Hello, I recently purchased a GTX 1080ti Gigabyte Aorus Extreme edition. I decided to play around with it and see how it overclocks. I should probably mention now that I'm fairly new to overclocking and I'm learning and getting experience as I go. Not far in to my overclocking endevaor i ran in to some suspicious readings in MS Afterburner.
At default, stock settings the GPU performed quite normal. The clock was fluctuating in the 1800Mhz and 1900Mhz, usage and power appeared to be normal and the temperature was in the range of 70-75c.
Since there was some headroom in the temp department I decided to do a slight overclock and raise the power limit from 100% to 120% out of the available 150% for the Aorus card. After running the same benchmarks again (as I did at the start to establish the baseline readings) I started to notice some weird behavior. After a few minutes the power % went from 100% to about 70%-80%, the temps also went down to mid 60s, but the core clock spiked up the maximum of 1999Mhz with voltage doing the same ramping up to 1,080v (while before it stayed below 1v or slightly above it), both staying there indefinetly. There was no FPS loss and the usage seemed fine also. The PSU i have is 750w Corsair RMi series which is sufficient for the GPU.
I did some further experimenting with power limit and voltage, discovering the power drop started to occur after the total power % was raised beyond 105%. The benchmarks i used for testing were Valley and 3D mark Time Spy and Fire Strike. I also decided to ran some tests with different OC profiles on my CPU(7700k) in Realbench, discovering the CPU voltage and frequnecy were irrelevant in this case. I did however found out that the GPU power drop to 70% also occured in Realbench when the default, stock settings of 100% were in place. It was the same when i started testing games -with World of Warcraf- straight to 70% with clock and voltage spiking up and the temps going down to 60s. If i increased the power limit under this circumstance the power would drop even lover in to the 40s.
Today before writing this post, I decided to record some sample logs in MS Afterburner of this occurrence, but when I tried to get a log of the increased power limit in Firestrike, It crashed after a couple of minutes and the clocks locked down to 200Mhz (yikes!) which was fixed with a simple reset, but it gave me some scare nonethelss since that never happened before under those settings.
In conclusion. My suspicion to what would be the cause of this problem would be some sort of power throttling. GPU rolling back the power, due to high temperatures somewhere on the PCB. Unlikely that it is the GPU core since the temps do not raise beyond 70s, but maybe perhaps on the VRM or power delivery side? I haven't yet done any manual thermal testing with thermocouples, I know about it but i have never done it myself on any component, so there would be a bit of a learning curve before i dive in to it.
Other than above mention theory, I'm kinda clueless to what would be causing this issue otherwise, so I'm very much interested and in the need of outside help from people with more knowledge and experience on the matter. The card is still under warranty and i can have it RMA'd if the issue is beyond my ability to resolve it.
I will really appreciate all the feedback you can provide me on this matter, thanks.
At default, stock settings the GPU performed quite normal. The clock was fluctuating in the 1800Mhz and 1900Mhz, usage and power appeared to be normal and the temperature was in the range of 70-75c.
Since there was some headroom in the temp department I decided to do a slight overclock and raise the power limit from 100% to 120% out of the available 150% for the Aorus card. After running the same benchmarks again (as I did at the start to establish the baseline readings) I started to notice some weird behavior. After a few minutes the power % went from 100% to about 70%-80%, the temps also went down to mid 60s, but the core clock spiked up the maximum of 1999Mhz with voltage doing the same ramping up to 1,080v (while before it stayed below 1v or slightly above it), both staying there indefinetly. There was no FPS loss and the usage seemed fine also. The PSU i have is 750w Corsair RMi series which is sufficient for the GPU.
I did some further experimenting with power limit and voltage, discovering the power drop started to occur after the total power % was raised beyond 105%. The benchmarks i used for testing were Valley and 3D mark Time Spy and Fire Strike. I also decided to ran some tests with different OC profiles on my CPU(7700k) in Realbench, discovering the CPU voltage and frequnecy were irrelevant in this case. I did however found out that the GPU power drop to 70% also occured in Realbench when the default, stock settings of 100% were in place. It was the same when i started testing games -with World of Warcraf- straight to 70% with clock and voltage spiking up and the temps going down to 60s. If i increased the power limit under this circumstance the power would drop even lover in to the 40s.
Today before writing this post, I decided to record some sample logs in MS Afterburner of this occurrence, but when I tried to get a log of the increased power limit in Firestrike, It crashed after a couple of minutes and the clocks locked down to 200Mhz (yikes!) which was fixed with a simple reset, but it gave me some scare nonethelss since that never happened before under those settings.
In conclusion. My suspicion to what would be the cause of this problem would be some sort of power throttling. GPU rolling back the power, due to high temperatures somewhere on the PCB. Unlikely that it is the GPU core since the temps do not raise beyond 70s, but maybe perhaps on the VRM or power delivery side? I haven't yet done any manual thermal testing with thermocouples, I know about it but i have never done it myself on any component, so there would be a bit of a learning curve before i dive in to it.
Other than above mention theory, I'm kinda clueless to what would be causing this issue otherwise, so I'm very much interested and in the need of outside help from people with more knowledge and experience on the matter. The card is still under warranty and i can have it RMA'd if the issue is beyond my ability to resolve it.
I will really appreciate all the feedback you can provide me on this matter, thanks.
