The card is like 25% slower on all tests than similar cards with the same gpu. It runs very hot on full load (100C), no OC! and very loud (1 fan). I tested it on a different computer, got the same results. my other computer with hd6850 runs more quiet and gets comparable and even better scores in 3d applications.
Thermal and Power Specs:
99 CMaximum GPU Temperature (in C)
170 WMaximum Graphics Card Power (W)
500 WMinimum System Power Requirement (W)
Those are the specs. for the Ti operating wise.
The graphics card manual states that "it's better to keep temperature under 105C".
Is there a chance I can return it to manufacturer or get it replaced because of very bad performace for this type of card and overheating?
That's not a good card imo. 100C is not normal. Assume you did a clean install or used driver sweeper ? Otherwise RMA i think.
Clean install, lastest drivers. Not sure what RMA is.
So which video card is on which machine? The main or the 2nd comp. ??? And why did you only mention the brand of the 500w?
I ran it on 650w PSU, but I tested it on 500W one, the results were approximately the same. It's not the problem of the PSU. The brand of the 650W is "Highpower".
What do you mean by "card is 25% slower on all tests than similar cards with the same gpu"??? What similiar cards with same gpu??
I mean in comparison to the tests of 560ti cards I see online.
Is there a software to check if all of the card processing units are working correctly? I suspect there might be a hardware problem with dead units and additional load on working units, therefore certain death for the card.
I left memory clock default.
Surprisingly graphics score with 3D Mark11 at 0.95V, 800/1600Mhz is P3750 (still P650 behind average score), while at 1.00V 822/1650Mhz (manufacturer setting) it's only P3000 which is awfully low for this card, HD6850 got 3450 with manufacturer settings.
Do graphic cards in general have a tendency to display lower scores with higher voltage, can 100C temperature negatively affect performance? Maybe there a problem with cooling system.
High temps are known to degrade the performance of storage and memory controllers. This card shouldn't be 100c to begin with. Check the cooler for anything broken or missing and change the thermal compound. If no improvement it is possible that there is a broken thermal sensor or nvidia screwed up when they made this sample. If no warranty on this card you can try to remove the IHS (integrated heat spreader) from off the gpu and apply new compound on the gpu die then put the ihs back on. I had to do this on an e7200 before.