bit-tech.net

AMD Radeon R9 290 Review

Comments 26 to 50 of 65

Reply
teppic 5th November 2013, 10:22 Quote
60W or so more than a 7970 at absolute max load, it's not much extra heat.

The assumption that a 95C core temp is going to make a hotter case/room than a 60C core temp is what I'm talking about, it's nonsense.
GuilleAcoustic 5th November 2013, 10:40 Quote
Quote:
Originally Posted by teppic
60W or so more than a 7970 at absolute max load, it's not much extra heat.

The assumption that a 95C core temp is going to make a hotter case/room than a 60C core temp is what I'm talking about, it's nonsense.

Look at the thermanl imaging I linked .... it prove that the card doesn't spit hot air, but that because the vast majority of the heat is trapped inside the die. With cooler that will lower the core temp ... this removed heat will have to go somewhere, it won't just disappear.

The real choice is : Do you keep the heat inside the chip or do you dump it inside your case / room ?

To be honest, I wouldn't like to have the chip and PCB running that hot (cf thermal imaging link).
teppic 5th November 2013, 10:55 Quote
Quote:
Originally Posted by GuilleAcoustic
Look at the thermanl imaging I linked .... it prove that the card doesn't spit hot air, but that because the vast majority of the heat is trapped inside the die. With cooler that will lower the core temp ... this removed heat will have to go somewhere, it won't just disappear.

The real choice is : Do you keep the heat inside the chip or do you dump it inside your case / room ?

To be honest, I wouldn't like to have the chip and PCB running that hot (cf thermal imaging link).

The question is true of every GPU that exists, but the high core temp is to do with the cooler, not the heat produced. This produces 10% or so more heat than a 7970, it's not going to heat a room up.
The_Crapman 5th November 2013, 11:01 Quote
I've read other reviews of the 290x where they said the air coming out the back was hot enough to burn your skin. Not even the 480 was that bad.
AlphaAngel 5th November 2013, 11:02 Quote
Quote:
Originally Posted by teppic
High core temp doesn't mean more heat generated.

Did I just read that right?! Higher temperature doesn't equal hotter?!

You will need to explain that one I'm afraid.
Panos 5th November 2013, 11:04 Quote
A R9 290 @ £300-£320 is steal if compared to the GTX780 at £400. Because by spending ~£70 for MK-26 and two Enermax Magma fans, you can OC the beast quietly, on air and beat the crap out of overclocked 780s and Titans.

See the test results at OCUK having put the 290 at 1200 core!


However still tempted to wait for the Toxic around Xmas :D
teppic 5th November 2013, 11:06 Quote
Quote:
Originally Posted by AlphaAngel
Did I just read that right?! Higher temperature doesn't equal hotter?!

You will need to explain that one I'm afraid.

Take a 290X and run it with the reference cooler, you have a core temp of 95C. Replace the cooler with a Prolimatech custom one, and the core temp drops to 50-60C with exactly the same performance. Is the card producing less heat? It's using the same energy and is producing the exact same amount of heat, despite totally different core temps.
Harlequin 5th November 2013, 11:12 Quote
Quote:
Originally Posted by The_Crapman
I've read other reviews of the 290x where they said the air coming out the back was hot enough to burn your skin. Not even the 480 was that bad.



http://www.bit-tech.net/hardware/2010/03/27/nvidia-geforce-gtx-480-1-5gb-review/12

erm??
xaser04 5th November 2013, 11:17 Quote
Quote:
Originally Posted by Dogbert666
It may well be that in order to hit the good price-performance points we've seen from this series that this is where a sacrifice had to be made. I certainly couldn't see the cards being as cheap as they are were they to feature as high quality cooling as Nvidia's GTX 780, for example.

This is true, but I do have to wonder how much it would have cost AMD realistically to design a better cooler from the offset? An extra $20 or so on the cooler could easily have been added onto the cards retail price whilst still retaining the price/performance balance.

