Gtx 1070 OVERCLOCKING temps?

leclerc_maxime

Prominent
Aug 22, 2017
19
0
510
Hi there I am overclocking my 1070 to +175 mhz on the clock what are maximum safe temps for the card on load
Can i hit a max of 80c on load after an hour of gaming and call it fine? Also would adding a case fan be more effective at cooling the gpu (taking noise into account) than turning the gpu fans up?
And finnally what dan speed should i leave my gpu at since im currently at 50% and temps max at 80c
 
Solution


here's a review of the 1080 Zotac Mini and as you can see, it does quite well, far below the 82C throttling point, even OC'd. That's 15C below the trottling point., if the 1080 does that well, the 1070 should do better

https://www.techpowerup.com/reviews/Zotac/GeForce_GTX_1080_Mini/33.html

temp.png


Id say your problem lies with case cooling

Assuming 175 watt GFX, 115 watt CPU, 35 MoBo and 40 rest, that's about 375 watts

at 50 -75 watts per fan for 1250 rpm 120mm fans, that's 5 case fans. say 4 if they 140mm

1. All FE cards throttle. Some cards (EVGA SC) are essentially FE cards w/ same PCB non blower style cooler. Other FE cards so much better .... so if ya name the card, you can determine what temp to expect.

2. Not knowing what card ya have makes this harder ... but here's settings for the MSI Gaming X

http://www.guru3d.com/articles_pages/msi_geforce_gtx_1070_gaming_x_review,30.html

Core Voltage : +100%
Power Limit : 120 %
Temp. limit : 92C
Core Clock : + 105
Memory Clock : +700 MHz (=9400 MHz effective data-rate)

As we can see here, the MSI Gaming card hits 72C with max OC. The card throttles if temps exceed 82C.

https://www.techpowerup.com/reviews/MSI/GTX_1070_Gaming_X/28.html

3. Our rule of thumb for case cooling design is based upon system / case fans being inaudible (< 850 rpm). To do this:

One (1) 120mm case fan for each 50 -75 watts of heat being generated
One (1) 140mm case fan for each 75 -100 watts of heat being generated which should keep you 10C below the throttling point.

General idea is to select a good AIB card and provide proper case cooling. Why pay for a higher end card wand then let its performance get gimped by throttling ?

4. Most of my builds of late have been water cooled, custom loops or open loop type CLCs paired w/ GFX cards having pre-installed water blocks (MSI Seahwak EK X). However, with a decent AIB air cooled card, and proper cooling fan set up as per above, have never had a reason to change default fan profile.
 
Throttling is generally defined as not being able to hit advertised frequencies, not just base core frequency

https://www.techpowerup.com/reviews/MSI/GTX_1080_Gaming_X/28.html

Idle temperatures are excellent even with the fans turning off in idle. During gaming, the card also runs much cooler than the reference design, which avoids clock throttling above 82°C.

https://videocardz.com/60838/msi-geforce-gtx-1080-gaming-x-is-much-better-than-founders-edition

NVIDIA’s own reference design suffers from severe throttling just after few minutes. It probably wouldn’t be that bad if not the frequency spikes. While average clock is somewhere around officially stated boost clock, those spikes cause micro-stuttering, which negatively affects gaming experience.

GTX-1080-FE-clocks-over-time.png
 


here's a review of the 1080 Zotac Mini and as you can see, it does quite well, far below the 82C throttling point, even OC'd. That's 15C below the trottling point., if the 1080 does that well, the 1070 should do better

https://www.techpowerup.com/reviews/Zotac/GeForce_GTX_1080_Mini/33.html

temp.png


Id say your problem lies with case cooling

Assuming 175 watt GFX, 115 watt CPU, 35 MoBo and 40 rest, that's about 375 watts

at 50 -75 watts per fan for 1250 rpm 120mm fans, that's 5 case fans. say 4 if they 140mm

 
Solution
Yeah all that graph shows us that GPU boost assigns different clocks to certain voltage points based on load and tmep/ volt headroom. Exactly what it's designed to do and isn't related to classic thermal 'throttling'

Op - spend more time working on sufficient case cooling as regardless of your set of offset, GPU boost will do what it wants within preset boundaries. Decent temps and a good offset is a good approach. Whacking the of offset up isn't always the answer with boost 3.0
 
I gave you two links which say otherwise. Could add dozens more if ya like. If your performance is impacted by heat, it's throttling no matter how ya slice it.

Put it this way.... what card do you want ?

GTX-1080-FE-clocks-over-time.png


The one that sits at 1920 or the one that drops to 1680. Clearly, as was quoted above, that author describe this phenomenon as "throttling". So does TPU ... so does everyone one else.

http://www.anandtech.com/show/10325/the-nvidia-geforce-gtx-1080-and-1070-founders-edition-review/15

The "guaranteed minimum" is different from "dynamic clock". Dynamic clocking has been a feature of nVidia cards since the 600 series. But simply put if any device sees an increase in performance when you can lower temps, it's being thermally controlled ... hence "thermal throttling".

 
Ok,well we can mess with semantics. And yes, I understand the term, some of us on here are legit engineers without the need to put 'PE' after our names on a forum.

My point was more that to think of it is as throttling is a bit harsh - in the sense that even under good temps clocks will be dropped. Different to the more classic 'reducing clocks at Tmax'. This is more 'reducing clocks in line with a pre-defined thermal curve even when well under Tmax.