7970 WTF?

[buttons]

for whatever its worth, my MSI 7970 at 100% load overheats 90C+ at stock voltage/clocks. Now i have it running 1.043V at 1000mhz and 100% load is 71C with only 38% fan speed in a silverstone rv03

 

[J. Cyrus]

Seems too high to me, but then again you pushed it to the brink lol

 

[Yobbles]

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.

But I just posted proof that proves the Gigabyte cools better than the stock cooler.

 

[Spare-Flair]

But I just posted proof that proves the Gigabyte cools better than the stock cooler.

What proof? I also owned multiple 5870s with the stock cooler and they consistently maintained better temperatures and stability, not including the core but of all components such as ram and VRMs.

It seems to me that the fan profile burned into the Gigabyte bios was tuned for noise and not stability.

I currently have 3 of the Gigabyte cards bitcoin mining and at 100% load and with fans at 100% throttle, they are sitting in the upper 90° range. I also have 5 stock 5870s with blower fans mining in the same environment and at 100% load, they are sitting in the upper 70° range. If I let both cards use "auto" fan profiles, the stock cards will remain stable. The Gigabyte cards will often throttle or overheat. These are all in a controlled and open air environment.

I bet what's happening in your situation is that you have them in crossfire and the stock card is somehow receiving deficient airflow for it's requirements or there are differences in fan tuning.

 

[Yobbles]

Possible but either way it doesn't matter because lower noise is more impront to me. Gigabyte is also factory OC an extra 50mhz compared to stock but when I ran the test I had both cards set at 850mhz manually. Did you do that or was not your Gigabyte OC by 50mhz like mine is?

 

[marekfreak]

I've had my eye on this non-reference GPU for sometime now. I'm too impatient to wait for the Super-Clocked edition. But kudos for making me question my purchase now

I am getting my http://www.gigabyte.com/products/product-page.aspx?pid=4102#ov (GV-R797OC-3GD) tomorrow to replace my aging GB 5870 (with a custom overclock), which has the same Windforce custom cooling, but with only two fans. I'm only using a single monitor setup @1900X1200.

Generally, I've never had any problems at all with the Windforce custom cooler, except for the usual dust buildup every 6 months or so. Dust build-up is the only problem with this cooling solution and can cause temps to increase by 10 - 15 degrees Celsius (base on my personal experience). A 5 minute clean of the cooling fins and fans resolves any temp increase. IMHO it's the coolest and quietest air-cooling solution I've ever had, even for an over-clocked GPU.

The only frustrating problem(s) I had with GB is their firmware for their GPU range, but a quick flash of the GPU firmware and update resolved my previous problems.

Generally GB mobos/GPUs have been great for all the systems I've put together so I'm hoping I won't have issues with my new GPU.

I'll test my card and post my brief experiences tomorrow, if I'm not too busy enjoying my new card or resolving firmware issues again.

 

[Mr. K6]

I've had my eye on this non-reference GPU for sometime now. I'm too impatient to wait for the Super-Clocked edition. But kudos for making me question my purchase now

I am getting my http://www.gigabyte.com/products/product-page.aspx?pid=4102#ov (GV-R797OC-3GD) tomorrow to replace my aging GB 5870 (with a custom overclock), which has the same Windforce custom cooling, but with only two fans. I'm only using a single monitor setup @1900X1200.

Generally, I've never had any problems at all with the Windforce custom cooler, except for the usual dust buildup every 6 months or so. Dust build-up is the only problem with this cooling solution and can cause temps to increase by 10 - 15 degrees Celsius (base on my personal experience). A 5 minute clean of the cooling fins and fans resolves any temp increase. IMHO it's the coolest and quietest air-cooling solution I've ever had, even for an over-clocked GPU.

The only frustrating problem(s) I had with GB is their firmware for their GPU range, but a quick flash of the GPU firmware and update resolved my previous problems.

Generally GB mobos/GPUs have been great for all the systems I've put together so I'm hoping I won't have issues with my new GPU.

I'll test my card and post my brief experiences tomorrow, if I'm not too busy enjoying my new card or resolving firmware issues again.

[H] just reviewed the card and gave it high praise: http://www.hardocp.com/article/2012/02/08/gigabyte_radeon_hd_7970_oc_video_card_review . It sounds like an excellent part.

I'm also a Gigabyte fan myself. I've been building almost all my systems (about 20+) over the last 5 years with Gigabyte motherboards and currently have a Gigabyte Z68 mobo and a reference Gigabyte 7970 in my system; both are excellent parts.

 

[WorldExclusive]

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.

This.

Furmark needs to go into the digital trash cash.
AMD and Nvidia has built-in sensors to throttle the card when Furmark is detected.

It pushes cards beyond the rated TDP and is not an accurate indicator of temps or power consumption.
You could push your card 50-100W out of spec using Furmark, which doesn't represent daily use.