Titan/780 expense isn't really required, just something that can run alot quieter than the current one would be a job well done - even if it meant no temperature change.
AlphaAngel 5th November 2013, 11:36 Quote
Quote:
Originally Posted by teppic
Take a 290X and run it with the reference cooler, you have a core temp of 95C. Replace the cooler with a Prolimatech custom one, and the core temp drops to 50-60C with exactly the same performance. Is the card producing less heat? It's using the same energy and is producing the exact same amount of heat, despite totally different core temps.

I understand your point but the core it still very inefficient and a lot of power is lost as heat energy rather than something productive, the heat is only being dissipated better.

Whilst I agree the performance is nice, either AMD need to invest more in cooler technology or make the design more power efficent.
teppic 5th November 2013, 11:53 Quote
Quote:
Originally Posted by AlphaAngel
I understand your point but the core it still very inefficient and a lot of power is lost as heat energy rather than something productive, the heat is only being dissipated better.

Whilst I agree the performance is nice, either AMD need to invest more in cooler technology or make the design more power efficent.

It's not inefficient - it's a huge increase in performance over the 7970 for a relatively small energy increase. The core temp is irrelevant, had AMD put on a beefier heatsink the temp would be 70C and nobody would be talking about it being 'too hot', despite it using the same power and having the same efficiency.

It uses on average about 10W more than a 780, about 5-10W more than a 7970 GE.
AlphaAngel 5th November 2013, 12:00 Quote
Quote:
Originally Posted by teppic
It's not inefficient - it's a huge increase in performance over the 7970 for a relatively small energy increase. The core temp is irrelevant, had AMD put on a beefier heatsink the temp would be 70C and nobody would be talking about it being 'too hot', despite it using the same power and having the same efficiency.

It uses on average about 10W more than a 780, about 5-10W more than a 7970 GE.

Fair point. The question then has to be why has AMD shot themselves in the foot with an easily rectified issue.
teppic 5th November 2013, 12:02 Quote
Quote:
Originally Posted by AlphaAngel
Fair point. The question then has to be why has AMD shot themselves in the foot with an easily rectified issue.

The reference cooled models are aimed at people who put them under water or review sites, they won't be around for very long. It was the same with the 7970 - quite a cheap and noisy reference cooler that nobody wanted. There's probably some pressure from the partners too, as they sell their models based on their custom cooler designs mostly.
The_Crapman 5th November 2013, 12:18 Quote
Quote:
Exhaust air temp wasn't hot enough to burn skin. I did used to have my case on the floor facing backwards to keep my feet warm in winter though. Lol
Spreadie 5th November 2013, 12:26 Quote
Epic performance for the price, but too hot and too hungry.
GuilleAcoustic 5th November 2013, 12:29 Quote
Quote:
Originally Posted by teppic
Take a 290X and run it with the reference cooler, you have a core temp of 95C. Replace the cooler with a Prolimatech custom one, and the core temp drops to 50-60C with exactly the same performance. Is the card producing less heat? It's using the same energy and is producing the exact same amount of heat, despite totally different core temps.

So the heat vanished ? You'll have to explain that to me, then I'm going to trademark the material and build cars that doesn't heat up during summer. I'm going to be pretty rich.

How does heater works ? You heat a surface and it irradiates, thus heating the air, no ?

How does cooler works ? Could it be water, heatpipe or just bare alu heatsink ... you take the heat / wasted energy from the heating point and you take it far away with liquid / air. The heat doesn't vanished.

GTX780 thermal under load :

http://www.guru3d.com/index.php?ct=articles&action=file&id=7350

R9-290 thermal under load :

http://www.guru3d.com/index.php?ct=articles&action=file&id=7363

R9-290X thermal under load :

http://www.guru3d.com/index.php?ct=articles&action=file&id=7352

Take the same GPU, same speed, same load one with an efficient cooler another one with an inefficient one. If you have, let's say, a difference of 30°C inside the core temp it means that the efficient cooler took 30°C OFF the GPU die, transmitting it somewhere. For sure it WON'T heat up your room, but the area around the case (prolly under your desk) will rapidly heat up during load (just like an hairdrier won't heat your room but will heat a small area around itself).

EDIT: Higher core temp --> higher heat around the chip, just look at the thermal imagery I posted.
The_Crapman 5th November 2013, 12:31 Quote
Quote:
Originally Posted by teppic
It's not inefficient - it's a huge increase in performance over the 7970 for a relatively small energy increase. The core temp is irrelevant, had AMD put on a beefier heatsink the temp would be 70C and nobody would be talking about it being 'too hot', despite it using the same power and having the same efficiency.

