AMD Radeon HD 2900 XT overclocking results

[jabski]

yes, its also listed on AMD site of certified PSU. and hopefully you will get newer version with 2 6pin AND 2 8pin pci-e connectors. mine is on its way although NO one could confirm for me if it was the newer version.

sweet. Also gona be using water cooling. Just need to order second card now
Also which board would be best to use. I have a P5WDH , DFI RD600 or a P5K Dlx. I would think the P5K myself

[Daveb2012]

Crossfire HD2900Xt + Conroe @ 4ghz + Floppy Drive + DVD RW Drive + 6 fans + lights + ChillControl + X-FI and 3x 74GB Raptors in Raid 0. Game all day with this Silverstone 750W

but is the x2900xt OC'd?

[jabski]

but is the x2900xt OC'd?

yes i would like to know that too

[SSXeon]

Graphics card installed to test setup: QX6700, DFI RD600, pair of Corsair XMS6400-C3 and Enermax Galaxy 850 W.

Tests next week

Go Sampsa! Waiting for results.

[Sampsa]

First test with dry ice. Unfortunately R600 + Aluminium solid container and 1 mm copper cold plate between them is not the optimal configuration for this kind of heat load. Also adjusting the contact for perfect is really tricky.







Pretty lame clocks on Radeon HD 2900 XT so far: 950 / 1030 MHz. Positive thing is that it feels like my QX6700 week 32 ES is working ALOT better on DFI's RD600 motherboard compared to nF 680i SLI platform with 8800 GTX SLI.

[jas420221]

Those clocks are on DI? Wow, I at least expected 1k...

[VonD'orr]

Those clocks are on DI? Wow, I at least expected 1k...

read his sentence very well
heat transfer etc etc..

[jas420221]

read his sentence very well
heat transfer etc etc..

I saw that my man, dont worry!

[malik22]

what the better chipset for crossfire with these cards the 975 or P35.

[Evil]

Pretty lame clocks on Radeon HD 2900 XT so far: 950 / 1030 MHz.

First i have to say that your results always kick ass. But what i really wanted to know is why are you using only 1030mhz on the mem, i know it can do more than that because mine can do 1130 with AmdGpuClockTool easy. Does the score really decrease if you go pass 1030mhz or is there something else.

malik22, i think that 975 is faster but that´s just my opinion.

[BenchZowner]

975x...

[malik22]

How much of a performance loss does 16x4 have compared to 8x8?

[k|ngp|n]

First test with dry ice. Unfortunately R600 + Aluminium solid container and 1 mm copper cold plate between them is not the optimal configuration for this kind of heat load. Also adjusting the contact for perfect is really tricky.

I love that pic sampsa .

[railmeat]

For 24/7 usage?

Sure LN2/D.I./Cascade cooling is fun, but 24/7 gaming ready is what matters.

could not agree more,stability for gaming is a must vs benchmarks for like 30 seconds.8 hours of a kik ass gaming session is what u bought your video card for.just set it to safe clocks and actually enjoy it without blowing it up the second u get intro a jet at 2530x1536 and really put a load on it.u can bench on a 5200 ffs.not to many true gamers anymore around seems.i guess benching is cool but i would rather play games then try and roast my new hardware for months .set it up 2-3 days find stable\safe speeds,run some benchies for e-penis bragging rights then enjoy it by gaming on it?weird me.only seems like guys wanna bench everyday then game anymore,heh.

play some bf2 on it for 4 hours and see how "real world stable" it is.

[STEvil]

If you guys want to game, have fun.

I like watching drag races as much as a good dakar rally, myself.

[blind_ripper]

love the pink beatch toy sampsa
omg that quad off ure's really that bad ?

[Sampsa]

First i have to say that your results always kick ass. But what i really wanted to know is why are you using only 1030mhz on the mem, i know it can do more than that because mine can do 1130 with AmdGpuClockTool easy. Does the score really decrease if you go pass 1030mhz or is there something else.

Please keep in mind that hardware I'm testing are mostly early Engineering Samples, not retails products. 1030 MHz seems to be the limit for this card under stock cooling.

love the pink beatch toy sampsa
omg that quad off ure's really that bad ?

It's not that bad but in this first test the focus was on GPU contact mainly and to get a feeling how HD 2900 XT behaves under cold. I can't say contact was perfect. With nF 680i SLI platform I couldn't push my QX6700 more than 4,3 - 4,4 GHz under dry ice cooling and heavily overclocked 8800 GTX SLI configuration but again, I was using early Asus and EVGA review sample boards.

Same old phrase applies here, there is plenty to improve I'll keep posting here when something new happens.

I have two 8800 Ultras here and new Asus Striker Extreme board so I'll give them a try also

[LowRun]

could not agree more,stability for gaming is a must vs benchmarks for like 30 seconds.8 hours of a kik ass gaming session is what u bought your video card for.just set it to safe clocks and actually enjoy it without blowing it up the second u get intro a jet at 2530x1536 and really put a load on it.u can bench on a 5200 ffs.not to many true gamers anymore around seems.i guess benching is cool but i would rather play games then try and roast my new hardware for months .set it up 2-3 days find stable\safe speeds,run some benchies for e-penis bragging rights then enjoy it by gaming on it?weird me.only seems like guys wanna bench everyday then game anymore,heh.

play some bf2 on it for 4 hours and see how "real world stable" it is.

This is the HD2900XT OC results thread in the Xtreme 3D Team section on XS, not the relevant place to vent your desires for 4 hours of bf2

[Sampsa]

Also note that this run was with default voltage

[blind_ripper]

really strang how one week makes all the differens :p
week 33ES here 4.55ghz dice on P5W64 , should be easy 4.6+ghz on P5B dlx now .
what vcore do u need for the 4.3+ ?

[Dynasty]

Those clocks are on DI? Wow, I at least expected 1k...

I don't think he is even close in maxing out vgpu......he is probably using
default voltage there........


nice results Sampsa............

[jas420221]

I don't think he is even close in maxing out vgpu......he is probably using
default voltage there........


nice results Sampsa............

Appears so...

[malik22]

How much of a performance loss does 16x4(965 P35) have compared to 8x8(975 RD600) running these cards in crossfire?

[Kinc]

How much of a performance loss does 16x4(965 P35) have compared to 8x8(975 RD600) running these cards in crossfire?

A lot actually. ASUS Blitz p35 will have a PCI-E switch for 8+8 though

[Fr3ak]

A lot actually. ASUS Blitz p35 will have a PCI-E switch for 8+8 though

That is great news