• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Does memory timing really make a noticeable difference?

Kaido

Elite Member & Kitchen Overlord
Does cas latency really make a difference at all? Noticeably? I'm looking at putting 4 gigs into an Athlon 64 system; the Corsair ValueRam is pretty cheap at $67/1gb stick, but the cas is 3 rather than 2 or 2.5. I will not be overclocking this system. Will the timing offering any noticeable improvement in XP or am I better off just getting the cheap stuff?
 
Unless you are running WinXP64 you wouldn't see benefits using 4GB instead you will see a decrease in performance because with 4 sticks of ram you will have to lower ram timing from 1T to 2T. As for the CAS latency,it does make a little difference about 10% in performance between 2.0 and 3.0.
 
I am running 4 sticks of 512mb kingmax pc3200 super ram at 2T. when I ran 2 sticks at 1T there was not a 10% difference. The difference was 2 or 3% at best, I couldnt tell at all by merely using my computer for surfing and gaming....I am running a venice core 3000+ 64bit athlon, my wife is running a 3000+ winchester core 64 bit athlon with the same amount and type of ram...no memory controller issues with either one, although some would have you believe that a winchester core wont run with all 4 ram slots filled..you cant argue with success.... that being said, I would not go 4 gigs, that will cause slowdowns and 2 gigs is plenty, plus you still get dual channel with 4X512...now when Vista comes out, that may be a different story...time will tell...
 
Originally posted by: allanon1965
I am running 4 sticks of 512mb kingmax pc3200 super ram at 2T. when I ran 2 sticks at 1T there was not a 10% difference. The difference was 2 or 3% at best, I couldnt tell at all by merely using my computer for surfing and gaming....I am running a venice core 3000+ 64bit athlon, my wife is running a 3000+ winchester core 64 bit athlon with the same amount and type of ram...no memory controller issues with either one, although some would have you believe that a winchester core wont run with all 4 ram slots filled..you cant argue with success.... that being said, I would not go 4 gigs, that will cause slowdowns and 2 gigs is plenty, plus you still get dual channel with 4X512...now when Vista comes out, that may be a different story...time will tell...

Why will it cause a slowdown? I'm throwing around ideas for doing a ramdisk system using Gigabyte's ramdisk pci cards.
 
All I can tell you is at the same clock speeds, going from Corsair Value Select to the OCZ, Platinum Revision II gained me in of itself less than 5%. If you overclock, the benefits of better RAM increase, but not by much in most cases.
 
In terms of real-world performance, no. Synethetic benchmarks like Sandra may show a significant increase, but who cares about that?
 
Its good for cutting a few ns's (Nanoseconds) off of its access times (which is negligible), but apart from that its all about e-peen, and as you can see mine is quite large since I can run timings of 2-2-2-5 at about 215Mhz tops 😉 😛
 
In my previous post I posted two factors which will affect performance, i.e. lowering of the timing and also increase in latency. Although one poster mentioned a 2-3% decrease when going from 1T to 2T he never addressed the second factor, which is more perceivable, the increase in latency. Latency is like a delay inherent to rams. It's like comparing your reaction time like when stepping on the brake to stop a car. Except in ram's world, it's like of how quickly the ram can access the address where the data was requested. This address is accessed through rows and columns like a two dimensional X-Y plot. CAS is Column Access Strobe and then RAS which is Row Access Strobe. The higher the CAS latency, the longer it will take for the ram to deliver the data requested by the controller thus delay in performance. Usually, ordinary computer users may not perceive the performance difference. But for high end users it is a big deal. You can tell how big this is by going through the price difference on a CAS 2 ram and a CAS 3 ram of the same speed. The difference in price can be as high as 50% to 100%.
 
Originally posted by: aarondeep
If you are using it in a ramdisk, go with the value ram.

ramdisk and system ram...it'd be nice to go with the value ram, both because it's cheap and because I can swap the ram throughout the system to test to make sure they're good.
 
Wait so the OP is goanna get that Ramdisk device where you stick in DDR ram and it works as a hard drive? If so, he mind as well go on ebay and pick up some DDR200 1GB sticks as SATA can't take advantage of faster memory since the interface is too slow.
 
Originally posted by: t3h l337 n3wb
Corsair ValueSelect vs XMS

As you can see, there isn't much of a difference, even in synthetic benchmarks.

yup, n3wb is right. but you could always go for CAS2.5 ram over the CAS3 ram; it's like $3 more or something per gig. if you can't find it for corsair, definitely check out G.SKILL's ram. it's around $70 per gig for CAS 2.5 ram and it's decent at overclocking...then again you don't want to overclock. still, getting the CAS 2.5 for a few dollars more is worth it
 
Originally posted by: goku
Wait so the OP is goanna get that Ramdisk device where you stick in DDR ram and it works as a hard drive? If so, he mind as well go on ebay and pick up some DDR200 1GB sticks as SATA can't take advantage of faster memory since the interface is too slow.

Yeah, but it's like $67 for a new Corsair ddr400 stick of value ram, so what the heck. I think I'm going to wait on this project until they come out with a ramdisk card that supports SATA 2.0 though.
 
Originally posted by: t3h l337 n3wb
Corsair ValueSelect vs XMS

As you can see, there isn't much of a difference, even in synthetic benchmarks.

Great article, thanks for the post! Quote from the article:

The XMS memory only produced ~2% more memory bandwidth in SiSoft2005 and even less in PCmark2005.

Looks like the only real reason to get spiffy ram is for overclocking. Nice to know!
 
Back
Top