1 of 1 people found this helpful
No, the x2 cards are GPU0 being the GPU like any other, GPU1 being a reduced GPU doing nothing more than co existing. The same goes for 2 separate GPUs in CrossFire, the slave GPU is effectively just co existing and cant be used to display.
Sadly there isnt much in the designing that could be done either, having a whole PCIe bus trial leading to the 2nd GPU would be a PCB nightmare.
I'm sure you know but both GPUs heat pipes meet in the middle above the PCIe3.0 bus. So have 1.5 fans each, if you can some how push more air into that direction. I'm not sure why the heat pipes didn't just run the full length to each heat sink.
Water cooling may be an option, but being an old card I'm having trouble finding water blocks for it. Yet have a mill so if I can find the design of one.
Depending on your mobo, you may be able to move it to a lower x16 PCIe bus.
Last but not least, fashion some sort of air funnel from the bracket at the back leading to that fan. Likely looking rubbish.
I hear you though, that fan when in single GPU mode is noisy. But I found the heat would maybe hit 100oc, now with a bigger case I'm only hitting 90oc but still noisy. The thermal tolerance for these cards are very high, being the same specs as a laptop duel card setup. Smaller, better quality components to their single GPU counterparts.
Nice GPU btw
Very Very Great-and Thankful for this professional Answer.
And a very big Thx for the compliment about this Card. It's really a nice beast
It's also not a HD 7990 ; it is a HD 8990 OEM ( Rebranded 7990 )
I Forget to write my specs:
Alienware Aurora R4
Intel i7 4930k 6 Core @ 3,4 ( Factory OC to 4,1 Ghz )
32 GB Ram
Radeon HD 8990 OEM 6GB Dual GPU
256 GB SSD
1 TB HD
Recon 3D Soundcard
Wanna also Show 2 Pictures and have Questions about it:
Like I said: GPU 1 far away from Fans and allways has got higher temps.
Sometimes it is also overheating like for example in Far Cry 3, or GTA 5 or since I have a 144Hz Monitor, also in Diablo 3 with or without Vsync.
But that are only examples. Not in every Game it overheats, but GPU 1 always has higher Temps.
Under the GPU 1 is my Soundcard, but also withou it, it doesn't care Temps.
But Backplate is very very Hot ( always and more at GPU 1! )
Any solution here ?
Picture Number two is a simple Crimson Driver Question, but I don't understand this numbers:
Why is written in my Crimson Drivers: AMD Radeon HD 8990 #4 Primary
AMD Radeon HD 8990 #5 Linked
What are These Numbers ? Should where be written #0 and #1 ; or #1 and #2 ???
I'm very Thankful about every Answer and for especially a good way Solution for eventually cooling GPU 1.
Also found Watercooling System, but for this Price and this Stuff specially only for this Card, I would save up the Money and better would spend it on a new better Card.
Very much Thanks in Advance
1 of 1 people found this helpful
Its all good man. I only have the 7990, waiting for the Fury x2
Excuse the quality, my LEDs mess my camera up. Again I found the bigger case brought down the temps a good 10oc. Now they are both near par to each other...
But, you then have a computer that looks more like a fridge. And you need to explain it to people when they say "what is that". For airflow it solved every problem known to man. If you have the option, when I had a closed loop CPU cooler, I found putting the rad at the top helped a lot too, it seemed to restrict air from leaving the case slightly. And heated up the water. Now I'm external.
The GPU backplate. The 7990 is the only GPU I have ever seen to have its own "warning, hot surface" sticker. Mobo side. It its hot enough to cook on I'm sure, but it has a sticker, so I'm sure its okay.
It looks snug in there, you could have a small 80mm fan sitting at the far left below the GPU exhaust. I use one of these on my VRM and might be ideal https://www.arctic.ac/uk_en/f8-pro-pwm-pst.html#?bild0 but it looks like the case hinge needs the room though? You could also go old school case modding with good old duct tape and some plastic. Not ideal, maybe even frowned upon in the forum
It will take almost German measurements and neatness to work, and you will be pulling in dusty air from the back. But it will be cooler. If I remember doing it back in the Athlon days, extreme case mod cooling. It worked, looked rubbish, but wasn't worth the time for a CPU. GPU might make sense.
