Naah.. AMD would rather sell 6850s (for more and with cheaper pcb/cooler), as long as the chips qualify. So the price is set to fit the number of bad chips produced..
AMD: R9xx Speculation
- Thread starter Lukfi
- Start date
Naah.. AMD would rather sell 6850s (for more and with cheaper pcb/cooler), as long as the chips qualify. So the price is set to fit the number of bad chips produced..
Looks like it's down.
So it's 5830 all over again? Too high power consumption, too expensive, too slow. At least this time it got downrated to HD67xx series.16 ROPs, 6870 pcb, consumption on par with 6850, 149$
Though if it battles with the 460 768MB it seems it's still faster than HD5770, and potentially close to HD5830 (which is nothing to be proud of) - maybe the quite high clock is helping a bit. I wonder if the 16 ROPs still have the same problem and can't effectively use all the bandwidth.
Sometimes I just hate to be right (ok I wasn't really sure about it...).
NO, no, no no and no.
Hmm your answer is cryptic
.So it's not like 5830? Not high power consumption? Not too expensive? Not too slow? And the half-bandwidth ROPs fixed? Well that would be good news. I have trouble believing ALL of this though given the supposed performance and price.
Hmm your answer is cryptic
So it's not like 5830? Not high power consumption? Not too expensive? Not too slow? And the half-bandwidth ROPs fixed? Well that would be good news. I have trouble believing ALL of this though given the supposed performance and price.
no, this cards potential and performance is beyond what the 5830 could ever aspire to be. (though i'm not talking about absolute performance, mind you.
itsmydamnation
Veteran
Look at the 8xAA performance :
Another 16 ROP 5830 fail card.
if it supports EQAA does it really matter? ie whats 4X EQAA perf like
The GPU looks like Barts and that one does not yet incorporate the redesigned ROPs allowing for EQAA.
It doesn't really matter if it doesn't support EQAA. And who buys a ~$120 video card to play at 8xAA? The fact that it beats the 550 at resolutions intended for this kind of card will be enough. If the significantly lower TDP figures I'm hearing are true I will take back my previous rant against this card.
Blazkowicz
Legend
users of smaller and older monitors would want to, also radeons have a sparse grid supersampling feature that actually works.
users of smaller and older monitors rarely use resolutions above 1280x1024, so this ROP/bandwidth bottleneck probably won't show up for them most of the time.
And users who want to use SGSSAA in more demanding games should go for something more powerful than a 6790 anyway.
In our tests, it didn't make a difference if it was left at default or the checkbox cleared and manually set to 64x max. Seems like AMD has yet to include profiles, which I still hope they will not enable by default.
The control is a win for everyone as it is: AMD has no more hassle with it, needs not to create profiles and cannot possibly being accused of trying to lessen the workload for their cards. Users can use it a their discretion. BTW: From what I've tested in 3DMark 11's last Game Test, slower cards do not profit more from lessen Tessellation than a HD 5870, also 6900 doesn't profit as much als 5870 as well.
HD 6790 being apparently a harvested part, TDP seems to be that high to incorporate a larger fluctuation of leakage into a single TDP-range, thus making AMD able to sell more chips into this SKU. So, IMHO chances are that most cards won't hit this generous TDP envelope by a wide margin. Cannot say more until NDA lifts though.
itsmydamnation
Veteran
there is defenity a profile for heaven in cat 10.4 preview, i was mucking around and had tried setting from 64X all the way down to 8X tess factors i couldn't notice a difference in IQ, min frame rate improvement was big. when i set it to use AMD optimsed i could see IQ drop in some areas most noticable was on rope.
i have a 6950 modded and OC to 1000core 1415 memory.
Sounds like placebo effect or you just didn't notice with forced one, I think it was already documented somewhere that it's bugged at the moment, so if you set it to for example 8x max, and then check AMD optimized instead of your own forced selection, it will stay at 8x (or whatever you had chosen before)
That updated is documented in the updated Cat 11.4 Preview notes.
Hmm in retrospect I'm not quite sure what you meant. This card looks to me the same fail as HD5830 for all the same reasons - performance (much) closer to HD5770 than HD6850, but price closer to HD6850 instead of HD5770, with still the same crippled rops. Add the comparatively low power efficiency (both compared to HD6850 and HD5770, though it's not as bad as the TDP suggests) and you have a card noone should be interested in buying. If the point was to beat the GTX550 though then yes it's a success (though this one seems to have got cheaper - if the HD6790 can match its price it wouldn't be too bad).
I'm thinking though AMD probably isn't really interested in making this a worthwile card to buy. They might just not have many chips which really need to be crippled more than for HD6850, so better sell HD6850 for a bit more (after all those certainly will at least not cost more to produce, depending on design they might even be a bit cheaper).
Similar threads
