AMD's Henri Richard delayed ATI R600

[tuteja1986]

http://www.beyond3d.com/#news39176

====================
Multiple reports hit the net tonight reporting that at today's San Francisco launch of their new 690G chipset AMD decided it was also time to talk a little turkey publicly regarding their recently delayed R600 GPU. Taking advantage of press presence, AMD publicly demoed two R600s coupled to an AMD "Barcelona" CPU and working on a stream computing workload. AMD's Henri Richard took the blame for delaying the R600 launch, but assured everyone that "R600 is doing very well" --an obvious bid to quiet rumours of a hardware problem with the GPU being to blame for the launch being pushed back. Further, AMD reported that the launch would only be delayed "by a few weeks" rather than the months some observers have feared. At least one source is reporting directly that the delay is in order that "AMD can roll out a full suite of graphics chips covering multiple market segments for the latest Microsoft DirectX 10 applications programming interface."

On the specifications front, several tidbits were pointed at to give enthusiasts some tea leaves to stir. In the configuration demoed in San Francisco, each R600 is said to have used 200W, be comprised of 320 "multiply-accumulate units", and between the two units on display be capable of a total of one teraflop of math power. While it's not clear yet which bits are definite AMD marketing language and which are reporters trying to invent language to report with, it would appear to us likely that AMD will also be joining the scalar bandwagon, and that to reach 1 teraflop of compute power with 320 ALUs x 2, would most likely imply a clock of roughly 800MHz operating on MADD units. Going well back to early rumours and reporting of R600 last spring, and adding today's 320 ALUs, this would begin to look like a modified/improved version of Xenos at 4 arrays of 16, with the 5D ALUs improved to be scalar and fully MADD. However, AMD's generosity of information today did not extend quite so far as that, and some actual poking and prodding would be required to confirm this.



=======================


http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=197700269

 

[LittleNemoNES]

MAYBE...MAYBE

All the problems AMD SEEMED to be having were much ado about nothing. If a company wants to have widespread availability of their product, it's their decision.

Ruis hints that the delay will be no more than 2-4 weeks.

 

[Kromis]

What a wonderful year 2007 is...

And a hellish school year too... :|

 

[zephyrprime]

Originally posted by: tuteja1986
While it's not clear yet which bits are definite AMD marketing language and which are reporters trying to invent language to report with, it would appear to us likely that AMD will also be joining the scalar bandwagon, and that to reach 1 teraflop of compute power with 320 ALUs x 2, would most likely imply a clock of roughly 800MHz operating on MADD units. Going well back to early rumours and reporting of R600 last spring, and adding today's 320 ALUs, this would begin to look like a modified/improved version of Xenos at 4 arrays of 16, with the 5D ALUs improved to be scalar and fully MADD. However, AMD's generosity of information today did not extend quite so far as that, and some actual poking and prodding would be required to confirm this.

1 MADD unit is not the same thing as 1 ALU. If it has 320 MADDS and is single-scalar, then it has 320 ALUs. If it is dual issue, then it is 320/2=160 ALUS. If it is Vector4 + scalar, then it is 320/5=64 ALUs.

I also would not be surprised if the ALU were significantly more than 800mhz.

 

[apoppin]

Originally posted by: tuteja1986
http://www.beyond3d.com/#news39176

====================
Multiple reports hit the net tonight reporting that at today's San Francisco launch of their new 690G chipset AMD decided it was also time to talk a little turkey publicly regarding their recently delayed R600 GPU. Taking advantage of press presence, AMD publicly demoed two R600s coupled to an AMD "Barcelona" CPU and working on a stream computing workload. AMD's Henri Richard took the blame for delaying the R600 launch, but assured everyone that "R600 is doing very well" --an obvious bid to quiet rumours of a hardware problem with the GPU being to blame for the launch being pushed back. Further, AMD reported that the launch would only be delayed "by a few weeks" rather than the months some observers have feared. At least one source is reporting directly that the delay is in order that "AMD can roll out a full suite of graphics chips covering multiple market segments for the latest Microsoft DirectX 10 applications programming interface."

