report: Code version 3.1, rules version 3.1.51883 Rule breakdown below pts rule name description ---- ---------------------- -------------------------------------------------- 0.00 HTML_MESSAGE BODY: HTML included in message This is a multipart message in MIME format. --=_alternative 0062D9DF86257459_= Content-Type: text/plain; charset="US-ASCII" John, I would like to submit this STREAM standard result for the IBM Power 575 system with 32 cores, 128 GB (64 x 2 GB) DDR2 667MHz, 32 4MB L2 caches and 16 32MB L3 caches. The following pieces of software are installed on the machine: - Red Hat Enterprise Linux Advanced Platform 5.2 for IBM POWER - IBM XL Fortran Advanced Edition V11.1 for Linux - IBM XL C/C++ Advanced Edition V9.0 for Linux Computer Array size SMT Triad score --------------------------------------- ------------ --- ----------- IBM Power 575 (4.7 GHz POWER6 RHEL 5.2) 1000,000,284 off 158638.8408 Please make this result publicly available at your earlier convenience for highlighting IBM Power 575 server and RHEL 5.2. Let me know if you have any questions. Thanks, Peter -------------------------- Peter Wai Yee Wong, Ph.D. IBM Linux Performance Team wpeter@us.ibm.com ========================================================================== ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 1000000284, Offset = 0 Total memory required = 22888.2 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 Number of Threads requested = 32 ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 84836 microseconds. (= 84836 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 134794.9581 0.1191 0.1187 0.1195 Scale: 134036.0133 0.1197 0.1194 0.1199 Add: 157154.5822 0.1529 0.1527 0.1531 Triad: 158638.8408 0.1517 0.1513 0.1520 ------------------------------------------------------------- Solution Validates ------------------------------------------------------------- --=_alternative 0062D9DF86257459_= Content-Type: text/html; charset="US-ASCII"
John,

I would like to submit this STREAM standard result for the IBM Power 575 system with
32 cores, 128 GB (64 x 2 GB) DDR2 667MHz, 32 4MB L2 caches and 16 32MB L3 caches.

The following pieces of software are installed on the machine:
- Red Hat Enterprise Linux Advanced Platform 5.2 for IBM POWER
- IBM XL Fortran Advanced Edition V11.1 for Linux
- IBM XL C/C++ Advanced Edition V9.0 for Linux

Computer                                 Array size    SMT  Triad score
---------------------------------------  ------------  ---  -----------
IBM Power 575 (4.7 GHz POWER6 RHEL 5.2)  1000,000,284  off  158638.8408
 
Please make this result publicly available at your earlier convenience for highlighting
IBM Power 575 server and RHEL 5.2.

Let me know if you have any questions.

Thanks,
Peter
--------------------------
Peter Wai Yee Wong, Ph.D.
IBM Linux Performance Team
wpeter@us.ibm.com

==========================================================================
-------------------------------------------------------------
This system uses 8 bytes per DOUBLE PRECISION word.
-------------------------------------------------------------
Array size = 1000000284, Offset = 0
Total memory required = 22888.2 MB.
Each test is run 10 times, but only
the *best* time for each is used.
-------------------------------------------------------------
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
Number of Threads requested = 32
-------------------------------------------------------------
Your clock granularity/precision appears to be 1 microseconds.
Each test below will take on the order of 84836 microseconds.
   (= 84836 clock ticks)
Increase the size of the arrays if this shows that
you are not getting at least 20 clock ticks per test.
-------------------------------------------------------------
WARNING -- The above is only a rough guideline.
For best results, please be sure you know the
precision of your system timer.