It uses on average about 10W more than a 780, about 5-10W more than a 7970 GE.
I beg to differ. You say 'on average it uses 10w more', what are you basing this on? Using more power for less performance than the 780 would conclude it is less efficient.

A good way to settle this argument (i hope you're listening here bit-tech), would be to strap an aio watercooler on the gpu of each card and set the fans on it to a set speed. That way the cooling solution for each card is as identical as can be.

Then, run a benchmark on the 780 and record it's score and max temp. Set up the 290 and x on the same benchmark, setting it's target temp to whatever the max temp of the 780 was and see what score they get.

That would be the fairest apples to apples comparison you could get that i can think of. So why do you say BT, you game?
teppic 5th November 2013, 12:42 Quote
I'm basing it on benchmarks of power consumption. I don't just make things up...

http://tpucdn.com/reviews/AMD/R9_290/images/power_average.gif

Real gaming power use and maxed out stress testing always produce hugely different results.
The_Crapman 5th November 2013, 12:45 Quote
Oh how silly of me. I forgot crysis was a stress test not a game....
teppic 5th November 2013, 12:49 Quote
Those averages in that benchmark were taken running Crysis 2.
The_Crapman 5th November 2013, 13:02 Quote
Yes I've seen the tpu figures and also know that their testing methodology is very different. If bt get round to doing my proposed test, they could also include average power consumption. We also don't know how long they run the benchmark for, as when it heats up it throttles back to keep the temps in check, reducing the power consumption. Put on a better cooling solution and it'll keep the clocks and power up, consuming more.
teppic 5th November 2013, 13:22 Quote
Techspot records much lower power use (300W full system) for Crysis 3, same resolution but even higher settings. With the Titan it pulls 298W, but no 780 figure is provided. The 290 slightly outperforms the Titan though.

The main point I'm making is that the heat produced isn't related to the core temp. It's not going to heat up the case, it's not going to heat up the room - any more than a similarly powered blower card, that is.
rollo 5th November 2013, 13:39 Quote
Core temp does not = heat produced by a card.

My friend had his 480s under water and the room when those cards were under load was boiling. they were all sub 60c under water. The difference between his 480s and my 580s under water was huge. In the summer months I could game quiet happily. He could not without some form of Air con. All the gpus were within the same temperature bracket.

Not just blower cards that heat up rooms much to popular thinking. Even under water that heat has to go someplace ( to the rads basically) Those rads have fans on which remove heat which goes into the room shockingly enough.

new process would indicate this card is sub 28nm its not so dont talk nonesence. 7970 was 28nm.

Its some new architexture but both have hit the limits of 28nm if this was released in the summer I wonder if peoples oppinions would change. Not everyone has a thermal limited room to run the cards in. On a warm summers day these cards would surely downclock themselves at record pace.

780ti will be no better in this regard either.

AMD have made 2 good cards at a good price, Question is will anyone buy one not saw anyone on this site looking at either card. I know someone wants 3 780ti.

When 680 and 7970 were released was threads on both with people buying them at launch. Not saw that level of popularity for either the 780 / 780ti or the 290x 290.

Seen alot of vocal support but I wonder if any of those vocal people will actually be owners of the cards they seem to support. As someone once said we speak with our wallets.

Me personally waiting on 20nm tech and maxwell most likely.
GuilleAcoustic 5th November 2013, 13:39 Quote
The PCB exceeds 90°C in some area, and more than half of its surface is above 80°C .... do you really think it won't acts as a little heater ?
teppic 5th November 2013, 13:48 Quote
Quote:
Originally Posted by GuilleAcoustic
The PCB exceeds 90°C in some area, and more than half of its surface is above 80°C .... do you really think it won't acts as a little heater ?

No, since if you just stick a big cooler on it which lowers those recorded temps, it'll still be giving out the same amount of heat. Use a quieter cooler with lower core temps that exhausts into the case instead and all of the heat will go into the case, rather than out of the back.
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums