Very long 22+ page article - nVidia GTX 480 (825/1100 MHZ) vs. ATI HD 5870 (975/1300 MHz) at http://alienbabeltech.com/main/?p=16997
- ATi Catalyst 10-3a; highest quality mip-mapping set in the driver, Catalyst AI set to “Standard”
- NVIDIA GeForce 197.13 WHQL for GTX 280 and 197.17 beta release drivers for GTX 480; High Quality
- Windows 7 64-bit; very latest updates
- DirectX February 2010
Overclocking and Temperatures
Each of our competing video cards was overclocked as far as it could go without exceeding the core voltage or thermals so as to throttle the respective GPUs. We got 100% stability from each of our GPUs at the noise-expense of running their cooling fans near 90%.
We used MSI’s Afterburner to overclock our reference Diamond HD 5870 to 975/1300 MHz – a solid overclock of +125 MHz on the core and +100 MHz on the vRAM. Correspondingly, we overclocked the GTX 480 with EVGA’s tool to 825/1100 from 700/924 MHz – another solid overclock which was coincidentally also +125 MHz on the core; however, we managed +176MHz on the GTX 480’s vRAM. We did not attempt to adjust the GTX 480 shader clock which would give us a further performance increase. Here is the EVGA overclock tool showing stock clocks; note the fan speed.
OC ToolSTOCK GTX 480 (825/1100 MHZ) vs. HD 5870 (975/1300 MHz), Overclocked Performance Analysis, Part 2
Now we see our maximum overclock and note the fan speed is now set at 90% as we finally settled on 1650/2200 MHz (825/1100 MHz) with stock voltage. We had to do the same thing with HD 5870’s fan when we overclocked it to 975/1300 MHz and upped its core voltage to 1.35V. The fan noise of either video card is intolerable and we might suggest water cooling or a better aftermarket air cooler if you are going to push your GPU to a near-extreme overclock.
OCD GTX1 GTX 480 (825/1100 MHZ) vs. HD 5870 (975/1300 MHz), Overclocked Performance Analysis, Part 2
Temperatures remained well within the each card’s specifications and no throttling was noted. Both fans were set at 90% each to insure thermal stability and most of the testing proceeded in a warm room with ambient temperatures of 76-80 F; rather warm. Each card was stable at their respective overclock and no issues were noted with either card.
Both cards are very noisy with fans set at 90%; it would be intolerable under gaming conditions without headphones. However, we wanted to test a near-maximum overclock of each card. The Diamond HD 5870 actually overclocked further than our PowerColor HD 5870 PCS+ (1325/1300 MHz) because the PowerColor is a non-reference design with a shorter and wider PCB than the reference, but it also lacks voltage adjustments.
We eventually settled on 1.35V for the Diamond HD 5870; the highest voltage allowed by the MSI Afterburner overclocking tool; we could not pass beyond 975 MHz on the core, but it is still a good overclock over the stock core clock of 850 MHz. 1300 MHz seems to be the upper limit of the GDDR5 that is used in both of our HD 5870s.
No voltage adjustments were done on our GTX 480 as it already ran pretty warm with stock voltage. You could probably push it further with a better cooling solution than the reference fan but it still managed a nice overclock from 700 MHz on the core to 825 MHz. The GTX 480’s vRAM was able to overclock from 924 MHz to 1100 MHz; not bad!
Make sure you check out the revised Power Usage section later on in this same review which revisits and explains how NVIDIA measures the GTX 470/480 TDP and perhaps explains the apparent discrepancy between the published specification of 250 W “maximum board power” and what we and other reviewers observed and measured.
Ok, let’s look at our test bed for both of these overclocked cards.
hmmm could be a bit cpu limited with Intel Core i7 920 reference 2.66 GHz and overclocked to 3.8 GHz); Turbo (21X multiplier for 3.97 GHz of a single core) though.