On the specifications front, several tidbits were pointed at to give enthusiasts some tea leaves to stir. In the configuration demoed in San Francisco, each R600 is said to have used 200W, be comprised of 320 "multiply-accumulate units", and between the two units on display be capable of a total of one teraflop of math power. While it's not clear yet which bits are definite AMD marketing language and which are reporters trying to invent language to report with, it would appear to us likely that AMD will also be joining the scalar bandwagon, and that to reach 1 teraflop of compute power with 320 ALUs x 2, would most likely imply a clock of roughly 800MHz operating on MADD units. Going well back to early rumours and reporting of R600 last spring, and adding today's 320 ALUs, this would begin to look like a modified/improved version of Xenos at 4 arrays of 16, with the 5D ALUs improved to be scalar and fully MADD. However, AMD's generosity of information today did not extend quite so far as that, and some actual poking and prodding would be required to confirm this.



=======================


http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=197700269

they couldn't demo it running any games because it crashed each time


:Q


who knows?

working on a stream computing workload

B-O-R-I-N-G

so for 2-D it uses only 200w ... pretty good for A13

:laugh:


show me the money

 

[Creig]

Originally posted by: apoppin
they couldn't demo it running any games because it crashed each time


who knows?

Certainly nobody here. So why make such a dumb statement in the first place?

Originally posted by: apoppin

working on a stream computing workload

B-O-R-I-N-G

so for 2-D it uses only 200w ... pretty good for A13

Doesn't stream computing use pixel shaders for its computational power? That would make it a 3D function, not 2D.

 

[jim1976]

Originally posted by: zephyrprime

Originally posted by: tuteja1986
While it's not clear yet which bits are definite AMD marketing language and which are reporters trying to invent language to report with, it would appear to us likely that AMD will also be joining the scalar bandwagon, and that to reach 1 teraflop of compute power with 320 ALUs x 2, would most likely imply a clock of roughly 800MHz operating on MADD units. Going well back to early rumours and reporting of R600 last spring, and adding today's 320 ALUs, this would begin to look like a modified/improved version of Xenos at 4 arrays of 16, with the 5D ALUs improved to be scalar and fully MADD. However, AMD's generosity of information today did not extend quite so far as that, and some actual poking and prodding would be required to confirm this.

1 MADD unit is not the same thing as 1 ALU. If it has 320 MADDS and is single-scalar, then it has 320 ALUs. If it is dual issue, then it is 320/2=160 ALUS. If it is Vector4 + scalar, then it is 320/5=64 ALUs.

I also would not be surprised if the ALU were significantly more than 800mhz.

Yeap...
For the consumption part I'm almost assured now that ~200W will be the norm for R600..
I've heard and seen so many things about DAAMIT strategy lately that almost made me loose faith in their statements..
I really hope things will go well with R600. For the sake of the competition..

 

[apoppin]

Originally posted by: Creig

Originally posted by: apoppin
they couldn't demo it running any games because it crashed each time


who knows?

Certainly nobody here. So why make such a dumb statement in the first place?

Originally posted by: apoppin

working on a stream computing workload

B-O-R-I-N-G

so for 2-D it uses only 200w ... pretty good for A13

Doesn't stream computing use pixel shaders for its computational power? That would make it a 3D function, not 2D.

yes ... of course .. you're probably right

why?

because AMD is making a SERIES of stupid moves and even stupider excuses

i am just *disgusted* ... and especially disappointed

they canceled a hard launch and now appear to be directionless in their scramble to "explain" their *marketing decision*

we can only *hope*

if i had to "guess", i'd say reducing the power consumption of r600 is one of their top priorities

 

[dreddfunk]

Delaying the launch may be a bad move, or it may not. It certainly annoys the enthusiasts, which comprise a decent part of the AT audience but only a fraction of the overall graphics market.

Let's not confuse "bad for me" with "bad unequivocally." The two are a bit different. The delay may be annoying a lot of folks on this board but I doubt that is the case for the majority of PC gamers, let alone the majority of PC users.


Truthfully, given the recent track records of nVidia and ATI/AMD, I'd be very happy if they both change their practices just a bit: launching high-end and mid-range cards at least in the same quarter, if not at the same time.

