• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

7970 WTF?

Zarathustra[H]

Extremely [H]
2FA
Joined
Oct 29, 2000
Messages
42,371
So I ordered one of those Gigabyte triple fan 7970 OC cards.

First thing I always do when I get a new video card is to test baseline temps etc.

So at stock speeds (for this OC card that's 1000Mhz core) I turn the fans up to 100% manually and check how hot it gets in Furmark.

I immediately noticed that the GPU was throttling itself. So i went to CCC and altered the Power Control Settings to +20%, leaving voltage and core frequency at their stock 1000Mhz.

With all three fans pinned at 100% I restarted Furmark.

The line chart in Furmark shows the GPU getting hotter and hotter and its increasing at about a 45 degree angle. I expect it to eventually taper off (like it has with every other video card I've ever owned) but it never does. Last temp I see is 98C, then the computer crashes.

My philosophy has always been to never overclock past where Furmark temps are safe. With this video card Furmark temps aren't even safe at stock...

So either this Gigabyte 7970 can't even handle Furmark at it's stock settings, even with its big ass cooler, or I have a defective board.

Which is it? Opinions?
 
No offense but loop heaven or something. Furmark is junk ass shit that never touches my cards since 2 generations ago.
while I'm here... why would anyone let a $550 card right out of the box get to 98c first run and then scratch their hear in confusion. Why didn't you stop it at something like 80c??

I would never subject my card to such high temps. Furmark isn't even a decent indication of any kind of stability anyway...

stick your fan on 50% and loop heaven a couple times and report back.
 
Last edited:
Get a water block and cool your HD7970 proper.
 
Reference HD7970 here... mine doesn't break 85C in Furmark. That's at 1070MHz core and only 70% fan speed.
 
Reference HD7970 here... mine doesn't break 85C in Furmark. That's at 1070MHz core and only 70% fan speed.

Is it throttling at all?

If you run furmark, and watch the Afterburner charts, does GPU usage stay at 100% or does it drop? Does core stay at 1070 where you set it or does it drop too?
 
Is the force voltage control checked in afterburner?
 
Zarathustra[H];1038339805 said:
Is it throttling at all?

If you run furmark, and watch the Afterburner charts, does GPU usage stay at 100% or does it drop? Does core stay at 1070 where you set it or does it drop too?

Definitely not, I had MSI AB open when I did the benchmark run and it stayed pegged at 99%.

I do have +20% set in Powertune.

Here's a record of my test... I scored 5062:

http://www.ozone3d.net/benchmarks/furmark_192_score.php?id=30902
 
3 reference 7970s here and barely break 85*C under full load driving 99% usage.

Those gigabyte cooler cards are junk, like most gigabyte products.
 
3 reference 7970s here and barely break 85*C under full load driving 99% usage.

Those gigabyte cooler cards are junk, like most gigabyte products.

Gigabyte has been a reliable brand for me over the years, especially in the MB department.
 
Gigabyte has been a reliable brand for me over the years, especially in the MB department.

Motherboards maybe, look at some of the feedback on Gigabyte video cards on newegg though. Heatsink and fans falling off, cards DOA, Cards dying after a couple uses, ect...I wouldn't touch a Gigabyte card.
 
Im running 3 Reference gigabyte 7970s in Tri-Fire all at 1100/1500 Stock voltage with CCC controlling the fans and in Heaven Benchmarks they dont go above 78C, Bf3 they are all at 75C or under. Glad i didnt wait for the Overclocked cards, One thing i like is the Hot Air blowing outta my case not in it.
 
Those gigabyte cooler cards are junk, like most gigabyte products.
My Gigabyte 7970 @ 1400MHz core disagrees with you. So do the 20+ systems I've built with Gigabyte motherboards over the last five years without a single failure.

Anyway,

OP, Furmark is a power virus and an absolutely useless program. It's possible Gigabyte's reference OCP doesn't function in line with AMD's drivers, and may not correctly prevent Furmark from running (as they do now, both AMD and NVIDIA have block Furmark in their drivers). Anyway, go play a game or run a real benchmark and see how the card performs.
 
