After-market coolers for your 6xxx?

Exactly. Hell, even at 50% that fan is too damn loud. And 40-42C at load? That's VERY unusual. Most people can't even achieve that with aftermarket coolers. Your case must be ridiculously well cooled/ventilated, along with a cool room.
 
Not really, it's downright easy with the fan at 100%. At normal fan speeds (maximum is about 40-45%) I see temps of around 85C under load, which is about right.
 
shantd said:
Exactly. Hell, even at 50% that fan is too damn loud. And 40-42C at load? That's VERY unusual. Most people can't even achieve that with aftermarket coolers. Your case must be ridiculously well cooled/ventilated, along with a cool room.
Click to expand...

Or he lives in Alaska :)
 
yeah...not my experience...these things are shrieking on a warm day to keep temps in check...though I am mining 100% 24-7
 
shantd said:
Wow, that's great. And the good news is that there is no game out there which will tax your 6970 to the extent that Furmark does. So you're probably looking at high 60s during gaming. Can't really ask for more than that. I'm a bit surprised you can get that much cooling with your fans topping out at 2,000 RPM. My stock 6970 fan can hit 6,000RPM and still won't cool that well.

- How long had you been running Furmark at that point?
- Where'd you order the twin turbo II from, and how much was it?
- Did you have any issues removing the shroud, or any issues installing the fan?
Click to expand...

- Not sure exactly how long it had been running. I let it go until the core temperature had basically flat-lined.
- Ordered the Twin Tubo II from FrozenCPU for $52: Link
- No issues at all removing the stock heatsink or installing the new one.

I should note that all of the Accelero GPU heatsinks with multiple fans do not report RPM correctly. Arctic Cooling spliced together the RPM sensor wires from both fans, which means the card receives two RPM readings at once with a fluctuating offset (totally useless reading). In theory, cutting the RPM sensor wire coming out of the second fan should allow correct RPM monitoring without messing with anything else, but I haven't tested this.

Also keep in mind that the Twin Turbo II uses two-part thermal epoxy to affix the RAM and VRM heatsinks to the card. This is much, much better than the old thermal tape that they Twin Turbo Pro used on its Ramsinks (which tended to fall off the card after a few days), but it's also semi-permanent. They claim the thermal epoxy they use is easy to remove...but I wouldn't try. lol
 
Last edited:
shantd said:
Exactly. Hell, even at 50% that fan is too damn loud. And 40-42C at load? That's VERY unusual. Most people can't even achieve that with aftermarket coolers. Your case must be ridiculously well cooled/ventilated, along with a cool room.
Click to expand...

I thought I'd point out - in the majority of cases, stock coolers will actually perform better than any aftermarket cooler at 100% fan speed. The much lower maximum fan speed of aftermarket coolers means that while they usually do their job better at low speed/noise, they can't reach the levels of performance the stock coolers can. It's rare to see stock coolers have a higher max speed than 2500rpm (Arctics are usually only 2000). So 6000 on a stock HD6970 cooler? It's not hard to see why it's so effective. Adjoining the VRM and memory sinks into the main cooler also helps keep them cool too.
 
samuelmorris said:
I thought I'd point out - in the majority of cases, stock coolers will actually perform better than any aftermarket cooler at 100% fan speed. The much lower maximum fan speed of aftermarket coolers means that while they usually do their job better at low speed/noise, they can't reach the levels of performance the stock coolers can.
Click to expand...
That's if you use the fans that come with the aftermarket solution. I'd like to point out that the stock "dual 92mm" combination fan/shroud on the Twin Turbo II un-clips from the cooler, allowing you to mount dual 120mm fans of your choice.

You can push significantly more air through that aftermarket heatsink than the stock heatsink (with its squirrel cage fan) could ever hope to achieve...and it would likely still come out quieter than the stock heatsink with the fan ramped up to 100%.
 
samuelmorris said:
The stock cooler at 6000 is pretty powerful, i've not managed to get load temps much over 40-42C with it.
Click to expand...

Lol. I'm lucky to keep the top card at 86-87C fully loaded.... crossfire though. Bottom card loads at around 76. You must have pretty cold ambient temps.
 
