AMD Responds to R9 290 Series Performance Variance

Bet THG forgot to update the retail cards to the new driver... assuming retail cards were already shipping with it. (47% fan speed).

V. funny if that turns out to be the case.
 
ok then, kudos for doing that. But is there any performance decrease with stock fan setting after hours of playing ? or do the fan must be set above 50% to maintain gaming performance? also I'm a bit worry about the thermal wear of the GPU and will it affect the thermal wear of surrounding components inside the case, like PCIE slot, the MB PCB etc?

I would suggest going back and reading the review a bit closer because we do specifically address the fan settings and and temps what we experienced.

If you are worried about "thermal wear" I would suggest that the first thing you look into is airflow in your case. If you are not properly exhausting your case, you are basically leaving all the heat in it that is produced by nearly every component it contains.
 
My personal opinion is that AMD is pushing the thermal limits of this card to beat Nvidia, plain and simple. And they're putting alot of pressure on their board partners to cool and support this card once the RMAs start coming in.

Which I have no problem with pushing boundaries. I would rather have a much better performing card that I can then find my own cooling solution for. I would wait and see what the actual RMA numbers are. I don't think the issue was that the cards did not work, it was more they weren't working as well as the press card. If the performance issue can be fixed with a driver update or better cooling methods, then I don't think its that big of a deal. If the issue turns into cards dying, that is a whole separate issue.
 
I understand perfectly how it works.

Which is precisely why i have concerns how those radeons will work in the more mainstream enviroment of less than perfectly cooled cases.

It's not the first time a card was run at this temperature before. I don't see why people think the "sky is falling." We've had hot cards before and the 400/500 series was not that long ago mind you. I ran a 470 in case that's not ideal for it (Antex P180.)
 
AMD cherry picking review samples?...say it ain't so...then again I'm sure they're not the first...that being said hopefully this 'variance' patch/driver is for real otherwise we could start seeing some chinks in the armor of all the goodwill AMD has been getting recently from the 290 series
 
its a 290x, there was no change in the fan cap for that card. just the 290 pro.

hes most likely stuck in quiet mode.
 
It seems that Guru3D looked at the issue, and they confirmed it, though it exist only in "silent mode" and not in "uber" mode (and who of [H] would run it in silent, eh? :D) Worst drop was core clock to 840 mHz, and it seems that it's downclocking as fans can't go louder and card has somehow deal with the heat.

The performance drop means in 3D Mark FireStrike drops from 9808 to 9698. It seems that once the non-reference cards hit the shelves in late November, the custom cooling will deal with heat, and the problem will disappear.

Source:
http://www.guru3d.com/news_story/are_retail_samples_of_the_r9_290x_clocking_down_too_much.html
 
It seems that Guru3D looked at the issue, and they confirmed it, though it exist only in "silent mode" and not in "uber" mode (and who of [H] would run it in silent, eh? :D) Worst drop was core clock to 840 mHz, and it seems that it's downclocking as fans can't go louder and card has somehow deal with the heat.

The performance drop means in 3D Mark FireStrike drops from 9808 to 9698. It seems that once the non-reference cards hit the shelves in late November, the custom cooling will deal with heat, and the problem will disappear.

Source:
http://www.guru3d.com/news_story/are_retail_samples_of_the_r9_290x_clocking_down_too_much.html

I don't see how this confirms anything. This is exactly the behavior that is expected. Quiet mode limits the fan speed to 40% which results in the card downclocking as needed. Exactly as it was designed to do. I'm confused.
 
I don't see how this confirms anything. This is exactly the behavior that is expected. Quiet mode limits the fan speed to 40% which results in the card downclocking as needed. Exactly as it was designed to do. I'm confused.

Yeah exactly, results will vary based on conditions such as ambient temp.
 
It seems that Guru3D looked at the issue, and they confirmed it, though it exist only in "silent mode" and not in "uber" mode (and who of [H] would run it in silent, eh? :D) Worst drop was core clock to 840 mHz, and it seems that it's downclocking as fans can't go louder and card has somehow deal with the heat.