No offense but loop heaven or something. Furmark is junk ass shit that never touches my cards since 2 generations ago.
while I'm here... why would anyone let a $550 card right out of the box get to 98c first run and then scratch their hear in confusion. Why didn't you stop it at something like 80c??

I would never subject my card to such high temps. Furmark isn't even a decent indication of any kind of stability anyway...

stick your fan on 50% and loop heaven a couple times and report back.

Got anything to qualify your comments? I have been using Furmark for years and do not see any indication of your claims.
 
Got anything to qualify your comments? I have been using Furmark for years and do not see any indication of your claims.

I'll vouch for his opinion on furmark. Do some reading and you'll see quite a bit about "powervirus" talk. Powertune & Nvidias version of throttling were created because of programs like Furmark.

The OP and his card sounds like a different story though. Sounds like bad contact with the HSF, IMO.
 
EVERY gigabyte product I've ever used has failed in some way over the past 20+ years of system building. Asus, on the other hand, has been rock solid. I'd take an MSI product over Gigabyte product any day (with Asus being #1 of course).
 
I'll vouch for his opinion on furmark. Do some reading and you'll see quite a bit about "powervirus" talk. Powertune & Nvidias version of throttling were created because of programs like Furmark.

IMHO, you are SUPPOSED to stress the GPU (and CPU for that matter) as hard as possible to determine stability. Furmark is the way to do this.

Its a matter of safety margin. If it can pass this, it can pass anything without crashing.

IMHO, Nvidia's powertune and AMD's power throttling are cheating in this regard. It's like saying "yeah, you get the greates we have to offer, as long as you don't run anything really demanding on it, then it will scale back and run like a 4 year old product" :rolleyes:

The OP and his card sounds like a different story though. Sounds like bad contact with the HSF, IMO.

Yeah, I'm thinking so to. I am going to return it, but right now I don't have a freaking clue what I am going to replace it with :(
 
Zarathustra[H];1038341349 said:
IMHO, you are SUPPOSED to stress the GPU (and CPU for that matter) as hard as possible to determine stability. Furmark is the way to do this.

Its a matter of safety margin. If it can pass this, it can pass anything without crashing.

IMHO, Nvidia's powertune and AMD's power throttling are cheating in this regard. It's like saying "yeah, you get the greates we have to offer, as long as you don't run anything really demanding on it, then it will scale back and run like a 4 year old product" :rolleyes:
All this post does is clearly demonstrate how you don't understand how hardware works. A power virus program like Furmark tests the ability of your cooling solution, nothing more. That's why you see countless threads across this forum an others with "Furmark stable but crashes in games!" Furmark overlapped with stability testing when GPUs were much simpler. This isn't the case anymore.
Zarathustra[H];1038341349 said:
Yeah, I'm thinking so to. I am going to return it, but right now I don't have a freaking clue what I am going to replace it with :(
Have you tested it in a game yet or not?
 
Zarathustra[H];1038341349 said:
IMHO, you are SUPPOSED to stress the GPU (and CPU for that matter) as hard as possible to determine stability. Furmark is the way to do this.

Its a matter of safety margin. If it can pass this, it can pass anything without crashing.

IMHO, Nvidia's powertune and AMD's power throttling are cheating in this regard. It's like saying "yeah, you get the greates we have to offer, as long as you don't run anything really demanding on it, then it will scale back and run like a 4 year old product" :rolleyes:



Yeah, I'm thinking so to. I am going to return it, but right now I don't have a freaking clue what I am going to replace it with :(

You couldn't be more wrong.
Here, go do this.
Go rename furmark.exe to Oblivion.exe and run it. And enjoy watching your so-called "super stable" system CRASH. What are you going to do then? Buy a Playstation?
 
I would give uniengine a lot of runs to see how stability is.

