Reviews - AMD Radeon R9 290

[scaryjim]

I wonder if AIB partners will be a bit peeved by this, since it eats into the headroom for making their 'pre-overclocked' cards?

Doubt it - I can't see AMD running this on the edge of clock capabilities, just their cooling capabilities. Keep this cooler and I suspect it will clock a lot higher (depending on power draw and temps, of course).

I know people are worried about the temperature of the chip, but actually the *heat* shouldn't be too bad. I think a lot of people are missing the fact that the die area is relatively small for such a powerful chip, so there's an issue of heat density making the silicon reach a higher temperature. It can't kick out that much more heat than any other high end card, as it doesn't draw that much more power than any other high end card.

That said, I'm a bit disappointed that AMD have pulled driver and frequency related shenanigans to boost the card's performance prior to release. I'd love to see a full set of performance figures with the beta 5 drivers to see what level AMD originally intended this card to perform at before nvidia cut their prices...

[Domestic_Ginger]

Any chance of ASIC (?) quality and undervolting tests? Preeeease.

Toms stuck a arctic cooler on their card and it worked much better.

[MrJim]

I wonder if AIB partners will be a bit peeved by this, since it eats into the headroom for making their 'pre-overclocked' cards? Either way I agree with previous comments that this architecture badly needs partner boards' custom cooling before it becomes properly tempting.

Another kick up the backside for NVidia though! I feel more price drops coming on

The 290 is crying out for a better cooler to such an extent, I don't think this will be an issue. All AIB manufacturers need to do is shove a decent cooler on the card, & they'll fly off the shelves, even if the clock speeds remain the same as the reference design.

[kalniel]

That said, I'm a bit disappointed that AMD have pulled driver and frequency related shenanigans to boost the card's performance prior to release. I'd love to see a full set of performance figures with the beta 5 drivers to see what level AMD originally intended this card to perform at before nvidia cut their prices...

http://hardocp.com/article/2013/11/0..._card_review/7

They test old and new for that bench, and also show that increasing fan speed further made no difference.

Anand has a wider comparison:
http://www.anandtech.com/show/7481/t...9-290-review/2

[scaryjim]

Yeah, I just had a read of Tom's Hardware too - they have a variety of different testing setups. Perhaps most telling is what happens when you strap an Accelero Extreme III to it. If that's genuinely indicative of the potential of Hawaii, there's no wonder AMD are waiting on the release of 780 Ti before unleashing their partners on the chip...

[TaintedShirt]

So watercooling would make this an excellent buy?
Now we just have to wait for a block to come out.


Edit
just looked on coolingconfigurator.com and it appears that the 290x blocks fit the 290. Is this right?

[scaryjim]

Edit
just looked on coolingconfigurator.com and it appears that the 290x blocks fit the 290. Is this right?

AFAIK the 290 and 290X reference designs are basically hardware identical (to the extent that Hexus didn't even bother with hardware shots/teardown for this review). So, that is almost certain to be correct.

However, I don't have first hand experience of either card, so there's always the risk that there's a minor difference or two that would cause issues.

But, chances are it'll work.

EDIT: two caveats: 1) Tom's don't include temperature or - more importantly - power consumption figures for the DIY custom-cooled card. It's faster and quieter, but it might be chewing through vast amounts more power to do that. 2) It's reasonable to assume the 290X will benefit equally from extreme cooling, so if you can afford the initial outlay you'll get a faster card (that is guaranteed to fit a 290X waterblock )

[mrsqueezy]

Is there any intention for you guys to get a retail 290x and review it? I'm asking because Chris At Tom's got lower results with a retail 290x than the sample 290x that was sent by AMD . Anyway, Great job AMD. Well done!

This has made me worry a little about my choice of the R9 290 for my new build.

Consistency issues coupled with it being hot and noisy.

I hope they hurry with the non-reference coolers so that at least it can be inconsistent, cool, and relatively quiet.

[Jingee]

Waaaaaaay too hot and loud. And very power hungry at that! Was considering upgrading, but may wait 'til the next lot of cards. These are almost as bad as the GTX 400 series cards :/

[AETAaAS]

Love how nVidia fanboys swing from one side to the other. Temps and noise will go down with custom coolers and when Fermi initially hit, oh no, heat and power were of no concern. Now the 290X is using 20W more than a Titan and all of a sudden its a big priority. Even compact florescent bulbs use more than 20W, and with the PSUs most enthusiasts have today, 20W is only in the single digit percentages of available power.

Very aggressively priced card the 290 is for the performance but I'd wait for the custom cooled versions to hit.

[MrRockliffe]

It feels to me that they've just designed a chip to run at 95 degrees rather than a completely new, more efficient and quicker chip.

[MrRockliffe]

Yeah agreed, the performance is all good and there is good performance per £ but it is totally ruined for me when the noise is excessive and the card runs so hot I wonder about it's longevity. I often play a lot of strategy or open world games with quiet times, I don't want to sit next to a hair-dryer... AMD need to get a better reference cooler, but probably their partners can sort this easily enough.

It isn't that noisy. Look up a decibel scale to give you an idea of how loud 43 decibels is. For someone like me who lives next to a busy road it wouldn't make a difference - and I game with headphones on.

[kalniel]

It feels to me that they've just designed a chip to run at 95 degrees rather than a completely new, more efficient and quicker chip.

And yet, if you look at the reviews talking about the architecture, the new, more efficient and quicker design becomes obvious as well Look at the die size compared to nVidia's chips for example.

[Noxvayl]

I eagerly await the custom cooled versions of the 290 series. I am eyeing one as a christmas present, just got to convince the family to contribute to it. I'll have it water cooled so am interested to see what the stock settings are like with a better cooler. The DIY Arctic Cooling heatsink shows impressive overclocking results so at so I know I'll have the extra performance available when needed.

[scaryjim]

Waaaaaaay too hot and loud. And very power hungry at that! Was considering upgrading, but may wait 'til the next lot of cards. These are almost as bad as the GTX 400 series cards :/

Not again, I've already eloquently debunked the "it's another GTX480" myth in a previous thread.

Interestingly, if you read around the web you'll find some rather varied power draw figures. Tom's hardware measure *just* the graphics card using some funky equipment and they've got the 290 below the GTX 780 and the Titan in gaming power draw. And it's as fast as a GTX780 in many circumstances. Uses less power, performs as well? Doesn't sound like much of a fail to me...

[kompukare]

EDIT: two caveats: 1) Tom's don't include temperature or - more importantly - power consumption figures for the DIY custom-cooled card. It's faster and quieter, but it might be chewing through vast amounts more power to do that. 2) It's reasonable to assume the 290X will benefit equally from extreme cooling, so if you can afford the initial outlay you'll get a faster card (that is guaranteed to fit a 290X waterblock )

But computerbase.de did measure power usage for their Arctic Xtreme III mod:
http://www.computerbase.de/artikel/g...90-im-test/10/

Looks like three fans do use up more power than whatever power leakage is saved by having cooler silicon.

Well maybe... From the 7V run it looks like the jump from 70°C to 81°C comes with a cost of an extra 16W. Well that's assuming everything else is the same. Pity hardware.fr or ht4u.net didn't join in on the DIY aftermarket cooler bit as they both generally run tons of tests.

EDIT: actually reading the computerbase text, they do attribute those extra 16W to just the higher leakage at those higher temps