I realize dealing with all of the manufacturing issues of having overlapping launches must be a pain, but this 'release the high end now and make the mainstream gamer wait' approach from both sides has me really annoyed. IMHO ATI/AMD has been worse in this area than nVidia. IF (and this is a big 'if') they are actually delaying the high-end parts so that they can release a line-up of cards that addresses a broader market, then I applaud the move. We'll see.

Another thing I'll be curious to see is if nVidia will go ahead with the 8600 launch prior to ATI/AMD's launch. If they do, they will expose themselves ever so slightly to the potential of ATI/AMD counter-launching a faster midrange product.

 

[A554SS1N]

Makes me think the AMD /ATi merger has been very bad for ATi - they would have done better separately. :/

 

[dreddfunk]

Possibly for ATI, though they could sure use the resources AMD had to offer. I think there will be a couple of different ways to judge the merger: from the perspective of enthusiasts, or from the perspective of the mainstream. I think there is a decent chance that the merger won't be much of a success from the enthusiast's perspective. Probably not because ATI/AMD will cease putting out competitive high-end GPUs (I think they will, if just to push the R&D envelope for lower-end products), but rather because they'll subordinate their GPU launch cycles to overarching business concerns, which now includes AMD's CPU business. That's likely to continue to annoy early adopters in the enthusiast market who spend a lot of time anticipating new technology once it's been leaked.

From a mainstream perspective, this isn't such a big deal, and the potential for increased competition in the integrated/low-end GPU markets is good in this area. If AMD can use ATI's GPU experience to get more of the 'platform' market (chipset/CPU/GPU combinations for low-end desktops and notebooks), they'll consider that a win. Heck, I would really love a small, lightweight laptop that could do a little gaming, but wouldn't drain the battery in 60 seconds (an exaggeration), and didn't cost a ton.

I think what we're seeing here is the new AMD management subordinating the R600 launch to larger concerns. My guess is that AMD/ATI (and I've changed the order now for a reason), wants to hit back on several fronts at once, releasing Barcelona, the R600, and R6xx mid-range derivatives, in the same general time frame. IF (and again, this is a big 'if'), they can simultaneously put a CPU out there that competes well with Core 2, and GPU products that compete well across the spectrum (high-end, mainstream, low-end), then they've got an advantage over their rivals when it comes to what they can offer OEM's (a chipset/CPU/GPU combination all from one vendor, likely at a reduced cost).

To sum up: ATI has access to a lot of interesting resources now but it comes with baggage, while nVidia has flexibility that comes with being focused on a core product but doesn't have access to as many resources. It will make for an interesting time in the GPU market.

 

[apoppin]

it's way too soon to tell the long-term effects of the merger

they all have an adjustment period ... like a marriage


an *who* said anyhing about 300w?


240w was the oft-quoted figure for x2900xtx ... a 20% reduction is nice
:thumbsup:

 

[jim1976]

Dreadfunk you make good and valid points that I think many rational users do understand..
What you seem to be ignoring is the fact that AMD is losing market and stock share every day..
AMD is already delayed in the CPU market and now it is delayed in the GPU as well.. That's not good news for the investors and certainly not for the shareholders..
Despite that I do believe that this merging is good and will benefit both sides.. They just have to convince the financial market as well..

 

[gsellis]

Originally posted by: Creig

Originally posted by: apoppin
they couldn't demo it running any games because it crashed each time


who knows?

Certainly nobody here. So why make such a dumb statement in the first place?

Give him *1* nVidia card and he turns into a fanboi

I read Henri Richard as saying, "unlike every release in the past, we are going to try to give our partners a little more time to ship on the same date." That is a change that they will like.

 

[SlowSpyder]

Makes me think the AMD /ATi merger has been very bad for ATi - they would have done better separately. :/

Yea, ATI's last few product launches have been timely and problem free.

 

[dreddfunk]

Jim - I'm not really ignoring those points at all. I actually think sacrificing a little market share right now is a good strategy for AMD/ATI, but I don't think they're sacrificing a whole lot at the moment (mainly because of the Vista effect, or lack thereof).

Simply put, there isn't a really compelling reason to get Vista and no compelling reason to get a DX10 GPU right now. AMD's market share isn't going to immediately plummet just because Intel's Core 2 architecture has taken the performance crown. Similarly, ATI's fortunes aren't going to plummet just because nVidia has taken the performance crown. I don't think market share is that fickle. I think ATI has seen their fortunes diminish at the hands of nVidia mainly because the latter has been very savvy about the timings of their launches.

