I thought some of you might like another installment in my series of stock v. OC benchmarks, this time focusing on the Sapphire R9 290 Tri-X. I compare it here to an EVGA GTX 780 SC and a reference EVGA GTX 780 Ti, with benchmarks for reference clocks, as-shipped clocks, and overclocked clocks on each card, where applicable.
Note that unlike the GeForce cards, the R9 290 has meaningful voltage adjustment (on the GeForces, extra voltage often results in LOWER clocks due to the aggressive power cap on those cards). That being said, I was not interested in taking the voltage over +.05, due to the extra heat and power use. There likely is more on tap, but you'll have to turn to other reviewers for the extreme Radeon overclocks.
Another two notes on my testing:
(1) Catalyst 14.6 beta was causing me lots of blackscreens at idle. Not good for extended benchmarking sessions, so that got scrapped. Used 14.4 instead, which was perfect.
(2) I did not bench BF4 with Mantle. Treason? No, just being reasonable. First, Mantle does not produce the same quality, in my opinion, but more importantly, it led to significant graphical anomalies, like swimming under ground. It's just not doing the same thing as DX11, so it's not an apples-to-apples comparison. No matter - the R9 290 was awesome in BF4 anyway.
With that out of the way, here are the benchmarks:
3DMark is much better at comparing cards within one family than cards across families or manufacturers. I use it to establish stable overclock settings (along with Crysis 3). Anyway, you can see here that the R9 290 performs well (and scales well).
This game is somewhat CPU-limited at 1080p, and yet the R9 290 pulls way ahead. A sign of things to come with other CPU-limited games???
Again, at 1080p, Hitman is very CPU-limited, but the R9 290 puts up huge numbers, beating even the OC'd 780 Ti. Impressive.
Uh, oh, here's an AMD title and the R9 290 gets cut down to size. Not quite keeping up with the GTX 780. Ironic!
Another AMD title, and another loss for the R9 290 - AMD should stick to making GPUs rather than supporting game titles! By the way, I don't include minimums here because the benchmark doesn't report correct figures for that due to scene changes.
Excellent performance here, going head-to-head with the more expensive GTX 780. Note that at 1080p, the game is CPU-bottlenecked, even on a highly-overclock i7, and yet the R9 290 is pulling ahead even there, other than in minimums. Focus on the 1440p results to get a feel for the actual performance of the cards.
Wow, wow, wow! No artificial Mantle-enhanced performance here. This is pure DX11 domination. Note that this is the single-player game, which avoids lots of inconsistency and CPU-bottlenecking, but this is also BF4, after all, which has been patched, patched, and patched again. I'm hoping DICE does less on the single-player side, but to the extent the engine has been optimized, the newer benches on the R9 290 may not be directly comparable to ones I generated previously for the GeForces.
The difference in this game was so significant that I actually played better in the game - it was that much smoother, especially at 1440p.
[UPDATE:] At stock volts (1.07V), this card was actually quite efficient, about matching the GTX 780. But when overvolting, power ramps up. Just to get to 1100MHz (a 153MHz OC over stock), I had to accept a big increase in power consumption due to the extra voltage (1.12V). The card runs perfectly at 1075MHz without extra voltage, and honestly, that's how I'll run it other than for benching.
----------------------
Overall impressions:
(1) The R9 290 was a watershed card when it was "released" at $400. Now that it's actually available around that price with a great cooler, it is simply awe-inspiring.
(2) The Tri-X is the quietest gaming card I've ever used. Temps never went above 75C even with a big overclock and extra voltage, and yet it wasn't any louder than my case fans.
(3) The R9 290 doesn't use nearly as much power as I thought it would based on reviews - it's actually fairly efficient, as long as you don't touch the voltage.
(4) Performance is fantastic, but it's very game-dependent. When people say "OMG, the R9 290 beats an OC'd 780 Ti!", you have to take it with a grain of salt. Overall, the OC'd R9 290 does in fact match a stock 780 Ti, but so does an OC'd 780. Now, I only took my 290 to 1100MHz (a 16% OC over reference 290s), but the voltage required to go much above that means only certain users will be interested in tapping that 780 Ti-beating performance. And the 780 Ti actually overclocks better.
Note that unlike the GeForce cards, the R9 290 has meaningful voltage adjustment (on the GeForces, extra voltage often results in LOWER clocks due to the aggressive power cap on those cards). That being said, I was not interested in taking the voltage over +.05, due to the extra heat and power use. There likely is more on tap, but you'll have to turn to other reviewers for the extreme Radeon overclocks.
Another two notes on my testing:
(1) Catalyst 14.6 beta was causing me lots of blackscreens at idle. Not good for extended benchmarking sessions, so that got scrapped. Used 14.4 instead, which was perfect.
(2) I did not bench BF4 with Mantle. Treason? No, just being reasonable. First, Mantle does not produce the same quality, in my opinion, but more importantly, it led to significant graphical anomalies, like swimming under ground. It's just not doing the same thing as DX11, so it's not an apples-to-apples comparison. No matter - the R9 290 was awesome in BF4 anyway.
With that out of the way, here are the benchmarks:
3DMark is much better at comparing cards within one family than cards across families or manufacturers. I use it to establish stable overclock settings (along with Crysis 3). Anyway, you can see here that the R9 290 performs well (and scales well).
This game is somewhat CPU-limited at 1080p, and yet the R9 290 pulls way ahead. A sign of things to come with other CPU-limited games???
Again, at 1080p, Hitman is very CPU-limited, but the R9 290 puts up huge numbers, beating even the OC'd 780 Ti. Impressive.
Uh, oh, here's an AMD title and the R9 290 gets cut down to size. Not quite keeping up with the GTX 780. Ironic!
Another AMD title, and another loss for the R9 290 - AMD should stick to making GPUs rather than supporting game titles! By the way, I don't include minimums here because the benchmark doesn't report correct figures for that due to scene changes.
Excellent performance here, going head-to-head with the more expensive GTX 780. Note that at 1080p, the game is CPU-bottlenecked, even on a highly-overclock i7, and yet the R9 290 is pulling ahead even there, other than in minimums. Focus on the 1440p results to get a feel for the actual performance of the cards.
Wow, wow, wow! No artificial Mantle-enhanced performance here. This is pure DX11 domination. Note that this is the single-player game, which avoids lots of inconsistency and CPU-bottlenecking, but this is also BF4, after all, which has been patched, patched, and patched again. I'm hoping DICE does less on the single-player side, but to the extent the engine has been optimized, the newer benches on the R9 290 may not be directly comparable to ones I generated previously for the GeForces.
The difference in this game was so significant that I actually played better in the game - it was that much smoother, especially at 1440p.
[UPDATE:] At stock volts (1.07V), this card was actually quite efficient, about matching the GTX 780. But when overvolting, power ramps up. Just to get to 1100MHz (a 153MHz OC over stock), I had to accept a big increase in power consumption due to the extra voltage (1.12V). The card runs perfectly at 1075MHz without extra voltage, and honestly, that's how I'll run it other than for benching.
----------------------
Overall impressions:
(1) The R9 290 was a watershed card when it was "released" at $400. Now that it's actually available around that price with a great cooler, it is simply awe-inspiring.
(2) The Tri-X is the quietest gaming card I've ever used. Temps never went above 75C even with a big overclock and extra voltage, and yet it wasn't any louder than my case fans.
(3) The R9 290 doesn't use nearly as much power as I thought it would based on reviews - it's actually fairly efficient, as long as you don't touch the voltage.
(4) Performance is fantastic, but it's very game-dependent. When people say "OMG, the R9 290 beats an OC'd 780 Ti!", you have to take it with a grain of salt. Overall, the OC'd R9 290 does in fact match a stock 780 Ti, but so does an OC'd 780. Now, I only took my 290 to 1100MHz (a 16% OC over reference 290s), but the voltage required to go much above that means only certain users will be interested in tapping that 780 Ti-beating performance. And the 780 Ti actually overclocks better.
Last edited:
Facebook
Twitter