*** Official AMD 7990 Owners Thread ***

[Gerard]

Don't forget that test reviews they run the system on an open chassis. this negates any bias from a particular case on the GPU.

Adding A PCI fan won't help remove the warm air from the case it will just circulate it within the case faster. I added an extra fan drawing fresh air from outside the and case mounted it on the window at the same height as the fans drawing the air into the card. The 7990 blows the air into the case and only exhausts some out of the rear of the case.

If the air coming out of the exhaust case fans is warm or very warm then there is insufficint flow through the case. That FT02 isn't the biggest case going but has plenty of fans, just make sure you have the correct setup where there are enough exhausting as there is supplying the case, especially if they run at different speeds.

As sgavan says, undervolting - this is an excellent way of lowering temps. I couldnt undervolt much before it became unstable.

Do you have your mobo controlling your fans or run them stock off the PSU?

They're controlled by a switch on top of the pc to set them to low or high, at low they're barely noticeable at high they're pretty audiable.

Undervolt your card when you do to keep temps down. 1.125v-1.150v should be plenty for clocks of 1000/1500 on a 7990.

Well I've actually put in place a temporary fix for it, been trawling about the place and seen several people with the same issue, that being that gpu1 (closest to the pci back plate) was overheating. Seen quite a bit of discussion saying that it was basically due to a combination of a stupid heatsink design by amd and its further made worse by cases like mine which have a funny layout. Essentially the front 2 fans seem to create a dead zone where gpu 1 is almost starved of cooler air with the front fan getting most of it and the secondary fan getting much less which leaves the third fan with near bugger all.

To remedy this I've placed a fan at the back of my case intaking air onto the back gpu which as of now seems to have taken down temps approx. 13-14c.

That's running crysis 3 at Eyefinity 5760x1080 at stock volts on the gpu, considering before even 5 or so minutes of that would rocket the temps to over 100c its a hell of an improvement

 

[LtMatt]

Good work, but regardless of whatever you do, you really need my bios and to undervolt. Save yourself a lot of hassle and get better (no throttling) performance. Guess what? Your screenshot tells me that gpu1 is throttling. Not because of heat, but because of the ridiculously low TDP of the stock bios.

 

[Gerard]

Will do over the weekend, might be finally installing windows 8.1 so will do it then.

 

[JamesM]

Well looks like i'm leaving the club.

Switched my 290s to my main system and will be selling my 7990s

Still can't find the bleedin reference coolers so will have to sell them with water blocks, doh

 

[Gerard]

Well looks like i'm leaving the club.

Switched my 290s to my main system and will be selling my 7990s

Still can't find the bleedin reference coolers so will have to sell them with water blocks, doh

You'll have to unsubscribe from our newsletter, hand in your badge and disavow any knowledge of the secret handshake. Otherwise the boys will be paying you a visit

 

[Gerard]

Well after much faffing about i'm sending the card back to get looked at, tried different bios versions, tried a different case and tried undervolting the card and still it climbs into the 90's with the fans howling like banshees. At one point the card would routinely get into the 100c range and start to throttle back on the clocks, some screens of that here and a sensor log:

http://homepage.ntlworld.com/gerald.marley/overheating/


Needless to say pretty irritated, ever moreso due to the fact that just about every review seemed to get some kind of magic card that seemed to run almost 30c lower with quiet fans, after looking around online a quiet, cool 7990 seems to be the exception rather than the rule. I can only assume that amd started wanging every 7970 core they had onto these boards to get rid of what would then be considered "old stock" just to get rid of them instead of picking cores that ran at lower temperatures and voltages. I can accept some variances in reviews to retail but going from cards that ran at 75c to almost 105c is just extracting the urine to put it mildly and is totally unacceptable nomatter what way you look at it.

Frustrating but I've tried everything I can and ultimately I'm po'd with having to fart ass about with it.

 

[_lecro_]

Thats a real shame m8, ive just about got mine working as it should now after lots of messing with drivers and bios files and water block fitting.
It used to go above 95 -105c when i first got it but now it maxes out at 50c on one core with the other not going over 42c .

