From ssha001 at fiu.edu Mon Jul 9 16:46:59 2012 From: ssha001 at fiu.edu (Shi Sha) Date: Mon, 9 Jul 2012 19:46:59 -0400 Subject: [Hotspot] Why PTscalar power trace stay the same? Message-ID: Hello, all I am using PTScalar to generate the power trace. However, no matter what workload I use I got the same power trace as following: starting SimpleScalar sim: ** starting performance simulation ** FIG starts 0.00484 0.01301 0.00484 0.01368 0.00702 0.01368 0.00702 0.01301 0.00484 0.01301 decode 0.00484 0.01228 0.00484 0.01301 0.00794 0.01301 0.00794 0.01228 0.00484 0.01228 branch 0.00702 0.01301 0.00702 0.01456 0.00795 0.01456 0.00795 0.01301 0.00702 0.01301 RAT 0.00702 0.01456 0.00702 0.01591 0.00927 0.01591 0.00927 0.01456 0.00702 0.01456 RUU 0.00795 0.01361 0.00795 0.01456 0.00927 0.01456 0.00927 0.01361 0.00795 0.01361 LSQ 0.00927 0.01467 0.00927 0.01523 0.01107 0.01523 0.01107 0.01467 0.00927 0.01467 IALU1 0.00927 0.01412 0.00927 0.01467 0.01107 0.01467 0.01107 0.01412 0.00927 0.01412 IALU2 0.00927 0.01357 0.00927 0.01412 0.01107 0.01412 0.01107 0.01357 0.00927 0.01357 IALU3 0.00927 0.01301 0.00927 0.01357 0.01107 0.01357 0.01107 0.01301 0.00927 0.01301 IALU4 0.00484 0.01368 0.00484 0.01457 0.00702 0.01457 0.00702 0.01368 0.00484 0.01368 FPAdd 0.00484 0.01495 0.00484 0.01591 0.00702 0.01591 0.00702 0.01495 0.00484 0.01495 FPMul 0.00927 0.01523 0.00927 0.01591 0.01107 0.01591 0.01107 0.01523 0.00927 0.01523 IntReg 0.00484 0.01457 0.00484 0.01495 0.00702 0.01495 0.00702 0.01457 0.00484 0.01457 FPReg 0.00795 0.01301 0.00795 0.01361 0.00927 0.01361 0.00927 0.01301 0.00795 0.01301 ITLB 0.00484 0.00968 0.00484 0.01228 0.00805 0.01228 0.00805 0.00968 0.00484 0.00968 IL1 0.00794 0.01228 0.00794 0.01301 0.01107 0.01301 0.01107 0.01228 0.00794 0.01228 DTB 0.00805 0.00968 0.00805 0.01228 0.01107 0.01228 0.01107 0.00968 0.00805 0.00968 DL1 0.00000 0.00968 0.00000 0.01591 0.00484 0.01591 0.00484 0.00968 0.00000 0.00968 L2_left 0.00000 0.00000 0.00000 0.00968 0.01591 0.00968 0.01591 0.00000 0.00000 0.00000 L2_bottom 0.01107 0.00968 0.01107 0.01591 0.01591 0.01591 0.01591 0.00968 0.01107 0.00968 I use the following command:./sim-outorder -config ../spec95-little/go.cfg ../spec95-little/go.ss 2 8 This benchmark works well on sim-wattch, and sim-profile. I got no idea why the power trace stay the same whatever the workload I use. Looking forward for your advises. Thank you and have a good day. Best regards, Shi July 9, 2012 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120709/60d094c3/attachment.html From skadron at cs.virginia.edu Mon Jul 9 18:13:46 2012 From: skadron at cs.virginia.edu (Kevin Skadron) Date: Mon, 09 Jul 2012 21:13:46 -0400 Subject: [Hotspot] Why PTscalar power trace stay the same? In-Reply-To: References: Message-ID: <4FFB81CA.7070906@cs.virginia.edu> I'm afraid we haven't worked with PTScalar, so we can't help you. You might try contacting them. --Kevin On 7/9/2012 7:46 PM, Shi Sha wrote: > Hello, all > > I am using PTScalar to generate the power trace. However, no > matter what workload I use I got the same power trace as following: > > starting SimpleScalar > sim: ** starting performance simulation ** > FIG starts > 0.00484 0.01301 0.00484 0.01368 0.00702 0.01368 0.00702 0.01301 0.00484 > 0.01301 > decode > 0.00484 0.01228 0.00484 0.01301 0.00794 0.01301 0.00794 0.01228 0.00484 > 0.01228 > branch > 0.00702 0.01301 0.00702 0.01456 0.00795 0.01456 0.00795 0.01301 0.00702 > 0.01301 > RAT > 0.00702 0.01456 0.00702 0.01591 0.00927 0.01591 0.00927 0.01456 0.00702 > 0.01456 > RUU > 0.00795 0.01361 0.00795 0.01456 0.00927 0.01456 0.00927 0.01361 0.00795 > 0.01361 > LSQ > 0.00927 0.01467 0.00927 0.01523 0.01107 0.01523 0.01107 0.01467 0.00927 > 0.01467 > IALU1 > 0.00927 0.01412 0.00927 0.01467 0.01107 