I think timing is a more pertinent issue here, especially for the GPU market. nVidia has been kicking ATI's fanny in this area for the last couple of years: they've consistently had a response ready to spoil any of ATI's moves. I think the current situation is a setup: nVidia is out of the gates first, pushing the pace with the 8800. They want ATI to respond piecemeal with the R600. nVidia can then spoil the launch of a competitive high-end part with the launch of their own mid-range parts (8600, where more of the money really lies)--moreover, the launch of ATI's high-end part gives nVidia a bar to go after. And then they can follow it up with a refresh (8900) that takes back the performance crown (or darn close).

As I said on another thread. I think it's insanity for ATI to do this again. nVidia already did it to them once with the 7xxx versus x18/x19xx. As the old saying goes: the definition of insanity is doing the same thing over and over, expecting different results.

What's going to matter this time around is timing: who has the best performing cards--across all market segments--with the least issues *when 'must-have' DX10 games hit the shelves*.

Most folks don't buy just to have the latest and greatest equipment; they buy because their old equipment *can't* do something they want. We won't get to the *can't* stage until DX10 games start arriving (and even then they'll have DX 9 implementations).

Being the baddest kid on the block will mean a lot more then than now. If ATI pushes out R600 and, just as DX10 games arrive, nVidia trumps it with the 8900, that's a big problem. Pushing R600 out the door now tells nVidia exactly what they have to do in order to take the performance crown when it really matters.

Being first to market is important but, and here's the kicker, the DX10 market hasn't arrived yet.

 

[jim1976]

I don't disagree I told you that from my last post.... BUT..
You can't expect the financial market to be aware of these facts.. A shareholder or an investor sees that atm, AMD does not have a "competitive" product neither in the cpu nor in the gpu market.. That's what they know.. You can't expect them to have good and sufficient knowledge of this market and thus critical thinking on what it is important..Economy doesn't work this way..And this is depicted in the continuous fall in stock price..

Look at this

And look at the behavior of the stock in this interesting chart

It's not always about you and me.. It's all about what shareholders and investors believe.. Some things precede the consumer market.. If AMD does not make careful strategic moves in the next months, they might put themselves in a very risky position..

 

[apoppin]

Originally posted by: gsellis

Originally posted by: Creig

Originally posted by: apoppin
they couldn't demo it running any games because it crashed each time


who knows?

Certainly nobody here. So why make such a dumb statement in the first place?

Give him *1* nVidia card and he turns into a fanboi

I read Henri Richard as saying, "unlike every release in the past, we are going to try to give our partners a little more time to ship on the same date." That is a change that they will like.

*Give* me a nice video card like a 8800GTX and i just might do that

i *Still* have my x1950p and IF i don't get $220 for it, back goes the 7800GS and i *will* buy a 800+w PS next week and stick it back it.
:Q

with the x1950p BACK in my rig, my PoV will not change - a bit

just call me a "former ATi fanboy" that is disappointed as hell with the BS and FUD pouring forth from AMD's marketing as they lamely try to *explain* the delay.
:brokenheart:

that way you can attempt to relate to what is happening in my mind
:roll:

 

[apoppin]

Originally posted by: jim1976
I don't disagree I told you that from my last post.... BUT..
You can't expect the financial market to be aware of these facts.. A shareholder or an investor sees that atm, AMD does not have a "competitive" product neither in the cpu nor in the gpu market.. That's what they know.. You can't expect them to have good and sufficient knowledge of this market and thus critical thinking on what it is important..Economy doesn't work this way..And this is depicted in the continuous fall in stock price..

