• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

GTX 680 VRAM cooling

Associate
Joined
6 Aug 2011
Posts
480
Hi guys,

How good is the VRAM cooling on GTX680 cards? They are cooled with the plastic cover (under heat sink) contacted via thermo pads

The question: is the reference GTX 680 VRAM cooling efficient? I felt like it might have been better if just use air cooling instead of covering the chips with some plastic which is bad heat transferring material.


some observations:
KFA2 EX OC - cooled with air (fan can blow the air through)
Gigabyte Windforce - cooled with main heat sink (sharing with GPU)
Reference 670 - Not sure about the front VRAM but the back ones are left at the mercy of case air flow.

There is a review showing the VRAM of reference GTX 670 reached 80-90 degrees.
 
http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_680/images/cooler2.jpg

lol the reference 680 has better vram cooling than most direct cuii coolers from asus, which leave vram cooling entirely up to air and fate. blackwhite is right, that's probably aluminium with aluminium fins above it. there's no way that's plastic, would make no sense. but even if it were it would technically be better than air
 
Last edited:
Asus DCU II has aluminium heatsink as well on VRM's , and it well cooled, as it has direct blow of two fans on it.
Without heatsink card will fried up after couple of seconds.
 
not vrm but vram. http://www.techpowerup.com/reviews/ASUS/GeForce_GTX_670_Direct_Cu_II/4.html
the 7970 direct cuii v1 has vram heatsinks but have contact issues on those according to many reports. while accordng to reports, the gtx 680, like the 670, has no dedicated vram heatsinks. not all cards have vram heatsinks anyway, but i've heard vram stability and overclocking is a concern with asus dcuii cards for the reasons above
 
Cooling VRAM isn't much of an issue these days. I wouldn't worry about it, just concentrate on GPU and MOSFET cooling.
 
"Hynix spec says the memory can run at up to 115C max operating temp. So at 82C, you’re nowhere near the max temp the modules are rated for. If the modules were overheating you’d be seeing artifacts and other stability issues." - NVIDIA PR

Also taken from a test bench with no airflow.
 
Also taken from a test bench with no airflow.

I kinda think an open system actually will have better air cooling then an enclosed case.

Yes a case will have some air flow but it's not like blowing directly toward the VRAM chips. The heat will accumulate and eventually an equilibrium is reached.

As for the open system, convection will do the work - hot air will rise and disperses the heat into the environment.


Not arguing, just some thoughts....
 
a closed case where air is blown to and from the graphics card would do better than a situation where the air simply dissipates the heat through convection. it's like how you feel colder when air is blown towards you. you're having the heat taken away by the air.

as for a closed case, you'll always have the same amount of air going in and out of the case at any one time so i don't buy this whole "heat will accumulate" thing.
 
a closed case where air is blown to and from the graphics card would do better than a situation where the air simply dissipates the heat through convection. it's like how you feel colder when air is blown towards you. you're having the heat taken away by the air.

as for a closed case, you'll always have the same amount of air going in and out of the case at any one time so i don't buy this whole "heat will accumulate" thing.

Again, not trying to argue, just some thoughts to add into this conversation.

I agree with your on the first point and the example of one feeling colder in a windy condition.

Though, I don't think the case is a perfect air duct so there will be local vortices and residual air. At least not my case I guess...
 
a closed case where air is blown to and from the graphics card would do better than a situation where the air simply dissipates the heat through convection. it's like how you feel colder when air is blown towards you. you're having the heat taken away by the air.

as for a closed case, you'll always have the same amount of air going in and out of the case at any one time so i don't buy this whole "heat will accumulate" thing.

Most case/ fan set ups do not generally allow for that much air flow around the graphics card so I would say most GPU's would run cooler on a test bench where any hot air will immediately rise away (as the gpu will be orientated on its side) rather than a typical tower case where hot air can accumulate under the GPU with little or no airflow to cool it.
 
See this review http://www.legitreviews.com/article/1925/14/

The VRAM chips reached 80C and other part reached even 99C. This is way I'm concerned.

back-temp.jpg

memory-temp.jpg

Armchair engineers are awesome:

Armchair Engineer: "Nvidia, uhm, your card gets warm, thats bad!"

Nvidia: "Yes they get warm, as thats how we've designed them, as its well within spec."

Armchair Engineer: "That's bad though."

Nvidia: "Why?"

Armchair Engineer: "Dunno, Google says so."

Nvidia: "Ok, it must be true then."

Armchair Engineer: "F U Nvidia I'm going to google some more to prove you wrong."
 
Armchair engineers are awesome:

Armchair Engineer: "Nvidia, uhm, your card gets warm, thats bad!"

Nvidia: "Yes they get warm, as thats how we've designed them, as its well within spec."

Armchair Engineer: "That's bad though."

Nvidia: "Why?"

Armchair Engineer: "Dunno, Google says so."

Nvidia: "Ok, it must be true then."

Armchair Engineer: "F U Nvidia I'm going to google some more to prove you wrong."

I have to say I am an armchair engineer. But I did have some experience designing computer and PCB (non commercial and not on the scale of today's high end graphic cards).

I think this is Hynix's specs: http://www.skhynix.com/inc/pdfDownload.jsp?path=/upload/products/gl/products/dram/down/GDDR.pdf

It doesn't list the exact number for commercial temp but
1. Other docs (eg DDR3 specs) on hynix's website indicate 70C and 85C
2. The industrial temp is 85C and we know it's always in the order of military > industrial > commercial.

81C is very close to 85C. From an engineering point of view, I would like too see more headroom here.

I could be wrong, please let me know.
 
I have to say I am an armchair engineer. But I did have some experience designing computer and PCB (non commercial and not on the scale of today's high end graphic cards).

I think this is Hynix's specs: http://www.skhynix.com/inc/pdfDownload.jsp?path=/upload/products/gl/products/dram/down/GDDR.pdf

It doesn't list the exact number for commercial temp but
1. Other docs (eg DDR3 specs) on hynix's website indicate 70C and 85C
2. The industrial temp is 85C and we know it's always in the order of military > industrial > commercial.

81C is very close to 85C. From an engineering point of view, I would like too see more headroom here.

I could be wrong, please let me know.

You are typical of most armchair engineers, where you Google docs or conversations you don't quite fully understand but you manage to glean one piece of information that sounds bad to you, so you repeat it on the internet only for others to do the same and rumours quickly spread.

Ddr5 is not ddr3, so you cannot use specs from one and other, simple as that.
 
You are typical of most armchair engineers, where you Google docs or conversations you don't quite fully understand but you manage to glean one piece of information that sounds bad to you, so you repeat it on the internet only for others to do the same and rumours quickly spread.

Ddr5 is not ddr3, so you cannot use specs from one and other, simple as that.

I heartily accept your criticism. Though it is not my intention to repeat on the internet nor making up rumours. I'm only trying to understand the matter and discuss with others here.

Re DDR3 vs DDR5, I probably wasn't clear. It's the GDDR part numbering decode specs. So it should be the correct specs to use.
 
Last edited:
I heartily accept your criticism. Though it is not my intention to repeat on the internet nor making up rumours. I'm only trying to understand the matter and discuss with others here.

Re DDR3 vs DDR5, I probably wasn't clear. It's the GDDR part numbering decode specs. So it should be the correct specs to use.

Let me put it another way, GDDR5 has been used under heatsinks that have a GPU heating the same sink up to 80-90C+ for years, and we've been fine.

You'll be fine, and if you aren't, that's what warranties are for.
 
Back
Top Bottom