DDR 400@ 2:3 or DDR 600@ 1:1

[scaryclownsarefun]

I am looking at buying an athlon 64 3200 DTR chip, a DFI Lanparty UT 250, and either two sticks of A-DATA DDR 400 CAS 2.5-3-3 and running it at a 2:3 ratio (keeping it at DDR400) or take the plunge and get some OCZ Enhanced Latency Rev. 2 (2x512MB) which can run DDR600 at CAS 2.5-3-4.

I was just wondering how much a performance hit i would take at the same clock speed on the athlon 64 from going down to DDR400 and losing the 1:1 ratio.

Could anyone do an experiment on this and try underclocking your good memory using a lower ratio? Make sure the timing remain practically the same.

 

[scaryclownsarefun]

Does anyone have some OCZ Platinum Rev. 2 that is overclocked to DDR600 @ 1:1? could you try putting it at 2:3 so that it is DDR 400 and turn the timings down to what it was at DDR 600?

 

[iwantanewcomputer]

not enough to warrent the cost unless you are a professional overclocker, whick i will assume not cause of the 754 system. truth is athlon 64's do well without fast memory

 

[scaryclownsarefun]

i am a very serios overclocker, and this system will be using a home-made watercooling system i have been working on quite some time now. the point is, i want to know how much performance increase i will get from running DDR 600 synchronously vs. DDR 400 A-synchronously.

 

[Comparisonman]

Definitely go with the 600mhz RAM. You should notice a substantial difference from 400mhz.

 

[scaryclownsarefun]

what do you base that on?
have you seen benchmarks?

 

[scaryclownsarefun]

comparisonman, do you have any links to web pages that compare athlon 64 systems running on 1:1 ratios vs a-synch (such as 2:3)?
or any links that compare difforent speeds of ram on the athlon 64?

 

[stelleg151]

Yeah I am interested in this as well, and if you look at anandtech's memory review, they use different FSB's with different multipliers, but no a-sync. With the cpu speed at 2.4, there was only a 1-2 percent difference in gaming that came from going from ddr400-ddr533, hope that helps, but I agree, it would be nice to see some benchmark comparisons with 1:1 and 2:3, as this is an important price factor for peopple who want to oc the 3000+.

 

[scaryclownsarefun]

yeah, i am projecting a 10% performance increase going from two 512MB sticks of A-DATA PC 3200 (a-synch) to 2 512 sticks of OCZ EL Rev. 2 DDR 400 @ DDR 600 (synch). that isn't bad when it only raises the price of my computer by $130 (about a 8% increse in total computer cost).

 

[ginfest]

Check out these forums

xtreme




Mike G

 

[scaryclownsarefun]

Hmmm.. i couldn't find anyone posting about the difforence betweet a-synch and synch modes on the athlon 64, nor anyone compairing DDR speeds impact on performance on the athlon 64.

if you have seen one please give me the link.

 

[WebDude]

It's rather difficult to answer your question unless you specify more precisely what you mean by a "performance hit". There are different ways to measure "performance" on a PC system. Do you mean memory bandwidth, cpu arithmetic functioning, or overall performance measured by some synthetic benchmark like WinBench, or something else? Running the memory slower while keeping latencies and the cpu freq. the same will impact different aspects of PC functioning to a greater or lesser degree, depending upon how those particular functions use system memory.

 

[Goi]

IIRC running your memory at a non 1:1 ratio with the FSB doesn't incur as huge a performance hit as they did with the K7 series, because the memory controller is on chip. I do not have links to performance differences however.

 

[Gamingphreek]

Originally posted by: Goi
IIRC running your memory at a non 1:1 ratio with the FSB doesn't incur as huge a performance hit as they did with the K7 series, because the memory controller is on chip. I do not have links to performance differences however.

You are correct in that. Also if you are serious in reaching those extreme speeds with the DDR memory i would encourage you to get the OCZ memory booster, as it will allow you to attain higher speeds like you want and will give more "clean" power to the memory without doing volt mods.

Try to keep the processor 1:1 as long as possible then when it maxes out keep going on the DDR and see how much higher it can go.

Shouldn't see much of a performance hit at all especially if you run the memory at DDR600 as that would be incredibly fast and easily negate and overcome the 1% (if that) "hit" you would incur.

-Kevin

 

[scaryclownsarefun]

I have also heard that the athlon 64 (K8) archetecture takes less of a performance hit by running a-synch than other archatectures. As for what "performance" i am measuring: i am mostly into gaming, especially with texture intensive and ultra high gaming setting, such as Doom 3 @ 1600x1200 @ high-max settings. i have heard that caching the textures to system memory does take a pretty big hit when you start lowering memory speeds. the problem i am having is justifying spending an extra $130 on some OCZ EL Platinum Rev. 2 instead of some genaric DDR 400 stuf. so i ask anyone that has any info regarding memory and the athlon 64 platform, to please post you Gaming benchmark results.