The performance drop means in 3D Mark FireStrike drops from 9808 to 9698. It seems that once the non-reference cards hit the shelves in late November, the custom cooling will deal with heat, and the problem will disappear.

Source:
http://www.guru3d.com/news_story/are_retail_samples_of_the_r9_290x_clocking_down_too_much.html
Exactly i am waiting on the none ref cards to come out before buying one.
 
Wasn't that purpose of quiet mode anyway ? it values silence over performance so what is the prob here ? if i were amd i would change that method cause card is loud anyway at least get rid of other probs and complants
 
I don't see the issue either with the quiet mode complaints. If I don't need trifire I disable a card, take the fans from 1400 rpm to 900... Basically the same thing this card does in quiet mode but automatic.
 
AMD has updated us *November 7th*

http://www.hardocp.com/news/2013/11/07/update_on_amd_responds_to_r9_290_series_performance_variance

Hello, We've identified that there's variability in fan speeds across AMD R9 290 series boards. This variability in fan speed translates into variability of the cooling capacity of the fan-sink. The flexibility of AMD PowerTune technology enables us to correct this variability in a driver update. This update will normalize the fan RPMs to the correct values.

The correct target RPM values are 2200RPM for the AMD Radeon R9 290X "Quiet mode", and 2650RPM for the R9 290. You can verify these in GPU-Z. If you're working on stories relating to R9 290 series products, please use this driver as it will reduce any variability in fan speeds. This driver will be posted publicly tonight.
 
Seems like something's wrong with the cooler. Just more evidence that this is probably a quality control issue. Sucks.

sounds more like the reference heatsink absolutely sucks but then again we've all known that for a while when it comes to AMD's reference cards. oh well hopefully the custom pcb/cooled cards come out before christmas.
 
sounds more like the reference heatsink absolutely sucks but then again we've all known that for a while when it comes to AMD's reference cards. oh well hopefully the custom pcb/cooled cards come out before christmas.

It's a real shame AMD didn't put more effort into the cooling solution. I easily would've paid at least $50 more. It was fine for the 7xxx cards but the R9 series needs something more robust.
 
It's a real shame AMD didn't put more effort into the cooling solution. I easily would've paid at least $50 more. It was fine for the 7xxx cards but the R9 series needs something more robust.

Hopefully they learn from all the backlash with the next gen. At the very least, could've not restricted the AIB's from initial release.
 
It's a real shame AMD didn't put more effort into the cooling solution. I easily would've paid at least $50 more. It was fine for the 7xxx cards but the R9 series needs something more robust.

So much this. I wanted to like the 290 series more than I did, I was kinda rooting for AMD. But that stupid cooler. Jeez. Wouldn't even be that bad if fan speed weren't so directly tied to card performance - the 7970 fan was the same but didn't require 47% fan speed for max performance. I remember the 7970 fan was definitely audible past 45%.

That said, I still think they will sell a shit load of 290s because of sheer value. For 400$ it performs amazing if you don't mind 47% fan speed - It is just too bad AMD didn't put a better cooler on the thing.
 
After everything is said and done on the cooler side with the board partners' solutions, I'm most curious about seeing Mantle play out. Considering the big 3 consoles use AMD graphics chips, the low level API access on the desktop side might prove to be a very powerful tool.
 
...we have to start thinking of clock speed differently, the old ways no longer apply...

I am picturing an old , bald, chinese guy with a long gray haired (thin) beard, in a kimono saying this... He's hanging out in some old lost castle in some forest.
 
We are now praising a company for providing uncharacteristically performing product to reviewers? So it would be ok for Subaru to provide test drivers with a WRX that performs better than the one you buy retail as long as they said "they think they know what the problem is"? What has this fanboy world come too? Nvidia would be thrashed right now...

