very casual question about GTX 1070 or similar temperatures

BonzaiDuck

Lifer
Jun 30, 2004
16,333
1,888
126
There are various issues at play in my question: for instance, whether I'd chosen to go "custom-water" or "strictly air-cooling."

I've run all the benchies and stress tests on my system for the graphics card -- a Gigabyte GTX 1070 Mini OC.

Running Furmark at a room ambient of 75F, that card equilibrates to 70C after 10 minutes and doesn't exceed it.

Of course I know these are acceptable temperatures, no less for using one of Giga's Extreme Engine profiles.

Those profiles don't seem to much vary in either voltage or speed. That is, the differences aren't going to make it much hotter or cooler. Only manual tuning would do that.

How good, how "average" etc. are those temperatures? I would think any GTX 1070 without water-cooling would be a decent basis for comparison. If your case cooling strategy is inferior to mine, I'd think my temperatures would be better. But I'm just fishing for an idea of the Furmark results.
 
Last edited:

narsnail

Junior Member
Jul 30, 2010
21
0
61
Mine idles at 30C, and seems to max around 65C. I have a dual fan card, your temps seem pretty normal if not really good for a card with lots of [components] jammed on a small PCB. ID be happy with those temps. I have a huge full tower case with a ton of fans and have a somewhat large overclock so you're def where you should be.

Profanity isn't allowed in the technical forums.
-- stahlhart
 

BonzaiDuck

Lifer
Jun 30, 2004
16,333
1,888
126
Mine idles at 30C, and seems to max around 65C. I have a dual fan card, your temps seem pretty normal if not really good for a card with lots of shit jammed on a small PCB. ID be happy with those temps. I have a huge full tower case with a ton of fans and have a somewhat large overclock so youre def where you should be.

I can only say that Furmark's stress test is about as hot as they come. For gaming -- hold on a minute . . . . OK -- GRID2 is mild, but like I said in another thread, I like to drive stoned. 61C was the max temperature running the Shelby GT through a couple laps in Barcelona.

People will say the 69-year-old has lost connection with reality, but I fuss and fret over minutiae once I've got an idea of something I want to try. I could post pictures later.

The mobo is the Sabertooth Z170 S, case is an old CM Stacker 830. Instead of the Mark I sabertooth, I wanted to build my own duct-plate or duct-panel to work with the CM Crossflow fan as exhaust. The lower part of the duct plate juts up to within an inch of the forward end of the GTX 1070 Mini. I can modify the plate to add a box with two slots: one for the existing 1070, and a slot that has a cover -- available for a second card. Since the duct-plate is maybe 1.5" above the mobo PCB or just clears the top of the RAM modules, it's already sucking air from the 1070's forward exhaust.

It would be an easy thing to do, but I'm not sure it will make any more difference in the 1070's temperatures.

I must be having a second childhood. . . . .
 
Last edited:

bystander36

Diamond Member
Apr 1, 2013
5,154
132
106
My MSI GTX 1070 Armor OC hits as high as 62C, which is likely a result of a decent fan profile. It also is not audible over the sound of my case, which is pretty darn quiet. I'm happy with it.

I'd be happy with 70C on a mini. The noise levels are the only thing that I'd worry about at those temps.
 

BonzaiDuck

Lifer
Jun 30, 2004
16,333
1,888
126
My MSI GTX 1070 Armor OC hits as high as 62C, which is likely a result of a decent fan profile. It also is not audible over the sound of my case, which is pretty darn quiet. I'm happy with it.

I'd be happy with 70C on a mini. The noise levels are the only thing that I'd worry about at those temps.

Can you see the lexan motherboard duct?


Installed%20duct-plate%202.jpg



The airflow to the left and under the heatpipe cooler is tapered, drawing air from the 40mm intake of the I/O plate. [Note that the mobo, cooler and duct of this CM Stacker 830 has been replaced by a Sabertooth Z170 S and TR LG Macho with TR Macho custom accordion duct. But the duct utilizes the surfaces of cards, bottom fins of the heatsink and its own design features to control the airflow.

Now look at the bottom left in the photo. The Stacker has a plastic inner door that will fit 4x 140mm fans. I blocked off the upper half with 140mm-square Lexan plate. You can imagine the graphics card (Mini OC) in the slot position that is obscured by a SuperMicro SATA card in this picture. The lower half of the plastic door is fitted with two LED BitFenix Spectre Pros. The fan on the left forces air down on the motherboard and graphics card, feeding the GPU fan, and sucked out by the barrel fan aligned with the black rectangle in the center.The barrel fan is also sucking air through the vent of the motherboard pan, drawing it from the flair of the lexan duct-plate at the top, and from the 12" motherboard forward side.

I don't know how many fans I can eliminate, but this is quite possible. If I had a vertical lexan panel perpendicular to the duct-plate and jutting up close to the right vertical side of the left-hand fan, all that air will be forced through the card, pushed out the case on the left, and pushed toward the duct exhaust on the right.

This is astounding, though. I've started to play with the Xtreme Tuning Engine interface, to get an idea of how I might overclock the card using Afterburner -- which I'd prefer. The Gaming, Eco and OC presets provide a lot of information, and I started fiddling with the manual "linear" setting. I bumped it up 50 Mhz just for starters, and ran Furmark, followed by Valley Benchmark.

Depending on which test you run, it is variously reported at 1986 and 2020 Mhz respectively peak core speed. Rock solid though, because this is really a minor incremental adjustment without changing the voltage.

Furmark shows about 57C @ 74F ambient, and Valley pushes it to about 60C.