My BEDD doesn't pass Furmark either, overclocked or not. However, I've NEVER seen a game come remotely close to producing that much heat. It clearly does some odd stuff that you won't see in real gaming situations. Crysis 2 was the most stressful I've seen, with temps in the low 70s with my custom fan profile.
 
I would give uniengine a lot of runs to see how stability is.

My BEDD doesn't pass Furmark either, overclocked or not. However, I've NEVER seen a game come remotely close to producing that much heat. It clearly does some odd stuff that you won't see in real gaming situations. Crysis 2 was the most stressful I've seen, with temps in the low 70s with my custom fan profile.
Again though, I believe the BEDD is anon-reference design, and therefore probably isn't affected by the driver-level control AMD employs. It might even be dangerous to run such a power virus program now.
 
That's why you see countless threads across this forum an others with "Furmark stable but crashes in games!" Furmark overlapped with stability testing when GPUs were much simpler. This isn't the case anymore.

Maybe this is because both AMD and Nvidia now throttle Furmark? :rolleyes:

Of course - a throttled furmark is not going to challenge the GPU, and then you might see other applications crashing.

I agree that Furmark is a bit overkill, but again, that's kind of the point. It gives you a worst case.

In the last three years I have used and overclocked the following video cards (in order of appearance):
- GeForce 9400GT
- Radeon 5750 (passive, fanless)
- Geforce GTX470 (standard cooler)
- Geforce GTX580 (standard cooler)
- Geforce GTX460 (gigabyte twin fan)
- Radeon 6970 (Asus Direct CUII Overclock)
- 2 Radeon 6970/s (Asus Direct CUII Overclock) in crossfire
- and now Radeon 7970 (Gigabyte triple fan OC)

In all of them I disabled the TDP/Furmark throttling function and made sure Furmark was running at full speeds unthrottled.

All passed with safe temps except the current 7970.

It also idles at ~35C with fans at ~30% load (in a 66F room, in a case with excellent air flow).

To me there is definitely something wrong with this card.
 
I see you have a Silverstone RV03. Some of the heatpipe coolers on these graphics cards are designed by monkeys and don't work in anything but the normal motherboard orientation.

I have an RV02 and have to be careful what I buy too. Supposedly the Twin Frozr cards work in these cases, but there are a number of aftermarket cooling solutions that just don't.
 
I see you have a Silverstone RV03. Some of the heatpipe coolers on these graphics cards are designed by monkeys and don't work in anything but the normal motherboard orientation.

I have an RV02 and have to be careful what I buy too. Supposedly the Twin Frozr cards work in these cases, but there are a number of aftermarket cooling solutions that just don't.

I have heard this too, but I read elsewhere that this is fallacy, and not to worry about it.

Heatpipes don't actually contain water in the fluid form, so gravity and orientation has little to no impact on them.
 
I do have +20% set in Powertune.

Powertune increases the TDP of the card but doesn't necessarily increase load voltage to what you need and it also increases the threshold by which temperature related down-throttling occurs.
 
Again though, I believe the BEDD is anon-reference design, and therefore probably isn't affected by the driver-level control AMD employs. It might even be dangerous to run such a power virus program now.

That's why they throttle it, it represents a completely unrealistic load when it's plowing full bore. The cards are undoubtedly made to perform under heavy real world load, but Furmark goes further yet oh fuck every subsystem is 100% code red reactor core critical fuck fuck fuck
 
My 'windfail' 6970; a ram chip gradually degraded to a death to start off with. Fin arrangement meant that large portions of the power circuitry end exhaust air blew onto the tantalum caps' metal skin, heated the back of the board so bad, that the bare power circuitry chips which already run hot were so scorching, burnt instantly on contact.

6970 DC2 was an absolute pleasure to have, cool, quiet and rock solid. How a mid high end card should be.

Windforce 7970 appears to have corrected this mistake. I went with the reference gigabyte 7970 because I'm modding to accelero for as fast of an air cooled, relatively microstutter free experience as possible. Aiming for the moon, I reckon most of these 28nms can get there with good cooling, a little volt modding and coaxing.
 