Just ran furmark with fans at 100%, both cards maxed out at 61C. That's with an ambient room temperature of about 25-26C. Under normal load and on cooler days they're usually a fair bit lower than that. Today's been quite a warm one :)

When I was testing the DOA (unstable) 4GB 5970 with the modified accelero cooler (2500rpm) on it, I recall it topping out around 68-69C, in a 21C room, though while that was a dual-GPU card, it was not in crossfire, so experienced much better airflow.
 
I've never been as lucky as you apparently when it comes to stock cooling. Aftermarket coolers get far better results with far lower rpms because they have far bigger/better heatsinks and far larger fans which push more air.

As for leaving the fan at 100%, it's moot. Unless you're deaf, who can leave a stock 6970 running at 100%?
 
samuelmorris said:
Just ran furmark with fans at 100%, both cards maxed out at 61C. That's with an ambient room temperature of about 25-26C. Under normal load and on cooler days they're usually a fair bit lower than that. Today's been quite a warm one :)

When I was testing the DOA (unstable) 4GB 5970 with the modified accelero cooler (2500rpm) on it, I recall it topping out around 68-69C, in a 21C room, though while that was a dual-GPU card, it was not in crossfire, so experienced much better airflow.
Click to expand...

You have made a couple contradictory statements, first you said your cards under load do not go above 42c, now you say they max out at 61c. Your anecdotal evidence goes against just about every other persons experience with using an aftermarket cooler vs stock.
 
The 42C measurement was on a colder day and was certainly not while running furmark. I am also missing one of the side intake fans I used to have, as I haven't got round to replacing it yet. This is also of course at 100% fan speed.
 
Ok, so throw out the 42c figure on the stock cooler since it wasn't "Furmark Load" (which is what we've all been comparing).

61c would be impressive with the stock cooler, if it weren't for the fact that you had to run the fan at 100% to get it there, and we all know an HD 6970 is insufferably loud with the fan at 100%.

I'm going to pop some 120mm fans on my Twin Turbo II tonight and see what happens. By bet is that I will easily see Furmark load temperatures below 61c with FAR less noise than the stock cooler at 100%.
 
I never said it was furmark load. I just said load. To me, 'load' implies normal load, i.e. games.
If it's furmark, either I'll explicitly say furmark, or 'burn test' / 'stress test'

I'm sure you will see better temps with uprated fans on your cooler, but that doesn't really say anything. My initial argument was that aftermarket coolers, while better heatsinks, use slower fans, and thus, can't achieve the same cooling potential as the stock cooler at 100%. If the fans were more powerful, they would. It's pretty straightforward.
 
samuelmorris said:
I'm sure you will see better temps with uprated fans on your cooler, but that doesn't really say anything. My initial argument was that aftermarket coolers, while better heatsinks, use slower fans, and thus, can't achieve the same cooling potential as the stock cooler at 100%. If the fans were more powerful, they would. It's pretty straightforward.
Click to expand...
So you're just stating the obvious? That's not an argument...

Using a ridiculously loud fan to push a ridiculous amount of air through a tiny stock heatsink can yield cooler temperatures than using a quiet fan pushing a small amount of air through a large aftermarket heatsink... Duh? You're overcoming the inefficiency of the stock heatsink by throwing CFM at it. Thing is, nobody likes gaming next to a leaf-blower, so that's not really an everyday-usable solution for overcoming the inefficiency of the stock heatsink.

Solution: Put a larger fan on the aftermarket heatsink, since everything about it is more efficient, and they've equipped it with less-than-fast fans in order to meet the noise numbers they wanted. The cooling potential is there, and obviously better than the stock heatsink.
 
Last edited:
It was more in response to this statement:
Most people can't even achieve that with aftermarket coolers.
Click to expand...
implying that typical aftermarket coolers can reach levels of cooling that stock coolers can't, which is not often the case unless you retrofit them with bigger fans.

I'm well aware of the limitations of the stock heatsink design, but given the issues that arise in crossfire/SLI with other types, they are still clearly the best design for manufacturers to fit.
 
samuelmorris said:
I'm well aware of the limitations of the stock heatsink design, but given the issues that arise in crossfire/SLI with other types, they are still clearly the best design for manufacturers to fit.
Click to expand...
Huh? Most of the aftermarket heatsinks I've seen don't interfere with crossfire/SLI at all (they don't extend past the edge of the PCB, so the normal crossfire/SLi bridge still fits fine). There are some insane ones like the Shaman, but that's more the exception than the rule.

