vacancies advertise contact news tip The Vault
Don't miss the latest HEXUS competitions! [x]
facebook rss twitter

Review: HIS AMD ATI Radeon HD 2900 XT

by Tarinder Sandhu on 14 May 2007, 05:00

Tags: AMD (NYSE:AMD), ATi Technologies (NYSE:AMD), HiS Graphics

Quick Link: HEXUS.net/qaipp

Add to My Vault: x

HEXUS.bang4buck, overclocking, power



HEXUS.bang4buck

In a rough-and-ready assessment of the cards' bangs per buck, we've aggregated the average 1920x1200 4xAA 8xAF and 2560x1600 4xAA 8xAF framerates for the three games, normalised them* and listed the cards' price. But there are more provisos than I care to shake a stick at. We could have chosen three different games, the cards' prices could have been derived from other sources and pricing is such that it can fluctuate daily. So, to reiterate, the graph below highlights a metric that should only be used as a yardstick for evaluating comparative performance with price factored in. Other architectural benefits are not covered, obviously.

Graphics Cards HIS Radeon HD 2900 XT 512MiB Sapphire Radeon X1950 XTX 512MiB NVIDIA GeForce 8800 Ultra 768MiB NVIDIA GeForce 8800 GTX 768MiB ASUS EN8800GTS/HTDP/640M 640MiB ASUS EN8800GTS/HTDP/640M 320MiB
Actual aggregate marks at 1920x1200 4xAA 8x(16x)AF 222.74 232.12 366.81 329.85 238.64 231.94
Aggregate marks, normalised**, at 1920x1200 201.26 206.06 273.4 254.92 209.31 205.97
Price - at time of writing £249 £225 £469 £350 £242 £192
HEXUS.bang4buck score 0.808 0.916 0.583 0.728 0.865 1.073
Acceptable framerate (60FPS av.) at 1920x1200 4xAA 8/16xAF Yes Yes Yes Yes Yes Yes


Graphics Cards HIS Radeon HD 2900 XT 512MiB Sapphire Radeon X1950 XTX 512MiB NVIDIA GeForce 8800 Ultra 768MiB NVIDIA GeForce 8800 GTX 768MiB ASUS EN8800GTS/HTDP/640M 640MiB ASUS EN8800GTS/HTDP/640M 320MiB
Actual aggregate marks at 2560x1600 4xAA 8x(16x)AF 143.25 141.82 224.48 199.97 128.99 114.44
Aggregate marks, normalised**, at 2560x1600 123.04 122.73 202.24 185.84 103.49 81.66
Price - at time of writing £249 £225 £469 £350 £242 £192
HEXUS.bang4buck score 0.494 0.545 0.431 0.531 0.428 0.425
Acceptable framerate (60FPS av.) at 2560x1600 4xAA 8/16xAF No No Yes Yes No No


** - The normalisation refers to taking playable framerate into account. Should a card benchmark at over 60FPS in any one game, the extra FPS counts as half. Similarly, should a card benchmark lower, say at 40FPS, we deduct half the difference from its average framerate and the desired 60FPS, giving it a bang4buck score of 30 marks. The minimum framerate, then, can be 20FPS, as that will score 0.

As an example, should a card score 120FPS we count it as 90FPS (120 - (120-60)/2) as only half the framerate above 60FPS is counted for the bang4buck. Similarly, should it score 30FPS we count it as only 15FPS (30 + (30-60)/2).

The reasoning behind such calculation lies with playable framerates. Should card A score 110FPS in a benchmark and card B 160, then card B would normally receive an extra 50 marks in our bang4buck assessment, even though both cards produce perfectly playable framerates and anything above 60FPS is a bonus and not a necessity for most. However, the bang4buck total would be identical if in another benchmark card A scored a smooth 70FPS and card B an unplayable 20FPS, as both aggregate to 180 marks, yet the games-playing experience would be vastly different. You would, on balance, say that card A was better because it ran smoothly in both games. In our revised aggregation, card A would receive 150 marks (85 + 65) and card B 100 (100 + 0).

In effect, we're including a desired average framerate, in this case 60, and penalising lower performance whilst giving higher-than 60FPS framerates half as much credit as the framerate up to 60FPS. If that doesn't make sense or you have issue with it, please hit the HEXUS community.

And here is the HEXUS.bang4buck in pictorial form, highlighting normalised performance divided by street price.





What the graph shows you is whether, on framerates alone, spending more is worth it. So, if the GeForce 8800 Ultra and Radeon HD 2900 XT had an identical HEXUS.bang4buck, the GeForce's higher price would be reflected in performance that was equivalently higher, too.

The 1920x1200 graph highlights that the HIS Radeon HD 2900 XT's pricing absolutely needed to be where it's currently positioned - £249 - for it to be attractive. It's not going to compete against the 8800 Ultra and GTX models on performance alone. The 2560x1600 graph is far more even, suggesting that, again, the pricing has been well thought out. On the evidence of our limited trio of games, then, the Radeon HD 2900 XT could not have been priced much higher for it to succeed. Of course, this graph blatantly ignores other benefits such as a SM4.0 architecture over say, the SM 3.0 Radeon X1950 XTX and various non-3D-related features.

Power

Who chews through the most, then?

Graphics Cards HIS Radeon HD 2900 XT 512MiB Sapphire Radeon X1950 XTX 512MiB NVIDIA GeForce 8800 Ultra 768MiB NVIDIA GeForce 8800 GTX 768MiB ASUS EN8800GTS/HTDP/640M 640MiB ASUS EN8800GTS/HTDP/320M 320MiB
Idle system draw at mains 170W 119W 185W 166W 154W 153W
Idle draw at system, assuming 80% PSU eff. 136W 95W 148W 133W 123W 122W
Peak system draw at mains 316W 271W 331W 293W 251W 253W
Peak system draw at system (80% PSU eff.) 253W 217W 265W 234W 201W 202W


Run at stock speeds and according to our measurements, the HIS Radeon HD 2900 XT's power profile sits somewhere between the GeForce 8800 GTX and Ultra models; pretty much what we'd expect.

Overclocking

We managed to push the HIS' stock 743/1586 frequencies to 844/1908 with ease. We re-ran the overclocked card in our regular 1920x1200 SC:CT HDR test.



The extra boost brings it to near-GeForce 8800 GTX stock levels.