Message boards :
Graphics cards (GPUs) :
Power usage vs TDP and Fermi vs Kepler power efficiency
Message board moderation
| Author | Message |
|---|---|
skgivenSend message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I did a bit of testing a year or so ago and found that most GPU's used around 75% of the TDP when crunching here. Although I had read in a review that the GF600's were different I was still slightly surprised to see a GTX660Ti purporting (Gigabyte OC Guru and GPUZ) to use 93% of it's 150W reference TDP when crunching. It's at 57°C so temperature wouldn't be an issue. The 1.175V might on the other hand be worth looking at. However, it auto boosts to 1188.9MHz which is 21.3% over the 980MHz reference boost. When you factor that in it's almost spot on that 75% guide-mark at 76%. I'm not really surprised at that as modest OC's usually improve performance fairly linearly compared to the Wattage of the GPU. How much of the TDP are your GPU's using? Just for reference, my GTX660Ti on W7x64 is 30% faster than my GTX470 (FOC) on XP and in terms of efficiency it's 50% more efficient in terms of throughput/Watt. FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help |
|
Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
GTX660Ti + Win7 64, using 69% of TDP at 1254 MHz, 1.175 V. Apparently it's got about a 15% higher TDP than stock cards, though. MrS Scanning for our furry friends since Jan 2002 |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
My Gigabyte GTX 670 OC (@1150MHz, 1.175V, 3 fans are running at 70%, 3600RPM, GPU temp is 55°C; +1 core of a Core i7-970 @ 4.16GHz) is using 150-160W (measured at the 230V AC outlet, the PSU is and Enermax Maxrevo 1500W), according to MSI Afterburner this GPU is using 75% power. The good half of my GTX 690 (@1019-1045MHz, 1.137-1.150V, fan is running at 83%, 2600RPM, GPU temp is 65°C; same config) is using 130-140W, MSI Afterburner shows 93-102% power usage. The GPU usage is 98% on both cards. |
skgivenSend message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
MrS, it seems your system gets through tasks around 3.5% faster than my similar rig. 4.2GHz/1188 (now varies from 1175 to 1201). Differences explained by clocks. Was 290W at the wall when crunching 6 CPU tasks and 1 GPU task. I added a 470 @656MHz, and the 660Ti is 33.9% faster. I didn't see any performance drop when moving from PCIE3.0 x16, to what I presume is PCIE3.0 x8 (for the 660Ti) and PCIE2.0 x8 for the 470. 96% and 97% GPU utilization. 43% and 27% memory controller loads. Idle is 90W at the socket, and crunching (6 CPU + 2 GPU) is 450W. Corsair HX750 PSU (@91% efficiency). FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help |
|
Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Wow, that are some consistent numbers! My idle power draw is 60 W, load is ~215 W with no BOINC project running besides GPU-Grid. MrS Scanning for our furry friends since Jan 2002 |
©2025 Universitat Pompeu Fabra