Dirk Meyer leaves AMD... with immediate effect

Waaaaat? :(

Does anyone know if this has anything to do with the AMD product manager job posting?
 
Sandy Bridge...and Bulldozer estimated perf...hmmm...timing...I hope Bulldozer isn't a dud.
 
Sandy Bridge, nvidia jumping into ARM computing and investors are clearly loving it, weak follow up to 5800 series. The party keeps going (south).
 
Man I hope Bulldozer is awesome and blows Intel out of the water, we need AMD to do well. With all the focus being on Win8 for ARM and how it affects Intel, I think AMD are a lot more vulnerable. Fusion looks very nice though, I'm thinking of getting a C-50 based netbook or maybe an E-350 based ultraportable. Atom is definitely getting old now and AMD can probably corner the whole low-end low power market if they play it correctly. A whole bunch of us at the office are looking closely at how successful Zacate turns out, it could help turn the company around.
 
Sandy Bridge...and Bulldozer estimated perf...hmmm...timing...I hope Bulldozer isn't a dud.
Since reading articles about Bulldozer I'm not optimistic.
They will have 2 "simplified" cores which can run on higher freq but with "80% effective lower speed per core"
So they'll need 4GHz in order to equalize 3.2GHz SB.
After years talking about Fusion, AMD is late to Intel. As good as Zacate looks, Intel need not to worry - they already have SB parts with 25W TDP which have good enough video chip.
On GPGPU there is just 1 undisputable leader.
And Nvidia is pushing into ARM-space, which is direct threat for AMD. And few years ago AMD sold their mobile chip division in order to pay for ATi and go Fusion.
With all the combined force of AMD and ATi, they lost from Intel, and in same time NV managed to make Tegra starting from scratch.
Who can say this is good execution?
Is there a single field where AMD succeeded? Except selling foundries that is.
 
Last edited by a moderator:
The problem with cornering the low end of the market is low ASP. Even then AMD does not have the volume of chips Intel sells.
I am wondering if buying ATI has been beneficial to AMD. At the moment I would say no.
 
Doubt this has anything to do with Bulldozer.

AMD bet almost everything on ATI and Fusion then simply failed to deliver, pushing back their fusion products node after node. Now we get to Llano and it's late and using massively outdated K10 cores. Right at the time the first Fusion product arrives Sandybridge pops up and steals all the thunder.

Fusion was a great idea and is the way forward. Quite how many generations of chips AMD thought Intel would wait before responding I don't know.

By the time Llano arrives it'll be as out of date on the CPU end as using a K8 based core would have been in 2008. AMD lost the initiative and it's hard to see what they've gained in return.

Since reading articles about Bulldozer I'm not optimistic.
They will have 2 "simplified" cores which can run on higher freq but with "80% effective lower speed per core"
So they'll need 4GHz in order to equalize 3.2GHz SB.

Link?

If Bulldozer can run a module at 4 gHz for the same power consumption as a multithreaded SB core then that'll be okay. BD will be AMDs first 32nm part though, and up against chips made on a very mature Intel process.

After years talking about Fusion, AMD is late to Intel. As good as Zacate looks, Intel need not to worry - they already have SB parts with 25W TDP which have good enough video chip.

The 9W Bobcat part looks amazing, and the 18W parts will be cheaper than a Sandybridge part.
 
The 9W Bobcat part looks amazing, and the 18W parts will be cheaper than a Sandybridge part.
But they are not very fast. Faster than Atom - yes. Nothing more, a SB derivative at 1.5GHz will wipe the floor in same TDP. Being bit bigger is not problem for Intel - if they want the market for themselves.

As for speed of BD cores, there was article on realworldtech and posts from AMD employee claiming that on average, the 2 int-cores which share resources, 80% as fast as 2 of same cores which do not share resources.
 
As for speed of BD cores, there was article on realworldtech and posts from AMD employee claiming that on average, the 2 int-cores which share resources, 80% as fast as 2 of same cores which do not share resources.
You got it wrong - the second core is going to run at 80% of theoretical peak - a single core residing inside the module. So that's going to be 90% as fast on average, comapring to two "regular" cores, with 2MB L2 cache and doubled FPU.

Regarding Dirk Meyer, Anand has some speculation:
I was talking to Ryan Smith earlier tonight when the news broke and he mentioned something that caught my interest:
“Mark my words, at some point they're going to try to sell the company to ATIC. Let ATIC deal with fighting Intel, and the board/investors can pocket a nice profit. Certainly if you were trying to sell, you'd sack Dirk. He doesn't seem like the kind of guy that would let them do it.”
http://www.anandtech.com/show/4123/what-a-day-amds-ceo-dirk-meyer-resigns
 
I don't recall it being mentioned that the latest cross-licensing agreement between AMD and Intel allowed for AMD to be purchased without voiding the agreement.
 
But they are not very fast. Faster than Atom - yes. Nothing more, a SB derivative at 1.5GHz will wipe the floor in same TDP. Being bit bigger is not problem for Intel - if they want the market for themselves.

No one doubts that the low-voltage SB SKUs are significantly faster than Zacate. However, the price classes are completely different. The cheapest low-voltage SB starts at $250 (reference: http://techreport.com/discussions.x/20193) while you should be able to buy laptops with Zacate in the $400+ range indicating that Zacate is likely priced at well below $100, I would estimate in the $50 range. Zacate is at half the die size of a dual-core SB (75mm-sq vs 150mm-sq) and should be cheap to manufacture. So I don't think comparing SB and Zacate makes much sense.
 
I don't get many of AMD choices, it likes they didn't know about swot sometime. Which opportunity they think they had in fighting Intel in the high end? It's a bit of the same for GPUs even though nobody agree with me they should have given up on the high end too.
From CPUs to GPUs more projects and teams that they can afford, they are late or products fail to realize all the promises, when they do they suffer from the software.

Strength / Weakness / Opportunity /Threat, it's obvious they got wrong at this stage.

Strength: second best X86 expertise out there, great GPUs hardware teams
Weakness: process, software, financial resources
Opportunity: low power processing on X86
threats: Intel on every other X86 segments, Intel have the resources to react quickly, ARM and Nvidia

How it comes they do what they do? I don't know.
 
Dave's promotion and this aren't related, just oddly coincidental believe it or not. :p

There was me thinking that they had recognised that Dave can save anything and promoted him right to the top! :p

Anyway what is ATIC? The people who bought Global Foundries? If so I guess this could be a good thing as it seems that AMDs main issue is they don't have two pennies to rub together to spend on extra investment or growth. Which of course meant that as a company they slowly starved in such an investment heavy industy.
 
But they are not very fast. Faster than Atom - yes. Nothing more, a SB derivative at 1.5GHz will wipe the floor in same TDP. Being bit bigger is not problem for Intel - if they want the market for themselves.

The main trick is that its a APU. When it will use GPU with CPU together it could be a lot faster than SB on same clocks.
 
An investment group owned by the emirate of Abu Dhabi.
It supplied the funds for the spinoff and is the majority owner+piggy bank for GF.
If I recall correctly, AMD itself has gotten funding from them as well.

I *think* they own a chunk of AMD.
 
Back
Top