 

[Zarathustra[H]]

This.

Furmark needs to go into the digital trash cash.
AMD and Nvidia has built-in sensors to throttle the card when Furmark is detected.

It pushes cards beyond the rated TDP and is not an accurate indicator of temps or power consumption.
You could push your card 50-100W out of spec using Furmark, which doesn't represent daily use.

Intel Burn Test doesn't represent daily use for CPU's either, yet it is still the best stability tester as it stresses the CPU to its absolute max...

 

[WorldExclusive]

Intel Burn Test doesn't represent daily use for CPU's either, yet it is still the best stability tester as it stresses the CPU to its absolute max...

Intel is Intel. I use LinX and Prime95 by the way, never used IBT for anything.

Furmark isn't AMD/Nvidia.

AMD and Nvidia throttles any third party benchmarking software that exceeds specifications, and doesn't recommend it for benchmarking their cards.
You can google it.

 

[Zarathustra[H]]

Intel is Intel. I use LinX and Prime95 by the way, never used IBT for anything.

Furmark isn't AMD/Nvidia.

AMD and Nvidia throttles any third party benchmarking software that exceeds specifications, and doesn't recommend it for benchmarking their cards.
You can google it.

Intel is not the maker of IBT. it's just a free software frontend for Linpack.

In other words Nvidia and AMD cheat, and ship their stock clocks unable be stable when stress testing.

It's amazing that anyone is OK with this.

It's almost as if AMD or Intel would ship CPU's that are only stable unless you run certain heavy load programs, and if you do they throttle themselves.

 

[Mr. K6]

Intel Burn Test doesn't represent daily use for CPU's either, yet it is still the best stability tester as it stresses the CPU to its absolute max...

Intel is not the maker of IBT. it's just a free software frontend for Linpack.

In other words Nvidia and AMD cheat, and ship their stock clocks unable be stable when stress testing.

It's amazing that anyone is OK with this.

It's almost as if AMD or Intel would ship CPU's that are only stable unless you run certain heavy load programs, and if you do they throttle themselves.

It seems you don't understand how these programs work nor the differences between them. Intel Burn Test isn't a power virus, Furmark is. Intel Burn Test is based on Linpack and puts the CPU at 100% load, tasks it with calculations, and checks the results. It isn't daily use, but it is a realistic load and furthermore is validated. Furmark does none of this.

 

[p3sty]

furmark hated my 6950 it didd the same thing as the op

 

[Yobbles]

So, that HardOCP review of the Gigabyte 7970 proved what I said is true about the Gigabyte cooling solution compared to stock. It cools better than stock. Don't argue with people who know more than you next time.

 

[xorbe]

In other words Nvidia and AMD cheat, and ship their stock clocks unable be stable when stress testing. It's amazing that anyone is OK with this.

It seems you don't understand how these programs work nor the differences between them. Intel Burn Test isn't a power virus, Furmark is.

The fact remains that CPUs are designed to function flawlessly at 100% load by any sequence of opcodes, and today's GPUs are clearly designed to handle less than full load, which wasn't always the case. Zarathustra is spot on.

 

[jaredpace]

It's amazing that anyone is OK with this.

It's almost as if AMD or Intel would ship CPU's that are only stable unless you run certain heavy load programs, and if you do they throttle themselves.

It's not all about the megahertz buddy. Dont hate the damn gpu if your video game doesnt use all the compute units simultaneously. Don't hate the miner or cruncher software if it doesn't use the max memory frequency or bandwidth.

Furmark and occt do the same simple thing to the gpu to load every shader/compute unit/streamprocessor/cude core simultaneously - something that doesnt happen in real world usage mining crunching gaming or benching. So dont hate your gpu or furmark or ATI/NV's marketing, just realize you have a misconception about what and why.

If you wanna max ur card in furmark all day, just use a 4870 or 9800gtx or earlier card.

 

[Comte]

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.

Of the thousands of video cards they ship, you see 10 or so complaints, but never see the other thousands working fine because the people arent bitching, they are playing their games. You need to take any comments on sites like that with a huge grain of salt...

 

[Comte]

Then you did something stupid to kill it. 3/3

 

[Comte]

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).

Then you are so young you missed the ASUS bad capacitor problems? Maybe your experience isnt so vast or valid after all.

http://www.badcaps.net/forum/forumdisplay.php?f=6

 

[tikiman2012]

My original XFX 6950 had a similar problem. I removed the heatsink to inspect it & found that the factory TIM pad was folded over & doubled up on part of the core. I cleaned & applied new TIM but the card eventually failed anyway.

 

[SonDa5]

Just return the card to the retailer or RMA it. It runs hot.

I have no problem testing Furmark or Kombuster with my GTX470 or GTX560ti 448 cards.