Any info on Radeon 256 DDR-II Overclocking??

[Hellbinder]

I looked briefly again through the initial reviews, but i am not seeing anything about what that DDR-II on the 9800pro is capable of. Since it is officialyl released now. Would someone who is testing one, or perhaps an ATi Gent, Comment on what they have seen the Ram on the 256 get pushed to???

I find it a little strange that this is not a big topic of discussion. The DDR-II has the same potential for overclockability that the 9600pro core does.. No?

Thanks y'all.

 

[Joe DeFuria]

]I find it a little strange that this is not a big topic of discussion. The DDR-II has the same potential for overclockability that the 9600pro core does.. No?

No, the potential isn't the same. (It may be better or worse). I'm guessing worse, but we don't know.

For one thing, the DDR-II model an entirely different PCB than the 9800 DDR. The PCB itself plays a significant role in overclockability.

Also, there are more memory chips on the 256 MB version. The more chips you have, the more likely you will have problems. (Will only overclock as high as the weakest link.)

In any case, it is odd that no one appeared to even try overclocking one...

 

[MuFu]

I bet it's not that much over 400MHz. Firsty, all the TL and I/O settings are probably configured to keep power consumption down. Should be quite easy to tweak for higher clockspeeds though if the PCB is up to it (it'll just use more power for termination). Secondly, the memory controller seems to fail pretty quickly once you start hitting ~450Mhz core.

MuFu.

 

[Hellbinder]

Right about the potential. But heck even if you can get 50-70MHZ out of it... along with a core tweak to 400 or higher. It would be interesting to take a new look at the numbers.

I heard, that the Ram used is on these is rated a lot higher than the actual speed on the card.

I mean if you could run this all day at 400/400 with 256mb ram you would have yourself one heck of a card. And i personally thik the feeling that you are getting what the 9800pro *Should* have been in the first place.

 

[MuFu]

It's worth remembering that they take big signal integrity hits compared with the 128MB board through the use of stub termination and the crazy trace density. This is likely one of the reasons they opted to use GDDR2, despite the similar clockspeed.

Any idea how many layers the A094 PCB has? Notice that the boards that have gone out are actual production models - not samples.

MuFu.

 

[ZenBearClaw!]

Samples Non-Samples? Huh?

Sorry MuFu i'm missing your point. Please explain what you meant by your comment about it being a non sample board? Does that make it closer to the limits or further from them? I'm just confused please inform me.

-sloan

 

[TempestX]

http://firingsquad.gamers.com/hardware/ati_radeon_9800_pro_review/page5.asp

 

[MuFu]

Sorry MuFu i'm missing your point. Please explain what you meant by your comment about it being a non sample board? Does that make it closer to the limits or further from them? I'm just confused please inform me.

-sloan

It's nothing to do with the subject of this thread, just a casual observation - the 256MB 9800s we've seen (in B3D's pictures, for example) are not samples, they are production boards.

This is encouraging because it means the design has been finalised (production qualified) and the product has already entered the mass-production phase. Hence the "".

MuFu.

 

[Brent]

I think I got 50Mhz oc on it? from 350 to 400? I don't remember exactly

I can tell you when I get home this afternoon and look at what i got

i know i didn't get as high as i thought it would go

 

[Evildeus]

Nous avons pu atteindre de manière stable et sans qu´aucun artefact n´apparaisse une fréquence de 405 MHz pour le VPU et 384.75 MHz pour la DDR-2 l´accompagnant, soit des gains respectifs de 6.6 et 9.9%.

Sous Quake III Arena en 1600*1200 avec AA 4x et Aniso 8x, on passe de 121.9 à 132.2 fps, soit un gain de 8.4%

http://www.hardware.fr/articles/464/page6.html

 

[Brent]

I got the card to 391.50 Mhz memory, from its default of 351Mhz

i'll be doing some more testing with that though a bit later

 

[ZoinKs!]

Based on the FiringSquad results from the link above, ati should have clocked the 9800pro higher. Fx 5900Ultra and 9800pro end up being roughly equal at default clocks. A higher clock on the 9800pro-256 would have taken a lot of steam out of the 5900Ultra.

EDIT: I'll refer you to the 16x12 aa+af UT2003 botmatch results at the bottom of this page. The 5900Ultra doesn't look too great in that graph.

 

[ZoinKs!]

Based on the FiringSquad results from the link above, ati should have clocked the 9800pro higher. Fx 5900Ultra and 9800pro end up being roughly equal at default clocks. A higher clock on the 9800pro-256 would have taken a lot of steam out of the 5900Ultra.

EDIT: I'll refer you to the 16x12 aa+af UT2003 botmatch results at the bottom of this page. The 5900Ultra doesn't look too great in that graph.

2nd edit: Looking again at the bottom graph, the 9800pro-256 seems too good to be true. Is it really getting 67 fps or did they get it mixed up with something else?

 

[WaltC]

I got the card to 391.50 Mhz memory, from its default of 351Mhz

i'll be doing some more testing with that though a bit later

Not too bad...I've been stretching my 9800P's (128mb version) legs a bit and can run at a sustained 438/371 without artifacts or other problems--stock. The ram is pretty much topped out there as pushing it a bit more produces artifacts. The core I don't know about--I used the oc program for 440 and it auto-adjusted to 438.xMHz--I may see if I can go further with it later on. That's a significant improvement over my 9700P which I got to 350/333 stock config.