Geforce GTX285 overclocking thread

[Biker]

Hi all .

I just was curious how ppl were getting on with their GTX285's....

I was hoping that they would be voltage adjustable via Rivatuner like several other cards have been found to be over the last week or two but it turns out things are not so simple.

Hopefully it is just a case of finding the right register to write to but things have scaled down from the 280's 4 voltage state registers to just 2. Whether other registers can be opened up to increase the range is at the moment unknown (well at least it is by me!) but right now it seems all 285's have a max voltage of 1.15 before you break out the soldering kits (it was 1.1875 in the 280) so lets see how we get on with these limitations....

I was initially going to fork out for an EVGA SSC but I had a feeling that most of these cards would clock more or less the same so I could not justify paying almost £100 more for a card that might be already maxed out so I settled for a Zotac AMP! edition which arrived today...

First thing I did was back up the bios but I was very reluctant to flash anything without checking its integrity with Nibitor. The latest available version 4.8 does not support the 285/95 series yet so I was stuck with Zotac's own "Firestorm" tool which provides all the basics: Core / Shader / Ram and Fan speed sliders.

Having read reports of ludicrous clocks I was tempted to go for broke but have decided to take it relatively easy for now.

I have just finished a 3 hour gaming session at 720 / 1620 / 2772 and all was perfectly stable / artifract free so I am going to eventually go for more but at this speed I can finally do rediculous things like play Crysis Warhead and 2560x1600 with all detail maxed out (DX9).

I was also supplied with an evaluation copy of Nibitor 4.9 to test and can happily report that it works flawlessly...

How is everybody else getting on?

[Blkout]

eVGA 285 vanilla here. These are actual clocks, as Nvidia sets clock speeds in increments. Stock cooler, auto fan profile. 100% stable and artifact free.

729c/1620s/1458s

From my testing, Crysis/Warhead benchmark and 3DMARK06 loops are the best tests to check for artifacts and crashes when overclocking.

[Talonman]

Some 285 numbers...

The memory does pritty darn well...

http://www.evga.com/forums/tm.asp?m=...55983;�

[Envydia007]

Mine is just at SSC speeds. Awesome vanilla.

[Blkout]

Some 285 numbers...

The memory does pritty darn well...

http://www.evga.com/forums/tm.asp?m=...55983;�

Lots of those numbers aren't ACTUAL clock speeds though. Apparently most people don't realize that Nvidia clocks their cards in increments.

For example, I can set the the shader clock to 1638, but actual is only 1620. Once I set the shader clock to 1639, actual clocks become 1656. Same goes for the core clock speed but they move in smaller incremements.

It's a little frustrating to compare clock speeds when people aren't using ACTUAL clocks.

[Envydia007]

Regardless, the 285's are awesome clockers.

[NapalmV5]

whatever clocks you think youre gonna post biker are gonna be invalid till you share nibitor 4.9 with the rest of us

[dengyong]

whatever clocks you think youre gonna post biker are gonna be invalid till you share nibitor 4.9 with the rest of us

+1

[Blkout]

Regardless, the 285's are awesome clockers.

Who are you speaking to? Who said the 285's didn't overclock well?

[bito]

I have found that whilst 3dmark and other benchmarks run relatively easily, including Furmark, actually playing Crysis Warhead will test the card far better.

[Brahmzy]

I have found that whilst 3dmark and other benchmarks run relatively easily, including Furmark, actually playing Crysis Warhead will test the card far better.

Me too. I always use Crysis Warhead loops to test my overclocks. I'll either get artifacting or black screens from it.

[pinto]

My two Evga SSC stock cooling 100%:

[Blkout]

I have found that whilst 3dmark and other benchmarks run relatively easily, including Furmark, actually playing Crysis Warhead will test the card far better.

It's a lot easier to see artifacts in 3DMARK06 during the forest and canyon tests. Both are also VERY stressful on a video card when run at 1920x1200, 4xAA. You can easily spot artifacts using no more than 3 loops of each test.

After I pass those test, then I move to Crysis Warhead at 1920X1200 and usually I'm pretty close to where I want to be. Artifacts are harder to spot, but crashes are easy using Crysis Warhead.

I use BOTH tests, not just one or the other, they work well together for different overclock testing.

[bito]

Excellent testing regime Blkout
Crysis Warhead will heat up and stress the cards like no other game, including FC2.

[G4h4o8s6T]

My vanilla eVGA 285 on water..........

771-1620-1476

30min Furmark\Warhead looped 10x

http://img8.imageshack.us/img8/2579/...rk30minzy5.jpg

http://img5.imageshack.us/img5/3476/...ad10loogu3.jpg

[Blkout]

My vanilla eVGA 285 on water..........

771-1620-1476

30min Furmark\Warhead looped 10x

http://img8.imageshack.us/img8/2579/...rk30minzy5.jpg

http://img5.imageshack.us/img5/3476/...ad10loogu3.jpg

Are you using the MCW60 for water cooling?

[G4h4o8s6T]

Are you using the MCW60 for water cooling?

Yeah, MCW60 + Ram/Mosfet sinks

[pinto]

Still stock cooling:

[Blkout]

Yeah, MCW60 + Ram/Mosfet sinks

Cool, that's what I thought. Your core speed is much higher than mine on stock cooling but your shader and memory speed isn't any better so that's why I was wondering. I'm going with a FC block because I want more memory speed as well, but I'm just waiting to see some reviews from those that have tried them first. I would buy Danger Den without question, but they don't have a block yet.

Got any pics by any chance?

[G4h4o8s6T]

Cool, that's what I thought. Your core speed is much higher than mine on stock cooling but your shader and memory speed isn't any better so that's why I was wondering. I'm going with a FC block because I want more memory speed as well, but I'm just waiting to see some reviews from those that have tried them first. I would buy Danger Den without question, but they don't have a block yet.

Got any pics by any chance?

The ramsinks on my card dont even get that hot, so not sure if a FC block would help mem clocks too much.

Here are a few pics, but I haven't gotten a chance to do any cable management, so still messy

http://img8.imageshack.us/img8/8625/dscn0178iu0.jpg

http://img8.imageshack.us/img8/8887/p2080005aq2.jpg

[Blkout]

The ramsinks on my card dont even get that hot, so not sure if a FC block would help mem clocks too much.

Here are a few pics, but I haven't gotten a chance to do any cable management, so still messy

http://img8.imageshack.us/img8/8625/dscn0178iu0.jpg

http://img8.imageshack.us/img8/8887/p2080005aq2.jpg

Cool, thanks.

[dengyong]

Lots of those numbers aren't ACTUAL clock speeds though. Apparently most people don't realize that Nvidia clocks their cards in increments.

For example, I can set the the shader clock to 1638, but actual is only 1620. Once I set the shader clock to 1639, actual clocks become 1656. Same goes for the core clock speed but they move in smaller incremements.

It's a little frustrating to compare clock speeds when people aren't using ACTUAL clocks.

Do you have a chart ?

[techguy2k5]

Just ordered an XFX vanilla 285, anyone else have one of these? I'll post my results next weekend, but hoping to get an idea what to expect beforehand. Also wondering how CPU-limited I'm going to be with an E8400 @ 4GHz (8x500). Considering an upgrade to a QX because I figure that's the only quad I'd have a chance of hitting 4GHz on air.

[Glow9]

Rivatuner ickkkkkkkkkkkkkkkkkkkk EVGA precision all the way

[dengyong]

Rivatuner ickkkkkkkkkkkkkkkkkkkk EVGA precision all the way

They are the same, precision just has a simple gui.