Did ATI hold back on R520 to make room for R580?
- Thread starter mrcorbo
- Start date
neliz said:
I also believe that ATI would want to contain costs and use 1.2ns memory if this part is to ship in volume and without immediate competion from G71 (and Orton did indeed claim that they are expecting OEM wins for R580). Besides, they will very likely need to keep some powder dry for NVDA's launch, when those cherry-picked R580 cores can ship with 900MHz modules. I don't see how else they will respond if the 7900GTX manages to clock to 700MHz+ as speculated.
Unknown Soldier said:
Now I am not sure about IQ. Or why they only test with 8xAF..
What really stands out to me is how seemingly superior G70 is to the R520. ATi gets simply hammered in all the basic tests and Nvidia pretty much holds their own with 4XAA+8xAF. The GTX512 even bests a seriously over clocked R520.
Now I know there are some questions about quality and texture filtering shortcuts that Nvidia is doing, but man what raw power. I am beginning to wonder if ATi is going to even be competitive with the R580 once the G71 is released.
I am still excited about my X1800XT that arrives tomorrow. It will run my native resolution of 1280x720 with 4xAA+16XAF at really high FPS. Which is all I really need.
It does seem that ATi has lost the beat with where they were with the 9700. ATi seems to have abandoned what made them strong and Nvidia has embraced it. Not surprisingly the results show the error of that reasoning.
Erm. Is there a non OC-ed version of HIS X1800XT ? (just a silly question, I might have ordered this one, not even knowing it is OC-ed ...)
However, look at this particular test: FEAR
Without Soft Shadows, HIS X1800XT OC is fastr than the 512 GTX, what is strange because the Asus X1800 XT Top with similar clocks is a lot slower. Is that and editor's error, or such scenarios could really happen ?
Look at the Soft Shadows enabled, even the standard GTX is faster than a OC-ed X1800 XT. PCF soft shadows ?
However, look at this particular test: FEAR
Without Soft Shadows, HIS X1800XT OC is fastr than the 512 GTX, what is strange because the Asus X1800 XT Top with similar clocks is a lot slower. Is that and editor's error, or such scenarios could really happen ?
Look at the Soft Shadows enabled, even the standard GTX is faster than a OC-ed X1800 XT. PCF soft shadows ?
Last edited by a moderator:
I feel this way, too, only NV30 seemed a bit too backward-looking (except for double Z), while R520 seems too forward-looking. Seems like NV "integrating" the texturing unit with the pixel shaders is a bit like "unified" pixel shading, allowing for greater flexibility, whereas ATI's decoupled texture units seem unnecessary and perhaps even a limiting factor with current games. R520 & co. seem like a stepping stone to SM4, while G70 & co. seem firmly rooted in SM3.boltneck said:
Given the leaked R580 benches, both from ATI and independent sites, a G71 @ 700MHz--be it 24 or 32 pipes--seems like it'll just walk all over it in most current benchmarks.
If G71 is priced sky-high and ATI shows lower prices across the board (thanks to their more efficient use of their various bits and pieces), tho, then I'll certainly understand their design goals. It's just that they're harder to fathom when you look at, say, X1600TX vs. 6800GS at current prices.
I was thinking more along the lines of how they structure their pipelines.
R300 was wide and shallow. Everyone saw that this was indeed better than narrow and deep which Nvidia was doing at the time.
Now Ati has seemingly completely abandoned that, and gone with narrow and deep Vs Nvidia who has embraced wide and shallow.
How can you be sitting there at Ati and knowing your won history actually do this and not realize how dumb it is? ATi is abandoning Pixel Fill rate way to soon in favor of Shader fill rate. What’s worse their Shaders which are supposed to be great totally get hammered by the G70's
This looks to only get worse with the R580 and G71.
Ati needs to get back to kicking some raw power but and not "clever design" themselves out of competition.
R300 was wide and shallow. Everyone saw that this was indeed better than narrow and deep which Nvidia was doing at the time.
Now Ati has seemingly completely abandoned that, and gone with narrow and deep Vs Nvidia who has embraced wide and shallow.
How can you be sitting there at Ati and knowing your won history actually do this and not realize how dumb it is? ATi is abandoning Pixel Fill rate way to soon in favor of Shader fill rate. What’s worse their Shaders which are supposed to be great totally get hammered by the G70's
This looks to only get worse with the R580 and G71.
Ati needs to get back to kicking some raw power but and not "clever design" themselves out of competition.
Unknown Soldier
Veteran
Behind the scenes with ATI - Eric Demers interview
*edit* also posted here
http://forum.beyond3d.com/showthread.php?p=680064#post680064
PH!: What top clock speeds are you expecting for the X1800 XT with air cooling and with extreme cooling? How far can the X1800 go without a die shrink?
