Does Martin Ashton drive an Aston Martin? 😛 Anyway, pretty weird to see both companies trading employees like this. Though, no offense to him but I think Raja was more valuable to AMD than Martin was to Intel.

schmidtbag:

Does Martin Ashton drive an Aston Martin? 😛 Anyway, pretty weird to see both companies trading employees like this. Though, no offense to him but I think Raja was more valuable to AMD than Martin was to Intel.

How do you know? What do you base that on?

yasamoka:

How do you know? What do you base that on?

Well for one thing, Raja was VP at RTG, he was the graphics director at Apple, and is now the chief architect for visual computing at Intel. He held both of the former positions for multiple years. Martin confuses me a little - I can't tell if he was VP of all of Intel, or VP of just visual computing. If it's the latter, that's not all that impressive, considering Intel's GPUs were never a selling point. PowerVR (the other big company he worked for), meanwhile, has never really been especially successful. Though, I think PowerVR's success was at its peak when he was still working there. Again, I'm not saying there's anything wrong with him or that AMD got themselves a bad employee. I'm just saying that he's not quite as prolific as Raja, and to my understanding, it is his shoes that are being filled in at AMD.

schmidtbag:

Well for one thing, Raja was VP at RTG, he was the graphics director at Apple, and is now the chief architect for visual computing at Intel. He held both of the former positions for multiple years.

I don't think Raja covered himself in glory while he was VP of the RTG. Under his tenure as VP the RTG failed to deliver Vega on time and then it fell short of everyone's expectations.

Herem:

I don't think Raja covered himself in glory while he was VP of the RTG. Under his tenure as VP the RTG failed to deliver Vega on time and then it fell short of everyone's expectations.

I can see why you'd say that, but that wasn't his fault. I heard how his team was split up to give priority to Sony and their PS5 graphics. Vega failed because he couldn't use his own resources.

Raja failed to deliver period. If you cannot deliver with the resources given to you, you out source or attain the needed resources. There are road maps followed and the critical time line junctions are highlighted. If a critical junction passes and nothing was done in time to alleviate the issue, it is on the managing director. If one fails to plan, they plan to fail. Why was Keller brought back into the fold for the cpu division? The necessary resources are acquired if they do not exist already internally. Like was stated, Raja failed. Why he was kicked to the curb. Execs at that level are not fired. They are told to resign or they will be fired.

Clouseau:

Raja failed to deliver period. If you cannot deliver with the resources given to you, you out source or attain the needed resources.

I think you misunderstood what I said... he had resources, but they were taken from him... So, what you're saying is kind of the equivalent of "I don't understand how you're poor, just have money!"

Do not be silly. Companies are not like you and me. Like I stated in my post, The necessary resources are acquired if they do not exist already internally. Companies do not tank their projects. The directing managers do. One cannot always be successful at whatever they do. That outing, he failed bad enough to get kicked to the curb.

Clouseau:

Do not be silly. Companies are not like you and me. Like I stated in my post, The necessary resources are acquired if they do not exist already internally. Companies do not tank their projects. The directing managers do. One cannot always be successful at whatever they do. That outing, he failed bad enough to get kicked to the curb.

Except that's not how it worked. Frankly, your speculation doesn't matter, because there's already evidence of what actually happened. Read the first paragraph: https://www.eurogamer.net/articles/digitalfoundry-2018-amd-next-gen-graphics-tech-built-for-ps5-report Again, Raja had the resources, but the Sony/PS5 team at AMD didn't, so, they acquired more resources: from him. You can't just magically buy more when your company is already in serious debt. Even if they were willing to put themselves in further debt, I can't imagine it'd be easy to convince engineers to work for AMD for [probably] unimpressive pay, and, when AMD's reputation has the potential to tarnish the engineers' too. Anyway, AMD probably prioritized Sony because they were an immediate source of income. Remember, Vega was still in development before Ryzen was released. The situation isn't so black and white. I know we want someone to point fingers at for how disappointing the GPU market has been the past 2 years, but I really don't believe he is the one to point at.