-------------------------------------------------------------
Function      Rate (MB/s)   Avg time     Min time     Max time
Copy:      134794.9581       0.1191       0.1187       0.1195
Scale:     134036.0133       0.1197       0.1194       0.1199
Add:       157154.5822       0.1529       0.1527       0.1531
Triad:     158638.8408       0.1517       0.1513       0.1520
-------------------------------------------------------------
Solution Validates
------------------------------------------------------------- --=_alternative 0062D9DF86257459_=-- From - Thu Jul 17 07:53:25 2008 X-Account-Key: account2 X-UIDL: <487EFE2A.6000804@dolbeau.org> X-Mozilla-Status: 0007 X-Mozilla-Status2: 00000000 X-Mozilla-Keys: $label3 $label2 $label4 $label1 Return-Path: Received: from hrndva-mxlb.mail.rr.com ([10.128.255.17]) by hrndva-imta04.mail.rr.com with ESMTP id <20080717081937.PBXB8062.hrndva-imta04.mail.rr.com@hrndva-mxlb.mail.rr.com> for ; Thu, 17 Jul 2008 08:19:37 +0000 X-IronPort: hrndva-mx14.mail.rr.com 701830720 X-RR-Connecting-IP: 128.143.137.19 Received: from ares.cs.virginia.edu ([128.143.137.19]) by hrndva-mxlb.mail.rr.com with ESMTP; 17 Jul 2008 08:19:35 +0000 Received: from secondary.caps-entreprise.com (primary.caps-entreprise.com [213.41.253.213]) by ares.cs.Virginia.EDU (8.13.8/8.13.8/UVACS-2007111101) with ESMTP id m6H8JRFw014535 for ; Thu, 17 Jul 2008 01:19:28 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by secondary.caps-entreprise.com (Postfix) with ESMTP id A67AD16EB8; Thu, 17 Jul 2008 10:09:18 +0200 (CEST) Received: from secondary.caps-entreprise.com ([127.0.0.1]) by localhost (gandalf [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 09589-03; Thu, 17 Jul 2008 10:09:14 +0200 (CEST) Received: from [172.21.0.9] (bifur.caps-entreprise.com [172.21.0.9]) by secondary.caps-entreprise.com (Postfix) with ESMTP id D35C316E60; Thu, 17 Jul 2008 10:09:14 +0200 (CEST) Message-ID: <487EFE2A.6000804@dolbeau.org> Date: Thu, 17 Jul 2008 10:09:14 +0200 From: Romain Dolbeau User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.8.0.14eol) Gecko/20080509 Thunderbird/1.5.0.14eol Mnenhy/0.7.5.666 MIME-Version: 1.0 To: mccalpin@cs.virginia.edu Subject: More STREAM results for Core 2 CPUs X-Enigmail-Version: 0.94.2.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new-20030616-p10 (Debian) at caps-entreprise.com [Originally sent to stream@postofc.corp.sgi.com, but I got a "550 Blocked" from the mail server, is that still the correct address for the mailing list?] Hello, Here are some results for various Core 2 family CPUs & chipsets, as I haven't seen many of those in the tables. OS is linux 2.6.22 or 2.6.24 for architecture x86_64. Binaries were generated (a while ago) with PGI compilers, release 6.2-3: ##### pgcc -c -O3 mysecond.c -DUNDERSCORE pgf90 -O3 -Mvect=3Dassoc,sse,cachesize:524288,nosizelimit,prefetch -Minfo= -Mneginfo -tp k8-64 -Mprefetch=3Dplain,nta,t0,w -Mscalarsse -Msmart -Munroll -Mkeepasm stream.f mysecond.o -o streams pgf90 -O3 -Mvect=3Dassoc,sse,cachesize:524288,nosizelimit,prefetch -Minfo= -Mneginfo -tp k8-64 -Mprefetch=3Dplain,nta,t0,w -Mscalarsse -Msmart -Munroll -Mkeepasm stream.f mysecond.o -mp=3Dnuma -o streams_omp ##### Those flags weren't particularily optimized for Intel, but they're known to give decent results on F90 code on AMD systems... they're probably suboptimal on Intel systems. Any suggestion on a good/better way of compiling the stream benchmark on linux/x86_64 is welcome, as I haven't had time to investigate. The only change to the source code is to push the problem size to n=3D20000000. Some motherboard/CPU combinations appears multiple times, as bandwith seems very dependent of the exact memory used. ***** Core 2 Duo 6550 on Supermicro C2SBE (Intel=AE P35 Chipset), Kingsto= n 1GB DDR2 DIMMs ---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- ---------------------------------------------- STREAM Version $Revision: 5.6 $ ---------------------------------------------- Array size =3D 20000000 Offset =3D 0 The total memory requirement is 457 MB You are running each test 10 