how could P35 and 965 benchies be so different??

[vois2]

This is the tale of two very similar setups, and my frustration on trying to understand how a P35 chipset could possibly outbench a 965 chipset in the manner below.

First, let me mention that this is with benchmark program PassMark Performance Test 6.1 . All arguments aside about the value of real versus synthetic benchmarking, let me say that I have used Perf Test for a number of years and prefer it because of its consistency and accessible sub-categories of benchmarks.

Also, I can say that I know (by examining sub-benchmarks) that the problem with my friend's machine is his CPU scores and CPU performance. I just don't know how to tell him how to fix it !!! Help?!

So, my friend and I have ...

(me) Q6600 stock / abit IP35 Pro / 4GB RAM(6400) / 7900GT
(him) Q6600 stock / GB GA-965-ds3 / 4GB RAM(6400) / 8800 GTS 320

I thought for sure his would outbench mine, or that they would be reasonably close. Nope:
(me) 1247 <-- with 7900 GT !!!
(him) 820 <-- with 8800 GTS !!! Big problem I have with that !!!

The only other difference worth mentioning is the configuration of RAM. He has 4x1GB and I have 2x2GB, but in MemTest86+ the transfer rates (in MB per second) for L1, L2, and Memory are almost identical.

In his BIOS (Gigabyte GA-965-DS3) we have turned OFF "EIST" ... turned OFF "Thermal Monitoring" ... turned OFF "Virtualization" ... and turned OFF "E1C HALT"

Just to check for consistency, I put a GeForce 6600 GT into my build and it rolled up at benchmark 1183, which is consistent with my 7900 GT score of 1247.

So ... what is going on with my friend's machine? How could the discrepancy be so large, what setting in BIOS affecting his CPU have I missed?

Thanks so much for reading and for any help.

 

[tno]

You probably shouldn't shutdown TM2; it only activates when your chip is about to fry itself. The speed of the chipset's I/O subsystems are a huge contributor especially in synthetics like perftest. From what I've heard the P35 chipset is fairly improved, add to that other contributing factors like the hard drive speeds and memory timings, and two systems that look similar can actually pretty different.

tno