R9 290 non(X) overclocking results

[Matson]

Does a 2 fps difference between stock and 1100/1400 sound right or completely wrong aka throttling? I guess it would be down to the 14.2 power limit bug.
only got 80 points more in valley between stock on my crossfire 290 tri x cards and 1100/1400.

 

[LtMatt]

Does a 2 fps difference between stock and 1100/1400 sound right or completely wrong aka throttling? I guess it would be down to the 14.2 power limit bug.
only got 80 points more in valley between stock on my crossfire 290 tri x cards and 1100/1400.

Use 13.12. 14.2 are no good for clocking/benchmarks.

 

[Litmoose]

Yea I can confirm is the drivers. Rolled back to the official ones and no throttle now

 

[LtMatt]

Does a 2 fps difference between stock and 1100/1400 sound right or completely wrong aka throttling? I guess it would be down to the 14.2 power limit bug.
only got 80 points more in valley between stock on my crossfire 290 tri x cards and 1100/1400.

Use 13.12. 14.2 are no good for clocking/benchmarks.

Rolled back to the official ones and no throttle now

 

[Litmoose]

http://www.3dmark.com/fs/1853681

My best yet at like 1165/1367.5

Could do more if i Oc'ed cpu and ram some more.


PS whats the safest max temp i should let VRams get to?

 

[Rangerjr1]

http://www.3dmark.com/fs/1853681

My best yet at like 1165/1367.5

Could do more if i Oc'ed cpu and ram some more.


PS whats the safest max temp i should let VRams get to?

Didnt know 290ps had VRAM sensors, mine doesnt atleast. And since when could you increase RAM in .5 increments? lol

 

[DMF1986]

Didnt know 290ps had VRAM sensors, mine doesnt atleast. And since when could you increase RAM in .5 increments? lol

Use HWInfo64 for temps on VRMs

 

[Litmoose]

Didnt know 290ps had VRAM sensors, mine doesnt atleast. And since when could you increase RAM in .5 increments? lol

Effective speed = 5470 / 4 = 1367.5


although realistically it probably rounds it up or down

 

[LtMatt]

Use HWInfo64 for temps on VRMs

Yeah i use HWINFO64 to monitor VRM and Vram temps. I found vrms sit exactly 10C cooler than whatever the core temp is.

 

[Rangerjr1]

Use HWInfo64 for temps on VRMs

VRM = / = VRAM...

 

[Rangerjr1]

Yeah i use HWINFO64 to monitor VRM and Vram temps. I found vrms sit exactly 10C cooler than whatever the core temp is.

Is there such a thing as VIDEO RAM SENSORS?

 

[Matson]

Over 1000 points more in firestrike on 13.12 compared to 14.2... Heaven and Valley I get exactly the same though. So I guess it wasn't throttling hardly in those with the power limit bug.

 

[elJarmen]

Throttling is solvable with the Radeon Pro fix on 14.2. For some reason it doesn't work with BF4 though (maybe because Mantle).

 

[LtMatt]

Is there such a thing as VIDEO RAM SENSORS?

There must be a way of reading it as HWINFO has and has done for all my Radeon reference gpu's. Incidentally my HIS 7950's did not have vrm or vram sensors.

 

[WalderX]

I've got my reference 290X on ebay, currently going for £355

Before I post it off on Monday I'm gonna stick it in my machine and run a few crossfire benches - I assume this will work with a 290 and 290X? I've never tried crossfire before you see!

 

[Matson]

Is it safe to max out the voltage to +200 for benchmarking? the card stays under 70 degrees. I wouldn't run it at that 24/7 as I just leave it at stock but im trying to see how high I can push the overclock.

 

[WalderX]

Is it safe to max out the voltage to +200 for benchmarking? the card stays under 70 degrees. I wouldn't run it at that 24/7 as I just leave it at stock but im trying to see how high I can push the overclock.

Yeah +200 is safe for benchmarking I have run mine at that many times for benching - think temps hit 75. Still 20 degrees lower than its max operating temp

 

[Matson]

what overclock did you manage to get using +200 since we have the same card so im curious what you got too..
I just had it at 1150/1450 with +150 but I was getting artifacts so I don't think id be able to push it much higher..at least on the core.

 

[WalderX]

what overclock did you manage to get using +200 since we have the same card so im curious what you got too..
I just had it at 1150/1450 with +150 but I was getting artifacts so I don't think id be able to push it much higher..at least on the core.

Do you know what memory your card has? Artifacts are usually related to too much core speed / not enough voltage but I imagine too high mem clocks also cause them.

I can get to 1190 core and 1750 mem with +200mv and +50 power limit. If I set 1200 core I get artifacts, and if I set 1800 memory I get a blackscreen crash. I have Hynix memory, if you have Elpida it is unlikely you will get above 1400 mhz memory in my experience.

So try leave memory at stock 1250, set power to +200, power limit to plus 50% and try 1200 core. If you get artifacts try 1190, then 1180 etc... Once core is stable set memory to 1300 then 1350, and so on unitl you get artifacts or a crash again.

Hope this helps

 

[Matson]

Do you know what memory your card has? Artifacts are usually related to too much core speed / not enough voltage but I imagine too high mem clocks also cause them.

I can get to 1190 core and 1750 mem with +200mv and +50 power limit. If I set 1200 core I get artifacts, and if I set 1800 memory I get a blackscreen crash. I have Hynix memory, if you have Elpida it is unlikely you will get above 1400 mhz memory in my experience.

So try leave memory at stock 1250, set power to +200, power limit to plus 50% and try 1200 core. If you get artifacts try 1190, then 1180 etc... Once core is stable set memory to 1300 then 1350, and so on unitl you get artifacts or a crash again.

Hope this helps

Yeah its the Hynix memory.
Is it best to try and see how far you can push the core first and then do the memory afterwards... With my mem at 1450 get no artifacts but pushing core up to 1150 even with +150 on the voltage I get artifacts.