• 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.

1080 Ti throttling issues

geordiejc

geordiejc

geordiejc

Associate
Joined
1 Nov 2013
Posts
860
So according to a youtuber (AdoredTV) + tomshardware and computer base there appears to be a thermal throttling issue on the 1080.

Here we see the 1080 hitting 81- 85 degrees at the 4 minute mark.

screenshot_61.png


At the 4 minute mark (in blue) we see the clock speed reducing to ~1650

screenshot_61.png


This table shows the clock after 20 mins.

screenshot_61.png


This is indicative that the cooler is pants ! forget overclocking this card if it cant maintain basic boost due to thermals.
 
P8

Already posted several times over in other threads. The issue is the fan profile doesn't ramp IP enough. Nvidia probably thought they could get away with a quieter fan profile than the 980ti since the 10ip uses decently less power but the cooler design is just poor it can't keep up with the lower fan RPM.
 
D.P. said:
Already posted several times over in other threads. The issue is the fan profile doesn't ramp IP enough. Nvidia probably thought they could get away with a quieter fan profile than the 980ti since the 10ip uses decently less power but the cooler design is just poor it can't keep up with the lower fan RPM.
Click to expand...

Kaap won't admit it because he thinks physics isn't you know, a thing, but this chip is 320mm^2 and uses 180W, what 220W or so at it's max power setting. The 980ti is a 610mm^2 chip with a 250W tdp. The 1080 has drastically higher W/mm^2 than the 980ti. That means the 980ti has W/mm^2 output of 0.401W per mm^2 while the 1080 is 0.563W per mm^2(using 180W stock figure), or just over 40% higher heat output for it's size. Temp and cooling ability end up defined by surface area and ability to transfer heat. Double the power but also the size of the chip and it's ability to transfer that heat is also massively improved. Also the temp of the cold side and the hot side dictates how well temp can transfer. A hotter chip will transfer heat more quickly than a colder chip if the heatsink is the same temp in both scenarios.

Basically for a 1080 to maintain the same temp as a 980ti due to having a much higher W/mm^2 output it needs a lower heatsink temp which means better heatsink or faster fan.
 
Last edited:
drunkenmaster said:
Kaap won't admit it because he thinks physics isn't you know, a thing, but this chip is 320mm^2 and uses 180W, what 220W or so at it's max power setting. The 980ti is a 610mm^2 chip with a 250W tdp. The 1080 has drastically higher W/mm^2 than the 980ti. That means the 980ti has W/mm^2 output of 0.401W per mm^2 while the 1080 is 0.563W per mm^2(using 180W stock figure), or just over 40% higher heat output for it's size. Temp and cooling ability end up defined by surface area and ability to transfer heat. Double the power but also the size of the chip and it's ability to transfer that heat is also massively improved. Also the temp of the cold side and the hot side dictates how well temp can transfer. A hotter chip will transfer heat more quickly than a colder chip if the heatsink is the same temp in both scenarios.

Basically for a 1080 to maintain the same temp as a 980ti due to having a much higher W/mm^2 output it needs a lower heatsink temp which means better heatsink or faster fan.
Click to expand...

That is definitely true and part of the conundrum of going to smaller processes nodes and smaller chips.

I'm sure that is why GP100 is so big, it's never going to be a consumer GPU, so with massive margins you can worry less about yields and chips per wafer. Instead a big chip with lower density improves electrical and thermal issues, which is Important for an industrial applications where it could be running at 100% 24-7-365.
 
Maybe people are finally starting to realize that turbo speed is, well, just a turbo. If they could sustain it, they'd call it the base.
 
Lets see how people get on with the actual retail cards shall we? Blimey I think this issue can be resolved with driver update to change the fan profile. Temp of 83-85 I don't think is an issue as the last few cards I've owned have maxed out around this(780 TI SC for example, 690) - just need enough cooling to allow the boost to run higher permanently. Cards not even released yet and bound to be new drivers available on day 1...or 2 :D yet people going on and on about it.

Oh,and the turbo maximum is never guaranteed anyway as h4rmony points out. Maybe someone should do a review showing benchmarks of other cards after an hour of gameplay (or benching), that would be interesting to se and more "real life". Really, that's an issue with all of these reviewers! Heat soak can make a large difference yet they probably boot up the machine from cold then straight into a benchmark for 5-10 mins.
 
Last edited:
Sweet, another thread. Question to the OP. Do other reference cards throttle? Do non ref cards throttle? What was the fan curve? What fan speeds were they throttling at?

In all my time using reference cards, they have all throttled, so just wondering why this is something a lot of people are worrying about?
 
I think the issue being Greg is that they are throttling to below the advertised boost clock. I can't speak from experience because anytime I have used a reference card I have always made a more aggresive fan profile, but have the reference Titan X's for example throttled to below their advertised boost clock on a regular basis using stock profile?

If so, then I guess it is a bit of a non story, but it is nonetheless disappointing that Nvidia is releasing a product that is unable to maintain at least the advertised boost clocks for 20 minutes!
 
You must log in or register to reply here.
Back
Top Bottom