ED: Well, with air cooling in the 90GT, we've seen core graphics clocks hit up to 800 MHz. With more extreme cooling, we've surpassed 1 GHz!! So there's a lot of headroom in these chips. On the memory side, I think we've gone well above 800 MHz, but the X1800XT now ships with 800 MHz memory, so, often, the limiting factor ends up being the dram speed. Board design also contributes to memory clock limitation. A few changes on the current board with faster memory could do significant improvements to memory. Perhaps a future product J
PH!: We suspect that the architecture first shown with the R520 has lots of reserves. The former “big one”, the R300 has doubled in pipeline count and clock speed throughout its lifespan. We assume the R520 will do similar… will we ever see a 32 pipeline (or rather say, 8 quads) R520 running around 1 GHz or will the unified shader architecture wash it away beforehand? Does it make sense to speak about unified shaders at all when we have Ultra-Threading?
ED: Well, I won't comment on unannounced products, but there's a lot invested into a new generation. About 110 man years for the R5xx generation. So, trying to maximize the number of parts we can get from it is important, to justify all the investment. The R5xx series was designed to be more flexible than previous architectures, since the metrics of yesterday have become less meaningful. 2 years ago, it mattered more how many “pipelines” you have, perhaps with some notion of the number of Z's or textures per pipe, but the basic metric was that. Today, we have moved away from that paradigm. Today, applications don't use fix function pipelines anymore, but create powerful shader programs to execute on the HW. It's not “how many pixels can you pop out per second?”, it's “what is the throughput of your shader”?. Our R5xx architecture has moved away from simply scaling of pipelines, to now scaling in terms of ALU operations, texture operations, flow control, Z operations as well as more traditional raster operations (all of this bathing in a design that can maximize the work done by each part). So will there be a 1 GHz 32 pipeline R5xx part? Well, we've ceased to measure things that way, so it won't be so easy to describe. But, yes, we will have more parts from this generation
*edit* also posted here
http://forum.beyond3d.com/showthread.php?p=680064#post680064
Last edited by a moderator:
That’s nice, but they seem to be wrong.
It looks to me like its still very much about "pipelines and Z per clock". If you are going to say all that fancy stuff about "Shader throughput" then maybe your product should actually deliver it.
They are even getting beaten handily in Source. Which is supposed to be their ace in the hole. Or at least I thought it was.
It looks to me like its still very much about "pipelines and Z per clock". If you are going to say all that fancy stuff about "Shader throughput" then maybe your product should actually deliver it.
They are even getting beaten handily in Source. Which is supposed to be their ace in the hole. Or at least I thought it was.
Karma Police
Regular
With ATi possibly using 1.1ns GDDR3 ram instead of the 1.2ns, they may be holding back the R580 instead, and waiting for nVidia's release to bump up the core & memory.
boltneck said:
Faster than the 7800GTX 256mb, slower than the 512mb GTX
X1800XL =500
7800GT =500
7800 GTX=600
X1800XT=750
7800 GTX 512mb=850
I expect ATI thinks just increasing the clock speed will yeild a big enough increase in multitexturing fill rate (they lead in pure fill rate)
Jawed
Legend
That seems reasonably true. And for most of its life R300 was stymied by a complete lack of decent SM2 games, so NV30 was happily competitive in DX8 games.Pete said:
Now it looks like R520 is in the same position, with a reliance upon games that use per pixel dynamic branching and lots of ALU operations.
No, it's simply the count of texturing units that's holding R520 back. Oh, and the lack of double-rate Z - which will plague R580 too.
Shadowmapping (as opposed to stencil shadow volumes) will favour R580 as the lack of double-rate Z will no longer be such a dominant factor in games.
I have my doubts - texturing bandwidth is simply going to strangle a card with a theoretical texture fillrate of 16800 or 22400. G71's memory will be no faster than GTX-512's.
Of course there's always the possibility that G71 has a far superior memory architecture (like R520 over R420) - but I doubt it.
Why compare those when X1600XT is priced against a 256MB 6600GT? Shouldn't you be asking how an 8-pipe 7600 is going to compare with X1600XT?
Jawed
Luminescent
Veteran
I'm almost sure RV350 has double z capability, are you sure R580 and R520 don't?Jawed said:
Jawed
Legend
ATI architectures have double-rate Z for AA, but RV530 is the only one with double-rate Z for non-AA (and quad-rate Z for AA).
This seems to be the long-standing reason for ATI cards' disadvantage in D3 and Chronicles of Riddick, both of which rely upon non-AA Z fill-rate for stencil shadow volumes.
We can't be sure about R580, yet, because it's not quite released. But the "16-1-3-1" spec looks ominous in comparison with RV530's "4-1-3-2".
Jawed
This seems to be the long-standing reason for ATI cards' disadvantage in D3 and Chronicles of Riddick, both of which rely upon non-AA Z fill-rate for stencil shadow volumes.
We can't be sure about R580, yet, because it's not quite released. But the "16-1-3-1" spec looks ominous in comparison with RV530's "4-1-3-2".
Jawed
Similar threads