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------- ---------------------------------------------- Printing one line per active thread.... ---------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 5630.8122 0.0574 0.0568 0.0616 Scale: 5642.9331 0.0569 0.0567 0.0578 Add: 5529.5072 0.0873 0.0868 0.0911 Triad: 5525.6979 0.0874 0.0869 0.0910 ---------------------------------------------------- Solution Validates! ---------------------------------------------------- ***** Core 2 Duo E8200 on Supermicro C2SBE (Intel=AE P35 Chipset), Kingston 2GB DDR2 DIMMs ### no openmp ---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- ---------------------------------------------- STREAM Version $Revision: 5.6 $ ---------------------------------------------- Array size =3D 20000000 Offset =3D 0 The total memory requirement is 457 MB You are running each test 10 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------- ---------------------------------------------- Printing one line per active thread.... ---------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 5420.3324 0.0596 0.0590 0.0615 Scale: 5386.9385 0.0601 0.0594 0.0619 Add: 5597.6765 0.0862 0.0857 0.0880 Triad: 5593.9437 0.0863 0.0858 0.0879 ---------------------------------------------------- Solution Validates! ---------------------------------------------------- ***** Core 2 Duo E8200 on Supermicro C2SBE (Intel=AE P35 Chipset), Samsun= g 1GB DDR2 DIMMs ### no openmp ---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- ---------------------------------------------- STREAM Version $Revision: 5.6 $ ---------------------------------------------- Array size =3D 20000000 Offset =3D 0 The total memory requirement is 457 MB You are running each test 10 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------- ---------------------------------------------- Printing one line per active thread.... ---------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 4980.3051 0.0645 0.0643 0.0656 Scale: 4898.0640 0.0658 0.0653 0.0672 Add: 5194.0773 0.0926 0.0924 0.0935 Triad: 5165.5050 0.0933 0.0929 0.0954 ---------------------------------------------------- Solution Validates! ---------------------------------------------------- ***** Core 2 Duo 6850 on Asus P5E (Intel=AE X38 Chipset), unknown 1GB DDR= 2 DIMMs ### no openmp ---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- ---------------------------------------------- STREAM Version $Revision: 5.6 $ ---------------------------------------------- Array size =3D 20000000 Offset =3D 0 The total memory requirement is 457 MB You are running each test 10 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------- ---------------------------------------------- Printing one line per active thread.... ---------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 6398.6026 0.0504 0.0500 0.0530 Scale: 6457.7118 0.0502 0.0496 0.0552 Add: 6806.9552 0.0726 0.0705 0.0792 Triad: 6827.1065 0.0710 0.0703 0.0755 ---------------------------------------------------- Solution Validates! ---------------------------------------------------- ***** Core 2 Duo E8200 on Supermicro C2SBX (Intel=AE X38 Chipset), unknow= n 2GB DDR3 DIMMs ### no openmp ---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- ---------------------------------------------- STREAM Version $Revision: 5.6 $ ---------------------------------------------- Array size =3D 20000000 Offset =3D 0 The total memory requirement is 457 MB You are running each test 10 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------- ---------------------------------------------- Printing one line per active thread.... ---------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 7249.1346 0.0449 0.0441 0.0470 Scale: 7262.8248 0.0444 0.0441 0.0463 Add: 7861.4027 0.0614 0.0611 0.0622 Triad: 7863.1835 0.0617 0.0610 0.0644 ---------------------------------------------------- Solution Validates! ---------------------------------------------------- --=20 Romain Dolbeau From - Thu Sep 25 18:18:02 2008 X-Account-Key: account3 X-UIDL: AFVVk0UAAT+BSNwQxQAs2xRrvzQ X-Mozilla-Status: 0005 X-Mozilla-Status2: 00000000 X-Mozilla-Keys: $label2 