Cold versus Hot R9 290X Results
The first set of results are going to showcase the performance differences between the "cold" and "hot" runs of Crysis 3 at each selected fan speed. Through these graphs, you'll see noticeable performance deltas (between the two runs) at the lower fan speeds and minimal deltas at the higher fan speeds.
With the maximum fan speed set at 20%, the R9 290X shows a very significant performance difference between the card at initial start up and after just a handful of minutes in the game. Average frame rates go from ~29 to ~24 which is equal to more than 20%. Clearly the game play experience would be affected.
At a 30% maximum fan speed setting, there are still very dramatic performance drops from the first run in Crysis 3 to the second run. We see basically the same 20% drop in average frame rate.
At 40% maximum fan speed we are looking at the out of box results you would get if you bought an R9 290X today. While clearly more consistent than the 20% or 30% results above, there are still performance differences between runs. Going from an average frame rate of 29 FPS to 26 FPS is a drop of 10% and, if you look at the Frame Times plot, you can see that in fact the delta between Run 1 and Run 2 actually expands over the 60 second window.
A 50% fan speed is ALMOST what you have with the 290X in the "Uber" fan mode and we clearly see a big advantage here. Performance in Run 1 is nearly identical to the performance in Run 2 with very little change in average frame rate and no change in the Frame Times graph.
At 60% we see no change from the 50% results above; moving the maximum fan speed up another 10% did us no good for expanding performance.
Alright, call me crazy, but
Alright, call me crazy, but is there anything at all scientific about setting a fan speed at 40%?? Not one, just simplicity. How about setting the fan RPMs or the fan decibels for something actually performance related. 40% is arbitray between two different fans. The size and blade pitch could dictate wildly differing CFM values.
Regardless, its a pretty crazy comparison, when its clear AMD put almost zero effort into the reference cooler. They planned to give consumers a good value, not the best of everything. Its loud, we all get it, but how about an actual useful review of someone strapping a water block on it, or another cooler? How many first adopters of $500 hardware leave well enough alone? If you buy 290X cards you know what you are buying and likely have water cooling, so why pay more money for a great cooler you will take off?
In this way it is also configurable performance in FPS, temps and sound levels.
I am not sure what to make of
I am not sure what to make of my R9 290 to be honest. My waterblock arrives tomorrow so my opinions are likely to change, however from the testing that I have done with this card over the last two weeks I can say that overclocking is totally pointless with stock cooling. My average clocks in BF4 with fan on auto is around 820mhz with a low of 720mhz… I quickly hit 94C. The latest 13.11 beta 9.4 catalyst drivers have no performance tab on my test machine (known bug), so I’m not bothering to test with power limit increased until this is fixed and my waterblock is installed.
I am sure that once adequately cooled this problem will disappear.
Still I determine that AMD should be blamed for shoddy marketing and a lousy stock HSF.
They should have been honest enough to state that the clocks on this card are 800mhz boosted to 950mhz (depending on heat and power)…. Just my opinion.
I sold my GTX 780 for a stock R9 290 and believe that this was a good deal. I got $500 for my GTX 780 and this allowed me to buy the R9 290 and a new EK Waterblock. Fully cooled I will be able to get better performance from the R9 290 and as I only play BF4 I am excited to see what Mantle brings to the table later in December.
Nice review as always.