- Aug 9, 2004
- 4,458
- 4
- 81
The 5850 doesn't look too bad also compared to the new cards, especially at your clocks.
Thanks for the review. :thumbsup:
This whole GTX5xx and HD69xx series will forever be remembered as the "refresh" generation in my eyes. The performance increase from GTX4xx and HD58xx series just isn't there. I am looking forward to 28nm GPUs to bring at least 50%+ performance increase over the HD6970/GTX580 at 2560x1600 4AA/16AF.
There needs to be some new games to force us to upgrade.
It was a robust little sucker once it got its clocks. With a 45% overclock it still retained ~35% more performance over stock, which is decent scaling in my book. The higher clocks gave the card a lot of horsepower, but its older design showed in the newest games.Thanks for taking the time with the benchies. The 5850 doesn't look too bad also compared to the new cards, especially at your clocks.
Thanks, that's why I posted it. IMO, this is a true enthusiast's review as it shows the real performance of each card in capable hands. I like to read them but there are so few posted at each product cycle so I wanted to contribute. :thumbsup:Very interesting review. Rarely do we see fully overclocked comparisons. I'd say the 6000-series is not a compelling upgrade from the 5000-series, but it looks like you're having fun with it, so more power to you.
I actually had 5850CF before and sold one of the cards. To me, it played almost the same as a single 5850, despite the FPS being much higher. I'm very sensitive to microstuttering, especially in AFR, and if the smoothness isn't there I'm not happy. I don't write off multi-GPU and try to give it a chance every generation or so (had a GTX295 before), but I have yet to come away impressed.Is there a reason you didn't go with an HD5850 crossfire setup? I know you mentioned this card approaches such a setup, but I doubt it actually equals it.
I agree that the performance jump isn't much of a boost, although I stress that it's more in actual gameplay: it's probably closer to +20% as opposed to the 15% shown in these benchmarks, and also there's more smoothness and quicker loading times. All in all, at 2560x1600 everything counts, especially when I like to stick to single GPU solutions.
I actually had 5850CF before and sold one of the cards. To me, it played almost the same as a single 5850, despite the FPS being much higher. I'm very sensitive to microstuttering, especially in AFR, and if the smoothness isn't there I'm not happy. I don't write off multi-GPU and try to give it a chance every generation or so (had a GTX295 before), but I have yet to come away impressed.
Also, some of you might say "then how come you didn't get a GTX 580?" and the simple answer is the cost simply isn't worth it. This upgrade will end up costing me about ~$100-120 after selling my 5850 for ~20% more performance, maybe more once newer games hit (thanks to better DX11 performance, tessellation, 2GB or RAM, etc.). If I had bought a GTX 580, even on a wicked sale, it'd be another ~$250 for at most another 15% in performance, and that's if I get a good GTX 580 that'll hit 1GHz stable on water. I'm all about diminishing returns :thumbsup:.
Is there a reason you didn't go with an HD5850 crossfire setup? I know you mentioned this card approaches such a setup, but I doubt it actually equals it.
While FPS is the final performance measurement, consistency in frame times is just as important. FRAPS benchmarks will measure frame times, but most websites don't use FRAPS and custom game runs to benchmark cards.I have a theory about this. What if fps isn't the right way to measure gameplay experience? What if the true measure is something like milliseconds between frames? If you have 59 frames in the first 5/6 of a second and only 1 frame in the last 1/6, your experience would be worse than 60 frames spread evenly in a second. And the critical measure would be maximum interval rather than average. Anyone ever seen a way to measure this?
Thanks! :thumbsup:Great job, I like this review. Two thumbs up!
1GB of VRAM is an issue, but that wouldn't kill FPS so much as cause stuttering. There are some games where 5850 CF just doesn't scale well - Metro 2033 is one of them.The problem with Xfire 5850's is the lack of vram for higher resolutions. Two 5850's are undoutbedly faster than a 6950 but they run out of memory at 2560x1600. I saw that with a single 5850 and I'm sure two would be just as bad.
I'm curious as to why you would think that. What are their prices like? While all three are decent cards, both the 6870 and 6970 have significantly less "potential" to be unlocked compared to the 6950.In OZ, either the 6870 or 6970 make sense, the 6950 seems pointless.
Thanks and my pleasure. I'm glad it's still useful :thumbsup:.Nice job and a nice read, thanks!
Hardly a gamble; it's practically guaranteed when done properly. I can't remember an instance of a respectable member of the community who had issue with just unlocking the shaders. Furthermore, if the 6950 is not a deal at $100 more, why is the 6970, which is more expensive with a worse price/performance ratio at stock, a better deal as I believe your first post implied?unlocking, oclocking, modding=CTD.
6950 is about a $100 more than a 6870, and hardly seems worth it unless you want to include the oclocking/unlocking gamble.
As Silverforce11 mentioned, the 6950 is clock for clock is 3% slower. This is while being 25% cheaper. So even without unlocking it's still a much better deal. The 6870 is a great card for the price, but it's a completely different tier.6870=affordable and powerful.
6970= 2 speed bumps over a 6870 and much cheaper than the moderately faster 580.
I have a theory about this. What if fps isn't the right way to measure gameplay experience? What if the true measure is something like milliseconds between frames? If you have 59 frames in the first 5/6 of a second and only 1 frame in the last 1/6, your experience would be worse than 60 frames spread evenly in a second. And the critical measure would be maximum interval rather than average. Anyone ever seen a way to measure this?
That doesn't look promising at all.. $250 for that card and you have a 10% increase over the previous Gen?
Garbage, thats what I say...
The 6970 is only 10-15% faster than the 5870 so what did you expect from the 5850 vs 6950?
The 69xx is a refresh.....