 

[Woodsta888]

Feel for ya Gerard,

I have a fan I modded into the window of my case blowing air to the fans on top to draw through heatsinks. I saw mine getting into the low 90's height of summer when ambient room temps are higher. But never over 100.

I do have a Temjin TJ10 case which is quite big with 5 120mm fans changing the air. 3 exhausts and 2 intakes but these are scythe gentle typhoons.

What happens to your temps if you leave the side panel off the case?

 

[Gerard]

Thats a real shame m8, ive just about got mine working as it should now after lots of messing with drivers and bios files and water block fitting.
It used to go above 95 -105c when i first got it but now it maxes out at 50c on one core with the other not going over 42c .

Its a joke how they are out of the box, the temps skyrocket yet practically every review say they run cool and quiet. Like I said it just seems amd got to a stage where they just started using all the 7970 chip inventory on these regardless of temps instead of using the lower vcore variations. Seen loads of posts with people having mad temperature issues, I've yet to see any user who can claim he's getting the magical 75c load temps that some reviews claimed was possible. Maybe in the middle of the arctic circle.

Feel for ya Gerard,

I have a fan I modded into the window of my case blowing air to the fans on top to draw through heatsinks. I saw mine getting into the low 90's height of summer when ambient room temps are higher. But never over 100.

I do have a Temjin TJ10 case which is quite big with 5 120mm fans changing the air. 3 exhausts and 2 intakes but these are scythe gentle typhoons.

What happens to your temps if you leave the side panel off the case?

Temps go down a bit but still end up with a very loud set of fans droning away so its just as irritating. Was playing bf4 earlier and within about 10 minutes it was climbing to 98c or thereabouts (and that was undervolted to 1125). I run eyefinity so obviously that puts more pressure on the gpu's but it was tested in reviews at eyefinity and 4k res and they still claimed all these low temps with barely any fan noise.

Cherry picked review samples doesn't begin to say it, from what I've seen on the web comparing the review temps and fans sound to the retail boards is like a night and day difference. 75c for some reviews, 103c for some end users, they generally seem to hover around the 90's but that's still around a 20c or thereabouts bump over what the review samples got along with an assload more fan noise.

 

[Gerard]

Well, cards going back tomorrow, as far as I know it goes to ocuk first for testing then to asus so they can assess it, god help me. The amount of threads I've seen with asus literally screwing people over claiming "physical damage caused by customer" is worrying and having been on the receiving end of it myself in the past it doesn't inspire me with confidence in the outcome.

 

[BuzzLloydyear]

Seems an age since I last trawled through here. I had the exact same problem, and it's still a problem to an extent, but I had a temporary fix, which is now going to stay until I get new cards.

I use a stick to prop up the end of the card the level it out. It seems the card kinks and heatsink for gpu 1 loses contact with the gpu. So using the stick levels it out. It's still about 15 degree's hotter than gpu 2, but it's been well within usable range (no longer going up to that dreaded 102 degree's). It's unsightly to have the stick there, but it does a job and will continue to do the job, until a card comes out that forces me to replace it.

When I opened up my gpu some of the fins had been bent in the middle of the card, I assume that's from the flexing of the card, it's quite a substantial flex and doesn't seems to effect every card equally, but the stick trick put the temperature difference between gpu 1 & 2 from 30+ degree's at worst case scenario, to 15-20, it also made it, so that I no longer get into that silly throttled mode when you hit 102 degree's. I guess time will tell if summer temperatures put pay to that though.

 

[JamesM]

tbh my 7990s never got that hot under the stock fans.

They used to hit around 70-80 undervolted and on an open bench with a space between them. I seem to recall using one in a case and the temps never reaching 90 degrees while mining.

That said under water they are fantastic, two cards in a loop along with CPU and while mining they run at 42 degrees on each core.

On a side note, I decided to keep them and run them in my main system. Quadfire works ok for BF4 with Mantle and even works fairly well with War Thunder so i'm happy.

Will just sue them to mine with when i'm not gaming as before

 

[Gerard]

Heres a clip showing what it does

With all the horror stories I've heard about asus im not exactly keen to send it back, seems they seem to love claiming "cid" (customer induced damage) on products to invalidate the warranty even if you can find nothing wrong with it.