This is backwards, probably. THG seems the exception. However, I thought I had read that THG didn't receive a review card and so bought one on the open market...? That's interesting in itself. At any rate, the issue seems to revolve around nothing except fan speeds, and all the reviews without exception picked up the fact that R9 reference-design performance is sensitive to fan speeds. Perhaps THG picked up an early-market sample with earlier drivers, or else slightly different fan controllers. At any rate, 200 points in 3dMark is an almost worse than meaningless distinction...;)

Also, there is no "cherry picking" involved. That phrase pertains to the old days when IHVs (in particular, nVidia) would send out over-clocked samples to certain review sites (in particular, THG) which would be compared to stocked-clocked competitors (like 3dfx)--but months later when the card shipped (which is what it usually took after the press demos--months), it would ship at a clock speed of 15%-25% less than the "review products," and Dr. Pabst-blue-ribbon over at THG at the time, would say nary a word about it. Now *that's* what you call cherry-pickin' and fanboyism...;) [H] doesn't qualify in such a category.

Very minor fan-speed variances on brand new gpus never-before-seen are what we call driver issues, in no way related to cherry picking. Driver issues are always expected in brand-new products. Always. Yawn.
 
Is there anyone like me who cares not for stock cooling?

I have always replaced the stock cooler for something more robust. More voltage room and overclocking......

However, does the stock nvidia coolers hold back overclocking potential? Would there be gains if the cooling was upgraded?
 
Is there anyone like me who cares not for stock cooling?

I have always replaced the stock cooler for something more robust. More voltage room and overclocking......

However, does the stock nvidia coolers hold back overclocking potential? Would there be gains if the cooling was upgraded?

I use to be interested in replacing stock heatsinks. Now I have a wife, house payment, kids, and a job and all I care about is that it works out of the box. I keep my old watercooling gear though with the thought that I'll get back into it....someday.
 
Is there anyone like me who cares not for stock cooling?

I have always replaced the stock cooler for something more robust. More voltage room and overclocking......

However, does the stock nvidia coolers hold back overclocking potential? Would there be gains if the cooling was upgraded?

Ati stock cooling has had many failures in years past.

Replacing the heatsink is going to add to the cost, something that consumers should factor in. Not to mention the possibility of voiding the warranty or damaging the card. Also looking on newegg.com, these r290x's are closing in on $600 in price, likely a supply/demand issue. Couple that with extra heatsink costs, I can't justify the hassle plus buggy driver history vs a 780ti. Though myself will wait for an AIB to make a 4gb 780ti before I take the plunge. Just give both of them a few months, prices should come down on both sides, hopefully see a 4gb 780ti, and R290x's with enhanced cooling solutions by then.

The Nvidia stock coolers have been better for years imho.

I should upload a pic of my modded Radeon 9700 pro... it's more heatsink than video card. voltage tweaked. Only got another 5% out of it. Was a fun project though, added another $100 to the cost. [H] for sure). Friends cards all burned up the fan on that shitty slim heatsink they had, one of them in as little as 3 months. (Might be thinking of the 9800 on that one.)

My opinion of this go-around, is that ATI wanted to make a splash and have the fastest card for once. Which it was for about a week. And still has good performance for the price even after the 780Ti came out. But they've had to push their silicon really hard for several generations to be competitive with nvidia. 95C attests to that. And even if the GPU itself can handle those temperatures, the surrounding components on the video card will suffer shortened lifespans. Likely see the pixel-artifacting in cards maybe a year old, when GDDR chips get overcooked. (Cards appear to all have 3 years warranty on newegg).

I am glad they have a competitive card, keeps nvidia's prices in check :)

The 3 year warranty will be put to the test.
 
So much this. I wanted to like the 290 series more than I did, I was kinda rooting for AMD. But that stupid cooler. Jeez. Wouldn't even be that bad if fan speed weren't so directly tied to card performance - the 7970 fan was the same but didn't require 47% fan speed for max performance. I remember the 7970 fan was definitely audible past 45%.

That said, I still think they will sell a shit load of 290s because of sheer value. For 400$ it performs amazing if you don't mind 47% fan speed - It is just too bad AMD didn't put a better cooler on the thing.

I see it as a really nice feature personally, but the cooler itself needs to be improved. Ideally the throttling should be there to aid overclockers or to safely operate the video card until a faulty cooler or case setup gets fixed etc.