The reason AMD and Nvidia are sticking to the blower-fan design is because it pushes the majority of the heat generated by the card directly out the back of the case. This keeps all your other components cooler, and prevents the video card from sucking on its own exhaust.

They can't assume all cases will have great airflow, so a blower is a safer choice than a heatsink that dumps all of its heat into the case.
 
That's not the issue - when placing two of the cards in crossfire, the top card is starved of airflow in many cases unless it uses the standard rear-blower design. It's a commonly reported issue.
 
samuelmorris said:
That's not the issue - when placing two of the cards in crossfire, the top card is starved of airflow in many cases unless it uses the standard rear-blower design. It's a commonly reported issue.
Click to expand...
Again, not sure what you're on about.

Any motherboard with slots close enough together to starve a non-insane aftermarket cooler of airflow would also starve a blower fan of airlfow.
I assure you, the top heatsink in the below configuration isn't breathing very well, especially since AMD doesn't notch the area around the fan like Nvidia does.
Edit: In a situation like this, I would have recommended one dual-GPU card rather than two single-GPU cards.

uOOJJ.jpg



Now lets take a look at a motherboard with a sane slot layout. Plenty of breathing room for the stock heatsink, or any of the Accelero heatsinks.

hWb05.jpg



Edit: and as for triple-card setups, the current method is extremely poorly designed. Three blower fans, two of which are constantly starved of airflow no matter what. They should include the option to replace the fan module on each card with an air channel (effectively a hollow tube that leads to the butt-end of the card). You could then mount a 120mm or 140mm fan to the end of the three cards, which would blow through all three heatsinks and right out the back of the case. Such a fan configuration would also make it trivial to duct air from a front-intake on one's case, directly into your video cards. The air that cools the tri-card setup would never actually have to interact with the rest of the case.

llF4G.jpg
 
Last edited:
Two reference HD6970s (rear-blower design):
Bitcoin temperatures: 69C and 71C with fan speed at 60% (3600rpm) - ambient temperature 28C

Two non-reference VTX3D HD6970s (twin-fan design):
Bitcoin temperatures: 105C and 109C with fan speed at 100% (3500rpm) - ambient temperature 24C

Having a good board layout does help - but really, these VTX3D cards were fine by themselves - stick them close together in crossfire, and they're useless. The same is true with most non-reference coolers.
 
I say again, why put them that close together with ANY type of heatsink when there are better options?

With them that close, you either end up with stock heatsinks operating at intolerable volume, or aftermarket heatsinks running on the warm side. Neither one of those choices is optimal...so why not just get a motherboard where the first and second PCIe slot are one slot farther apart? Both types of heatsink will work better.

If you're forced to have the slots that close together, you'd probably be better off with an HD 6990 with an aftermarket heatsink rather than two individual cards.

Edit: I also hold those bitcoin temperatures suspect. I've never seen the stock heatsink on the 6970 work that well, even with just a single card.
 
Last edited:
Those are the temps I got. Bear in mind this was with running four 120mm 1900rpm slipstream side intake fans at the cards, and two more on the top exhausting, in addition to the normal 140 rear and 230 front on the HAF.
 
Doesn't mean the reading is accurate. And those VTX3D coolers you compared against aren't the greatest in the world either.

Worth noting: I have a GTX 260 mounted directly below my HD 6970.

Load temperature on the HD 6970 with the stock heatsink at stock clockspeed got as high as 102c
Load temperature on the HD 6970 with the Twin Turbo II and the card overclocked: tops out at 76c.

Both heatsinks were sucking on the back of another full size graphics card. The stock heatsink doesn't always win in this situation.
 
Last edited:
Geforces though, do have slower stock fans than Radeons, typically they max out at 3500rpm (same as my XFX non-ref HD4830) instead of AMD's 5000 (or 6000 for the 6970. Not sure about 6990)
 
What does that have to do with anything? Those were temps from the 6970, not the GTX260 :confused:
 
samuelmorris said:
My initial argument was that aftermarket coolers, while better heatsinks, use slower fans, and thus, can't achieve the same cooling potential as the stock cooler at 100%.
Click to expand...

