Subpar Sandra memtest scores

[stncttr908]

Alright, here are the specs of my brother's system. All components have never been OC'd.

Athlon 1700+
MSI KT3 Ultra (KT333)
2x256MB OCZ PC2700 CL2 (1 rev 3.1 & 1 rev 3.2)
40GB Maxtor 7200rpm
MSI GF4 ti4200 64MB
Enermax 330w PSU
Liteon 52X CDROM
SB Audigy

Well, I've got the timings at 2-3-3-5-1T as of now. They refuse to go any lower. I've also got the VIA hyperion 4.46 drivers installed. Even with everything tweaked in this way, the Sandra scores are only around 1975/1860, whereas the comparison KT333 CL2.5 PC2700 score is around 2100/1950. I know that this isn't overly important, but I just want to make sure that he's getting every last drop of performance out of his machine that he should be. Any suggestions would be grealy appreciated.

Also, I haven't updated the BIOS recently. Should I? The rig isn't connected at the moment or I'd post some screenies of the results. CPU multimedia tests and everything are where they should be.

[Skynet]

The difference is the scores you are comparing to are probably running at 166 FSB you are at 133 if you didn't overclock anything. The best thing you can do for better performance is lower the muliplier and raise the FSB.

I do believe you will have to mod your CPU to change the multiplier on the motherboard though.

Rough estimates for memory bandwidth in Sandra are:

133 FSB: 1900-2100
166 FSB: 2300-2500
200 FSB: 2800-3400

This on the AMD platform of course. CPU speed does play a factor as well but less so than the FSB.

[stncttr908]

It really shouldn't come into play since I'm sure the Sandra comparison is at stock speeds. I have the mem running at 166mhz as well (HCLK+33) so I *should* be getting performance at least inline with the given score.

[ragejg]

kt333's vary from *just over* SiS 735 perf. to *just over* kt400 performance... it depends on the make/model of mobo, and what you had for breakfast this morning...

[FastM]

Disable all the USB ports in the system BIOS and try again...