0.01467 0.01107 0.01412 0.00927 > 0.01412 > IALU2 > 0.00927 0.01357 0.00927 0.01412 0.01107 0.01412 0.01107 0.01357 0.00927 > 0.01357 > IALU3 > 0.00927 0.01301 0.00927 0.01357 0.01107 0.01357 0.01107 0.01301 0.00927 > 0.01301 > IALU4 > 0.00484 0.01368 0.00484 0.01457 0.00702 0.01457 0.00702 0.01368 0.00484 > 0.01368 > FPAdd > 0.00484 0.01495 0.00484 0.01591 0.00702 0.01591 0.00702 0.01495 0.00484 > 0.01495 > FPMul > 0.00927 0.01523 0.00927 0.01591 0.01107 0.01591 0.01107 0.01523 0.00927 > 0.01523 > IntReg > 0.00484 0.01457 0.00484 0.01495 0.00702 0.01495 0.00702 0.01457 0.00484 > 0.01457 > FPReg > 0.00795 0.01301 0.00795 0.01361 0.00927 0.01361 0.00927 0.01301 0.00795 > 0.01301 > ITLB > 0.00484 0.00968 0.00484 0.01228 0.00805 0.01228 0.00805 0.00968 0.00484 > 0.00968 > IL1 > 0.00794 0.01228 0.00794 0.01301 0.01107 0.01301 0.01107 0.01228 0.00794 > 0.01228 > DTB > 0.00805 0.00968 0.00805 0.01228 0.01107 0.01228 0.01107 0.00968 0.00805 > 0.00968 > DL1 > 0.00000 0.00968 0.00000 0.01591 0.00484 0.01591 0.00484 0.00968 0.00000 > 0.00968 > L2_left > 0.00000 0.00000 0.00000 0.00968 0.01591 0.00968 0.01591 0.00000 0.00000 > 0.00000 > L2_bottom > 0.01107 0.00968 0.01107 0.01591 0.01591 0.01591 0.01591 0.00968 0.01107 > 0.00968 > > > I use the following command:./sim-outorder -config > ../spec95-little/go.cfg ../spec95-little/go.ss 2 8 > This benchmark works well on sim-wattch, and sim-profile. > I got no idea why the power trace stay the same whatever the workload I use. > Looking forward for your advises. > > Thank you and have a good day. > > Best regards, > > Shi > > July 9, 2012 > > > > > > _______________________________________________ > HotSpot mailing list > HotSpot at mail.cs.virginia.edu > http://www.cs.virginia.edu/mailman/listinfo/hotspot > From nastaran_motavali at yahoo.com Mon Jul 9 23:45:42 2012 From: nastaran_motavali at yahoo.com (Nastaran Motavalli) Date: Mon, 9 Jul 2012 23:45:42 -0700 (PDT) Subject: [Hotspot] Why PTscalar power trace stay the same? Message-ID: <1341902742.69391.YahooMailNeo@web160102.mail.bf1.yahoo.com> ? ?Hi, To get a true power trace from PTScalar, u should manipulate the sim-outorder source code as follows: In "sim_main(void)" function, define a text file with "write" mode and at the end of the function body, call the "collected_power_output" fucntion to write the power tarce into the mentioned?text?file. For example, if you want to get the power values of functional blocks every 10000 cycle, you can use this code: if { ???? collected_power_output(fp);(sim_cycle % 10000 == 0)} -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120709/f36220b5/attachment.html From skadron at cs.virginia.edu Wed Jul 11 07:13:40 2012 From: skadron at cs.virginia.edu (Kevin Skadron) Date: Wed, 11 Jul 2012 10:13:40 -0400 Subject: [Hotspot] Why PTscalar power trace stay the same? In-Reply-To: <1341902742.69391.YahooMailNeo@web160102.mail.bf1.yahoo.com> References: <1341902742.69391.YahooMailNeo@web160102.mail.bf1.yahoo.com> Message-ID: <4FFD8A14.30708@cs.virginia.edu> Thanks for replying! We're thrilled when we see peer to peer support in this community. That's what really makes it a community! --Kevin On 7/10/12 2:45 AM, Nastaran Motavalli wrote: > // > // Hi, > To get a true power trace from PTScalar, u should manipulate the > sim-outorder source code as follows: > In "sim_main(void)" function, define a text file with "write" mode and > at the end of the function body, call the "collected_power_output" > fucntion to write the power tarce into the mentioned text file. For > example, if you want to get the power valuesof functional blocks every > 10000 cycle, you can use this code: > if(sim_cycle % 10000 == 0) > { > collected_power_output(fp); > } > > > _______________________________________________ > HotSpot mailing list > HotSpot at mail.cs.virginia.edu > http://www.cs.virginia.edu/mailman/listinfo/hotspot > From indirarawat at gmail.com Wed Jul 11 22:40:35 2012 From: indirarawat at gmail.com (Indira Rawat) Date: Thu, 12 Jul 2012 11:10:35 +0530 Subject: [Hotspot] help Message-ID: Sir, I want to use hotspot-5.0 for temperature estimation of a 3-stack structure. i am using the command ./hotspot hotspot.config -p core.ptrace -o core.ttrace -model_type grid -grid_layer_file ind.lcf where core.ptrace is a file for power trace, lcf is also specified, but not getting any result, only details of a thermal simulator are coming. Kindly help. with kind regards, indira From nastaran_motavali at yahoo.com Thu Jul 12 01:09:04 2012 From: nastaran_motavali at yahoo.com (Nastaran Motavalli) Date: Thu, 12 Jul 2012 01:09:04 -0700 (PDT) Subject: [Hotspot] Why PTscalar power trace stay the same? Message-ID: <1342080544.33995.YahooMailNeo@web160104.mail.bf1.yahoo.com> ? Hi again, To generate an appropriate power trace for Hotspot, you should define a new function in power.c and power.h source code, I defined it as follows: void my_power_output(FILE *fp) { ? int i; ? for(i=0; i < TOTAL_MODULES + MODULE_L2_EXTRA - 1; i++) ??? fprintf(fp, "%2f\t", power_vector[i]); fprintf(fp, "%2f\n", power_vector[i]); } and also you need to add this lines to the sim-outorder source code (sim_main(void) function) to write the functional block names on top of the power trace file: char my_module_names[TOTAL_MODULES + MODULE_L2_EXTRA ][32] = { ? "decode", "branch", "RAT", "RUU", "LSQ", "IALU1", "IALU2", "IALU3", "IALU4", "FPAdd", "FPMul", ? "IntReg", "FPReg", "ITLB", "IL1", "DTLB", "DL1", "L2_left", "L2_bottom", "L2_right" }; int j; for(j=0; j < (TOTAL_MODULES + MODULE_L2_EXTRA)-1; j++) ??? ??? fprintf(fp, "%s\t", my_module_names[j]); ??? fprintf(fp, "%s\n", my_module_names[j]); Then you call "my_power_output" function instead of "collected_power_output" function to generate the suitable power trace. Note that the functional blocks names must be the same as which defined in the ev6.flp (floorplan file) in Hotspot directory. If they are not matched, you should make them the same. Good Luck -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120712/9f7da70f/attachment.html From vadrao at utu.fi Thu Jul 12 02:47:28 2012 From: vadrao at utu.fi (Vaddina Kameswar Rao) Date: Thu, 12 Jul 2012 12:47:28 +0300 Subject: [Hotspot] help In-Reply-To: <8003b28512fd4b329b46d5ea4e8b3ece@EXCH-CAS-01.utu.fi> References: <8003b28512fd4b329b46d5ea4e8b3ece@EXCH-CAS-01.utu.fi> Message-ID: Could you also paste the error message in here? It is difficult to diagnose or judge anything without the error text and may be the files which caused that error. /Kamesh On Thu, Jul 12, 2012 at 8:40 AM, Indira Rawat wrote: > Sir, > > I want to use hotspot-5.0 for temperature estimation of a 3-stack > structure. i am using the command > ./hotspot hotspot.config -p core.ptrace -o core.ttrace -model_type > grid -grid_layer_file ind.lcf > where core.ptrace is a file for power trace, lcf is also specified, > but not getting any result, only details of a thermal simulator are > coming. Kindly help. > > with kind regards, > indira > _______________________________________________ > HotSpot mailing list > HotSpot at mail.cs.virginia.edu > http://www.cs.virginia.edu/mailman/listinfo/hotspot > -- University of Turku (Turun Yliopisto), Department of Information Technology, Kameswar Rao Vaddina, Joukahaisenkatu 3-5, 20014 Turku, Finland. Web : http://users.utu.fi/vadrao E-mail : vadrao at utu.fi Phone : +358 2 333 8664 Mobile : +358 440992996 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.cs.virginia.edu/pipermail/hotspot/attachments/20120712/68b59e8a/attachment.html