The LAST R600 Rumours & Speculation Thread

Status
Not open for further replies.
So we're talking like A15 silicon? Isn't that like +12 what most chips ship at? :oops:

I believe they (ATI) start at A11. A13 is the 3rd spin, A15 would be the fifth.

I can't imagine its come to that though as I thought A13 was suposed to be looking good, at worst it would be A14 now.
 
As I posted in the duplicate thread, I think this is definitely a possibility. AMD wanting to showcase their GPU with a competitive CPU compared to what is currently available from Intel. If we hear RD790 for 1207(+) is coming in June (codenamed "WAHOO"), which it may as I imagine they'll push a new chipset with the new CPU for 4x4 and roadmaps say to expect it around then, I would chalk it up as the likely scenario.

I can see this a possibility if R600 supported AMD CPUs only.
 
All this stuff about waiting for Barcelona is nonsense. Intel is in the CPU lead right now and NVIDIA the GPU. If AMD could launch a product to recapture the GPU lead in March they would.

They have been planning this for months and months and they have known what they are up against with the G80.

The only logical answers to this in IMO are substantially subpar performance to the 8800 (not likely since why change now) or there is a technical problem serious enough to delay it in a major way.

That's definitely part of what I see as the 'bigger picture'. Perhaps performance IS sub par or even right now with the 8800, but while holding the launch for any of the aforementioned reasons (and certainly perhaps not all of them, but allowing for some if not all of them because of the time frame) it gives them time to perfect drivers especially in areas they are hurting. Remember, for instance, ATi is expected in the 7.5 Catalyst drivers to release adaptive MSAA (obviously improved over EATM/ASM), perhaps similar or better to Nvidia's CSAA. Also, by that time their new OGL driver should be substantially improved, all of which would look good to have when the initial product launches and the first reviews come in. 7.5 is in May, around when we now expect R600. See what i'm saying? I think everything is just falling into that time frame to put out the best launch possible on all fronts.

@spin revision - I don't know what the number is, but I do think one has just come back or is coming back, hence the delay for stockpiling for hard-launch.
@zeal - It's not about supporting Intel cpus, it's about demonstrating your GPU on your own CPUs that isn't sub par to the competition, as the aging K8 certainly is to Core2, and if they don't show it off on Barcelona, it'll be the aging AM2 or K8 FX 4x4. Do you think they want to push having a R600 in an Intel box over a 4x4 Barcelona box? Again, like I said, it's just one possibility of something that MAY fall in that time frame. I imagine they will do it if feasible, but I agree they won't unnecessarily hold it for it's release. I think they are just going for a launch when the most things will have fallen into place that can in that point in time...Be it any mix of driver maturity, Rv610/Rv630, stock-piling R600 for multiple SKUs, Barcelona 4x4, etc etc.
 
Last edited by a moderator:
@zeal - It's not about supporting Intel cpus, it's about demonstrating your GPU on your own CPUs that isn't sub par to the competition, as the aging K8 certainly is to Core2, and if they don't show it off on Barcelona, it'll be the aging AM2 or K8 FX 4x4. Do you think they want to push having a R600 in an Intel box over a 4x4 Barcelona box? Again, like I said, it's just one possibility of something that MAY fall in that time frame. I imagine they will do it if feasible, but I agree they won't unnecessarily hold it for it's release. I think they are just going for a launch when the most things will have fallen into place that can in that point in time...Be it any mix of driver maturity, Rv610/Rv630, stock-piling R600 for multiple SKUs, Barcelona 4x4, etc etc.

I can certainly visualize a big press extravaganza of AMD releasing their most powerful CPU and GPU in history simultaneously. It isn't far fetched. I just think the timing is wrong. Where is that PowerPoint presentation giving an explanation for the delay? Better yet, isn't there a Gandolf-like chap that roams these woods who seems to know all things ATI and could give us an explanation?
 
The bad binsplits/yields possibility for R600 could be consistent with what we see, and it could be spun to match AMD's corporate line.

If R600 is delayed, then the coincidental timing that the other variants will be ready for release could be made to look like they are doing this willingly.

If the hard-launch plan for all product lines is true, then some things could be inferred.
If the smaller value cores based on R600 are present in volume but share the same base architecture, it would lend credence to issues getting R600 to proper yields, or that binsplits weren't good enough.

The yield issue is somewhat equivocal. The smaller chips would have similar problems if yields were bad, though being smaller designs would leave them less vulnerable than a big chip like R600.

The binsplit issue dovetails nicely from an outsider's perspective. Smaller, simpler chips are easier to clock higher and within power budgets.
It also matches the lack of any information on clock domains.
If R600 is trying to push the entire core to high clock speeds, then all that hardware is being pushed nearer the edge with regards to timings and heat than with a chip with clock domains.
If a chip is smaller, this is less of an issue, either because the performance target doesn't need to be as high and because it's pushing less transistors and wires overall.

Depending on the availability of lesser R600-based boards with reduced clocks, we might have a better idea if it's yields or binsplits. Yields would depress availability more overall, binsplits would lead to a good number of less than top chips.

(unfortunately, this could be masked by deactivating otherwise good quads on the lower clocked chips and making things ambiguous between binsplits and yields again)

There may be a good run on the value lines, but the top chips are not in sufficient quantity.