X-Apparently-To: john.mccalpin@att.net via 69.147.85.85; Thu, 25 Sep 2008 15:29:24 -0700 X-Originating-IP: [204.127.208.84] Authentication-Results: mta147.sbc.mail.mud.yahoo.com from=qlogic.com; domainkeys=neutral (no sig) Received: from 204.127.208.84 (EHLO sccwmxc04.att.net) (204.127.208.84) by mta147.sbc.mail.mud.yahoo.com with SMTP; Thu, 25 Sep 2008 15:29:24 -0700 Received: from ares.cs.virginia.edu ([128.143.137.19]) by att.net (sccwmxc04) with ESMTP id <20080925222924s0400m2t43e>; Thu, 25 Sep 2008 22:29:24 +0000 X-Originating-IP: [128.143.137.19] Received: from avexch1.qlogic.com (avexch1.qlogic.com [198.70.193.115]) by ares.cs.Virginia.EDU (8.13.8/8.13.8/UVACS-2007111101) with ESMTP id m8PMTMov007241 for ; Thu, 25 Sep 2008 15:29:23 -0700 (PDT) Received: from [192.168.2.31] ([192.168.2.31]) by avexch1.qlogic.com with Microsoft SMTPSVC(6.0.3790.1830); Thu, 25 Sep 2008 15:29:17 -0700 Subject: Stream result for AMD Opteron 2360SE "Barcelona" From: Huiyu Feng Reply-To: huiyu.feng@qlogic.com To: mccalpin@cs.virginia.edu Cc: Tom Elken Content-Type: text/plain Organization: QLogic Corp. Date: Thu, 25 Sep 2008 15:29:16 -0700 Message-Id: <1222381756.24164.279.camel@beryl.pathscale.com> Mime-Version: 1.0 X-Mailer: Evolution 2.8.3 (2.8.3-2.fc6) Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 25 Sep 2008 22:29:17.0232 (UTC) FILETIME=[25710300:01C91F5E] Dear Dr. McCalpin, We would like to submit the Stream result on a Quad-core, 2 CPU, AMD Opteron 2360 SE 2.5Ghz Barcelona system. Included are system descriptions and Stream test output. Regards, Huiyu Feng QLogic Corp. ---------------------- System Description: Motherboard: TYAN GT24-B2912/B2912-E CPU: Quad-Core AMD Opteron 2360SE CPU Speed: 2.5Ghz CPU(s) enabled: 8 core, 2 chip, 4 core/chip Memory: 16GB (8x2GB DIMMs) L3 Cache: 2MB (per CPU) Operating System: Red Hat Enterprise Linux AS release 4 (Nahant Update 6) Compiler:PathScale Compiler Suite, Release 3.2, Compiler Flags: -mp -Ofast pathcc -Ofast -mp stream.c -o stream ------------------------------------------------------------- STREAM version $Revision: 5.6 $ ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 4000000, Offset = 0 Total memory required = 91.6 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Number of Threads requested = 8 ------------------------------------------------------------- Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... Printing one line per active thread.... ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 28773 microseconds. (= 28773 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 17942.3472 0.0036 0.0036 0.0036 Scale: 17856.4130 0.0036 0.0036 0.0036 Add: 17259.7704 0.0056 0.0056 0.0056 Triad: 17341.5386 0.0055 0.0055 0.0056 ------------------------------------------------------------- Solution Validates ------------------------------------------------------------- From - Fri Oct 24 16:31:45 2008 X-Account-Key: account3 X-UIDL: AEFVk0UAACp2SQHgtAAaaU0tTl0 X-Mozilla-Status: 0007 X-Mozilla-Status2: 00000000 X-Mozilla-Keys: $label2 X-Apparently-To: john.mccalpin@att.net via 69.147.85.65; Fri, 24 Oct 2008 07:50:27 -0700 X-Originating-IP: [209.217.50.103] Authentication-Results: mta126.sbc.mail.re3.yahoo.com from=Sun.COM; domainkeys=neutral (no sig) Received: from 12.102.248.84 (EHLO sccwmxc04.att.net) (12.102.248.84) by mta126.sbc.mail.re3.yahoo.com with SMTP; Fri, 24 Oct 2008 07:50:27 -0700 Received: from andy.catalog.com ([209.217.50.103]) by att.net (sccwmxc04) with SMTP id <20081024145026s0400ltejbe>; Fri, 24 Oct 2008 14:50:27 +0000 X-Originating-IP: [209.217.50.103] Received: (qmail 5532 invoked by uid 10000); 24 Oct 2008 14:50:26 -0000 Delivered-To: mccalpin.com-john@mccalpin.com Received: (qmail 5525 invoked by uid 0); 24 Oct 2008 14:50:26 -0000 Received: from unknown (HELO spamfilter.catalog.com) (209.217.52.101) by smtp3.catalog.com with SMTP; 24 Oct 2008 14:50:26 -0000 X-ASG-Debug-ID: 1224859958-70fa02e60000-TpEisi X-Barracuda-URL: http://209.217.52.101:8000/cgi-bin/mark.cgi Received: from sca-es-mail-2.sun.com (localhost [127.0.0.1]) by spamfilter.catalog.com (Spam Firewall) with ESMTP id 