• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

6970 overheating - any suggestions ?

WT

Diamond Member
Sep 21, 2000
4,818
58
91
I picked up an HiS 6970 reference card earlier this year for cheap, with the understanding that its previous owner started having issues when the card was stressed. I currently use it in a workbench rig, so no gaming at all, just used for diagnostics on customer PCs.

It replaced a 4870x2, bought at the beginning of this year, AGAIN with overheat issues. Sensing a trend ? Check out GPU Temp #2 - I dare you to top that temp (its went over 122c before it rebooted):



Both cards were taken apart and cleaned, and I replaced both TIM and GPU thermal paste with the 'good stuff', forget the name of the TIM but it was $27 for a 4"x4" sheet, and Noctua NT-H1 was used on the GPU chip itself.

On to the 6970. The TIM looked terrible on it, as bad as the 4870x2:



Same mess on the GPU:



VRMs look meh ...



Added some Noctua NT-H1 to the GPU itself and made sure it seated cleanly:



High end poly TIM (I forget the brand but it is the correct thickness):



Put the card back together and then re-ran GPU-Z, anxious to see what my work had done:



Oh Hellz Ya !!! That's MUCH better !! That is HALF of what it had been running at) and those earlier readings were done sitting idle at the desktop).

OK, time to try and run the Vantage benchmark again, then if that completes its time for a Unigine Valley run .. but wait, what's this ??


Temps are yet again creeping up, but not as hot as it was. There is NO way that GPU Temp #2 should be that high. It's 20+ higher than GPU 1 and 3. Sitting idle at 80 again, which is better than 100c, but a run of Vantage had it back up to 119 again before I stopped it. What exactly is that GPU Temp #2 that GPU-Z is reporting ?

For reference, here's my CCC screen, with the card at default core and memory speeds. I raised the power limit setting from -3 to +20 just to see if it changed anything - no change that I can see. I also bumped fan speed up to 45%. Any higher and it can be heard easily outside of the case.

I know the first suggestion would be that I didn't have the TIM cleanly applied (I'm assuming that GPU Temp #2 is the VRMs) but if you look at the 6th pic down, I seated the HSF then removed it, and the pads stuck in place cleanly. The VRMs also get air from the fan and are not blocked by the HSF shroud like the XFX 7970 cards are. Any suggestions would be greatly appreciated.
 

Techhog

Platinum Member
Sep 11, 2013
2,834
1
26
Tried google. I would assume VRMs also at that temp.
Yeah, this. If so, those temps are fine, assuming it doesn't get much hotter than that. If you're really uncomfortable, you may just need to invest in VRM heatsinks or a new cooler.
 
Last edited:

KaRLiToS

Golden Member
Jul 30, 2010
1,918
11
81
The VRMS were overheating like hell mate. Probably crappy thermal pads or the wrong thickness.

Anyway, those coolers are not the best and by the way, it is NOT a reference card.

This is a reference card (PCB)



VS

HIS 6970 PCB (but with reference cooler)



_______________________________


I don't know what more you could do but there is one solution. Get some Fujipoly Ultra Extreme Thermal pads

They are suppose to be 1.5mm if I remember correctly, and according to this http://www.ekwb.com/shop/EK-IM/EK-IM-3831109835234.pdf (user manual)

Or just sell the card. VRMs can run hot and the card won't die.

My 2 cents.
 
Last edited:

WT

Diamond Member
Sep 21, 2000
4,818
58
91
KaRLiToS, those are the exact thermal pads I used ... Fujipoly Ultras. I can never remember the name but those are it. And they are in fact the 1.5mm pads I used on the VRMs on this card.

I assumed it was a reference card based on the cooler. Is it possible the card had its HSF replaced with another unit that doesn't cleanly contact the VRMs ? That's my only guess as to why it is running so hot. Looking at pics online it appears to be the proper cooler for an HiS 6970 card.

@Techhog - keep in mind those temps are at IDLE, and when I run a benchmark it will actually reboot the PC and the fan then runs at 100%. The PC will not boot until I power it off and then restart it. Those temps are not fine unless you want to fry an egg.
 

Techhog

Platinum Member
Sep 11, 2013
2,834
1
26
KaRLiToS, those are the exact thermal pads I used ... Fujipoly Ultras. I can never remember the name but those are it. And they are in fact the 1.5mm pads I used on the VRMs on this card.

