Originally posted by: Kuzi
I second that
Nemesis 1 is always one of our crazier tinfoil-hat members. Move along, nothing to see here...
Nemesis 1 is always one of our crazier tinfoil-hat members. Move along, nothing to see here...
Speaking of Sandy Bridge (or are you talking about Westmere?)...has anyone come across anything on this subject?
Are we agreed that in terms of marketplace timing and availability, like it or not, that Shanghai will be competing with Nehalem and that Bulldozer will be competing with Westmere?
Or are we thinking Bulldozer will be competing with Sandy Bridge?
This isn't a question solely directed at Nemesis, I'd be interested in what the forum's take on this.
I'd expect Bulldozer to arrive in between Westmere and Sandy Bridge, perhaps closer to the Westmere release.
basically...
late 2008: Nehalem
late 2009: Westmere (32nm debut)
late 2010: Sandy Bridge
I've read some rumors that Westmere will be in 6/12 core versions, but nothing remotely confirmed. What is confirmed is that the AVX ISA extension will debut with Sandy Bridge. AVX is a 256-bit extension to the current SSEx instruction sets (which are 128-bit), so it will double performance (theoretically) on code written to take advantage of it.
Info on AVX
If you are still around Viditor:
When you say you are unsure what the GPU in fusion will be, is AMD heavily redesigning the interfaces for ATI chips so that they'll work though hypertransport? You promised we would get more information on fusion a year and a half ago but I've seen little. Does AMD expect better communication from CPU-GPU to increase performance in gaming? Fusion seems aimed entirely at the low to mid market pc's and probably laptops.
When you say you are unsure what the GPU in fusion will be, is AMD heavily redesigning the interfaces for ATI chips so that they'll work though hypertransport? You promised we would get more information on fusion a year and a half ago but I've seen little. Does AMD expect better communication from CPU-GPU to increase performance in gaming? Fusion seems aimed entirely at the low to mid market pc's and probably laptops.
Thanks for the info on AVX. I read up some at Aceshardware too. It's hard for me to get too excited about expansions in the instruction sets only because it seems to take eons for these features to find there way into software programs. Still though, progress is progress so I am glad to see this part of progress is continuing.
Regarding the timeline. Since they are kind of staggered between the two companies, and they in turn stagger their own server/desktop releases, it probably makes sense to breakout the architecture releases by market segment...at least for Server and Desktop.
Consider the market availability lag between Harpertown and Yorkfield...call it 6 months if you are willing to call Q9450 and Q9550 "available now". (although it still seems quite rare to find Q9450's in stock, let alone Q9550's)
Barcelona and Phenom had a pretty sizable lag between them as well, although I'm not sure how much the TLB bug issue played into that.
So I agree Gainestown (server Nehalem) will likely be released Q4/08...and fight for server market share alongside Shanghai. What isn't clear to me is when Montreal (2 MCM'ed Shanghais) enters the scene. If it will be early enough to fight with Gainestown or if it will be late enough that it has to fight with Westmere.
For desktops I won't be at all surprised if Deneb and Propus (desktop shanghai) sees a desktop before Lynnfield and Havendale (desktop Nehalem) are finally shipping in volume. Sure there will be Bloomfield (Nehalem Extreme) out for desktops in Q4/08...but will Intel rush the non-extreme Nehalems to market and canabilize their Penryn sales? (current yorkfield situation says no they won't)
And remember that AMD is beholden to its shareholders to merely stop losing money, Intel is beholden to not let gross margins erode below 55%. Desktop Nehalems won't be cheaper to produce than desktop wolfdales/yorkfields (they will be more expensive to manufacture). Further eroding my confidence in their desktop timeline being anything like the server timeline at all.
The interface will be Direct Connect, not HT...
Certainly the first Fusion chips will be geared towards the laptops...
bryanW1995
Lifer
- May 22, 2007
- 11,144
- 32
- 91
intel would be foolish to try to rush nehalem into the mass-market. They'll intro the extreme editions/high end at $1500+, then they'll have plenty of time to get it right since penryn will continue to trounce shanghai. Worst case (for intel) scenario they intro a Q9750 or Q9850 and really ramp up the clocks, best case they just keep their current skus in place and offer the bloomfields for aigo and company.
I don't think that's a nitpick...that's actually my point. Intel stock takes a hit when the shareholders don't get their expected 55% or higher GM. You can bet they'd be priced about $2/share higher right now were their GM's in the 55-60 zone instead of the 50-55 zone.