If you believe everything you read, this discussion is pointless. Companies do not sabotage their bread and butter. That report was put out there to make excuses for his failure without directly pointing a finger at him. That is counter productive morale wise. My point is not hey I point the finger at this person. Creating a counter productive culture in the work place by back stabbing employees or airing the company's laundry in public would do just that. When the head of a division fails to such an extent they are kicked to the curb and excuses made so investors do not unload is what that linked story is about. It's too picture perfect to swallow. Remember, he went on sabbatical and then decided to leave and months later that story paints a story that it was not his fault. Buy it in its totality. This tangent is far off the op. My part in this is done.

Clouseau:

If you believe everything you read, this discussion is pointless.

No... what's pointless is to make speculations based on nothing but personal opinion. Say what you want about the sources, but it's better than nothing. As far as I'm aware, nobody else from AMD has disputed them. Besides, why couldn't I just take your advice and use it against everything you're saying?

Companies do not sabotage their bread and butter.

Except they didn't sabotage their bread and butter... they didn't have the money to do both, so, they catered to what gave them the bread first, and held off on the butter. Whether or not you believe the article, the point still remains: AMD didn't have the money to make Vega better. The specifics as to how that happened largely don't matter, but whether you believe Raja's team was cut down or not, he clearly didn't get the funding he wanted to make it work.

Personally, I think they are trading losers with losers. Raja failed at Vega and Martin failed at graphics for Intel. Although, all they can do now is tell each others company some "trade secrets" about what the future will hold.

thank you so much for adding the names under the pictures, I wouldnt ever tell which one is Ashton and which one is Wang 😛

schmidtbag:

I can see why you'd say that, but that wasn't his fault. I heard how his team was split up to give priority to Sony and their PS5 graphics. Vega failed because he couldn't use his own resources.

Vega Failed because it delivered one single improvement since Polaris. That's Packed Math/FP16. That 10% improvement... I call it BS, you do not need to add 30% more transistors to get 10% higher clock. They for sure eat energy and spit heat, therefore it is very costly change in terms of power efficiency. All other improvements which were still there on launch day seem to go under and away. Polaris delivered good things. People can see them in benchmarks. And with recent Polaris 30 rumors, seems like Vega is not worth it in consumer market. And AMD may want to get some representation with Polaris 30.

Fox2232:

Vega Failed because it delivered one single improvement since Polaris. That's Packed Math/FP16. That 10% improvement... I call it BS, you do not need to add 30% more transistors to get 10% higher clock. They for sure eat energy and spit heat, therefore it is very costly change in terms of power efficiency. All other improvements which were still there on launch day seem to go under and away. Polaris delivered good things. People can see them in benchmarks. And with recent Polaris 30 rumors, seems like Vega is not worth it in consumer market. And AMD may want to get some representation with Polaris 30.

For me Vega has lower clock per clock performance than Polaris, Vega is just a bit over Fiji and sometimes under https://www.bitsandchips.it/9-hardware/7334-tonga-vs-polaris-sfida-clock-to-clock?start=2 https://www.computerbase.de/2016-08/amd-radeon-polaris-architektur-performance/2/ https://www.Disney.net/guides/2977-vega-fe-vs-fury-x-at-same-clocks-ipc (interesting, post changes gamers-nexus link to disney lol) all an deception greetings

I would like to further add to my previous statements: If Raja was as incompetent as people made him out to be, why would Intel hire him for such a big project? Keep in mind his position went from VP of graphics to chief architect. To me, that sounds like he's more involved the architecture design, so, clearly Intel knows what he's actually capable of, given proper funding. Also keep in mind, he pretty much immediately transitioned from AMD to Intel. That means Intel either coaxed him to join, or, he talked his way into Intel. Whichever way you look at it, Intel agreed that he is valuable to them.

Fox2232:

Vega Failed because it delivered one single improvement since Polaris. That's Packed Math/FP16. That 10% improvement... I call it BS, you do not need to add 30% more transistors to get 10% higher clock. They for sure eat energy and spit heat, therefore it is very costly change in terms of power efficiency.