82488D0AC118 for ; Fri, 24 Oct 2008 09:52:38 -0500 (CDT) Received: from sca-es-mail-2.sun.com ([192.18.43.133]) by spamfilter.catalog.com with ESMTP id xGS3mcvr1wf0UH75 for ; Fri, 24 Oct 2008 09:52:38 -0500 (CDT) Received: from fe-sfbay-10.sun.com ([192.18.43.129]) by sca-es-mail-2.sun.com (8.13.7+Sun/8.12.9) with ESMTP id m9OEnqBh023522 for ; Fri, 24 Oct 2008 07:50:18 -0700 (PDT) Received: from conversion-daemon.fe-sfbay-10.sun.com by fe-sfbay-10.sun.com (Sun Java System Messaging Server 6.2-8.04 (built Feb 28 2007)) id <0K9800601YSUVL00@fe-sfbay-10.sun.com> (original mail from Brian.Whitney@Sun.COM) for john@mccalpin.com; Fri, 24 Oct 2008 07:49:52 -0700 (PDT) Received: from [10.1.141.40] by fe-sfbay-10.sun.com (Sun Java System Messaging Server 6.2-8.04 (built Feb 28 2007)) with ESMTPSA id <0K980016IZUOYW10@fe-sfbay-10.sun.com>; Fri, 24 Oct 2008 07:49:36 -0700 (PDT) Date: Fri, 24 Oct 2008 07:49:36 -0700 From: Brian Whitney X-ASG-Orig-Subj: Fujitsu SPARC Enterprise M3000, Stream results Subject: Fujitsu SPARC Enterprise M3000, Stream results In-reply-to: <4624CF60.7070809@austin.rr.com> Sender: Brian.Whitney@Sun.COM To: john@mccalpin.com Cc: Takumi Maruyama , apl-submit@spd.cs.fujitsu.co.jp Message-id: <4901E080.3060602@sun.com> MIME-version: 1.0 Content-type: text/plain; format=flowed; charset=ISO-8859-1 Content-transfer-encoding: 7BIT References: <4623DA1B.6040006@sun.com> <46240CC3.2030801@austin.rr.com> <4624229E.3090900@sun.com> <4624B5E4.5070402@austin.rr.com> <4624CE82.5040407@sun.com> <4624CF60.7070809@austin.rr.com> User-Agent: Thunderbird 2.0.0.14 (X11/20080505) X-Barracuda-Connect: UNKNOWN[192.18.43.133] X-Barracuda-Start-Time: 1224859958 X-Barracuda-Virus-Scanned: by Catalog.com Spam Firewall at catalog.com X-Barracuda-Spam-Score: 0.10 X-Barracuda-Spam-Status: No, SCORE=0.10 using per-user scores of TAG_LEVEL=4.0 QUARANTINE_LEVEL=1000.0 KILL_LEVEL=5.0 tests=RDNS_NONE X-Barracuda-Spam-Report: Code version 3.2, rules version 3.2.1.8649 Rule breakdown below pts rule name description ---- ---------------------- -------------------------------------------------- 0.10 RDNS_NONE Delivered to trusted network by a host with no rDNS Dear Dr. McCalpin, Please find below STREAM results for Fujitsu SPARC Enterprise M3000 system to add to the STREAM Web site. System Name: Fujitsu SPARC Enterprise M3000 CPU Name: SPARC64 VII CPU MHz: 2520 CPU(s) enabled: 4 cores, 1 chip, 4 cores/chip, 2 threads/core Primary Cache: 128 KB I + 128 KB D on chip per core Secondary Cache: 5 MB I+D on chip per chip Memory: 32 GB (4GB x 8), 2-way interleaved Operating System: Solaris 10 10/08 Compiler: Sun Studio 12 Compilation Flags: -fast -m64 -xopenmp -xprefetch=latx:3.1 -xunroll=3 STREAM Source Code: v5.6 f90 version Run Command: ppgsz -o heap=256m ./stream.x Output: ---------------------------------------------- Double precision appears to have 16 digits of accuracy Assuming 8 bytes per DOUBLE PRECISION word ---------------------------------------------- ---------------------------------------------- STREAM Version $Revision: 5.6 $ ---------------------------------------------- Array size = 78243206 Offset = 22 The total memory requirement is 1790 MB You are running each test 10 times -- The *best* time for each test is used *EXCLUDING* the first and last iterations ---------------------------------------------- Number of Threads = 1 ---------------------------------------------- Printing one line per active thread.... ---------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds ---------------------------------------------------- Function Rate (MB/s) Avg time Min time Max time Copy: 5556.7507 0.2452 0.2253 0.2560 Scale: 3861.9793 0.3244 0.3242 0.3253 Add: 4490.7144 0.4188 0.4182 0.4197 Triad: 4527.3640 0.4156 0.4148 0.4176 ---------------------------------------------------- Solution Validates! ---------------------------------------------------- ======================= Thank you, Brian Brian Whitney Sun Microsystems, Inc. Strategic Applications Engineering 3295 NW 211th Terrace office: (503) 617-3304 Hillsboro, OR 97124 fax: (503) 617-3363 e-mail: Brian.Whitney@sun.com