The following results were obtained from two different configurations of tuned Power Mac G4 AGP (or Sawtooth). The one was standard 100MHz FSB and CPU was overclocked to 500MHz. The other was overclocked FSB to 133MHz.
I used standard 601-optimized executable because PPC7400 prefers 601-optimized one than 604-optimized one. I also found that some early revisions of Power Mac G4 AGP are disabled speculative processing function. It makes lower the STREAM result drastically.
Michiro Isobe
-- Result 1 System descriptions Power Mac G4/500MHz (OC'd from 450MHz), L2 Cache: 1MB/250MHz, RAM: PC100 CL2@100MHz bus, MacOS 9.0.4, VM enabled, Speculative Processing enabled ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 400000, Offset = 0 Total memory required = 9.2 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Your clock granularity/precision appears to be 7 microseconds. Each test below will take on the order of 20409 microseconds. (= 2915 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) RMS time Min time Max time Copy: 558.0746 0.0118 0.0115 0.0127 Scale: 531.1203 0.0122 0.0120 0.0128 Add: 509.5271 0.0190 0.0188 0.0192 Triad: 529.7721 0.0182 0.0181 0.0184-- Result 2 System descriptions Power Mac G4/466MHz (OC'd from 450MHz), L2 Cache: 1MB/233MHz, RAM: PC133 CL2@133MHz bus, MacOS 9.0.4, VM enabled, Speculative Processing enabled ------------------------------------------------------------- This system uses 8 bytes per DOUBLE PRECISION word. ------------------------------------------------------------- Array size = 400000, Offset = 0 Total memory required = 9.2 MB. Each test is run 10 times, but only the *best* time for each is used. ------------------------------------------------------------- Your clock granularity/precision appears to be 8 microseconds. Each test below will take on the order of 15750 microseconds. (= 1968 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) RMS time Min time Max time Copy: 791.4915 0.0082 0.0081 0.0091 Scale: 673.9680 0.0096 0.0095 0.0098 Add: 783.4177 0.0124 0.0123 0.0129 Triad: 685.1270 0.0141 0.0140 0.0145
--
This archive was generated by hypermail 2b29 : Mon Jul 17 2000 - 04:46:00 CDT