I assumed it was a reference card based on the cooler. Is it possible the card had its HSF replaced with another unit that doesn't cleanly contact the VRMs ? That's my only guess as to why it is running so hot. Looking at pics online it appears to be the proper cooler for an HiS 6970 card.

@Techhog - keep in mind those temps are at IDLE, and when I run a benchmark it will actually reboot the PC and the fan then runs at 100%. The PC will not boot until I power it off and then restart it. Those temps are not fine unless you want to fry an egg.
Oh, idle. I see your point... My advice: underclock, undervolt, save up for a replacement (preferably an R7 265 or better). You need a new cooler for this card, but I don't think it's possible to justify the cost of a cooler.
 

WT

Diamond Member
Sep 21, 2000
4,818
58
91
Hey now, I think I just hit on the problem. I was under the impression that the 1.5mm pads were what I was to use (same thing that my 4870x2 uses) but I just found this thread over on [H] that says the 6970s use a .5mm pad, NOT a 1.5. Looks like my TIM is much too wide for the HSF to properly contact the card.

http://hardforum.com/showthread.php?t=1618547

It looks like this is what I now need to order -> http://www.frozencpu.com/products/17499/thr-181/Fujipoly_ModRight_Ultra_Extreme_System_Builder_Thermal_Pad_Blister_Pack_-_60_x_50_x_05_-_Thermal_Conductivity_170_WmK.html
 

Annisman*

Golden Member
Aug 20, 2010
1,917
89
91
Let us know how it goes :) I remember the days of 4870x2 vrm issues myself
 

Rvenger

Elite Member <br> Super Moderator <br> Video Cards
Super Moderator
Apr 6, 2004
6,282
5
76
Is it crashing at those temps? It's looking like a misread to me. Put an infrared thermometer on the back of the PCB to see if those temps are actually for real. I don't think the thinner thermal pads will help much because the fujipoly pads compress and spread instead of making the gap bigger.
 

WT

Diamond Member
Sep 21, 2000
4,818
58
91
It's actually worse than crashing, it restarted the PC after a run of Unigine's Heaven benchmark yesterday. I started the benchmark and walked away, came back 10 minutes later only to find the PC had rebooted - the fan was at 100% and the PC would not even POST !!

I bought one of these yesterday at Harbor Freight -> http://www.harborfreight.com/infrared-thermometer-93984.html so I will check on the card today. Obviously, I will only be able to read the backside of the PCB, but it will at least confirm that it is getting hot.

Rvenger, I am sure it's not a misread, as if you look at my second posted GPU-Z screenie (after I had replaced the TIM and paste) the card shows the temps dropping from 73c to 49c on GPU Temp #1, and Temp #3 dropping from 77 to 59. Temp #2 is where I had the biggest drop in temps, going from 122 (!!) all the way down to 67, but they started to raise as soon as I booted into Windows.

@Annisman - Sadly, my 4870x2 is not fully functional. The temp drops were HUGE and I thought I had resurrected that card, but it crashed in benchmarks as well. It now hangs on the workbench pegboard hutch as a display piece.

http://wtslair.com/Project4870X2/Reassembled.jpg
 

Annisman*

Golden Member
Aug 20, 2010
1,917
89
91
Well if multiple cards are crashing maybe look at other parts of your system for the instability. Is your processor overclocked ? And I wish I had kept all my old video cards for showpieces ! The 4870X2 was just a gorgeous card.
 

KaRLiToS

Golden Member
Jul 30, 2010
1,918
11
81
Do you have enough of airflow in yout case?

Does the back of the card, where the outputs are, have enough clearance for the hot air to exhaust?

I am almost 100% sure that it is not a false reading and the VRMs really gets hot.
 

WT

Diamond Member
Sep 21, 2000
4,818
58
91
I tried to be as thorough as possible in giving details on the issue, but neglected to detail the rigs that both cards were used in. Note that I bought BOTH cards with known overheating issues, and they were dirt cheap for a reason. Both cards have been tested in multiple rigs, with similar results in both.

