• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Increasing Vdimm and memory performance

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Captain Hilts

Member
Joined
Sep 5, 2002
Location
Toronto, ON
I've got a 256MB stick of CAS2.5 DDR 333 memory on my Abit KD7. I have been playing with a number of different bus speeds to see how far I can push the ram.

I have been able to run it stable at CAS2 with a 166MHz memory bus. It can also run stable at default timings with a 185MHz memory bus, but fails to completely boot when I set it at CAS2 at that speed. Increasing the Vdimm to 2.65 allows the memory to run at 185MHz though.

My question stems from my Sandra memory benchmark results. At 166MHz, the results were better as CAS2 than at 2.5, as expected. At 185MHz, however, I got better results when it was set at CAS2.5. Is this somehow the result of increasing the Vdimm? Do you sometimes lose some performance as a result of increasing the Vdimm, or is it maybe just screwing up the Sandra benchmarks somehow so they're not reporting accurately?
 

mirko_3

Member
Joined
Oct 27, 2002
Location
Italy
Sorry that I'm not able to answer your questio, but I had similar questions:
I have a stick of DDR 266 and it does:
CAS 2, DDR266
CAS 2.5 DDR400
From sandra I get a better score with DDR400 (of course)
but that's true also for unbuffered; shouldn't it be the opposite, or is the jump between 266 and 400 big enough to compensate?
 

hdj

Member
Joined
Apr 5, 2002
Location
new orleans
You are comparing 166 cas2 to 185 cas2.5 ? I am not really surprised that the latter gives you a higher score. You are running your memory faster...

I want to add that you have to run the sandra memory bench about 6 times in a row to get a reproducible (sp?) result.
 

deeppow

Senior Member
Joined
May 10, 2002
Location
Los Alamos, NM
First, hdj is right to suggest running Sandra several times. Make sure you exit and restart it or it'll just repeat the previous info without rerunning the test (at least my version does that).

Basically there is variation each time your run Sandra and you need to understand that variation to make sure that the 2 cases are showing an average difference.

To answer your original question: No, increasing the Vdimm won't cause the effect. Vdimm gives you stability thus allowing the memory to function at the higher fsb rate. Unless you are on the hairy edge of the necessary Vdimm needed, raising it higher won't help. Running Prime95 for a period of time often points out just how stable your memory (or cpu) are. Prime95 is a much better tool for this IMO than is Sandra.

You might want to go back to the more aggressive timings (CAS2 and whatever) and see how far you can push the fsb while leaving the Vdimm at as high a settings as you are willing to run. :cool:
 
OP
Captain Hilts

Captain Hilts

Member
Joined
Sep 5, 2002
Location
Toronto, ON
hdj said:
You are comparing 166 cas2 to 185 cas2.5 ? I am not really surprised that the latter gives you a higher score. You are running your memory faster...

I want to add that you have to run the sandra memory bench about 6 times in a row to get a reproducible (sp?) result.

Actually, I'm comparing 185 cas2 to 185 cas2.5 with a slightly higher Vdimm. In this case, Sandra produces slightly better benchmarks at cas2.5.

I will try running it a few more times under each to see if this is still the result.
 
OP
Captain Hilts

Captain Hilts

Member
Joined
Sep 5, 2002
Location
Toronto, ON
Okay I just ran the Sandra 2003 benchmarks three time in each situation. Here are the results

180MHz memory bus, CAS2.5, 2.55Vdimm:

RAM Int. 2643, 2644, 2642
RAM Float 2431, 2432, 2432


180MHz memory bus, CAS2, 2.65Vdimm:

RAM Int. 2550, 2549, 2549
RAM Float 2355, 2343, 2355


How does it get worse when set to cas2? All other settings are the same except for the Vdimm, which I have to increase in order to be able to set at cas2.