Look at this

And look at the behavior of the stock in this interesting chart

It's not always about you and me.. It's all about what shareholders and investors believe.. Some things precede the consumer market.. If AMD does not make careful strategic moves in the next months, they might put themselves in a very risky position..

that *risky position* means "borrow money" - at high interest rates

that means little money for research and development for future products ... or many projects just get cut or scaled back

that means firing less essential personnel and selling off parts of their company

closing fabs and plants

being acquired

of course AMD has been in arguably *worse* spots ... but not after taking on such massive debt with ATi.

i argued that the merger *could* be good for both AMD and ATi ... you can look up my posts if you can get past the search engine ... i was *very hopeful*

and i still am ... but my confidence in their judgment is fading

an do ANYONE of you *seriously* think my opinion depends on which brand video card is currently running in my rig?


that is just stupid

i have *both* and i really like them *both* ... IF i get $220 as planned for my x1950p, then i will sell it ... if NOT, then i will return the 7800GS and stick the x1950p BACK into my rig with a new PS that i was intending to buy next year.
:shocked:

Do you think i will suddenly say, "AMD marketing excuses now makes sense" ?...
. . . tomorrow
--just because there is a x1950p back in my rig
:thumbsdown:

 

[Zebo]

AMD are idiots. Not only were both NV and ATI cheaper two years ago NV was the same price as ATI two years ago ~4B. Which would you rather own considering NV's chipset and cellphone business and tight integration with AMD..the 6800 card had just came out so even graphics business was shinning. Plus they are both green Seriously I knew it was a mistake then and is showing it now. I think going into huge debt for a company with no new ideas since the 9700 which wasn't even their idea a shotty chipset business and paying a premium for it could crash AMD.


Now NV is worth more than ATI and AMD combined. Hows that for a deal.

 

[BFG10K]

So this is all Henri's fault? Burn him!

 

[Dadofamunky]

Originally posted by: Zebo
AMD are idiots. Not only were both NV and ATI cheaper two years ago NV was the same price as ATI two years ago ~4B. Which would you rather own considering NV's chipset and cellphone business and tight integration with AMD..the 6800 card had just came out so even graphics business was shinning. Plus they are both green Seriously I knew it was a mistake then and is showing it now. I think going into huge debt for a company with no new ideas since the 9700 which wasn't even their idea a shotty chipset business and paying a premium for it could crash AMD.


Now NV is worth more than ATI and AMD combined. Hows that for a deal.

Gotta admit, that boy's got some points there....

 

[apoppin]

in case you missed it in the other thread:


what about AMD's PARTNERS ??
:Q

http://www.digitimes.com/mobos/a20070302VL201.html

Graphics partners try to keep the faith as AMD pulls the rug on CeBIT

Ricky Morris, DigiTimes.com, Taipei [Friday 2 March 2007]

Reports from Taiwan's graphics card manufacturing industry are revealing that AMD's recently announced delay of its upcoming flagship graphics processing unit (GPU), the R600, was not just a shock and disappointment to fans of the former-ATI, who have been waiting for graphics cards based on the new chip, but also to AMD's partners, who were set to introduce products based on the chip at CeBIT in March.

AMD and its partners were arranging to launch R600-based products at CeBIT, according to industry sources, and live demos were being planned by individual partners. These have now almost all been canceled after the announcement by AMD, with only a few select few partners being allowed to show demo systems ? provided by AMD ? in private showings, while the rest will have to make do with showcasing old products or, at best, new variations of old products at one of the industry's largest trade shows, which starts in a little than over two weeks, on March 15.

But it's not the inconvenience that's causing problems for the card makers, but rather the uncertainty that the late change of plans casts over AMD's ability to deliver in the graphics card market and its hope of being able to compete with Nvidia following the AMD-ATI merger. Nvidia launched its current high-end GPU in November 2006.

According to graphics card makers, the R600 is AMD's hope of getting back on top and the delay only serves to highlight how far behind the company has fallen. The fear that many card makers now face is that supporters of the ATI brand who have been waiting months for the new product will finally decide enough is enough, the effect of which would be reduced confidence, and therefore sales, of all ATI-branded products.

However, despite the doubts, it still appears that graphics card makers are backing AMD and hope that the company will succeed. "We have our fingers crossed," said one maker while another asked everyone to "keep the faith." With the R600 launch now scheduled for early in the second quarter amid added speculation that mainstream (R630) and low-end (R610) parts will launch at the same time, the new plans could result in AMD having a complete top-to-bottom product lineup at launch, a rare occurrence in the graphics market, but one that many industry experts believe may prove successful for the company.

do i have to draw you a *picture*?

i *hope* AMD succeeds ... but they are LOST and drifting


*the above* is what *i* have BEEN saying ...

well put article