 

[JamesM]

100 degrees is mental.

Mine never reached that while gaming. I presume yours is at stock speeds too?

I presume the bug still exists that running more than one monitor means the GPU never down clocks?

Whats case cooling like?

 

[Gerard]

100 degrees is mental.

Mine never reached that while gaming. I presume yours is at stock speeds too?

I presume the bug still exists that running more than one monitor means the GPU never down clocks?

Whats case cooling like?

Coolings good I've tried it in my old fortress ft02, now its in a 750d and no change. Stock settings stock volts, even downclocking to 850 and running lower volts it still reaches 90 odds.

In hindsight i shoulda just rma'd the card instead of redoing the thermal paste though it made no difference to temps anyway. Ran just as hot before and after. Just remains to be seen if Asus still honour the warranty.

 

[JamesM]

Have you tried it with just one monitor?

 

[Gerard]

Have you tried it with just one monitor?

Runs a bit cooler but if its unusable for eyefinity then its really of no use. Im pretty sure even on my 6990 one gpu clocked a bit higher if it had 3 screens attached.

 

[JamesM]

It is strange, especially as you've already replaced the heat sink and fans once.

Have you toyed with the idea of putting it under water if you really want to keep it?

Only suggested the one monitor thing as providing the card was 'ok' you could simply try and sell it rather than go through the hassle of returning it and go down the two 290 route instead?

 

[Gerard]

It is strange, especially as you've already replaced the heat sink and fans once.

Have you toyed with the idea of putting it under water if you really want to keep it?

Only suggested the one monitor thing as providing the card was 'ok' you could simply try and sell it rather than go through the hassle of returning it and go down the two 290 route instead?

Watercooling I was never a fan of, its in the post as of about 45 mins ago so just hoping for a good outcome. Though from some of the stories I read last night I'm not holding my breath The card still works fine its just it runs far too hot which leads me to think that at some point amd stopped binning lower vcore capable Tahiti chips and just slung any remaining inventory onto the boards hoping that the 3 fans could cope with it.

The discrepancy of temperatures from review samples to retail cards is totally laughable, its like a totally different product with the reviewers getting the "cool and whisper quiet 75c load at 4k cards" and pot luck to anyone else. Very disappointing and its something I'll bear in mind in future and wait for people to get retail cards and hear back reports on them first, the reviews of this card in terms of temps and noise were totally contradictory to the retail boards.

 

[Gerard]

Seems an age since I last trawled through here. I had the exact same problem, and it's still a problem to an extent, but I had a temporary fix, which is now going to stay until I get new cards.

I use a stick to prop up the end of the card the level it out. It seems the card kinks and heatsink for gpu 1 loses contact with the gpu. So using the stick levels it out. It's still about 15 degree's hotter than gpu 2, but it's been well within usable range (no longer going up to that dreaded 102 degree's). It's unsightly to have the stick there, but it does a job and will continue to do the job, until a card comes out that forces me to replace it.

When I opened up my gpu some of the fins had been bent in the middle of the card, I assume that's from the flexing of the card, it's quite a substantial flex and doesn't seems to effect every card equally, but the stick trick put the temperature difference between gpu 1 & 2 from 30+ degree's at worst case scenario, to 15-20, it also made it, so that I no longer get into that silly throttled mode when you hit 102 degree's. I guess time will tell if summer temperatures put pay to that though.

I noticed that some of the copper fins on the vrm heatsinks were slightly closer together as well, on my fortress case though the card sat vertically with no sag and temps remained the same. Knowing asus they'll see some of the fins closer together and class it as customer damage and void the warranty.

I'm really not expecting them to honour the warranty given by what I've read and experienced in the past, it just seems they go out of their way to try and find some reason to screw the customer. And if you do get a replacement its likely to come battered, though if you send in a card like that they refuse it right away, so the question is how do they get these battered looking replacements in the first place?

Wish id opted for the gigabyte card now looking back, probably less hassle dealing with them and a closer to hand rma centre.