Unknown-One said:
Using a ridiculously loud fan to push a ridiculous amount of air through a tiny stock heatsink can yield cooler temperatures than using a quiet fan pushing a small amount of air through a large aftermarket heatsink...
Click to expand...


I really doubt even these statements are true. I had a 4850 that would get super hot so I had to leave it at 100% at all times while gaming (fortunately the noise was nowhere as bad as the 6970). I'd still reach 90s and occasionally break 100. Adding an aftermarket cooler brought me down to the 70s under load, completely silent.

As for the 6970, again, if I play Witcher 2 for an hour, I'll be in the 90s with a normal fan profile. If I left the fan at 100% I'd come down to the 80s at the lowest. An aftermarket cooler will do much better than that. As far as I'm concerned, an aftermarket cooler is superior to stock cooling in every conceivable way.
 
Then it must have been faulty. At 100% fan speed the HD4850s ran reasonably cool, the problem AMD had was that the default profiles let the cards get very hot because they did not force the fan speed up enough. There's probably something seriously wrong with your case airflow if you're getting temps in the 80s with 100% fan speed on a 6970.
 
shantd said:
I really doubt even these statements are true.
Click to expand...
It's going to come down to the two coolers and fans you're comparing, really.

Though what we said was true, it's possible for a small heatsink with a lot of air being forced through it to keep a component cooler than a large heatsink with very little air being pushed through it. You'll always be able to put a better fan on the latter heatsink and improve its abilities, though.

samuelmorris said:
There's probably something seriously wrong with your case airflow if you're getting temps in the 80s with 100% fan speed on a 6970.
Click to expand...
I saw temperatures in that range on my HD 6970 before switching to an aftermarket heatsink... and the aftermarket heatsink improved the situation drastically.

If my airflow were bad, then adding the aftermarket heatsink would have been detrimental to my cooling, since all the heat from the graphics card is now being dumped into the case rather than out the back.

My airflow is obviously just fine, the stock heatsink simply didn't seem to work very well.
 
Just wondering, is the Accelero Twin Turbo II and dual or triple slot cooler?
I'm planning on picking up the Accelero Xtreme Plus II for my new 6970 coming, but I'd rather go dual rather than triple slot if I can. I'm not planning on going nuts from an OC side, but I am going to CF with my unlocked 6950.
 
Unknown-One said:
It's going to come down to the two coolers and fans you're comparing, really.

Though what we said was true, it's possible for a small heatsink with a lot of air being forced through it to keep a component cooler than a large heatsink with very little air being pushed through it. You'll always be able to put a better fan on the latter heatsink and improve its abilities, though.


I saw temperatures in that range on my HD 6970 before switching to an aftermarket heatsink... and the aftermarket heatsink improved the situation drastically.

If my airflow were bad, then adding the aftermarket heatsink would have been detrimental to my cooling, since all the heat from the graphics card is now being dumped into the case rather than out the back.

My airflow is obviously just fine, the stock heatsink simply didn't seem to work very well.
Click to expand...

Even in crossfire I only see temps in the 80s on auto, which appears to set the fan speed no higher than 45% or so. Like I said, at 100% the temps barely exceed 60 in furmark, let alone normal games. There's nothing unusual about this setup, these are just the temperatures I've come to expect,. Sure, on auto most cards run in the high 80s, but at 100% that's very rare for stock coolers, even that on the 4850.
 
samuelmorris said:
Then it must have been faulty. At 100% fan speed the HD4850s ran reasonably cool, the problem AMD had was that the default profiles let the cards get very hot because they did not force the fan speed up enough. There's probably something seriously wrong with your case airflow if you're getting temps in the 80s with 100% fan speed on a 6970.
Click to expand...

My ambient room temperature is above average, which is one problem. My case's airflow could be better, but it's not completely lacking or anything. In any case, even if the noise wasn't an issue and I was willing to run my fan at 100%, my card would still run too hot. It was true with my 4850 and it's true with my 6970. With an aftermarket cooler however, there's no issues.
 
samuelmorris said:
Even in crossfire I only see temps in the 80s on auto, which appears to set the fan speed no higher than 45% or so.
Click to expand...
If you had taken the time to look up the fan ramp on the HD 6970, you'd know it will automatically run the fan at 100% if the card reaches 102c. This is a BIOS override; the card will blast up to 100% fan speed if it hits 102c even if you've forced a lower speed in CCC.