As for the GPU number thing...
Its simply where on the PCI bus the device is located. I think if you have 4 cards say, and 1 is not working correctly. You can pinpoint what GPU it is. Windows doing its usual NT thing.
Yeah them water blocks are outrageously expensive. I cant find second hand either.
I hope this is of help, not just an info overload of mad ideas and mutual love for a GPU. Keep firing questions and discussions if you wish. And keep us posted on your plan for cooling, intriguing.
Very competent Answer and many many Help and Info, which I never got anywhere else. Very Greatful Tomtalk24.
Guess I'm not the only one with this Problem.
The Overheating of GPU 0 only happens in Gaming very demanding GPU Games like Far Cry 3 or GTA 5 ( Teststanding a half hour in "Ultra" Grass, for excample ). And only if my GPU runs constantly at 99% usage.
First of all I wanted to find a cooling solution for GPU 0 without spending Money.
For Me: The best Solution was to crank up the PCI Fan to hold GPU0 between 80-90° C.
Since it runs at 100%, the Noise is a little bit irritating, but maybe I think to buy a quieter PCI Fan the next days. Maybe!
Your recommended Arctic Fan will be eventually my next Step to cool the System. Thanks again for the tipp by the way.
I also found out, that the Overheating only, or maybe faster happens, because I have set the PowerLimit to +20.
Somewhere here I read that it is recommended to set it to +20 to run at full clocks constantly. Games are running really better with +20 than with set to 0.
I've found out, if it's set to Powerlimit 0, the Cores aren't clocking at 99% and sometimes switching from 1000Mhz to 950Mhz and back.
Also GPU Graph in MSI Afterburner goes up and down, up and down. With +20 it stays allways up, Result: Games running better and stabler.
With Powerlimit +20 the clocks are constant 1000Mhz and are only dipping down, when Card is getting too hot ( First at 102°C throttle down to 950 Mhz constantly, Temps sinks back to 95°C and when if it rises 102°C again, it throttles down to 500Mhz and Gaming isn't fun anymore with FPS dips, stutter etc..)
So I try to hold the Temps.
Also another little idea and Question, would it bring something against High Temps, if I would plug in the Card in the lower PCI Slot?
Or is it better to stay the card at the higher
Another little thing next to the Main Topic:
I also read here in this Forum to disable ULPS in Registry ( if it's really necessary). But I don't find any Difference when on or off. Can anyone eventually also explain this to me ? I know what ULPS is, but I can't fell any Difference in Disabling it. GPU 2 ( or exactly called GPU 1 ) is allways at 0% when it is not in Use. Can't see a Difference between ULPS on or off.
Thanks in Advance.
Cooling this card and especially GPU 0 is for now and the next Days my main Fever.
Also if anyone has experience with Overdrive Settings and this card, I would be very thankful, what I should change?
Tomtalk24, you have for example 1025Mhz and 1575 Memory Clock, why or is the Card running better with this options ?
2 of 2 people found this helpful
I think in this situation its a balancing act, and a little physics.
You may or may not be familiar with the older Catalyst Control Centre. It was a little more visual with what effects settings had on the card. I think the graph idea was more to benefit Laptops. Higher clocks and more power will create more heat and use more power, lower the opposite.
+20% is merely the current the GPU can use, allowing more air into the wood burner causing a more vigorous burn. -20% would be closing the air vent, calm flame. Poor analogy but a nice way to look at it.
You could try lowering it by 10%, see how you get on. Just trying to find a sweet spot of heat generation and its ability to remove that heat in such a confined space. While not letting it over heat and throttle.
I think the higher slot is best, but no harm in giving it a go.
Disabling ULPS I think only benefits those with 2 separate GPUs. I have never had to disable it myself, even with 2 separate GPUs. My OCD cant cope knowing I'm using energy for no reason on idle. kingfish has some info about it here.
