R290X Crossfire V NVDA's SLI 780s/Titans from [H]

[BrentJ]

There seems to be a misconception about the [H] use of uncapped fan.

One isn't setting it at 100% - [H] just allowed (in the single card review) the fan to potentially spin @100% but according to them it never went above 65%.

correct

the way CCC power tune works now is that the fan speed slider is a "fan cap" slider, whatever percentage you set it to is the allowable percentage it can "go up to" that doesn't mean it will run at that speed though, the fan is still completely dynamic

therefore, i never saw the fan get to 100% fan speed when it was unlocked at 100% fan cap, 65-70% was the max I saw

apparently, there was a performance difference in FC3 going from 55%-65-70% fan speed, 55% was bottlenecking it, but 65-70% was not

the GPU will always try to be 95c unless you change the max temperature, which we didn't touch

 

[GaiaHunter]

I think the point he is making, is that what's the point of running 65% fan speed when it is going to give you a headache.

Between 55% of the uber mode and 65% there is 10%.

They never said the cards actually reached 65% or 75%, just that 65% and 75% is unbearable.

EDIT: Well I guess FC3 does need more fan for full performance then.

 

[toyota]

Between 55% of the uber mode and 65% there is 10%.

They never said the cards actually reached 65% or 75%, just that 65% and 75% is unbearable.

lol

 

[Rezist]

Kyle isn't infallible and his preferences and definition can be vastly different than anyone else. His or Brent's opinions aren't universal and never were, even when they thought SLI was much smoother. However, given that even PCPer's data shows the variance difference being so small, it is basically a moot point now. I just want to see AMD continue to improve upon this.

They got the lead now? But I'm sure the only way is up.

 

[GaiaHunter]

and can you read? I just showed you their comment saying that will go up that loud if it needs to and the POINT is that it does goes past the 55% so STOP saying it doesn't. what part of them complaining about noise at 65-70% can you not understand? jeez

And does the performance achieved and presented in the graphs need 65% fan or is it between 55% and 65% fan?

All of our tests on the gameplay pages are in "Uber Mode."

So, yeah.

lol

55%, which is audible but not intolerable and nothing like it has been said in this thread that it requires 100% fan to run crossfire.

 

[BrentJ]

The statement I'm about to make will clear things up.

In all the games we have played so far, at every resolution up to 4K, including single card and CrossFire, in order to guarantee that you are stabilized at a 1GHz GPU frequency at 95c, a fan speed of 65-70% must be maintained. This will ensure all your games are not bottlenecking by downclocking of the GPU while gaming. To achieve this, you must set the fan slider higher than "Uber Mode", setting it to 70% or higher will work. So far, this seems to be the golden spot for achieving a stable 1GHz clock speed in every game, single and CrossFire.

70% fan/95c GPU = Stable 1GHz while gaming

 

[GaiaHunter]

The statement I'm about to make will clear things up.

In all the games we have played so far, at every resolution up to 4K, including single card and CrossFire, in order to guarantee that you are stabilized at a 1GHz GPU frequency at 95c, a fan speed of 65-70% must be maintained. This will ensure all your games are not bottlenecking by downclocking of the GPU while gaming. To achieve this, you must set the fan slider higher than "Uber Mode", setting it to 70% or higher will work. So far, this seems to be the golden spot for achieving a stable 1GHz clock speed in every game, single and CrossFire.

65-70% fan/95c GPU/Stable 1GHz while gaming

But the benchmark results were achieved using the uber setting?

 

[wand3r3r]

Kyle isn't infallible and his preferences and definition can be vastly different than anyone else. His or Brent's opinions aren't universal and never were, even when they thought SLI was much smoother. However, given that even PCPer's data shows the variance difference being so small, it is basically a moot point now. I just want to see AMD continue to improve upon this.

So in a roundabout way you are saying the data backs up their opinion.

 

[BrentJ]

But the benchmark results were achieved using the uber setting?

Yes, Uber at 55%.

The only game I think this really affects, in our lineup, was Far Cry 3.

It may also affect BF4, not sure yet.

Not all games need that higher than Uber mode fan speed to get the full performance keep in mind. Some are just happy at 55%, IT DEPENDS SOLELY ON THE GAME.

 

[BrentJ]

My statement above, is generally speaking. To ensure in ALL games, if you are paranoid about downclocking.

But not all games behave the same. Some are JUST HAPPY at Uber 55%. Some, are not. Depends on the game.

MOST have been OK at Uber 55%.

 

[GaiaHunter]

Yes, Uber at 55%.

The only game I think this really affects, in our lineup, was Far Cry 3.

It may also affect BF4, not sure yet.

Not all games need that higher than Uber mode fan speed to get the full performance keep in mind. Some are just happy at 55%, IT DEPENDS SOLELY ON THE GAME.

Thanks.

 

[BrentJ]

This is why my opinion is that Uber mode should have just been set higher, or unlocked completely to 100%. Let there be the quiet mode, but then let users flip a switch and bam, no limits.

 

[toyota]

Brent just said in the comments that he has to maintain at least 65-70% fan speed to keep card from lowering clocks. so you people arguing they did not go over 55% were wrong. and again they already said 65-70% was headache inducing loud.

 

[GaiaHunter]

Brent just said in the comments that he has to maintain at least 65-70% fan speed to keep card from lowering clocks. so you people arguing they did not go over 55% were wrong. and again they already said 65-70% was headache inducing loud.

Sure, but even with lower clocks in some games it is roflstomping the competition - all the benchmark results are with 55% fan.

With a better cooler some games will be even faster.

