memory timings (What's better??)

lytalbayre

Senior member
Apr 28, 2005
842
2
81
2-2-2-5/1T at 200mhz & cpu @2.78 Ghz (ratio 3:4)
vs.
2.5-3-3-6/1T at 230mhz & cpu @2.78 Ghz (ratio 5:6)
vs.
2.5-4-3-7/2T at 275mhz & cpu @ 2.75Ghz (ration 1:1)
 

Appledrop

Platinum Member
Aug 25, 2004
2,340
0
0
will vary app to app, but id go for the 1:1..

edit: oh thats 2T and 2.75ghz.. soz just skimmed.. hrrm well probably the second one then.
 

RichUK

Lifer
Feb 14, 2005
10,341
678
126
Originally posted by: lytalbayre
2-2-2-5/1T at 200mhz & cpu @2.78 Ghz (ratio 3:4)
vs.
2.5-3-3-6/1T at 230mhz & cpu @2.78 Ghz (ratio 5:6)
vs.
2.5-4-3-7/2T at 275mhz & cpu @ 2.75Ghz (ration 1:1)

I would probably say the latter, i know people always say that the lower latencies are what the doctor ordered, but it think my system was more responsive running the higher frequency (when i used 280 HTT on 9x multi) .. the biggest diff in mem performance is when it comes to the CAS, the differenece in performance between CAS 2 and 2.5 is not as bad as 2.5 and 3, so i would say go with the 1:1 CAS 2.5 .. with the increase in RAM Mhz that will compinsate for the 2.5 CAS setting over 2.0 ..

Rich

 

imported_michaelpatrick33

Platinum Member
Jun 19, 2004
2,364
0
0
These forums have answers to this thread just do a search. If I remember correctly the timings made little difference on the AMD system. It made some, but not enough to justify the massive cost increase for lower latency ram in my opinion.
 

RichUK

Lifer
Feb 14, 2005
10,341
678
126
Originally posted by: Azzy64
will vary app to app, but id go for the 1:1..

edit: oh thats 2T and 2.75ghz.. soz just skimmed.. hrrm well probably the second one then.


lol i just skimmed aswell, didnt notice the 2T .. yeah go with second one what he ^^^ said
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
YOu should test it versus the apps you use though....

Check out this link.... http://forums.anandtech.com/messageview...atid=28&threadid=1591148&enterthread=y
Had Inot used auto settings and ran the 133 divider at cas 2-2-2-5 which it will do aggressively at 2.75v it should have won most all of the test versus the higher speed ddr...cas 2.5 at near 550ddr is pretty hard to beat but as I continue to OC the 200 will start tailing off and require cas 3 and even 2t to get stable at just 20-50ddr more so then 400ddr at cas 2-2-2-5 can still compete and make it a wash even at same clock speeds...


With 1:1 at 2t I think it made it closer to 1st option but I think the 30mhz more speed and the non 2t (can reduce performance 1-5% app depending) of 1st option would win this route if they are all truly stable

Try testing the 1:1 ram again at like 2.5-4-4-11 or cas 3-4-3-10 trying to hold 1t...It may work out to be the best...


However no way is the 60mhz more off ddr going to make up for the switch to cas 2.5 and trcd and trp being higher....2 almost never wins against 1...

 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Originally posted by: michaelpatrick33
These forums have answers to this thread just do a search. If I remember correctly the timings made little difference on the AMD system. It made some, but not enough to justify the massive cost increase for lower latency ram in my opinion.

Actually not right...Actually all ram tweaking made little difference, but timings made more of a difference then actual bandwidth.

Therefore buying decent 400ddr ram capable of low cas 2 ratings can be as good as 480-500ddr running cas 2.5 and even 550-600ddr running cas 3 or perhaps 2t......
 

Zebo

Elite Member
Jul 29, 2001
39,398
19
81
2-2-2-5/1T at 200mhz & cpu @2.78 Ghz (ratio 3:4)

vs.
2.5-4-3-7/2T at 275mhz & cpu @ 2.75Ghz (ration 1:1)

These two should be closest... but I'm afraid the former will win... Easyist way to check is run winrars internal benchmark... it's the best tool possible for real world memory configuations, as it translates even to games but less dramatic.
 

lytalbayre

Senior member
Apr 28, 2005
842
2
81
The only benchmark I used for preliminary comparison is 3dmark 03...

the 3rd option wins consistenly by about 50 pts. (14000 range)...

 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
50pts in 14000 is neglible .4%.....Try running superpi 4mb or greater. winrar, and a few gaming apps....
 

Zebo

Elite Member
Jul 29, 2001
39,398
19
81
I told you benchmark to use... quick and painless and accurate.
 

Leper Messiah

Banned
Dec 13, 2004
7,973
8
0
Originally posted by: lytalbayre
2-2-2-5/1T at 200mhz & cpu @2.78 Ghz (ratio 3:4)
vs.
2.5-3-3-6/1T at 230mhz & cpu @2.78 Ghz (ratio 5:6)
vs.
2.5-4-3-7/2T at 275mhz & cpu @ 2.75Ghz (ration 1:1)

why don't you bench it, and tell us?