Agressive Memory Timings Comparison

[jasonpetras]

All,

Has anyone found an article comparing the relative performance gain of more agressive memory timings? For example, compare Kingston's 2-2-2-5-1T with El Cheapo's 3-4-4-8-3T? If you work the numbers of 1GB of PC3200, the difference in price is around $70 to $80 more for the faster timed memory. But for that much, you could upgrade from a 9500 pro to a 9700 pro or another level of CPU.

Thanks in advance! - J

 

[squidman]

I can tell u the performance gain from personal experiences.
Me and my friend have ALMOST similar sytems. The only difference is that he has a 2500+ barton, and 2x256 Corsair XMS ram, as opposed to my samsung. So, at stock 166mhz, my memory was 7-3-3-2.5 at agressive, his - 5-2-2-2. In PCmark2002 i scored 5500 for memory, he - 5900.

 

[squidman]

However (prolly cuz of different CPU) i score almost a 1000 3dmarks more than him

 

[jasonpetras]

so, your faster cpu with slower memory is *slower* than his slower cpu with faster memory... interesting data point... Thanks! Anyone else?

 

[SnowBoarder9455]

No, he said that with equal CPU's, he is 400 points lower on 3D mark. THe only real way to tell how much it effects a system is to get good RAM, all the way to aggressive settings, then just relax the settings a lot and do the test again. That way there is no variable other than the RAM. But thats not useful to someone BEFORE they buy the RAM.

However, using that $100 on a better graphics card is a smarter choice.

 

[SnowBoarder9455]

No, he said that with equal CPU's, he is 400 points lower on 3D mark. THe only real way to tell how much it effects a system is to get good RAM, all the way to aggressive settings, then just relax the settings a lot and do the test again. That way there is no variable other than the RAM. But thats not useful to someone BEFORE they buy the RAM.

However, using that $100 on a better graphics card is a smarter choice.

 

[squidman]

My Cpu is faster, but my memory is slower than his. Memory can be a bottleneck on todays systems. So getting faster ram shall boost the memory subsystem quite significantly. But i specifically indicated the 3dmark 2001 SE score, to showe that its morew cpu dependent.
What u want to upgrade, and what u need to upgrade should be solely based on what u need from PC.
1) If games - then either video card or CPU
2) video editing, design - memory and/or cpu
3) just wanna score higher at PCmark 2002 - then memory.

 

[chocoruacal]

The difference in 3DMark01 on my system w/2.5-3-3-5 (most relaxed) and 2-2-2-5 (tightest) is maybe 500 points. There's no way anyone can feel that difference.

 

[jasonpetras]

Squid, that seems interesting that his faster mem gives him the advantage between the two cpus. Does he have a different video card or were you running different number of background? I understand your tests were probably very informal, but the results are interesting.

Choco, have you tried that test?

Tom's Hardware did a very very small snippet in testing timings. But with the huge push from the mem vendors, its proably time someone here should do their usual impeccible investigation.

 

[x3m]

I'll guess the biggest difference between cheap and expensive memorys is that you could overclock the exepensive a lot ..

But I bought this TwinMos PC3200 w/ Winbond chip, cost just a few $ more than ordinary DDR PC3200 RAM but they are really kick-ass. I've seen people running them with FSB 230+ and still good timings (2-2-2-5)

 

[RandySavage]

if you're overclocking, get expensive ram... especially on an athlon system...

if not, you're wasting your money... spend that extra money on a videocard.

 

[jasonpetras]

Thanks all for your input!

Also, thanks Evan Lieb for the great article you just posted!

J