After seeing the recent article, I decided to examine the performance of my own system, in particular, of SPECViewPerf.
My system is an Duron 800MHz in an ASUS A7V with 256 MB of 133MHz SDRAM at CAS3, running Windows 2000 Pro SP1. My video card is a Visiontek GeForce2 MX 32MB SDR.
Here are my results:
---------- SUM_RESULTS\AWADVS\SUMMARY.TXT
AWadvs-04 Weighted Geometric Mean = 49.88
---------- SUM_RESULTS\DRV\SUMMARY.TXT
DRV-07 Weighted Geometric Mean = 13.27
---------- SUM_RESULTS\DX\SUMMARY.TXT
DX-06 Weighted Geometric Mean = 15.95
---------- SUM_RESULTS\LIGHT\SUMMARY.TXT
Light-04 Weighted Geometric Mean = 5.154
---------- SUM_RESULTS\MEDMCAD\SUMMARY.TXT
MedMCAD-01 Weighted Geometric Mean = 13.30
---------- SUM_RESULTS\PROCDRS\SUMMARY.TXT
ProCDRS-03 Weighted Geometric Mean = 10.84
I found that in a number of test, my system was performing MUCH better than the THG test systems, especially for a Duron-based system. So I was curious as to why.
My conclusion is that while THG test system had only 128 MB of RAM, my slower system with 256 MB of (slower!) RAM ran much better. I think that THG results (particularly SPECViewPerf under Windows 2000) do not accurately reflect CPU/videocard scaling due to a lack of RAM. I think THG might want to look at those benchmarks with additional memory in the test system (after all, anyone running a CAD system, which SPECViewPerf simulates, will have AT LEAST 256 MB, if not 512 MB).
BAH!
My system is an Duron 800MHz in an ASUS A7V with 256 MB of 133MHz SDRAM at CAS3, running Windows 2000 Pro SP1. My video card is a Visiontek GeForce2 MX 32MB SDR.
Here are my results:
---------- SUM_RESULTS\AWADVS\SUMMARY.TXT
AWadvs-04 Weighted Geometric Mean = 49.88
---------- SUM_RESULTS\DRV\SUMMARY.TXT
DRV-07 Weighted Geometric Mean = 13.27
---------- SUM_RESULTS\DX\SUMMARY.TXT
DX-06 Weighted Geometric Mean = 15.95
---------- SUM_RESULTS\LIGHT\SUMMARY.TXT
Light-04 Weighted Geometric Mean = 5.154
---------- SUM_RESULTS\MEDMCAD\SUMMARY.TXT
MedMCAD-01 Weighted Geometric Mean = 13.30
---------- SUM_RESULTS\PROCDRS\SUMMARY.TXT
ProCDRS-03 Weighted Geometric Mean = 10.84
I found that in a number of test, my system was performing MUCH better than the THG test systems, especially for a Duron-based system. So I was curious as to why.
My conclusion is that while THG test system had only 128 MB of RAM, my slower system with 256 MB of (slower!) RAM ran much better. I think that THG results (particularly SPECViewPerf under Windows 2000) do not accurately reflect CPU/videocard scaling due to a lack of RAM. I think THG might want to look at those benchmarks with additional memory in the test system (after all, anyone running a CAD system, which SPECViewPerf simulates, will have AT LEAST 256 MB, if not 512 MB).
BAH!
Facebook
Twitter
Instagram