Forget G80 then, compare directly to R580 and you'll see R600 didn't hit performance targets (other than in synthetic tests).
RV770 is what R600 should've been. A year later than it already was.
AMD: R7xx Speculation
- Thread starter Unknown Soldier
- Start date
- Status
Forget G80 then, compare directly to R580 and you'll see R600 didn't hit performance targets (other than in synthetic tests).
RV770 is what R600 should've been. A year later than it already was.
Why would that be the case? VLIW has been there since R300.
It fits it more properly into the segment that R600 was supposedly meant to target, at least. R600 was too much of a power-burning furnace to not have been an attempt at the top.
It's a wild crazy monster at certain tasks, that's for sure. They just mis-targeted where games would be.
Silent_Buddha
Legend
Well, as I see it, the rumored specs of Rv770 is where games are ending up.
R600 was targeting a speculation of where ATI expected gaming trends to continue with a much higher math/shader useage and the same or slowly growing texture use.
Had game developement proceeded as ATI was gambling on, the payoff might have been huge. As it was. Nvidia with a more traditional architechture ended up being much closer to the mark.
Other than heat and process issues, I'm pretty sure R600 hit the mark ATI was aiming for.
Unfortunately the mark ATI was aiming for ended up not being on target as to where games would be when it finally came out.
So yes it ended up being a failure in that ATI guessed wrong all those years ago when developement of R600 started. However, it was a successful product for what they designed it for.
Unfortunately, that's only reflected in the professional market and not in the gaming market. Which isn't going to get you very far.
Luckily with Rv670 they were able to take the architechture and turn it into a rather attractive and fairly successful product by reconfiguring for the market segment its performance most closely matched.
Regards,
SB
R600 was targeting a speculation of where ATI expected gaming trends to continue with a much higher math/shader useage and the same or slowly growing texture use.
Had game developement proceeded as ATI was gambling on, the payoff might have been huge. As it was. Nvidia with a more traditional architechture ended up being much closer to the mark.
Other than heat and process issues, I'm pretty sure R600 hit the mark ATI was aiming for.
Unfortunately the mark ATI was aiming for ended up not being on target as to where games would be when it finally came out.
So yes it ended up being a failure in that ATI guessed wrong all those years ago when developement of R600 started. However, it was a successful product for what they designed it for.
Unfortunately, that's only reflected in the professional market and not in the gaming market. Which isn't going to get you very far.
Luckily with Rv670 they were able to take the architechture and turn it into a rather attractive and fairly successful product by reconfiguring for the market segment its performance most closely matched.
Regards,
SB
This being apparent in what particular scenarios?Let's not get into the AA debate, as that can be argued either way. In recent titles, the delta between the R600 and the R580 is less than what was customary with generational shifts?Really?Could you please direct me to proof of that?Mind you, not some pre-release beta-drivered investigation, but one including somewhat decent, post-release drivers.
Haha that has to be the fanciest way of saying "lowered the price" that I've ever seen
Oh SB, Xbit just posted an article comparing the GX2 to 3-way and 4-way crossfire....should give a good idea of how things stand in the multi-GPU arena at the moment.
http://www.xbitlabs.com/articles/video/display/zotac-9800gx2_16.html
http://www.xbitlabs.com/articles/video/display/zotac-9800gx2_16.html
well that only tells you about xfire performance, how does SLI compare to the gx2?
Battlefield 2142 doesnt have Nway enabled for anything beyond 2 way AFR. Probably because it wasnt tested yet. But you can enable 4 Way/3Way easily by changing a single numerical value in nvapps. Oh wait. Thats only a single 9800GX2. :S
right
i meant how you have to have a mix of instructions in order to maximize ALU thruput
I'm sure it would at least alleviate VRAM restrictions a bit in higher res as well as saving the transfer of Buffers between the individual GPUs.
That said, even a PCIe-v2.0-bridge on the X2 could help scaling a bit. But in some magazine i've read them being quite expensive at about 75 Dollars...
That said, even a PCIe-v2.0-bridge on the X2 could help scaling a bit. But in some magazine i've read them being quite expensive at about 75 Dollars...
Jawed
Legend
I disagree because the available bandwidth essentially goes to waste, which is why the drastic bandwidth cut for RV670 had very little impact on absolute performance in games (there may be some synthetics that show a big drop).
Jawed
Jawed
Legend
It might improve the handling of persistent resources:
http://www.techreport.com/articles.x/14284/2
Jawed
Sorry if I'm being dense, but why would that be the case?
Because on current reference boards both RV670-GPUs on an HD3870 X2 are communicating with each other using a 3-Port PCIe-1.1-compliant 48-Lane-Switch (Link).
Jawed
Legend
The same applies to R300 and Xenos. R600 is more flexible, so on average it will run at higher utilisation.right
Jawed
Yes, and?What's the precise limiting factor in that?
Jawed
Legend
A chip so big and fast that we don't need 2 of them on a single board?
Or, two won't fit on a single board?
I'm sceptical that ATI will abandon the X2 configuration - if nothing else there's still a marketing advantage in offering such a board. Although the "short life" of 7950GX2 and (apparently) 9800GX2 might offer a different perspective on the value of X2.
I dare say there won't be much surprise if the X2 board takes a while to appear after the single-chip board's launch - but I'm struggling to conceive of a reason for it never appearing. Unless it's superseded.
Jawed
- Status
Similar threads
