Bondrewd
Veteran
Yeah.
Besides, they have core markets to worry about.
The AMD Execution Thread [2007 - 2017]
- Thread starter overclocked_enthusiasm
- Start date
-
- Tags
- amd
- Status
Besides, they have core markets to worry about.
Even though Larabee failed one could say they evolved it into Xeon Phi, but as a hybrid competitor the Phi coprocessor/accelerator did not have much traction say compared to interest in GPU accelerators and so has been EOL'd last quarter.
Larabee started back in 2007ish, the Larabee derived Xeon Phi came to light as a card in 2012ish, then we have the modern Xeon Phis albeit with PCIe accelerator-coprocessor option ended Q3 2017.
Maybe they finally realise the coprocessor approach was not working that well when competing directly against hybrid GPU solutions, hence the new strategy with Raja - but this will impact top down through many tiers just as Nvidia does with their GPUs.
TBH I also think there is a financial opportunity in a few more years with an evolved and matured GRID solution as offered/being continually developed by Nvidia, another area that would interest Intel along with the higher margin performance consumer side.
Last edited:
Graphics, too. Larrabee also says hello.
And now if they laid off a lot of their graphics engineers I dunno how they plan to build discrete devices, much less high-end ones. Kadouri won't be able to pull the whole load himself. 
digitalwanderer
Legend
You're assuming he's merely human, a fatal flaw in your logic. <sagely nods>And now if they laid off a lot of their graphics engineers I dunno how they plan to build discrete devices, much less high-end ones. Kadouri won't be able to pull the whole load himself.
In fact I heard from trustworthy source that he actually hand-drew a full blown ultimate GPU on napkins while in India before the sabattical, but it required Intels process to be manufactured so he had to jump ships. Now they're already half way done making the new GPU from those napkins.
el etro
Newcomer
I just wish that the globalfoundries 7nm LP EUV2 is ready at 2019, for the 25x20 monster APU can be achieved.
Bondrewd
Veteran
Don't expect EUV insertion before 2020.
We lost around 50-60% of one of our R&D tech product team to another tech company collaborating with us when the senior lead engineer jumped to them and then targetted others involved in the development, they went from having nothing to 3 years later a product that competed moderately well to another 2 years after that being a serious competitive headache; this was an incredibly complex technology to develop from scratch - cannot say much more than this otherwise would start to become too specific.Graphics, too. Larrabee also says hello.
They managed this without breaking their IP/trade secret related obligations as well.
The key is who Raja will target, which may also include specific engineers within Nvidia and other companies as well.
Working on a new fresh project with high levels of resourcing and under someone like Raja will appeal to quite a few; look how Jim Keller prefers to operate and move around for the project interest.
Of course anyone joining will need to weigh up the history of Intel-the current CEO and commitment to such a project/division like some in this thread mentioned.
And will Xeon Phi (and associated team) in the shorter term also end up under new division/Raja or at least overlap/work with them.
Last edited:
el etro
Newcomer
I just expect this 2020 EUV insertion is the EUV2 from the Anandtech article. TSMC will already have this 7nm EUV2 ready at 2019....
Bondrewd
Veteran
That's on paper & pure promises.
Real world EUV is ass.
But hey, at least ASML announced the availability of pellicles for 250W sources.
well from many internet reports, Raja is described as acting for him self, not a team player...
Moreover, I'm curious to know what exactly are his achievements. At S3, he ended up with Savage 2000 broken T&L. Not very glorious... At Apple, he basically integrated Imagination PowerVR into iPhone SoC. Not a groundbreaking task... Finally at AMD, he gave up after Vega fiasco lunch. Not something I will be proud of... I don't know the guy, I have nothing against him, he's surely something special to get these high profile jobs, but from the cold outside facts, I don't see what the hype is about...
Well, he invented texture compression, which is part of DXTC standard today. He also came with interesting solutions to substitute broken or missing features. E.g. custom filters for anti-aliasing or CrossFire.
Result of this „not groundbreaking task“ was, that Apple's SoCs had faster GPUs than competitors'.
Define "fiasco", please. AMD is able to sell every manufactured Vega. Let me remind the reason, why AMD fired Rory Read: AMD had a lot of products with not really bad reviews, but nobody bought them. AMD and distributors had stock of unsalable GPUs, which lost value, AMD had to offer financial compensation and lost money. Vega and Polaris doesn't have great reviews, but it sells quite well. By the way, Koduri announced his decision just a few days after Kaby Lake-G announcement. AMD's GPU in Intel's CPU is also unprecedented achievement.
Anyway, I think, that Eric Demers should never leave AMD. I'm also convinced, that neither Demers nor Koduri are responsible for current situation. What we see is a result of decreased budget for AMD's GPU division, result of Read's administration.
Bondrewd
Veteran
Well, duh, Read was sorta right, GPU division was not that profitable even in 2009-2010 while being a moneypit.
And AMD was totally not in position of throwing money at it until it finally starts outselling nVidia.
There are also other reports mentioning he is a genuinely nice guy.
Part of the problem with Radeon is budget and resources available, and some other vocal "marketeers" and leaking/dripping of information such as Roy Taylor.
I agree as VP he could had run a tighter ship just like the CPU Division,but not convinced this will stop engineers being interested.
That said I tend to think it was right for the budget/resources focus to be on the CPU Division at the point they did, but the reality is this has a knock-on effect what can be achieved by Radeon.
Opinions will be split tbh on whether someone is very good/successful or not.
But the enticement is not just Raja but the potential of a new project/division resourced well by Intel (weighing up if they fully commit).
Last edited:
What would Polaris and Vega sales look like without cryptomining though?
Unless someone at AMD was able to successfully forecast Ether price spike begging in early 2017 (and if so, God bless them) “they sell all they make” seems a little disingenuous (and in case of Vega, with all the production volume rumors abound could mean not that much, really). For what these GPUs were actually intended, namely competing primarily in gaming and professional market, they seem to have fallen rather flat. The fact that he was bypassed for executive bonus in the face of increasing sales speaks volumes about the internal perception of his success/lack thereof, crypto-driven bonanza notwithstanding.
Bondrewd
Veteran
That's why one should never trust Wall Street monkeys for anything technology-related.
They are clueless, really.
- Status