This still leaves the question as to why they'd be pushing the clock so aggressively, which hints that without pushing the silicon near the top end of the envelope, R600 is not competitive or not competitive enough.

I'll add the caveat that I'm running on less than rumors. I'm running on the lack of rumors for the alternative scenarios.
I may very well be wrong. I'd hope so, since I want there to be a good leapfrogging of GPU tech every release. Things get stagnant when one competitor can only muster parity with difficulty.
 

why would that be? its not going to be largly bandwidth limited because of the bus width regardless.

I can certainly visualize a big press extravaganza of AMD releasing their most powerful CPU and GPU in history simultaneously. It isn't far fetched. I just think the timing is wrong. Where is that PowerPoint presentation giving an explanation for the delay? Better yet, isn't there a Gandolf-like chap that roams these woods who seems to know all things ATI and could give us an explanation?

Thats a stupid risk to take because Penryn is going to be coming out right around the same time. They'll look like asses if all the reviewers proclaim how much AMDs quad core sucks compared to Core 2 processors when it comes to gaming and the new card, let alone what impact Penryn might have. I still havent seen good things about Barcelona and AMD has been very tight lipped about its performance even at their own shows which usually is not a good sign.


The problem with alot of the spin put on the delay, such as the so called benefits that some people see in a family launch, is that you do this by giving up brand recognition. No company in their right mind would delay a product to launch it all together if it was ready because of this and thats also why flagship cards are the first out the door. The performance of these parts impacts sales all the way down the chain because people will link nVidia 8 series or ATi X2k series with great performance and/or industry leading features and they will buy parts based on this despite if they lead in their price segment or not. I think nVidia has proven this time and time again and did so very well with the launch of the 7800 and all cards that followed. It is good to be king of the hill, infact its key to winning sales. I can already tell you AMD/ATi are going to post craptastic sales this year in their graphics department compared to nVidia precisely because nVidia had this lead. Think about it a moment, how do you tell someone, who never bothers with reviews, that the product (any card in any price range) that just launched is just as good or better then the nVidia counterpart based off the highly acclaimed 8x00 series which has been proclaimed the best for gaming for close to a half a year. You just cant do it, you cant dent that wall in a short time frame and thats exactly what happens. This is bad for sales any way you cut it.
 
Last edited by a moderator:
why would that be? its not going to be largly bandwidth limited because of the bus width regardless.

This is just speculation, but it could have something to do with costs and yields of GDDR4 memory or it could have something to do with latency.
 
I just dont understand why they have to launch the fastest and most difficult to build (XTX) model first, cant they launch the XT/XL variants first and then the XTX one or two months later? Right now the 8800GTS is stretching and extending the anii of the x1950xt/x series, they could put an end to that with an XT/XL release.
 
why would that be? its not going to be largly bandwidth limited because of the bus width regardless.

Thats a stupid risk to take because Penryn is going to be coming out right around the same time. They'll look like asses if all the reviewers proclaim how much AMDs quad core sucks compared to Core 2 processors when it comes to gaming and the new card, let alone what impact Penryn might have. I still havent seen good things about Barcelona and AMD has been very tight lipped about its performance even at their own shows which usually is not a good sign.

I think you misunderstood me. I certainly don't think it's far fetched that they would hold off launching it simultaneously with the new AMD cpu, I just think it's highly unlikely that is the case. In summary, we are in agreement that they are probably not delaying the launch for the above reason. Only Microsoft can release stuff whenever they get around to it. ;)
 
I can't help feeling the Inquirer is going to lose much of what makes it special when Fud-o leaves them at the end of the month.

I was actually hoping theyd maybe become more reliable sans Faud. There's a place for a rumor site. Inq should be tellling us right now why R600 got delayed, for example. But they're not.
 
Has anyone considered the possibility that ATI has no launch partner for their flagship product ? If that speculation was true it it puts things in a little different perspective:

That is, the 600 does meets all its intended specs. (clock, performance, yield, power), but the bottom line is its not competitive and therefore without launch partners, it does not make economic sense to launch it?
 
Has anyone considered the possibility that ATI has no launch partner for their flagship product ? If that speculation was true it it puts things in a little different perspective:

That is, the 600 does meets all its intended specs. (clock, performance, yield, power), but the bottom line is its not competitive and therefore without launch partners, it does not make economic sense to launch it?

All AMD would have to do would be to adjust what it charges for each chip. If yields were acceptable, this wouldn't be an issue.

If an optimal version of R600 places an undue burden on board partners, what's few months going to change? The design would be flawed or AMD would be stupidly overcharging.
 
Has anyone considered the possibility that ATI has no launch partner for their flagship product ? If that speculation was true it it puts things in a little different perspective:

That is, the 600 does meets all its intended specs. (clock, performance, yield, power), but the bottom line is its not competitive and therefore without launch partners, it does not make economic sense to launch it?

Why wouldn't they ?
Sapphire, Powercolor, GeCube, etc.
The usual suspects are always there, and they must be anxious by now to have something to sell against the Geforce 8xxx, since they can't claim to be *fully* Windows Vista ready unless they have DX10 hardware, right ?
 
Status
Not open for further replies.
Back
Top