As for my settings, the size of my case and possible over the top airflow means I can run at higher clock without the card needing to throttle. I also play in 3D and even the slightest bit more speed from this GPU can add to the frame rate.
I think what may be best, until you improve your air circulation. Is reducing the clocks by about 50MHz, power control maybe 10%, tweaking those settings until you have the GPU working 100% and comfortable enough not to throttle. Trying to find that sweet spot between heat generation, heat removal and the performance you desire. On a external test bench that GPU can work all day at 100%, but in a small space it becomes like an oven. I haven't had to turn my room radiator on the past couple of years in the winter due to this PC
Another thing that may help, or should is adding a push fan to your rad. Better air flow. More of that hot GPU air out of your system. I have temp probes all around my system and with 25oc air coming in to the case, I get 45oc out at 100% GPU load. And thinking about air like water cooling. The grater the volume of air passing through the case, the more heat removed. Like an F1 car doing 2 laps in the same time a normal car would, twice as fast is twice that volume of air so twice the heat removed in the same time. (might need re wording if that don't make sense)
Read it already Mr. Kingfish ,
you are my Inspiration why I made this whole Forum Thread. Now it's out and I'm also greatful to you
It helped me a lot to know better new/old Things about my Card and about the Options.
@tomtalk24 also. Each Sentence from you helped me so much. You're are writing and sharing your Knowledge like reading my Soul for the last 2 Years.
By the way, you're still using Catalyst. I have the latest Version Crimson and because one issue is still not fixed, i let the Memory Clock @ Default:
When Changing Performance Memory Clock with my Card in Crimson, the whole Option will be deleted immediately. Memory Clocks stay at higher 1575Mhz and can't be changed back to normal except to de-and new install Crimson.
This bug is still not solved since first Crimson Driver came out.
Also send this issue to AMD but it's still not fixed. Hope for next Driver.
About the Fans:
Thanks first. I will give you Feedback. Maybe it also helps other HD 7990 or 8990 usern, because I'm not the only one with an Alienware Aurora R4 or an user with GPU 0 overheating Problems on this card.
Keep cool! with the cheapest best way is the next Mission
I literally just done a system restore as Crimson did the same to me with the Vulken Beta Crimson driver. If it has only started happening to you it might be worth a try. I'm back to idle GPU when idle not that 500mhz and full vRAM speed. Big bug.
Keep us posted with your progress.
1 of 1 people found this helpful
Hello Guys and HappY Easter
It past Time, but I finally found my cheapest solution to fix my Overheating Problem of GPU 1 in very demanding Games.
First of all I read in many Forums through People, that also wanted tips for their same ATX Case.
I also read a lot that a GPU like that it is pushing air in for freshing the gpus and putting it hot out on the "Side"!! Also on the back like it have to, but very more hot air on side (Handtest was enough).
The Case has a Black Shroud that covers the Card on the side. I thought why? The Shroud only helped against Overheating if I push up the PCI System Fan on the right Side for better airflow, but the System was very loud.
Next thought: If the Card is pushing fresh air in, so the soundcard under my GPU could take the place away for getting more air. Also I read anywhere else, that the Card is also sucking the hot air away from the other parts, so my guess with the Soundcard shouldn't be wrong.
I removed the soundcard, and installed my 5.1 Speakers on the Mainboard and let the Slots on the back opened.
Finally GPU 1 didn't overheat anymore. It never reached the temps to throttle down and also couldn't hit them again for throttle more down.
Now everything is fine, the only one sacrivice was my soundcard and it really had better Sound than the Onboard Card.
But I also have a solution for this for the next days: Just buying a pci to pci cable and install my soundcard anywhere on the back outside the Case.
Here's a Pic, to somehow seeing my solution :3
i'm french so sorry if my english is not very well.
I got a sapphire 7990 with reference design the car work as well in charge but not in idle mode
Problem one : the second gpu no entry in off mode on desk apps like chrome, word, excel ... (i haven't got the green led on the back of the card) but during the boot of windows this led is turned on and when windows is fully loaded the led turn off
Problem two : the gpu as 50/60 in idle mode and the slave is at 44° is same state 300Mhz @ 0.85v i've changed the thermal past and fully clean the radiator, no more effet.