I was hitting that mark with the stock cooler, the fan continuously bounced up to 100% as the load temperature flirted with 102c. Using an aftermarket cooler, I can't get temperatures to go over 76c no matter what I do. In my case, the aftermarket cooler (even with its slower, silent fans) seems to be doing better than the deafening stock cooler at 100% fan speed.

Nobody else in this thread seems to know how you're making the stock cooler work as well as you say it's working...
 
I am aware of the fan override at 102C. I have manually triggered it before to check it works (by forcing the fan speed to 24% under load).
The only issue people have taken with my temperatures was that of how low the temperature drops when forced to 100%. I'm pretty sure most people in here would take issue with your setup if it reaches the 100C mark under auto fan speed. Check the wide spectrum of reviews for the 6970, both by itself and in crossfire, and try and find one example where they get anywhere near those temperatures. I can't think of one.
 
Hey Unknown, which accelero model is supposed to be the best (for the 6970)? Assuming you have no issues with space (I only have the one 6970)...

I had tentatively decided on the Shaman. I was under the impression that it cools slightly better than the accelero series and is cheaper than the $52 you paid for yours, but I'm wondering if I compared the Shaman to an accelero model that wasn't top of the line...
 
samuelmorris said:
Check the wide spectrum of reviews for the 6970, both by itself and in crossfire, and try and find one example where they get anywhere near those temperatures. I can't think of one.
Click to expand...
A telling review is right under your nose, mate...

[H]'s own review shows the HD 6970 reaching 90c, with a single card, in [H]'s well-cooled test bench.

In a normal case with a second graphics card just below the HD 6970 (like the GTX 260 sitting below mine), a 10c increase sounds easily doable.

shantd said:
Hey Unknown, which accelero model is supposed to be the best (for the 6970)? Assuming you have no issues with space (I only have the one 6970)...

I had tentatively decided on the Shaman. I was under the impression that it cools slightly better than the accelero series and is cheaper than the $52 you paid for yours, but I'm wondering if I compared the Shaman to an accelero model that wasn't top of the line...
Click to expand...
I had to go with the Accelero Twin Turbo II due to space constraints. I have cards below my HD 6970 that prevent me from using the Shaman, and my drive cage prevents me from using longer coolers like the Accelero EXTREME Plus II.

The tri-fan Accelero EXTREME Plus II would be the best heatsink from Arctic Cooling for that card, and it comes with all the RAM and VRM heatsinks you'll need. You'll see even better temps than I'm getting with its dual-fan little brother.

The Thermalright Shaman is, hands-down, the best. It consistently scores the coolest temperatures and the quietest operation. The drawback is that you lose access to almost every slot on your motherboard because the thing is so gigantic. You'll also have to figure out your own VRM cooling solution.
 
Last edited:
Unknown-One said:
A telling review is right under your nose, mate...

[H]'s own review shows the HD 6970 reaching 90c, with a single card, in [H]'s well-cooled test bench.

In a normal case with a second graphics card just below the HD 6970 (like the GTX 260 sitting below mine), a 10c increase sounds easily doable.
Click to expand...

But I'm pretty sure this doesn't happen.
Test benches are very variable as well, as while they don't have any confinement, they don't have any ambient airflow either.
 
You asked for a review where a 6970 got close to 100c, I provided one from the same site you're currently posting on. You obvious haven't looked very hard.

You also have a good number of people in this thread scratching their heads over your oddly low temperatures... Something on your end just doesn't seem right.
 
[H] show the HD6970 reaching 89C at load, with a low fan speed, which it does do. This isn't 100C, at full fan speed, nowhere near it.
Additionally, as
http://www.hardocp.com/article/2011/03/07/amd_radeon_hd_6990_antilles_video_card_review/8
proves, in crossfire, the cards only reach 90C, and again, at auto fan speeds, not maximum - which concurs with what I have said. I claimed auto temps 'in the 80s' which is not distant from 90C exact - allow a few degrees change in ambient temperature and case ventilation and that's easily doable.
It is your temps that are off, not mine.
 
You must log in or register to reply here.
Back
Top