As things stand this feature is instead leaving performance on the table that can be traded for noise, it really doesn't need to be that way so it's better to wait for third party cooling solutions to surface before buying, unless you're buying to overclock of course.
 
Yeah it would be nicer if the implemented better cooling. I guess it is the cost/benefit analysis they do. I know a lot of people ditch the stock coolers anyways.

I just need to find a way to fit an AIO water cooler to my gpu......
 
I know what you mean about bios flashing as I have done 2 evga cards in the past with the last one being the 460GTX 768Mb for a fan speed update..

The new driver is adding DX 11.2 to the HD79XX cards also in release notes.

Also I think TH did a review of the R9 290 with an after market cooler install and it really woke the thing up to 1.1Ghz clock speeds.. place my order on my 290 today!!
 
AMD should follow up with the whole issue about some able to flash to 290x.
 
AMD should follow up with the whole issue about some able to flash to 290x.

It's most likely that some of the initial 290s were actually 290Xs that were castrated at the BIOS level to meet the initial demand ($400 cards sell a lot better than $550 cards) while later chip shipments were meant to be only 290s from the start and are laser cut. Those who received the ones that could unlock got extremely lucky while those who's don't got exactly what they paid for.
 
Is there an after market cooler that won't cost an arm and a leg ? I seriously hate my 290.
My old 680 lighing was noisless compared to this jet engine
 
Is there an after market cooler that won't cost an arm and a leg ? I seriously hate my 290.
My old 680 lighing was noisless compared to this jet engine

Not worth it. Can possibly void the warranty depending on manufacturer, costs an arm and a leg (accelero's generally cost 80-110$), and they do not properly cool the VRM and memory.

You may want to RMA the card to the point of purchase and either get a different card or wait for an aftermarket 290. I don't like the 290 reference design either, FWIW - yet buying a cooler will not be cost effective; by the time you spend 400$ + 80$ on a cooler, you could have bought an overclocked GTX 780 instead. Or you could just wait an buy an aftermarket 290 in 1-2 months for 430-450$ - with that being the case, i'd highly recommend not buying a 3rd party cooler.

In fact, the only AIB partner for AMD who DOESN'T void warranty for removing the stock cooler is MSI. To me, the voided warranty + ineffective VRM cooling makes the decision a clear one. RMA + wait a month for an aftermarket 290 or RMA + get an OC 780.
 
The thing is, i live in Argentina. Bought it, its mine. So either i go through the hassle of selling or put an aftermarket cooler. Gonna wait for one that has official support.

Card is at 93c @70% making me go crazy
 
Not worth it. Can possibly void the warranty depending on manufacturer, costs an arm and a leg (accelero's generally cost 80-110$), and they do not properly cool the VRM and memory.

You may want to RMA the card to the point of purchase and either get a different card or wait for an aftermarket 290. I don't like the 290 reference design either, FWIW - yet buying a cooler will not be cost effective; by the time you spend 400$ + 80$ on a cooler, you could have bought an overclocked GTX 780 instead. Or you could just wait an buy an aftermarket 290 in 1-2 months for 430-450$ - with that being the case, i'd highly recommend not buying a 3rd party cooler.

In fact, the only AIB partner for AMD who DOESN'T void warranty for removing the stock cooler is MSI. To me, the voided warranty + ineffective VRM cooling makes the decision a clear one. RMA + wait a month for an aftermarket 290 or RMA + get an OC 780.

Funny, according to the people who own the card and don't just bash it every chance they get like you, say it is worth it http://hardforum.com/showthread.php?t=1788658
 
Seems like review and retail cards have near identical performance in "uber" mode.

This is an important point to the folks at AMD, because it means the card-to-card performance variance we've cataloged can be eliminated quite literally at the flip of a switch. Owners of retail R9 290X cards will have to be willing to put up a non-default configuration that produces substantially more noise, but their cards should should be able to obtain the same performance that the 290X review sample achieved in "uber" mode in our review.

Source:
http://techreport.com/blog/25751/an-update-on-radeon-r9-290x-variance
 
Back
Top