From 2007: Intel: Gross Margins Aren't Heading Any Higher This Year
From 2008: Intel gross margin becomes NAND?s latest victim
Back when Intel announced they were going to compete with TI over HDTV technology (CES 2004) we didn't even break a sweat of worry over it because we knew what the gross margins were like in that industry (DLP) and we knew that as soon as the right people at Intel got a dose of the GM reality then they'd be jumping out of that market segment like it was netburst part duo.
Intel shows off giant screens
Intel delays first TV chip
Intel kills TV chip plans
What impressed me with Intel's foray into HDTV was that it seemed like someone didn't do the easy/obvious homework before they sunk a bunch of money into productizing the thing...in the end them coming to realize the "ROI" wasn't there is kinda silly to have be your hindsight and not foresight.
Back to gross margins...Intel's executive team must defend all business actions which take Intel towards a product mix that reduces GM's below 55%. They divested themselves of their mobile phone ambitions for identical reasons. To gain marketshare would have required them to downgrade ASP's to an unacceptable (to Intel) gross margin outcome.
I sort of agree with that, but with some caveats...
1. If their Revenue increased proportionally to their increased GM, then I think you are underestimating it...more like $5+/share.
2. If their Revenue remained the same, then that would have meant they lost marketshare during a period when they really shouldn't, and I think they might have dropped.
3. For the in-between models, I think you have it about right...
We've seen this kind of behaviour before from Barrett (and people complain about Ruiz...geez!). Remember when he started investing huge amounts into "Server Farms"? They lost a mint on that one too...
I agree...I have a feeling that Intel will be following AMD's lead and dumping their memory division in the not to distant future...
They already do...
I think what you meant to ask is will Fusion be competing in those markets...?
Certainly not in the first release, but it is more than possible that they will do so not too long after that, but it has to make sense financially for the customer (trying to sell a $1000 CPU/GPU to a gamer would have limited success no matter how fast it is)
The 2 most important areas for Fusion initially will be (as has been said) in the low end/low power systems (remember that Fusion will require much less power than a seperated system), and for the other end of the spectrum in the HPC area where they are already using CTM GPUs for high end computing.
Will we be getting a Fusion chip that kills at gaming? A very good chance, but not for awhile yet...
I know it isn't going to happen but I have long wished that the GPU market would become more like the CPU markets in that you'd have a GPU "socket" on the mobo alongside GDDR3/4/5 (just one obviously) dimm slots and an open market of GPU speeds and GDDR speeds/sizes.
Obviously the discreet video card sellers would prefer to sell you the package, just as Abit and Asus would love to sell you nothing less than a packaged mobo with soldered on CPU and soldered on pre-selected ram.
I wonder how fusion is going to handle the graphics ram end of the equation. Will the mobo have onboard GDDR? (hardwired or customer configurable)
bryanW1995
Lifer
- May 22, 2007
- 11,144
- 32
- 91
AVX will only double SSE performance if they build a 256-bit datapath. If they do what they did for SSE from its introduction in P3 up through Core, they'll run the 256-bit operations through a 128 bit pipe twice (P3 => Core did 128 bits in two 64-bit passes).
Originally posted by: bryanW1995
They already do...
Did amd just announce the phenom 10450 @ 3 ghz??? how'd I miss that???
What you missed m8 is that Intel still can't build an Enterprise server yet...
They really need Nehalem to work!
bryanW1995
Lifer
- May 22, 2007
- 11,144
- 32
- 91
hopefully intel will throw all they have at snagging the server/business market and amd will respond by making cpus that enthusiasts like us would love to have!
Naturally this cause-and-effect would likely not occur. (and I have no doubt you knew this when making your tongue-in-cheek comment
)If Intel decimates AMD's last major cash generator (GM's on the 8xxx opertons) then we could probably kiss goodbye the chances of seeing a 32nm node performance CPU from AMD that isn't targeted towards competing for Via's marketshare.
AMD needs cash over the course of 2009 to have any chance of investing and tooling up for 32nm in 2010. That cash has got to come from Shanghai and Montreal sockets in 2009.
Whatever cash the ATi brand is generating needs to be desparately reinvested back into furthering ATi GPU development and not cannabilized for subsidizing 32nm fab tooling or CPU development.
