That's a good catch - I'm surprised that more review sites did not hone in on the fact that this card essentially performs like a 128-bit card for AA situations.
AMD: R8xx Speculation
- Thread starter Shtal
- Start date
That's a good catch - I'm surprised that more review sites did not hone in on the fact that this card essentially performs like a 128-bit card for AA situations.
But the fps is very random. Its all over in different games.
.Too bad they didnt go for a 192bit 24rop card with 1536 MB. And they could use 6x2Gbit chips. They have now a card with almost a full 5870 PCB cost with a very random performance. I dont think this card will have a long future.
Babel-17
Veteran
Some benchmarks with 8x AA.
http://benchmarkreviews.com/index.php?option=com_content&task=view&id=469&Itemid=72&limit=1&limitstart=8
http://benchmarkreviews.com/index.p...sk=view&id=469&Itemid=72&limit=1&limitstart=9
http://benchmarkreviews.com/index.p...k=view&id=469&Itemid=72&limit=1&limitstart=10
http://benchmarkreviews.com/index.p...k=view&id=469&Itemid=72&limit=1&limitstart=11
Tom's has some too.
http://www.tomshardware.com/reviews/radeon-hd-5830,2564-7.html
http://www.tomshardware.com/reviews/radeon-hd-5830,2564-10.html
Interesting differences between Tom's, which uses DX 9 for Resident Evil 5, and Benchmark Reviews, which uses DX 10.
See how the 5830/ATI and the 260/nVidia swap places, depending?
Anyway, yeah, 8x AA numbers are something to consider, imo.
Though for people who use high resolutions and play newer, demanding, titles this might not be much of a deal breaker.
Edit: By that I mean 4x AA looks pretty darn good at high resolutions and I think with lots of other eye candy dialed in a lot of people wouldn't be thinking of using higher AA anyway, due to frame rates already being not very high.
Last edited by a moderator:
Whoops
This is a custom card, they'll blow the PCIe power spec and ATX form factor spec.
Welcome to the new age of ultra GPU's, where standards are meant to be broken.
Reminds me of the old Pentium 4 Prescott. Talk about efficiency per watt
The way things are headed, they need to rethink their goals. Perhaps Tile Based Rendering will get an renaissance as it seems to have currently in the mobile space.
Not too many of the people who would buy that kind of card still use ATX spec cases any way (no case with botton mounted PSU is ATX compliant). It's hardly the first card to break 9.6" either.
Jawed
Legend
So any reviewers going to compare this to GF100?
http://www.pcgameshardware.com/aid,705735/Asus-Ares-HD-5970-done-right-first-benchmarks/News/
http://www.pcgameshardware.com/aid,705735/Asus-Ares-HD-5970-done-right-first-benchmarks/News/
Silent_Buddha
Legend
I don't see why not, if it's available. Lots of reviews in the past have compared stock ATI cards to overclocked/AIB specific boards. And in the past users in forums have always compared stock ATI boards to overclocked/custom Nvidia boards, so this would be no different.
Doesn't affect me either way though, I always only look at stock comparisons.
Regards,
SB
FrameBuffer
Banned
I think it will be some time before you see a direct comparison against the 5900 of any flavor.. if history is anything to go by NV has most likely issues review guidelines to which potential reviewers must adhere. The guides usually contain a set of white list benchmarks to be used and card to compare against. Initial (pre-re)views would seemingly be according to such limits unless the reviewer is able to get ahold of a GF470/480 outside of NV directly.
I'm betting that we'll see lots of 470 vs 5850 and any GF480 vs 5870 will be limited to 3DMark, FC2, DarkVoid, Batman NFS:Shift and other nv-friendly TWIMTBP titles.
Do you really want to compare a 375 - 450 Watt mGPU AFR card to a 250 Watt single cpu card? I hope not...
For the top dog comparison every parameter is out of the window except performance itself.
FrameBuffer
Banned
Well that hasn't stopped many many many sites from comparing the GTX 295 (430W-ish) vs the 5870 (330w-ish) (see:computerbase.de charts)
Anyone thinking the ARES takes anywhere near 375-450W is out of their mind, having the PCIe powerplugs for it is different from actually using that. The cards TDP when using the "stock clocks" shouldn't be much over 300W considering how "little" higher the clocks are.
The extra plugin is for OCers, which are the only ppl who really will ever buy one.
Silent_Buddha
Legend
As has been noted before, at stock clocks it shouldn't use significantly more power than a stock 5900 running at 5870 speeds.
The card was built specificly with insane (IMO) overclockers in mind that often use LN2. So unless you are suggesting, that reviewers will be using LN2 based overclocking in comparisons, I don't see where those numbers even come into play.
Regards,
SB
Jawed
Legend
Silent_Buddha
Legend
Heh, and will come with a ton of converters...
Wonder if the ATI store will also sell additional converters or if there will be another place to get these. I'm guessing there's been a breakthrough for affordable converter cables as I don't see them shipping 2x of the expensive Apple type converters.
Regards,
SB
aaronspink
Veteran
The passive converters are pretty cheap to make, its the active ones that cost a small fortune. The passive ones can already be had for under $5. When the shipping costs are almost as much as the converter, its gotten cheap. The only problem is that the passives will only do up to 1920x1200@60Hz. For higher resolutions, you have to get an active converter and those will set you back ~$100 per.
Yes, previous boards are measured with a max of 3 simultaneous displays, this is done with 6.
Similar threads
