PDA

View Full Version : Memory Timings : Do They Make A Difference?



wookie
01-19-2004, 09:38 PM
I was always under the assumption that ram timings either make or break a computers performance. Then I read the recent ram timing article at toms hardware (not the best source but some decent information) and it did the comparison.

So my question is do you really need top of the line ram and run 2.0cas or can u get the average middle of the road with possibly a little more headroom and run 2.5cas and makeup the difference in the clock speed advantage. I am thinking the later is true but please if I misread or am missing a peice of information let me know....

Starting to make me really think about buying my next dual channel kit as to weather spend the extra and get Muskin Level 2 black or maybe just some OCZ PC3700GOLD rev B and go all out FSB speed.... Lemme know your thoughts on the subject.

Soulburner
01-19-2004, 10:38 PM
If you ask me, slowing down timings to increase clock speeds is like running in circles....you don't get anywhere.

STEvil
01-19-2004, 11:56 PM
Depends on system.

dim3z
01-20-2004, 07:31 AM
the cas timing doesnt mean all that much, so 2.5 is ok, the one that really hits performance decently relative to the rest is ras to cas, on a intel system counts for a good 500 3dmarks on a good system. thing is there is basically no ram i know of that will let you run ras to cas 2 at high DDR speeds short of BH5, get BH5 and be done with it, spend 50$ more cos in bang for buck terms BH5 is still a very worthy investment. oh and btw, mushkin l2 its great ram but mushkin have gone spastic on pricing. buy some cheaper BH5, KHX3200 non "A" is cheaper. available and the same stuff, and if u can get twinmos, buffalo or ADATA BH5 for even cheaper by all means get that instead u wont regret it.

wookie
01-20-2004, 08:41 AM
Thats the thing people always say it makes a difference and preach cas latency as the god of all timings.... but when you actually see benchmark comparisons it isnt all that much of a difference. If you read the review the Athlon FX-51 doesnt loose much ground no matter what timings it is using. The Intel does loose some ground depending on performance and the normal Athlon XP's dont seem to be affected very much either. Everyone always says to go high with P4 and go for good timings for amd this review has got me thinking the otherway around.

Now I know an extra 500marks in 3DMARK may mean the difference between 29500 or 30,000 which I can understand just saying that from what the review states the memory timings dont make as big a difference as I was under the impression of and I beleive many other people were as well.

Now this is XtremeSystems so I understand our sickness that we have to get the very best of the best and then when we do it still isnt good enough.... Just it surprised me the little impact it did have on overall benchmarks...

Games varied by like 5fps which when most are over 60fps now anyways it doesnt make that big of a difference. Anywho enough ranting and raving just curious on everyone else's thoughts.

longsiew
04-22-2005, 11:14 PM
I was always under the assumption that ram timings either make or break a computers performance. Then I read the recent ram timing article at toms hardware (not the best source but some decent information) and it did the comparison.

So my question is do you really need top of the line ram and run 2.0cas or can u get the average middle of the road with possibly a little more headroom and run 2.5cas and makeup the difference in the clock speed advantage. I am thinking the later is true but please if I misread or am missing a peice of information let me know....

Starting to make me really think about buying my next dual channel kit as to weather spend the extra and get Muskin Level 2 black or maybe just some OCZ PC3700GOLD rev B and go all out FSB speed.... Lemme know your thoughts on the subject.

I am currently ruuning my TwinMos SP (ch-5, i guess) @ 250mhz @ 2-2-2-5 at 3.5v. Actually, I can run my TwinMos at 1.5-2-2-5 or 1.5-2-2-2 or 2-2-2-2. (All at 250mhz). When I run sisofsandra 2005, I noticed that there is not significant improvement in memory bandwith for the above timings. So I stick to 2-2-2-5 (just because I noticed that most of us running at 2-2-2-5). Hope this will help.

texuspete00
04-24-2005, 08:10 AM
When sites like toms hardware says things like "timings don't matter" they are more talking to joe user. What matters to them is you pay out twice as much for good memories and then the difference isn't one you can really feel. While I know where they are coming from, how far do you take this philosophy? Is there a reason to buy anything other than a 3000+ A64? Why spend 50 more bucks for 200MHz on a 3200+ when you can't "feel" the difference? For power users, placebo effect or not, 1 thousand 3d01 points is 1 thousand 3d01 points. When you're in the boat that a lot of us are, where you'll switch cpus in the hope of finding one that clocks better, and actually be tickled pink to find a new 100MHz and flog it's older slower brother in the classifieds... timings don't matter is blasphemy. We're all here at xtremesystems.... not PracticalSystems.org

I guess for good mem you have to pay 2x the cheap stuff for smaller gains so some people have stood up and took notice. But at the end of the day when I got my DFI Ultra-D, mach1 Lian-li, WC'd 6800GT, OCZ 600w all ready and waiting... I want to top it off for the total package. Not throw in CL3 PC3200 joe user average bunk stuff.

That's really what the issue is. Regardless of platform, it's always moderately sizable. I mean people are most inclined to say this stuff about the A64 platform: oh use dividers, it's all about cpu speed blah blah. The difference is very sizable and real to an enthusiast. I should also drive a hyundai to save on gas.... No thanks!!!

Pivo504
04-28-2005, 03:26 PM
timings matter. Its like night and day at least thats the experience Ive had with the memories Ive run with.

Evil_Spork
04-29-2005, 07:39 AM
Depends on system.
on my A64, i noticed a BIG boost in memory bandwidth with tight timings. with my AXP systems i can actualy feel if i have very loose times compared to 2-2-2

TekXoID
04-29-2005, 07:55 AM
on my A64, i noticed a BIG boost in memory bandwidth with tight timings. with my AXP systems i can actualy feel if i have very loose times compared to 2-2-2
This thread is 1yr. 3mo. old. :)

saaya
04-29-2005, 01:27 PM
please use the search button next time, there are nice guides in the xb sub section