chavvdarrr
Veteran
This exactly brings bad news about Buldozer. One doesn't change its CEO over night, but over 2 moths looks likely
Since Jerry Sanders left, almost everything they tried went from bad to worse and late.
The AMD Execution Thread [2007 - 2017]
- Thread starter overclocked_enthusiasm
- Start date
-
- Tags
- amd
- Status
Since Jerry Sanders left, almost everything they tried went from bad to worse and late.
itsmydamnation
Veteran
wow can you be anymore negative?
on point one, you think they got silcon back pretty much the day they demo'd it?
yet you have anand reporting they had bulldozer silcon back in august(who knowns when they actually got it back from the lab) so if bulldozer was that bad it takes 6month to blame the CEO and get rid of him? I also seem to remember john fruehe around that time getting quite anoyed with people saying when they got silicon back and at launch he would say when they did and that a "few" people would be supirsed.
Remeber GF have had 32nm production ramp issues, if they didn't we would have likely seen both llano and bulldozer earlier then what we are.
on the second point. the only really bad execuition AMD has had is agena(what a fuck up
) . The big problem they had is they didn't have a road map or a plan to improve there cores/IPC in the time of the athlon 64/X2. that bit them in the arse big time and its taken time to get a new design (which is hardly based on there exsisting core at all), hopefully they will now maintain momentium and get bigger jumps in IPC and or clock rate with each new version compared to STARS.there hasn't been anything wrong with deneb, thurban, magny-cours, Istanbul or Shanghai . There have been some good improvements along the way like HT assist, but with AMD's limited resources it likely was to much of a strech to make major non transferable imporvements to STARS while at the same time designing bulldozer.
when P4 hit the wall intel where lucky they had another X86 design floating around that looked the goods (core), if Intel didn't have core it would have taken them longer to get to the position they where in with conroe, AMD wasn't so lucky......
Last edited by a moderator:
itsmydamnation
Veteran
i dont mean luck as undertermined/indescriminent, i mean in, its just the position intel was in, lots of money, lots of power, lots of R&D. I dont know how much of the stuff on the net is true but a lot of people within intel seem to have as small determined bunch of people to thank. from what is on the net as well, i wouldn't call it prior planning, remember intel was always talking about 10ghz processors, IA-64 was to replace X86. none of those plans came off to well.
Intel's large R&D budget allows them to have many concurrent teams working on the same problem.
Unfortunately AMD do not have that luxury meaning that when they get it wrong it takes a lot longer for them to turnaround with a new product.
Add to that the fact that processor developement (x86 especially?) takes many years.
Unfortunately AMD do not have that luxury meaning that when they get it wrong it takes a lot longer for them to turnaround with a new product.
Add to that the fact that processor developement (x86 especially?) takes many years.
Well lets ponder the two possible scenarios for BD.
1) It actually is largely competitive with Ivy Bridge, particularly in servers. I'm not sure that there's a chance in hell of this happening but we'll see. They bring out $1000 FX chips again. Server market share increases nicely.
2) It's the new Phenom/Phenom2/K6-2 and is forced to sell from $50 - $300. It's the budget "value" processor.
If it takes ~5 years to bring in a new architecture, is #2 pretty much the end of AMD at this point? They would probably not be able to make any significant inroads into the server space that is really the main point to BD. Can they continue for years essentially in the same position they are right now?
I don't see their other push, the netbook CPU/GPU or budget notebook CPU/GPU, being all that beneficial really. It's not particularly different in notebooks than what they've had for the past few years with 740G/780G. It's just the next step in marginal 3D IGP performance and I don't think that excites anyone other than tech site readers. 780G was "thrilling" and a "competitive advantage" at one time too, and so were 690G, Xpress 200, etc.
1) It actually is largely competitive with Ivy Bridge, particularly in servers. I'm not sure that there's a chance in hell of this happening but we'll see. They bring out $1000 FX chips again. Server market share increases nicely.
2) It's the new Phenom/Phenom2/K6-2 and is forced to sell from $50 - $300. It's the budget "value" processor.
If it takes ~5 years to bring in a new architecture, is #2 pretty much the end of AMD at this point? They would probably not be able to make any significant inroads into the server space that is really the main point to BD. Can they continue for years essentially in the same position they are right now?
I don't see their other push, the netbook CPU/GPU or budget notebook CPU/GPU, being all that beneficial really. It's not particularly different in notebooks than what they've had for the past few years with 740G/780G. It's just the next step in marginal 3D IGP performance and I don't think that excites anyone other than tech site readers. 780G was "thrilling" and a "competitive advantage" at one time too, and so were 690G, Xpress 200, etc.
Last edited by a moderator:
While they could technically continue (they aren't losing buckets of money or anything), the shareholders probably wouldn't allow it, they'd wind up getting bought out or split up or something.
Although I'm not sure everything is riding on bulldozer, a success in other area's could carry them for a while.
I certainly hope bulldozer is at least somewhat competitive.
Silent_Buddha
Legend
The group that bought Global Foundries would be one logical candidate. They don't particularly have a lot of talent in CPU or GPU design, but they certainly have a LOT of cash.
I'm not sure they would be any better at competing with Intel. And if they can't, then they may lose interest in sinking money into the corp. and lose interest thus relegating AMD to eventual oblivion.
But I kind of see that as a distinct possibility no matter who buys them. IBM and Motorola have already shown they really can't compete head to head with Intel. VIA and its parent company have just plain avoided going head to head with Intel.
There are a few upsides to the AMD CPU side. The Brazos platform is currently better than the Pinetrail or Ion2 platform and might be as good or better than the Oaktrail platform. Higher performance Fusion APUs could liven things up if compute shaders and/or OpenCL applications were more prevalent but unfortunately for AMD it's currently not there yet. At the least it may bring current gen gaming to everyone, at reduced but not castrated graphics settings. But it's hard to see the CPU side consistently competing well with Intel. Perhaps loads of cash would help... Bulldozer is an unknown at this point so no point discussing that.
The AMD GPU side is far more interesting. The prospects there when looking ahead are a bit brighter than the CPU. Nvidia and AMD's GPU division (god I still hate that they ditched the ATI brand) is basically neck and neck with AMD being more efficient and Nvidia having higher performance at a much larger die and power budget. It's possible a company many try to maneuver to get AMD to divest itself of the GPU division and purchase all rights related to that. This opens up the possibility of Intel getting in on the picture. They did have good relations with ATI in the past.
Imagine the former ATI with the level of R&D Intel could fund.
Regards,
SB
Well that's what Phenom did. The problem is that Phenom II wasn't enough of an improvement and couldn't keep pace with Yorkfield, Nehalem, etc. I think Phenom II is actually still sometimes slower per clock than Kentsfield and I'm not sure it scales higher in clock speed either. AMD has bumped the voltage for the top models and Intel hasn't exactly pushed the limits of their various processes in the past few years.
However, BD is a rather brand new and innovative design so there are bound to be more opportunities for improvement than with Phenom which was essentially a very refined Athlon K7.
Lets just hope that BD is not a disaster right now. It can't suck too much at version 1.0 or they are probably doomed.
Intel has a strange level of interest in graphics. Their tech isn't bad at all anymore but they don't seem to want to do more than build IGPs and deliver that basic level of performance. Competing with ATI or NVIDIA would probably not be much trouble for them if they cared to. For some reason they don't seem to want to really go there. I wonder if its corporate culture and CPU tradition getting in the way or what. I'm not sure they'd be interested in DAAMIT...
The biggest issue with AMD in the consumer side is that their main advantage in terms of product is directly in an area where most people don't understand why they actually need it, graphics. Hopefully we'll see some good numbers coming for them in the next quarterly report and hopefully we'll see Bulldozer as a high performance direct competitor to Intel because right now we need it.
Silent_Buddha
Legend
They do try from time to time to go head to head. But as yet haven't been all that successful. The i740 was originally meant to compete for top graphics card spots but was let down due to a variety of factors not the least of which was that AGP texturing was a performance and evolutionary dead end (ATI was also hurt by this as they had been banking on it too) combined with RAM prices dropping through the floor which allowed 3dfx (and others) to make accelerators with large local pools of graphics memory.
And they tried again with Larrabee, but weren't able to get that where they wanted.
Basically Intel haven't been as good as other graphics chip makers at predicting where the market will go. And when it doesn't pan out their attention wanders.
They do at least "appear" to have a renewed interest in the graphics sector as they've done a pretty good job of making robust drivers for the Sandy Bridge GPU. Whether this indicates they have higher asperations or not is unknown.
As well purchasing the AMD graphics unit would instantly give them a fairly performant GPU accelerated Compute Shader/OpenCL part.
But yes, I still rank this fairly low on the buyout chart. Not only due to it weakening AMD whom Intel would like to stick around as a low end CPU competitor, but it would also make it harder to sell the AMD CPU division.
Regards,
SB
No, that wasn't clear.
Honestly, that statement was so vague that even if it's somehow "true", it's probably best to disregard it entirely.
Hence the reason for asking if it panned out into anything of substance.
Not as far as I know. I doubt we'll get any concrete performance information until the actual launch, or very close to it.
- Status