Have you guys seen this interesting interview comments made by David Kirk, Nvidia's Chief Scientist? (I got this from an XS forums thread, credits)
He makes some very astute and valid points. AMD has got too much of the WWII Germany analog going on right now (simmer down now Nemesis
) in that they are literally fighting a two-front battle with bigger/stronger enemies who themselves are getting to focus all their (superior on their own right) resources on just fighting AMD on a single front. Complete the anlogy with it snowing early in Russia (the R600 was released to a thud), and misjudging where Eisenhower was going to land his forces in Calais and not suspecting Normandy (C2D anyone).It's a desparate battle for survival, but at this current time it is too straightforward to play-out the timelines and the resource requirements and be reasonably confident it is a losing battle in all but a few scenarios. Does AMD develop the atom bomb before Manhatten island, or will they be run out of time and cash in an anticlimatic ending?
(please people realize I am not comparing AMD to Nazi/Hitler Germany, just making comparisons to the strategic mistakes and bad luck that led to the outcome of WWII based on my crappy TV knowledge gleaned from watching a buttload of the History Channel
)
Kuzi
Senior member
- Sep 16, 2007
- 572
- 0
- 0
Reading the interview gives me a scary picture. ATI hasn't made any money for AMD since they bought them, actually they lost a bit ($130 million). How can they compete with Nvidia if they have no money for R&D. Supposedly ATI will release a new GPU (R700) in a couple of months, which is based on R600 architecture, but with more stream processors, higher clock speed, separate shader and core frequency (what Nvidia has been doing for a while), and GDDR5 support. That is all good but it's probably not enough to compete with what Nvidia is preparing to release around the same time.
On the CPU front, it doesn't look like Intel will give AMD a break at all, and AMD does not seem to have anything that might have a chance to compete on the high end till Bulldozer, which is still a way of. Shanghai/Deneb at best may be a good competitor to Core 2 not Nehalem.
So AMD won't get a break on both CPU and GPU fronts, can they make enough money to keep up and survive selling budget CPUs and GPUs?
I'm sure no one wants to see AMD/ATI go under.
On the CPU front, it doesn't look like Intel will give AMD a break at all, and AMD does not seem to have anything that might have a chance to compete on the high end till Bulldozer, which is still a way of. Shanghai/Deneb at best may be a good competitor to Core 2 not Nehalem.
So AMD won't get a break on both CPU and GPU fronts, can they make enough money to keep up and survive selling budget CPUs and GPUs?
I'm sure no one wants to see AMD/ATI go under.
bryanW1995
Lifer
- May 22, 2007
- 11,144
- 32
- 91
not true. There are many who would like to see daamit go under. Most of those people, however, will be less pleased when their westmere cpu/ gforce 11800gtx computer costs $10,000 to build...
Kuzi
Senior member
- Sep 16, 2007
- 572
- 0
- 0
From The Inquirer New AMD Roadmap info
They have this to say about Bulldozer:
"The most distressing part is Bulldozer. Much was made of it not being on the roadmaps last December, but that was a tempest in a tea kettle. The real problems have come out since then, and the only way to say it is that the architecture is a huge flop.
No, this isn't to say that it won't be fast or meet every spec that it was meant to, simply that it can't be done on a 45nm process. This means that it will be on the 32nm node pushing it out to late late 2010 best case."
Not sure how much truth there is to this, but it does make sense to say Bulldozer will only be made on 32nm process, considering it's supposed to be 8 to 16 cores (huge). Also if this is true it means AMD only has K10.5 to fight Intel for the next few years till 2011, so it doesn't look very good, even with a 6-core K10.5.
Any thoughts on this Viditor, what happened to the dual and quad core Bulldozers?
They have this to say about Bulldozer:
"The most distressing part is Bulldozer. Much was made of it not being on the roadmaps last December, but that was a tempest in a tea kettle. The real problems have come out since then, and the only way to say it is that the architecture is a huge flop.
No, this isn't to say that it won't be fast or meet every spec that it was meant to, simply that it can't be done on a 45nm process. This means that it will be on the 32nm node pushing it out to late late 2010 best case."
Not sure how much truth there is to this, but it does make sense to say Bulldozer will only be made on 32nm process, considering it's supposed to be 8 to 16 cores (huge). Also if this is true it means AMD only has K10.5 to fight Intel for the next few years till 2011, so it doesn't look very good, even with a 6-core K10.5.
Any thoughts on this Viditor, what happened to the dual and quad core Bulldozers?
Personally I take this to mean AMD is doing the much-needed sandbag thing and naturally they are starting the dis-information cycle with Charlie and friends as a way to thank him for his past services. (dancing in the aisles, 10% layoff's, etc...all things that didn't help)
Under-promise and over-deliver. I hope this is the plan.
bryanW1995
Lifer
- May 22, 2007
- 11,144
- 32
- 91
charlie isn't exactly the best source for info, but if this is true I don't think amd is sandbagging right now
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter