Overclock.net › Forums › Graphics Cards › AMD/ATI › XFX 7970Ghz Edition Double D throttling problem. Faulty card or PSU related?
New Posts  All Forums:Forum Nav:

XFX 7970Ghz Edition Double D throttling problem. Faulty card or PSU related? - Page 4

post #31 of 53
Thread Starter 
Quote:
Originally Posted by Taigatrommel View Post

I am not sure if it is completely off-topic. I am pretty sure there is a temperature problem going on here with the OPs case too. Two days ago my clockspeeds also were unconsistent after a longer Tomb Raider session. I checked GPU-Z (haven't logged though) and there have been some very small, yet noticable negative spikes in both GPU clock and voltage. The game also ran obviously choppier compared to when I fresh started playing. On the opposite I have a ~23 minutes log of Far Cry 3 (Crossfire enabled) yesterday which only shows two clock drops to 1000 MHz for a single second each as well as a drop to 501 MHz for a single second. That drop to 501 MHz also dropped the voltage to 2d numbers of 0.950v.
I think it would be interesting to find out wether there is (still) a problem with all of XFXs 7970 DD cards or if there is a specific fault with the setup -red- uses.

As for the cooling modifications: These would hardly make any use. As already said the primary cooling block of the DD version is pretty sealed of, no air can reach the PCB surface nor its components. So heat gets stuck in there, I am not sure what kind of use RAM heatsinks would have in such an environment. There still is a question about the VRM cooling situation, which I learned to be rather questionable. The VRM also seem to have some sort of heatsink (maybe integrated in that cooling plate on the PCB?), however what use is this passive heatsink if hardly any airflow touches it to remove the heat?


As for the answer XFX gave you, I call it BS. I have never heard about a 7970 or any other card, even GeForces, clock down during a game when GPU usage is low. Taking a look at the earlier mentioned FC3 log, I have situation where GPU usage sits well below 50% and none of the cards throttle down. Why should they? It doesn't make any sense. The only reason why this card ever should throttle down would be either 2d mode to save power or during games in 3d mode when a temperature problem appears to keep the card save. But not to save energy while playing...

The shift to the 2d clocks sounds indeed weird, the core dropping at 1000mhz is not a problem though. All gigahertz edition are clocked at 1000mhz, the 1050mhz clock they reach (and they are advertised at) is including the boost. Dropping to 1000 is nothing to worry about and actually, within specification. However if you are experiencing clock drops similar to those in the videos I posted, that's definitely not a boost feature, there's something else going on.

I do agree with you on the XFX answer, to me it sounded like the most nice answer to try to avoid a RMA. Everytime the throttling happens, I take a 5-10 fps hit, which is not very annoying when you are getting 100fps, but when you are playing at 40fps it's very annoying. That and the fact that it stutters everytime a clock shift occurs. I am asking around to other 7970 owners, just to make sure how normal this "intended behavior" is.
post #32 of 53
I finally got all the parts today, so installed an Arctic Accelero Hybrid cooler....

It's now silent, ran Furmark for over an hour, load temps are around 55-60c in a fairly warm room already, however, no thermal throttling occurs!!

The cooler is a heap of garbage, MOSFETS overheat because of the poor design.
post #33 of 53
Quote:
Originally Posted by BiGK View Post

I finally got all the parts today, so installed an Arctic Accelero Hybrid cooler....

It's now silent, ran Furmark for over an hour, load temps are around 55-60c in a fairly warm room already, however, no thermal throttling occurs!!

The cooler is a heap of garbage, MOSFETS overheat because of the poor design.
Hey, word of warning withe the Accelero coolers. They do an amazing job at cooling the core but they are mediocre at cooling the VRMs. Look at your VRM temps through GPU-z or HWinfo and make sure they don't go above 90c maximum.

Furmark is also a very dangerous program to use as it stresses video cards well above anything any game will ever realistically demand and puts VG components at risk of failure. I recommend you use Heaven 4.0 or sleeping dogs with extreme AA to test your cards thermals
Bender
(18 items)
 
  
CPUMotherboardGraphicsGraphics
Core i5-2500k @ 4.6Ghz Gigabyte Z68X-UD3H-B3 Sapphire AMD R9 290X Tri-X (1050/1300mhz)  EVGA GeForce GTX 1080 (How do you OC?) 
RAMHard DriveHard DriveOS
8GB 1600Mhz OCZ Vertex 3 240GB SSD OCZ Vertex 3 240GB SSD windows 10 64bit 
MonitorKeyboardPowerCase
Asus MG279Q Logitech G510  Corsair TX750 CM II 690 Advanced 
MouseMouse PadAudioAudio
CM Sentinel Advance  some big corsair one beyerdynamic DT 770 Pro 250 Ohm SoundBlaster Z Soundcard 
  hide details  
Reply
Bender
(18 items)
 
  
CPUMotherboardGraphicsGraphics
Core i5-2500k @ 4.6Ghz Gigabyte Z68X-UD3H-B3 Sapphire AMD R9 290X Tri-X (1050/1300mhz)  EVGA GeForce GTX 1080 (How do you OC?) 
RAMHard DriveHard DriveOS
8GB 1600Mhz OCZ Vertex 3 240GB SSD OCZ Vertex 3 240GB SSD windows 10 64bit 
MonitorKeyboardPowerCase
Asus MG279Q Logitech G510  Corsair TX750 CM II 690 Advanced 
MouseMouse PadAudioAudio
CM Sentinel Advance  some big corsair one beyerdynamic DT 770 Pro 250 Ohm SoundBlaster Z Soundcard 
  hide details  
Reply
post #34 of 53
Quote:
Originally Posted by th3illusiveman View Post

Hey, word of warning withe the Accelero coolers. They do an amazing job at cooling the core but they are mediocre at cooling the VRMs. Look at your VRM temps through GPU-z or HWinfo and make sure they don't go above 90c maximum.

Furmark is also a very dangerous program to use as it stresses video cards well above anything any game will ever realistically demand and puts VG components at risk of failure. I recommend you use Heaven 4.0 or sleeping dogs with extreme AA to test your cards thermals

Would love to monitor VRM temps... but can't! Not displaying in HWInfo, GPU-Z, HWMonitor, even AMD Overdrive. Nothing displays VRM temps!

I've done the next best thing.... place finger on VRM heatsink. No burnt fingers. Can't be that hot! lol

By the way, I am using the Accelero Hybrid, which is that watercooled one smile.gif

I always use Furmark because if it was going to go "pop", I would rather it do it now than later on smile.gif
post #35 of 53
Oh and anybody curious as to what the difference between the revisions is, it isn't the coolers, it's the PCB. One of the VRM's have been taken out/moved, the one to the top left of the GPU die,, the one that's normally on it's own!
post #36 of 53
I have 2 of the XFX DD 7970s with the reference PCB. They used to overheat so much I ended up watercooling both cards. Now they're great though I run both @ 1200MHz 24/7 for OpenCL tasks, never cracks 50C. Slight artifacting but I'm sure that's because of the overclock, plus my screen doesn't need to look pretty to do math biggrin.gif

Moral of the story, great PCB and solid power delivery, crappy cooling.
post #37 of 53
Yeah I should add, that I never get above 60c on LOW fan RPM (pull and push config) on a crappy Asetek 120mm radiator, set up so that already exhausts warm air. I seriously think card manufacturers should make cards with a standalone watercooling system pre-installed.

The reason being, CPU's have an average TDP of 90w but there is plenty of space for a MASSIVE heatsink. (Currently using a Scythe Ninja....fanless and never goes above 50c.

GPU's on the other hand though have around 200w TDP and can only have a relatively small heatsink.

Logic would dictate that graphics cards need watercooling rather than CPU's....doh.gif Wish Corsair would release a half decent set up, the Arctic Hybrid is a bit of a nightmare to install... even though the results are impressive!!
post #38 of 53
Well in XFX's defense if you read through the paperwork:

My fan is making noise or my card is hot. Can I fix it?
The normal temperature range of a video card is up to 95C underload. We
expect that in most cases a card will not break the high 80s but large cards
under high load can get as hot as 95C.

This temperature will not damage your video card and is considered normal.
If you need to check the temperature of your video card then we
recommend using GPU‐Z from http://www.techpowerup.com/gpuz/
This software shows the GPU temperature of your video card in the Sensors
tab. Using the Refresh while in the background or Log to file options will
allow you to go back to check the temperature when you cannot see the
current readings.

If your fan is making noise or your card is hot then it usually just needs
maintenance. Most coolers are designed to easily clean dust and lent from
the fan and heatsink using compressed air. Blowing compressed airin
through the fan and vents is usually sufficient to clean the cooler and brings
the temperature of your card down or resolve any debris causing problems
with the fan’s motor.

In more extreme cases or when a card is many years old the thermal paste
between the heatsink and GPU may stop being as effective as when the card
was new. Youmay need to remove the cooler and clean debris from the
heatsink from the inside and replace the thermal paste. Itis recommended
that you contact our support team before removing the cooler on any video
card to allow us to give you any information on removing the cooler.

Heat is not my problem the shuddering caused by the core switching is. Whether it's caused by heat or a BIOS defect I don't know but simply cooling it better isn't an issue the stock cooler keeps mine at 73C or under. Lets stay on topic, please only post if you having this problem or have a solution. Coolers and such can be discussed in other threads.
post #39 of 53
I think it's a combination of the VRM heat causing the core to clock down and an over protective BIOS. The combination of both cause stuttering when the core clocks up and down wildly. I know the graphics card is safe at 80-90c but mine got to 60c and then sounded like a wind turbine along with loads of stuttering. My card is brand new, fresh everything and no dust. The problems have largely been resolved since I changed the cooler, the heat from the GPU will not affect the VRM anymore thanks to the watercooling, which helped massively with the stuttering. biggrin.gif
post #40 of 53
Quote:
Originally Posted by BiGK View Post

I think it's a combination of the VRM heat causing the core to clock down and an over protective BIOS. The combination of both cause stuttering when the core clocks up and down wildly. I know the graphics card is safe at 80-90c but mine got to 60c and then sounded like a wind turbine along with loads of stuttering. My card is brand new, fresh everything and no dust. The problems have largely been resolved since I changed the cooler, the heat from the GPU will not affect the VRM anymore thanks to the water cooling, which helped massively with the stuttering. biggrin.gif

Can you provide proof? I'd rather not spend 80 bucks on a new cooler for a 500 dollar video card without knowing it's the solution. Also proving this would open a big can of worms with XFX, a manufacture defected cooling problem, possibly.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD/ATI
Overclock.net › Forums › Graphics Cards › AMD/ATI › XFX 7970Ghz Edition Double D throttling problem. Faulty card or PSU related?