Need help O/Cing my FX5900se

[cmitchellfly]

I just installed the 53.03 drivers.

At 431 core and 737 memory I get 15,750 3Dmark01se. Seems like I should be able to push the core and mem a little higher.

Anybody else get better with this card?

 

[Sirius]

Hey you go this is all you...
http://www.ocforums.com/showthread.php?s=&threadid=268784
If you dont want to do the BIOS flash then just OC the core 5mhz and the memory 10mhz at a time and when your score seems to drop or you start to see artifacts then come down 5mhz on both and you should have your stable speed.

 

[cmitchellfly]

431 and 737 are the highest that I have been able to go so far. When I go higher and click "Test Settings" in the "Clock Frequencies" tab it gives me the error msg.

 

[Buhammot]

look for a program called Powerstrip.. much better... and i have been able to do like.. 475/800 with stock settings.... should replace some thermal pads with as epoxy or somethin.. but too lazy

 

[cmitchellfly]

Ill check it out. Thanks

What benchmarks are you getting with your 5900se?

 

[cmitchellfly]

Thanks Buhammot.

Im using PowerStrip now.

At 448/758 I get 15,923 3d 2001. Any higher than this and 3d mark locks up momentarily on some of the tests. I guess its the thermal throttling slowing it down for a few moments - of course the scores are alot lower when it locks.

And you are getting 475/800 - thats great - Ill keep trying!

Not bad for a $188 card from NewEgg....

 

[Mr.Guvernment]

i did not think the video card had thermal throttling? i thought that was only in CPU's...

 

[Buhammot]

Well the throttling is only after the core gets over like... 120C or 140C, depending on what set of drivers you have...

but my highest 03 score was 5970, and my highest 01 score was 16864 i think....

 

[cmitchellfly]

I think you are right about not throttling back until 140c. So what would make some of the tests in 3dmark lock momentarily at 448/758? I check my GPU temps right after 3dmark ends and it reads 50 - 52c so I dont think I was getting anywhere near 140c.

 

[cmitchellfly]

In coolbits 140c (which is @285f) is the temp that the GPU is supposed to throttle back. This must be an error though 140c is darn near 300 F. They must have meant 140F which is 60C.

 

[Mr.Guvernment]

^^^^ iwould think so as @ 140c the plastic would likely get soft and close to melting.

 

[sLk]

I'm thinking about picking up a 5900xt (same as card as the 5900se) instead of a Radeon 9600xt.

Can anyone give me a reason not to do this?

 

[cmitchellfly]

The 5900se outperforms the 9600xt by a fairly wide margin (in almost all tests) from the reviews I have seen. This is (I think) largely due to the 5900's 256 memory interface vs 9600xt at 128.

The only fault some may have with the 5900se (or any NVIDIA card) is that supposedly their DX9.0 support is not as good as ATI.

Still, for me after seeing the reviews and price it was a no brainer.

 

[Mr.Guvernment]

my reasons you should NOT get an FX card

DX -9 support is not comparable to ATI - infact i dont even think the FX lines can be considered to have FULL DX9 support that is how abd it is an NO driver will fix this! it is a design problem with the cards.

NVIDIA's lying and cheating and many optimization in their drivers resulting in more FPS and losing out on quality - such things as entire shadows and lighting missing from sceense and god know what else - and these optimizations as far as i know are still present in their recent drivers - or @ least their last batch.

Finalyl once you start to turn on AA and FSAA to the max - using NVIDIA drivers that DONT cheat - ATI tends to come out on top in the end for over all quality.

 

[sLk]

my reasons you should NOT get an FX card...

So you would rather get a Radeon 9600xt over a FX5900xt?
Both are similar in price in my area.

 

[OperativeSix]

Try using rivatuner to overclock the card. You might also want to check to make sure your heatsink makes good contact with your card. Mine doesn't, and as such, idle temps are over 50 degrees celcius, and load temps shoot up over 70!

The problem is many of the stock coolers use low pressure mounts, and combined with thermal pads, makes for very poor cooling. I can actually shine a light through mine and see to the other side when looking between the sink and core. Non-stock cooling is definately needed on some of these 5900's to OC well.

 

[cmitchellfly]

Thanks OperativeSix.

Questions:

Is Rivatuner better than PowerStrip or Coolbits for this card?

Do you have the 5900se? What core and mem timings do you get?

Any suggestions on a good (and fairly quiet) GPU cooler. The stock cooler is at least quiet (my rig is very quiet)

 

[OperativeSix]

Is Rivatuner better than PowerStrip or Coolbits for this card?

In my experience, they have all worked well. I happen to use Rivatuner because it allows you to monitor temperatures, core frequency, and memory frequency on-the-fly. This allows you to see just when your card starts throttling itself (my 5950 soft mod'd 5900 does this around 70 degrees C). To be honest, they're all good programs. Use whichever one suits you best.

Do you have the 5900se? What core and mem timings do you get?

I currently have a GFFX5900 NON-se, and softmodded it to a 5950 ultra using the leadtek A380 BIOS. Before the mod, I could *bench* (I clock back from my maximums GREATLY) at 455/960. After the mod, I can loop a few runs at 498/1050 .

Any suggestions on a good (and fairly quiet) GPU cooler. The stock cooler is at least quiet (my rig is very quiet)

The only one I can think of off the top of my head is zalman's passive heatpipe cooler, or a modified CPU heatsink of come type. I am planning to watercool mine, which is quite a different foray. To be honest, if you REALLY want to watercool your GPU, you can probably do it for around $60 if you look in the right places for pumps and blocks.

 

[Buhammot]

will people stop complaining about cheating in drivers? they do that to make their cards better, and dont even think that ati doesnt do it either...

 

[Mr.Guvernment]

^^^^ ATI did it and then stopped - ATI is not doing it @ the level NVIDIA is and then denying it completely and then going out and encrypting their drivers. AI fessed up to cheating and admitted it - also the performance gains that ATI got were very small such as 2-3% where as NVIDIA is getting 10+% in eprformance increase - also ATI is not cutting out ENTIRE scenes and background, shadows and lighting from games.

So no i will not stop complaining until NVIDIA stop cheating
\

They do it to make their cards "appear" better but YOU the buyer are getting screwed! I guess you dont mind getting cheated out of a product.