Problem tree : when i look the card with atiflash i saw the gpu 0 is slave and the gpu 1 is master. (perhaps an issu of problem one ?) for this problem i try to flash master bios on gpu 0 and slave bios on gpu 1 but i got black screen after. Since i've re-flash the good bios to the good gpu and it work like before.
Please anyone can help me ?
1, ZeroCore wont get fixed. I have been waiting for years, reported many times. Seems to be an issue with .NET etc. Try disabling all your programs that auto start with Windows and enable one by one until you find the problem. Also report your issue to AMD.
2, Those temperatures look like your watching a video, check your GPU usage at idle on the desktop. Use lots of thermal paste on the GPU die and clean the heat sink (I'm sure you have). What driver are you using?
3, You should only need to flash this card to extend the overclocking range, normally on water. In ATIFlash it should have on the firmware what GPU it is for with a S or M at the end of the code. They shouldn't be mixed up. But again unless your on water, the default BIOS limitations are pushing that GPU anyway. You can always try a few combinations of master slave bios' and keep note of what goes where.
thank's to your answer, i do a text point by point for you:
1- I've desactivate all start program and i found the program who enable two gpu, it's logioptions.exe (UNICODE)
i trie to create a profile with crossfire disable for this exe (and ther under program) but it's still enable two gpu how can i say to crimson "for this program don't enable two gpu"
2- i change 2 times the thermal paste and it's a litlle better (50° in net navigation with box opened) the gpu usage in navigation is beggining at 0% and sometimes do a run to email@example.com and down to 0% @ 0.85v
3-i've allready flash the car on position "1" to the switch on the pcb with intervert the bios slave to gpu 2 et bios master to gpu one, since here the card won't boot. So i've switch ti bios "2" and boot to dos atiflash and switch anymore to "1" position to reflash the correct bios on gpus (mine card have gpu1 slave and gpu2 master and i don't knoxw if it's normal)
Now i don't know how to resolve my Zero Core Power with logioption.exe i take a screenshot to you, in the three profile in crimson with "logioptions" "logioptionsMsgr" "updater" program a choose "disable crossfire"
Thank's to your help
1: Sounds about right mate. I have a G700s mouse, and the Logitech Gaming Software triggers both GPUs to turn on. I have also tried creating profiles for these programs to not enable CrossFire. Again until AMD sort this in their driver, anything .net related (iCloud, SoundBlaster, many others) will trigger both GPUs to run when not needed. Stupid right? I now use my mouse with settings saved on my mouse and have disabled Logitech's software.
Fresh boot into Windows...
Load up my sound card software....
Nothing will resolve this, and its stupid as these are normal day to day programs with no need for more than 1 GPU. Its just the driver wont stick to 1 GPU. So...
1.1: If you want to save energy, heat and money you will need to disable any programs that trigger CrossFire on idle. There is no way to have these run on GPU1 only, they will always wake GPU2 until AMD sorts it out.
2: 40/45 idle on water for me on water. So yours look better now. Clock speeds will hit max occasionally, that's fine.
3: Yep, only flash BIOS 2. BIOS 1 is factory. You can always switch to BIOS 1 and save them to desktop if you need to flash back. Have a read about it here...
7990 Custom Bios - Remove Throttling - Overclockers UK Forums you may recognise a few members on the thread. Do's and Dont's are nicely pointed out, along with the limitations of these GPUs and what to expect.
I think this covers it, if I have miss understood or not made sense myself let me know
Ok thank's a lot to your help. I know now than it's impossible to have zerocore power with logioptions of logitech actived ans i think like you : it's very idiot
With my mx master i can't save program in my mouse, when i disable logioptions, all buttons come back to the default function, how do you save your parameters into your mouse ?
i've allready seen the thread on uk forums and i take my bios their to flash my 7990, i've try to push 1.2v on both core and up to 1125Mhz GPU + 1650Mhz Memory and it work with 3dmark : 4050 graphic points in firestrike ultra