CAS Latency

coolego1

Member
Jun 29, 2004
70
0
0
I was reading an article (can't remember where) that said that CAS Latency was only defined in integral values and was rounded up, meaning that 2.5 latency, without OC'ing runs as 3 latency... Correct me if I'm wrong. I'm looking into an AMD Athlon64 3000+ for a CPU and some OCZ Performance Series RAM with 2.5-3-3-6 latency...
 

imported_michaelpatrick33

Platinum Member
Jun 19, 2004
2,364
0
0
Uh, I have never ever seen anything like that. DDR memory has cas timings that run from 2.0 to 3.0 to 4.0 depending on the speed of the ddr memory. ddr2 memory has higher latency due to the higher speeds and starts at 4 but they are already planning on 3.5 CAS latency! About time ddr 800 CAS 2.5 memory comes out AMD will release the ddr2 integrated memory controller! The higher latency is why ddr2 at 400(200ddr2) is the same speed if not slower than ddr memory at 400 (200ddr)
 

Sonic587

Golden Member
May 11, 2004
1,146
0
0
Sounds like FUD to me. If it "rounds up" already then why would there be both a CAS 2.5 & 3.0 setting?
 

Cerb

Elite Member
Aug 26, 2000
17,484
33
86
and if it rounds up (for DDR), why do you get different scores?
 

SneakyStuff

Diamond Member
Jan 13, 2004
4,294
0
76
my Corsair Value RAM is CAS 2.5, I'm running 2 sticks of 512, and lovin' it :) my memory is like WAY up thur on AIDA 32's relative benchmark :)
 

imported_jediknight

Senior member
Jun 24, 2004
343
0
0
I don't buy the part of "rounding up". DDR transmits on both the rising and falling edge of the clock signal, so I'd think (assuming we start with the request - is that the right term? oh well - on the rising edge)
So, if the data arrives:
two full cycles later, on the rising edge, you've got CAS2
two full cycles later, on the falling edge, CAS2.5
three cycles later, CAS3

I'm not sure if I explained that well.. a timing diagram would make things much easier :->
 

coolego1

Member
Jun 29, 2004
70
0
0
OK, I just had to make sure that spending the extra $$ on some 2.5 was really worth it. By the way, is a 2.5-4-4-8 worth $25 over a 2.5-3-3-6 ? I don't know how important the other numbers(after the first) really are...
 

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
32,032
32,505
146
Timings aren't very significant to A64 gaming performance so I'd suggest buying economy ram unless you intend to overclock, then I would go high frequency over tight timing.
 

coolego1

Member
Jun 29, 2004
70
0
0
I am not overclocking.

The question remains about the second and third numbers. Actually I just want to know if some 2.5-4-4-8 is better than some 3-3-3... The reason I'm so concerned about price is that I'm buying 2 sticks of 512MB so any money I can save is very good because it counts double.
 

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
32,032
32,505
146
Originally posted by: coolego1
I am not overclocking.

The question remains about the second and third numbers. Actually I just want to know if some 2.5-4-4-8 is better than some 3-3-3... The reason I'm so concerned about price is that I'm buying 2 sticks of 512MB so any money I can save is very good because it counts double.
The performance difference between them is virtually non-existant, buy the less expensive ram.
 

imported_michaelpatrick33

Platinum Member
Jun 19, 2004
2,364
0
0
However, the AMD 64's are more latency sensitive due to the integrated memory controller but the two rams you are talking about will not be great. 1 or 2 fps. MMMMM 1000 FPS in Quake 3. My goal (sarcasm)
 

JBT

Lifer
Nov 28, 2001
12,094
1
81
There is a difference for me between 2.5 and 3 on my system on benchmark.... Anyways go for the cheaper stuff. Most of the timmings have little to no significant performance benifits.
 

Sonix7

Member
Jun 27, 2001
140
0
0
Originally posted by: jediknight
I don't buy the part of "rounding up". DDR transmits on both the rising and falling edge of the clock signal, so I'd think (assuming we start with the request - is that the right term? oh well - on the rising edge)
So, if the data arrives:
two full cycles later, on the rising edge, you've got CAS2
two full cycles later, on the falling edge, CAS2.5
three cycles later, CAS3

I'm not sure if I explained that well.. a timing diagram would make things much easier :->

As far as i know that is the reason, DDR memory works on two edges and due to the cas latency is the most important timing value sometimes the system performs better. In my own case i've reached better results with cas2 than cas2.5