Once the card reaches around 83°C, the clocks level out at around 1705 MHz, which is still 100 MHz higher than the base clock.
^^^^still higher than base clock, and those few mhz make basically no difference in the real world.
Here is one of my GTX 1070 machines as an example:
ETH:
GPU0 26.668 Mh/s - GPU0 t=82C fan=96%
GPU1 26.694 Mh/s - GPU1 t=74C fan=81%
GPU2 26.685 Mh/s - GPU2 t=69C fan=64%
GPU3 26.673 Mh/s - GPU3 t=82C fan=96%
GPU4 26.685 Mh/s - GPU4 t=81C fan=93%
All cards are within 0.03Mh or so......does that really look like a bunch of thermal throttling? I can post hundreds of other miner status lists showing the same thing.
As noted in the second part of the review, modern Nvidia cards begin to progressively throttle in 13 MHz increments as low 40C. Once the temperature exceeds 82C, there was a drop to the base clock. Which means modern cards DO in fact progressively throttle the core frequency and voltage according to the temperature, not just at the thermal limit cutoff.
We can see a linear trend that has clocks go down as the temperature increases, in steps of 13 MHz, which is the clock generator's granularity. Once the card exceeds 82°C (I had to stop the fan manually to do that), the card will drop all the way down to its base clock, but will never go below that guaranteed minimum (until 95°C where thermal trip will kick in).
This means that for the first time in GPU history, lower temperatures directly translate into more performance - at any temperature point and not only in the high 80s. I just hope that this will not tempt custom board manufacturers to go for ultra-low temperatures while ignoring fan noise.
Ethash is a memory dependent algorithm which is not affected as much by by a lower core clock. Especially given that for mining ETH, you should UNDERCLOCK the core to LESS than the default clock. For other core intensive algorithms such as Equihash, where you need to INCREASE the core overclock, the thermal throttling would affect the hash rate much more.
