AMD: R7xx Speculation

Status
Not open for further replies.
@CJ: I know (see my posting about RV700 ES), but the clk-speed rumors going round this time are wrong.

@INKster; According to AMD a RV770 XT is 30% faster than a 9800GTX, has UVD2, DX10.1, Powerplay, etc. so the price semms to be right.


Are you sure ?????:smile::smile::smile:
 
Rv770 has 800SP not 480;)
Woot, woot, finally someone who agrees with me. Although I'll admit I don't think your arguements are very convincing... ;) But hey, I'll just pretend that's good enough for me - I have yet to find anyone who doesn't think I'm crazy on this, so even if you were crazy too, that's still better than nothing... right? (I've been pushing this idea privately and in midly subtle terms publicly; what do you think my 625*800*2 comment was about? or my 40 TMU jokes? heh!)

Of course, it's not because I *think* such a thing that it's not a massive amount [censored]. It's not like I've never been wrong in the past... (*cough* G92 *cough*)
 
Woot, woot, finally someone who agrees with me. Although I'll admit I don't think your arguements are very convincing... ;) But hey, I'll just pretend that's good enough for me - I have yet to find anyone who doesn't think I'm crazy on this, so even if you were crazy too, that's still better than nothing... right? (I've been pushing this idea privately and in midly subtle terms publicly; what do you think my 625*800*2 comment was about? or my 40 TMU jokes? heh!)

Of course, it's not because I *think* such a thing that it's not a massive amount [censored]. It's not like I've never been wrong in the past... (*cough* G92 *cough*)

No way it's getting clocked at 625 though, that's obviously way too low.

800 sp's would be fine, except it would possibly mean only 16 TMU's, or some horrible physics thing. But hey, if it's 32 TMU's, and no physics, then 800sp's is fine with me, it'd just be overkill.
 
Well, yes, I'm just extremely skeptical of the 480 SPs stuff. What makes me even more dubious now is the supposed X2 delay. Seems like a perfectly phased rumour, given that it came slightly before you might have started to expect seeing board pic leaks or AIBs getting briefed... So yeah, call me crazy, but I'm still giving a 65%+ chance of RV770 being monolithic (350mm²+) *shrugs*. Might be usable for a X2 with some units disabled down the line though, who knows.

Once again, I could be horribly wrong here, but I wouldn't have been surprised at all if the everyone has been misled. This would hardly be unprecedented in the history of the industry either so I don't think it'd be horribly surprising...
 
The first problem is how to implement the NUMA configuration between multiple discrete chipsets.

ATi has found the solution for this with R700 AKA 4870 x2. They've ditched NUMA all together and gone to a unified memory space for multiple GPUs. This will provide performance gains in many scenarios. You can count on it.
 
ATi has found the solution for this with R700 AKA 4870 x2. They've ditched NUMA all together and gone to a unified memory space for multiple GPUs. This will provide performance gains in many scenarios. You can count on it.
So if that's such a big part of R700... Why isn't anyone talking about it? Why is no rumour mentioning it? Why is no 'leak' from ATI docs heralding it as the next best thing since sliced bread? I'm not saying it's not true; I'm just skeptical, and wouldn't be surprised if that rumour was, let us say, outdated.

And Lux_, uhhhh, I have no idea why you think it's even theoretically possible NOT to render his frame from scratch? (at least for the main framebuffer)
 
So if that's such a big part of R700... Why isn't anyone talking about it? Why is no rumour mentioning it? Why is no 'leak' from ATI docs heralding it as the next best thing since sliced bread? I'm not saying it's not true; I'm just skeptical, and wouldn't be surprised if that rumour was, let us say, outdated.

Was there ever a rumour? I don't recall anything remotely stating that beyond speculative forum posts.
 
ATi has found the solution for this with R700 AKA 4870 x2. They've ditched NUMA all together and gone to a unified memory space for multiple GPUs. This will provide performance gains in many scenarios. You can count on it.

OTing & nitpicking badly:
Unless they have managed to seamlessly (no bandwidth decrease) extended their internal ring bus, it is NUMA.

Hint: a widespread NUMA arch is dual socket opteron.
 
Anyway, so the bottlenecking point of 3870's were thought to be their texture fill rates, yeah?

3870 has a fillrate of 775*16 = 12400MT/s and a processing rate of 775*320*2 = 496GFLOPS

4870 has a fillrate of 850*32 = 27200MT/s and a processing rate of 1050*480*2 = 1008GFLOPS

So a 2.2x increase in fillrate and a 2x increase in math processing power; and 2x the memory.

I don't like the VLIW-oriented architecture of ATI's post-R500 chips because it looks like it's too dependant on the application; but actually 3870's performance was mostly consistent in much of the games, and I read somewhere that the compiler can generally extract 4 instructions from the code (that goes into the 5-shader clusters) so it's not that inefficient at all. But still I'd like them to do a major architecture rehash and go for executing the same instruction on groups of threads, like NVidia. Anyway, you guys have talked about this probably a hundred thousand times so I'll quit.

I've missed a lot of rumors about 4870's power, but it's probably doubled 3870; which makes it more than just %25 over 9800GTX. Which is cool, but for a price of $350 I don't think it's that very ground-breaking at all, considering 9800GTX is $250 and to hell with 9800GTX anyway, the Gigabyte Type-B 8800GT (which can go as high as 850MHz because of its inherent voltage modifiability) is just $160, which IMHO is really a ridiculously low price, considering its awesome overclockability and custom cooler. I mean, if 4870 is %25 better than 9800GTX, it's just %30 better than a $160 GT is, and it's more than two times the price.

I have a very strong feeling my next card will be a GTX-280; but that's a long time away, since there are absolutely no graphically intensive games on the shelves today, except Crysis, which I finished four times already. Stupid consoles! Every cool game is being designed for them, and their power just can't match even a $160 8800GT. With those new cards the performance gap between consoles and PCs will be even more gigantic. And for example, what about 2 years later, while the X360 and PS3 will still be reigning, and the PC probably five or six times faster than them? Will every cool game still be developed for those consoles?

Holy shit, that's a long-ass post for a second ever post on a board! Bye.
 
Status
Not open for further replies.
Back
Top