The workbench rig that I am using primarily for testing includes:
AMD Phenom II X4 965BE,OC'd (via multi) to 3.7 @default voltage.
MSI 790X-G45 ATX
XFX 650w XXX Edition PSU
HiS 6970 2GB video card
8GB Ripjaws DDR3 RAM (4GBx2)
Vertex 4 SSD, Seagate 1TB HDD
Chieftec Dragon Ultra full tower case (old skool, baby !!) - I still use this case as it is so large, it disperses any heat build-up simply due to the fact that hot air rises. The case has a 120mm Thermaltake Icage in the lower 3 drive bays, as well as 3 80mm fans (1 intake blowing over the SSD and HDD, 2 exhaust above the GPU)

Edit: added link to old pic of the case before it's last upgrade (hardware you see in the pic doesn't match what I have listed, which is what the case currently contains) -> http://www.wtslair.com/ProjectBlackEdition/itsgreentastic.jpg

The other rig is a bit vintage, offering up the following:
AMD Phenom 9750 B.E. (stock @ 2.4)
Asus M2N32-SLI Deluxe (Nv Nforce 590 chipset)
XFX 550w Core Edition PSU
Kingston 40GB SSD, WD 750GB HDD
Antec 830SX tower case - modded with 120mm fan VGA blowhole mod

Yea, trust me ... cooling should NOT be an issue for either. The workbench (top) rig had previously run a GTX280 with absolutely no issues. I upgraded to the 6970 for this rig simply to give AMD a try. Having bought Nvidia exclusively since the Riva 16MB TnT days, I could be considered a 'fanboi' only because of my purchasing history, but I desperately want to own and regularly use an AMD card, as I think they offer a better bang for the buck right now for the only game that I play - Company of Heroes 2.
 
Last edited:

WT

Diamond Member
Sep 21, 2000
4,818
58
91
Measured the back of the card after letting it sit idle for 20 minutes and it measures 83c. My problematic GPU Temp #2 that is always reading 10-20c higher is sitting on 80, so both GPU-Z and my digital thermometer are reading similarly.

I'll bet you a dollar that it ramps up over 100 if I launch any benchmark tho. TBH, I am not using the benchmark for anything other than a stress test, as I don't even game on that PC, but I want the ability to game if a friend stops over for some LAN action.
 

AsianriceX

Golden Member
Dec 30, 2001
1,318
0
0
I'm interested to hear the results of redoing the VRM TIM with 0.5mm.

I got a HIS 6970 when Amazon had them for just over $200 a while ago and it's been forever finicky. I was convinced the fan profile was just awful and it's more than happy to bake itself to death unless I manually shift it to 55% fan speed before gaming.

I had same symptoms you're having now. If I don't control fan speed manually, it'll crash the entire system with looping sound and garbage on the displays. Other random crashes will cause it to reboot to a blank screen and 100% fan speed. Reset button doesn't work and requires full power off to resurrect.

This may all be moot if I grab a GTX 970 but it'd be nice to stabilize this guy too.
 

WT

Diamond Member
Sep 21, 2000
4,818
58
91
Hey, a fellow HiS 6970 owner !! And to think that I at first thought about Crossfiring with another one ... *shudder*

I ordered the Fujipoly Ultra .5mm last night. In the meantime, my 7970 should arrive by Tuesday, so I have some card shuffling to do this week. The GTX580 in the main rig will be moved ... somewhere.

It feels very odd to be running AMD cards in 4 of the 5 rigs here. I admit to enjoying a pain-free driver experience over many years with Nvidia, and after reading the bugs with the fresh Catalyst 14.9s, I am kinda worried now.

Read yet another thread on the 4870x2 last night while on the topic of TIM, and in that thread they claimed that the 4870x2 used 1.0mm TIM, not the 1.50mm that I had read elsewhere. Whom to trust ?? Either way, the 4870x2 appears to have other issues, as I found a small portion of the PCB near the upper corner RAM chip that has discoloration on it. It may be a candidate for a baking, as I think GPU1 is bad, and you cannot disable GPU1 on the x2 cards, only GPU2.

http://wtslair.com/Project4870X2/HeatDamage.jpg
 

WT

Diamond Member
Sep 21, 2000
4,818
58
91
OK, got the Powercolor 7970 today (used, but with NO ISSUES to report !!) I know .. why'd I buy it if it just works ?!?! Very clean card - the HSF was pulled when the previous owner bought it and it lived under water until being sold (it supposedly was not used for mining, just gaming). As such, the HS is spotless:



Anyhoo, out went the HiS 6970 (until my .5mm TIM arrives) and in went the 7970. What a HUGE difference !! Look at those temps:



Everything stock so far as I run a few benchmarks. Keep in mind I have never used an AMD card in a main rig, and have no idea how to safely OC the card. In fact, I never OC a video card in my rigs, but the 7970 cards have such headroom built into them that I want to give it a go:



Just finishing a run of the Valley benchmark on the card, and that benchmark reports the cards temp climbing up to 72c at max (I have it running on a pair of 22" LCD panels in an Eyefinity setup). I will be bumping that default fan speed from 32% up to 42% to see how loud it is and how much it drops the temps. This would confirm that my workbench rig is stable and providing plenty of airflow, so it's clear that the 6970 has some major cooling issues, and may have other issues than just thicker TIM applied to it.
 
Last edited:

WT

Diamond Member
Sep 21, 2000
4,818
58
91
The Fujipoly Ultra TIM showed up today. Lemme tell ya, that blister pack was a LOT smaller than the 1.5mm pack I bought earlier this year, and the .5mm stuff is harder to work with, especially to peel off the backing sheet. I don't even have enough to do the card a second time, but I have 2/3 of the 1.5mm stuff left after doing 2 cards already.

Yet again, I was initially thrilled with the temps, as all three temp readings were in the 30s once I started GPU-Z back up. The card was now able to complete a run of Vantage and the temps seemed fine. A second run of the benchmark returned an odd error though:



Time for a run of Valley, which before my work on the TIM and paste replacement took the card to a sky high 120+c and eventual reboot of the PC. Even before I started it, I watched those temps climbing .. past the 50s, 60s, 70s .. 80s ... 90s ... but .. it did finish the Valley benchmark, where it topped out at 102c. Ouch.

Back out of the benchmark, and the temps soon dropped, but not anywhere NEAR the original temps at bootup. From 40 to 80c with no load on it means something is still going on with the card.



Not much more to say or do. I am at least partially satisfied, as the card will complete a run of both benchmarks, and no game I play will ever stress the card like Valley will. If it can play Company of Heroes 2 without crashing I will just leave it in the workbench rig and stop worrying about it.

Edit: I seem to recall reading about AMD cards having issues with a dual monitor Eyefinity setup (the card won't clock down to 2D speeds ??) , so I'm now thinking that those temps are being caused by my second display, and once I set up the second panel that's when the temps all but doubled from mid 30s to mid 60s. It'd be interesting to disable Eyefinity and see what temps I get when running only one display.
 
Last edited:

WT

Diamond Member
Sep 21, 2000
4,818
58
91
Holy crap, it was the dual monitor setup causing the issue the entire time !! If I disable the second display via the Win key+P keystroke combo, the temps drop rapidly from 82c to 48c with no change at all to the fan profile or core/memory speeds. What a kick in the @ss this is.

I'm running the newest 14.9 Catalyst driver and tried to downclock the memory from 1375 to 1000 to force the card into idle mode, but no change at at to the temps until I turned off the second display.

One thing about going AMD on this one .. I am learning about this quirky 'bug' which AMD continues to not feel is a bug. It is more a limitation of TDMS technology. Now, going Displayport will fix it as well, so I guess I can look at picking up Displayport adapters for my setup.

Edit: Looks like Nvidia is also having this issue, at least with the older 480 cards.
 

WT

Diamond Member
Sep 21, 2000
4,818
58
91
So I picked up a DVI KVM to act as a workaround to my matched monitors/unmatched inputs issue (one DVI, one VGA thru my VGA KVM). I figured that with a dual DVI connex coming out of the card into my 22" panels (both ID'd as Viewsonic 22s with 1680x1050 @ 60hz refresh), I could correct the issue.

Still no luck, as the card immediately jumped up to 70+c when I enabled the second display. I tried setting a custom 2D profile to downclock the core and mem speeds as low as the sliders would allow in MSI afterburner, but again had no luck in forcing the card to lower its speeds.

Might have to just sell the card and replace it with the 7970 I picked up while I worked on fixing the 6970. That card seems to work fine with both panels, and idles at much lower 2D speeds properly.

I'm guessing the KVM does something to throw off the refresh rate on the LCD that it is plugged into, and that may be why the card is still not throttling ?
 

ASK THE COMMUNITY