Mhz Vs. Cas - Guess Whats Faster...

[sierra_bound]

and its just p5b dlx?

gigabyte ds3 and dq6 are in this situation?

I didn't say "just p5b dlx". Read my post again.

Kunaak, thanks for the additional tests.

[Exedy]

I didn't say "just p5b dlx". Read my post again.

Kunaak, thanks for the additional tests.

i know what you said i just have a doubt and made a question

[sierra_bound]

Understood.

Regarding the hidden bios setting referred to in an earlier post, well, that would have to be one helluva tweak to make up 18 seconds in 8M (see first post in this thread). I have my doubts about that.

[lowfat]

wow, it blows me away the differences in 3d01 lobby and 3d03 the dogfight.

[bachus_anonym]

Awesome results with 3D01 Lobby and 3D03 WoF... High Bandwidth wipes the floor with 8-2-3-3.0! Difference is just huge! I'm pretty sure that TRCD=2 would help some for Fatbody, but still bandwidth rules, it seems
Thanks for your hard work, man!

[kiwi]

i know what you said i just have a doubt and made a question

DS3 is not as good as p5b dlx. And it has problems with fsb/ram above 500 or so.

[Revv23]

yeah GB board have problems with micron memory IIRC/

[NO1B4ME]

wow, it blows me away the differences in 3d01 lobby and 3d03 the dogfight.

Definitely true. What a difference.

[hipro5]

Nice comparo Kunaak......It seems that Conroe's need more fsb to compete with ram frequency......Of course there are a LOT more to consider...
The "subtimings" may be the same in bios for all configs, and just barries the 367MHz bandwidth.....(one of them).....

[hipro5]

Awesome results with 3D01 Lobby and 3D03 WoF... High Bandwidth wipes the floor with 8-2-3-3.0! Difference is just huge! I'm pretty sure that TRCD=2 would help some for Fatbody, but still bandwidth rules, it seems
Thanks for your hard work, man!

Yes actually (3) - 2 - 2 - 1(4) is much better......

[G H Z]

Good stuff Kunaak, I like the 8M's for a little more clarity as well as those eye opening 3D results. Is it possible we shal see hipro running 700Mhz Cas 5 in 3D now

[Revv23]

yeah TRCD makes biggest difference imo, you got those fatties kunaak?

maybe we can finnaly put it to rest on 965 at least.

[hipro5]

Good stuff Kunaak, I like the 8M's for a little more clarity as well as those eye opening 3D results. Is it possible we shal see hipro running 700Mhz Cas 5 in 3D now

There's NO WAY for me to run anything else than 1:1......

[Kunaak]

Hipro,

why not just try some benchmarks?
these I did, seriously only took about a half hour to do them all.
another 20 minues to post them all...

[hipro5]

Hipro,

why not just try some benchmarks?
these I did, seriously only took about a half hour to do them all.
another 20 minues to post them all...

As a matter of fact I'll try some out of couriosity.......I have done some in early DDR2 and Conroe rig but I'll re-do them.......

[fhpchris]

As a matter of fact I'll try some out of couriosity.......I have done some in early DDR2 and Conroe rig but I'll re-do them.......

George, just run 4:5 @ 500+ 4-3-2

[Orbitech]

Thanks for taking the time to produce those very important results Kunaak..

I bought Gskill 2GBHZ 2x1GB and with a first ran I can do 1000 4-4-4-12..
I'm curious to see if I can go further with the same timings but most importantly to ran with 5-4-4-8 and see how far can I go with 2.4V.. I hope I can do better than that..

[Kunaak]

Heres some more...
kinda to help those wondering, is running 367 at 3:5 better, then running 1:1 at tight latencies better...

7 x 367 Vs. 6 x 428.

[zakelwe]

Here's my tests with nvidia SLI 6800XT at 450/1130 in 2001 Conroe X6800 average of 3 runs on 975 motherboard

300x12 3-3-3 1:1 = 46800

300x12 4-4-4 1:1 = 46350

300x12 4-4-4 2:3 = 45800

360x10 4-4-4 1:1 = 47450

360x10 3-3-3 1:1 = 47700


I've averaged to 50 points but you get the drift. If you are running Sli on 975 then stay 1:1 and get the FSB up. If you have to go off 1:1 then you are flogging a dead horse and from my results memory bandwidth is not King for 3dmark2001 with Sli nvidia.

I think each person needs to suck it and see on their particular setup.

Regards

Andy

[nn_step]

I think each person needs to suck it and see on their particular setup.

Regards

Andy

Exactly, every Ram, Mobo, and CPU is different. Sometimes you have to feel out what your system will do

[Fr3ak]

And I already thought I am doing something seriously wrong.
1:1 is also slightly faster for me.

Nice comparisson Kunaak. Looks like Mhz wins on 965.

[sierra_bound]

1:1 generally works better with 975X platform because of the way the straps are set. With the P5B Dlx, if you boot into Windows above 400FSB you are on a looser strap. The same is not true for most 975X boards. The advantage of 965 boards is that they will hit higher memory speeds, with reasonable voltage. And as the tests in this thread have shown, high RAM speed will beat 1:1 and tighter timings, at least on the 965 platform.

I use both 965 and 975X. There are things I like about each. But you can't draw conclusions about the 965 platform, based on your experience with 975X. The opposite is also true.

[bachus_anonym]

1:1 being faster than 4:5 or 2:3 seems to be 975x "problem" then. I have not run 3D01 (I might try soon, though) but a while ago I compared SPi16M and 4:5, 2:3 was sure noticeably faster than 1:1

Comparison of memory ratios - SPi16M on ASUS P5B Deluxe (965P)

400x9, DDR2-800, 1:1  - 6:26.781 
400x9, DDR2-1000, 4:5 - 6:12.000
400x9, DDR2-1200, 2:3 - 6:05.000

So, like sierra says, what is true for 975x may not (and probably will not) be the same for 965P...