I'll vouch for his opinion on furmark. Do some reading and you'll see quite a bit about "powervirus" talk. Powertune & Nvidias version of throttling were created because of programs like Furmark.

OK, looked it up and AMD fixed the issue with Furmark in hardware since Cypress so it is safe to use. Even in older cards the issue was addressed in the drivers.

http://www.geeks3d.com/20090925/ati...on-against-power-virus-like-furmark-and-occt/

For Cypress based cards, AMD has hard-wired the protection and has implemented a hardware solution to the VRM problem, by dedicating a very small portion of Cypress’s die to a monitoring chip. This chip monitors the VRM. If the chip detects a dangerous situation (overload), the chip will immediately throttle back the card by one PowerPlay level. Once the dangerous situation has disappeared, the card goes back to its performance level. In the case of FurMark or OCCT, this can continue to go back and forth as the VRMs permit.

With this hardware protection, the Radeon HD 5870 is fully protected against FurMark or OCCT. One consequence is that a software protection in the driver is not needed anymore and you can use your favourite stress test tool to torture your brand new Cypress-based graphics card.
 
If anyone is curious what DVI video output looks like check out these two video's.

Its a huge issue in person, but due to the closeness of black and blue, not as apparent in the video's.

Watch in high res.

Gigabyte Video 01

Gigabyte Video 02

If I leave ity long enough the monitor eventually just says "signal out of range". It's repeatable across multiple monitors.

This is primarily why the card is going back.

Something really messed up with it.
 
I used Furmark to stress test for overclocking the 6850 I just got. I know that BF3 won't make the card run anywhere near as hot as Furmark, but I still stop at the temps that I'm comfortable at in Furmark. With Accelero L2 Plus attached, this card goes to 95ish in Furmark with 2xMSAA when running at 950 with 1.24 volts, and was going to require more volts to be stable. No thanks to that. I don't see what the problem is with Furmark. I stress test my CPU with Linpack, too, not just prime95. I think both Furmark and Linpack/OCCT are evolutions of stress testing. I don't see how Furmark would be outdated if it still puts the most stress on the card.
 
Those Gigabyte fans are useless. I would never buy one of those cards again. I had a couple of their 5870s with exactly the same fans. They suck at cooling and the card would go over 100° and crash as well.
 
That is sad to know. With 3 big fans you think 'nice, this will be quiet and cool!'. I almost bought one, there was a used 6950 for $210 shipped on Amazon. It's gone. Now I'm happy knowing there's a sucker out there somewhere. I'll stick to reference cards or proven oem-cooler-cards like IceQ and Frozr.
 
EVERY gigabyte product I've ever used has failed in some way over the past 20+ years of system building. Asus, on the other hand, has been rock solid. I'd take an MSI product over Gigabyte product any day (with Asus being #1 of course).

Only mb I have ever had fail was an MSI.

As for the Gigabyte fans, I have 5870 with those fans along with a stock 5870 in Xfire. I just ran Furmark and Gigabyte reached 74c while the stock card reached 79c. Who you gonna believe? Hearsay or fact?
 
Last edited:
Only mb I have ever had fail was an MSI.

As for the Gigabyte fans, I have 5870 with those fans along with a stock 5870 in Xfire. I just ran Furmark and Gigabyte reached 74c while the stock card reached 79c. Who you gonna believe? Hearsay or fact?

Fact. I've owned five of those exact Gigabtye 5870s. They are quiet but not great at cooling. Good overclockers though. If ambient temps are cool, they all did 1000MHz.
 
3 reference 7970s here and barely break 85*C under full load driving 99% usage.

Those gigabyte cooler cards are junk, like most gigabyte products.
Owned Gigabyte 4870 and 6950 for past years. Where are you getting your information from?
 
Back
Top