 

[BrentJ]

Brent just said in the comments that he has to maintain at least 65-70% fan speed to keep card from lowering clocks.

But keep in mind, not for ALL games. I'm just generally speaking, if you want to make sure that in every game you play there isn't downclocking. But a lot of games are perfectly happy at Uber Mode 55%. It depends on the game demand.

so you people arguing they did not go over 55% were wrong. and again they already said 65-70% was headache inducing loud.

In our testing, for the actual performance data, it was all at Uber Mode 55%.

We did crank up the fan slider to see how noise would be affected, hence our comments when the slider is at 100% and the fan is allowed to go up to 70% while gaming.

 

[ICDP]

Brent just said in the comments that he has to maintain at least 65-70% fan speed to keep card from lowering clocks. so you people arguing they did not go over 55% were wrong. and again they already said 65-70% was headache inducing loud.

He also said all their benchmarks were done at Uber which does not exceed 55%. So essentially custom cooled R9 290X cards will be able to widen the gap even further because they will eliminate throttling.

 

[Silverforce11]

Brent just said in the comments that he has to maintain at least 65-70% fan speed to keep card from lowering clocks. so you people arguing they did not go over 55% were wrong. and again they already said 65-70% was headache inducing loud.

I don't even know how you could verbal him like that, his statement is plain as day. Performance data is done with 55% fan cap in uber mode. Thus, it potentially is running throttled.

They just merely stated, to get maximum performance, release the limit above 55%.

These AMD ref cards are simply a PRELUDE to the real fun from AIBs, AMD has yet again gone with a crap default HSF so that their partners can be all creative with their own solutions. That crap ref HSF card is for people who go water, because waterblocks are generally designed after a standard PCB layout. Nobody sane would use these blowers on a day to day basis, above 55% would be ridiculously noisy.

 

[UaVaj]

The statement I'm about to make will clear things up.

In all the games we have played so far, at every resolution up to 4K, including single card and CrossFire, in order to guarantee that you are stabilized at a 1GHz GPU frequency at 95c, a fan speed of 65-70% must be maintained. This will ensure all your games are not bottlenecking by downclocking of the GPU while gaming. To achieve this, you must set the fan slider higher than "Uber Mode", setting it to 70% or higher will work. So far, this seems to be the golden spot for achieving a stable 1GHz clock speed in every game, single and CrossFire.

70% fan/95c GPU = Stable 1GHz while gaming

good info. :thumbsup::thumbsup:

 

[DooKey]

This is why my opinion is that Uber mode should have just been set higher, or unlocked completely to 100%. Let there be the quiet mode, but then let users flip a switch and bam, no limits.

Brent, aren't all of your tests done on a tray setup, with the cards open to the ambient air rather than in a case?

 

[BrentJ]

Brent, aren't all of your tests done on a tray setup, with the cards open to the ambient air rather than in a case?

All our test systems are in an open environment, sitting on the motherboard on a test bench.

 

[Will Robinson]

Hi Brent...thanks for posting your views in my thread on the Xfire results.
We all appreciate you dropping by and taking the time to expand on the [H] conclusions and results.
Unfortunately a few posters in this thread have gone beyond stating their objections to this card and have begun what is pretty close to spamming the thread by repeatedly posting these same objections over and over again.
No doubt you have to deal with this at the [H] forums as well and I would remind those posters that this thread is being monitored by the VC&G moderators who are prepared to take some pretty decisive action should it continue.
Thanks to all who have contributed so far.

 

[f1sherman]

I'm pretty sure Nvidia's recommendation to reviewers is to use enclosed casing.
Brent, can you disclose what are AMD's recommendations for 290X (Crossifre) benchmarking?

Hi Brent...thanks for posting your views in my thread on the Xfire results.
We all appreciate you dropping by and taking the time to expand on the [H] conclusions and results.
Unfortunately a few posters in this thread have gone beyond stating their objections to this card and have begun what is pretty close to spamming the thread by repeatedly posting these same objections over and over again.
No doubt you have to deal with this at the [H] forums as well and I would remind those posters that this thread is being monitored by the VC&G moderators who are prepared to take some pretty decisive action should it continue.
Thanks to all who have contributed so far.

Thread crapping your own thread... Nice




There was no need to add that comment. OP was speaking in general terms, and not to you.

You were the one in fact threadcrapping.


esquared
Anandtech Forum Director

 

[BrentJ]

I'm pretty sure Nvidia's recommendation to reviewers is to use enclosed casing.
Brent, can you disclose what are AMD's recommendations for 290X (Crossifre) benchmarking?



Thread crapping your own thread... Nice

This is all that AMD has said in regards to CrossFire, which is the same thing we posted in the review:

"For optimal performance in AMD CrossFire™ Technology mode on the R9 290X we recommend using the “Uber Mode” bios switch position. It is also advisable to use a motherboard that allows for some open space between your graphics card."

I agree with this recommendation.

 

[tviceman]

This is all that AMD has said in regards to CrossFire, which is the same thing we posted in the review:

"For optimal performance in AMD CrossFire™ Technology mode on the R9 290X we recommend using the “Uber Mode” bios switch position. It is also advisable to use a motherboard that allows for some open space between your graphics card."

I agree with this recommendation.

I'm under the very strong impression that an open test bench will be favorable in regards to temps and/or fan speed settings with a target temperature. Is this true?

 

[Rvenger]

I'm under the very strong impression that an open test bench will be favorable in regards to temps and/or fan speed settings with a target temperature. Is this true?

No. A case with good airflow is just as good if not better. Any enthusiast would know this.