I don't disagree with any of that, but I'm not sure I understand how that relates to what I said. It's merely the specifics as to how/why Vega failed. Vega was incomplete, so, one could say FP16 and HBM2 was the most progress AMD made on it before they ran out of time and resources. Anyway, those extra transistors still seemed to accomplish something, just apparently not much for gaming. There are some tests were Vega is underwhelming and horribly inefficient, and then there are others where it outperforms a 1080Ti (either in raw numbers or an efficiency standpoint).

I think you get too wrapped up in titles @schmidtbag . You make good points, but no one really knows what went on with Raja leaving and being rehired. We all can make a guesses on bit of info, and that is pretty much all.

schmidtbag:

I would like to further add to my previous statements: If Raja was as incompetent as people made him out to be, why would Intel hire him for such a big project? Keep in mind his position went from VP of graphics to chief architect. To me, that sounds like he's more involved the architecture design, so, clearly Intel knows what he's actually capable of, given proper funding.

He apparently has some operational knowledge and keeps his primarily compute aspirations. I think intel got him to make good compute GPU, not gaming one. Intel still keeps almost entire CPU server business. But they have entire GPU server business to take.

schmidtbag:

I don't disagree with any of that, but I'm not sure I understand how that relates to what I said. It's merely the specifics as to how/why Vega failed. Vega was incomplete, so, one could say FP16 and HBM2 was the most progress AMD made on it before they ran out of time and resources. Anyway, those extra transistors still seemed to accomplish something, just apparently not much for gaming. There are some tests were Vega is underwhelming and horribly inefficient, and then there are others where it outperforms a 1080Ti (either in raw numbers or an efficiency standpoint).

Even with limited resources, Vega delivered so few changes. Going from HBM to HBM2 IMC is not exactly hard task when AMD was making that design with jedec. There are quite a few hints telling if he left as he felt he is no longer needed in AMD or if he was given some easy choice and left as result. Now I consider it as 30:70 for left:made-to-leave. If we'll really see Polaris 30. It would be like 95% for him over-promising and under-delivering.

This kind of threw me off a bit with the name.

Embra:

I think you get too wrapped up in titles @schmidtbag . You make good points, but no one really knows what went on with Raja leaving and being rehired. We all can make a guesses on bit of info, and that is pretty much all.

Honestly, I'm not really trying to get wrapped up into it, but people keep questioning me, and therefore I answer. Do I need to? No; if I had things my way, I'd have had 5 fewer posts (including this one). But I'm not the type to have someone tell me my thoughts are silly with generalizations as the means to dispute my claims, and just walk away. Anyway, I agree that there isn't concrete evidence on what happened, and that guessing is all we can do. That being said though, this is why I spoke up in the first place - a crude guess was made about Raja, which I deemed misinformed given the evidence we do have (albeit, wavering), so, I spoke up.

Fox2232:

He apparently has some operational knowledge and keeps his primarily compute aspirations. I think intel got him to make good compute GPU, not gaming one. Intel still keeps almost entire CPU server business. But they have entire GPU server business to take. Even with limited resources, Vega delivered so few changes. Going from HBM to HBM2 IMC is not exactly hard task when AMD was making that design with jedec.

I agree with all of this. Though, I think you may be slightly underestimating the amount of changes that went into Vega. The reason I say this is because the driver stability was pretty atrocious on release day (specifically, for Vega FE). Before someone jumps in saying "hurr durr AMD always makes crappy drivers on release day" this was no ordinary situation. If an architecture is mostly the same, the drivers ought to have been much more stable, since I imagine most of it would just be copy+paste. But, maybe I'm wrong.

There are quite a few hints telling if he left as he felt he is no longer needed in AMD or if he was given some easy choice and left as result. Now I consider it as 30:70 for left:made-to-leave. If we'll really see Polaris 30. It would be like 95% for him over-promising and under-delivering.

I'm interested in what you feel those hints are. Knowing you, your thoughts tend to be more focused on evidence and context than most people. That doesn't mean I'm going to agree, but given when Embra said, all we can do is guess, so I see this as nothing more than a friendly debate.