Raja: AMD cannot compete

[Trixanity]

6 minutes ago, leadeater said:

Also qualitative vs quantitative assessment too. 

Of course. My point was basically that's it's a good idea to look at things taking into account multiple factors instead of looking at just one and saying "this is good" or "this is bad". That's what happens so frequently when you see people point to statistics with no context and make a statement from that alone. Often it's a lot more nuanced and sometimes it just becomes misleading.

This is all generally speaking but I think it applies here as well. 

[Doobeedoo]

He did work on Navi as last thing no. Full core Navi should perform good enough over Radeon VII though. But that GPU being released much later makes no sense. At least when we look the timeframe for "Next gen" is supposed to come too. 

[Trixanity]

16 minutes ago, Doobeedoo said:

He did work on Navi as last thing no. Full core Navi should perform good enough over Radeon VII though. But that GPU being released much later makes no sense. At least when we look the timeframe for "Next gen" is supposed to come too. 

Don't get your hopes up. It's still GCN.

[Doobeedoo]

5 minutes ago, Trixanity said:

Don't get your hopes up. It's still GCN.

I'm aware it's still a GCN derivative though should give a solid boost since it's still improved architecture wise. Radeon VII was just Vega on shrunk process and gave like around 25% performance just from that. So Navi should at least have the same improvement I take it. 

[Trixanity]

2 minutes ago, Doobeedoo said:

I'm aware it's still a GCN derivative though should give a solid boost since it's still improved architecture wise. Radeon VII was just Vega on shrunk process and gave like around 25% performance just from that. So Navi should at least have the same improvement I take it. 

A lot of it was clock speed though. So memory plus clock speed should be the primary drivers of performance versus Polaris. Beyond that I don't see any reason to believe it'll impress.

[Doobeedoo]

9 minutes ago, Trixanity said:

A lot of it was clock speed though. So memory plus clock speed should be the primary drivers of performance versus Polaris. Beyond that I don't see any reason to believe it'll impress.

I'd expect it will be whatever how much faster than Radeon VII but also with lower price. We'll see. Maybe it's not a true successor to it but a bridge-gap before so called "Next-Gen" is ready I guess. 

[thorhammerz]

7 hours ago, LAwLz said:

You seem to think that anyone who isn't sucking AMD's dick is against them. It's not critique to point out that statements such as "AMD contributes more to open source than Intel" is incorrect.

Zealots are generally not known for their objectivity. Especially this one .

[thorhammerz]

8 hours ago, Stefan Payne said:

You shouldn't criticize AMD for doing a decent job, with the Resources they have, you should criticize the ones that don't although they have a pretty big R&D Budget but act rather Egomaniac...

The world of business cares about results - quite different from grade school where "effort" gets condolence grade points.

[Silentprototipe]

Raja is Traitor 

[Tenelia]

Well, if Raja were that good... Lisa won't be seen as such a superb leader now. Let's see what Raja actually contributes at Intel in the next 5 years, especially in light of all the fumbles by Intel's team on their major R&D projects.

[samcool55]

The last time he did a bald statement it didn't go well did it?

Anybody remember #poorvolta? That ended well... /s

[maartendc]

13 hours ago, Nowak said:

But what about Via Technologies? There's a third company with the x86 license out there, remember.

Wait do these companies still pay IBM for a license on that x86 stuff? I had no idea...

[maartendc]

Well that is embarrassing, Intel must be paying him a LOT of money to basically say:

"Yeah, you know all that stuff I used to make decisions on / be in charge of? That was all trash!!"

[yian88]

15 hours ago, DrMacintosh said:

It's the Intel money talking I think. It's best not to listen to Raja until he can come to his senses. 

He might be referring to something specific in his mind but generalizing in his blabbering after intel threw some $ at him, as AMD increases in market share and competitiveness and new products launching this year that will blow intel away they have to do something, sadly they used Raja and ruin his reputation.

Raja might be referring to intel's compiler framework and other cpu optimization/library software intel has. It cant be talking about graphics, intel has little on the graphics front compared to AMD and nvidia is light years ahead of both.

 

[XenosTech]

6 hours ago, Humbug said:

The other rumour was that he was upset that Lisa was pulling all his engineers off from the Vega project to work on Navi which he blamed for Vega being sub-par. So he decided he was not gonna stick around once Vega was done. But that's all we have, rumours...

 

Obviously the sabbatical was fake, in reality it was cause he was joining Intel.

I heard it the other way around since Navi was his love child as vega was already in the works by the time he was put in charge of RTG and vega needed all hands on deck cause it wasn't looking too pretty

[Humbug]

5 hours ago, XenosTech said:

I heard it the other way around since Navi was his love child as vega was already in the works by the time he was put in charge of RTG and vega needed all hands on deck cause it wasn't looking too pretty

That's true but he still had to take responsibility for Vega's poor performance. As he had been at AMD a few years by then. So it reflected very poorly on him.

[moriel5]

12 hours ago, leadeater said:

Also qualitative vs quantitative assessment too. You can be submitting a lot of work but if it nets little benefit or use to the wider world or only helps your own products then how much value do you actually bring to the open source ecosystem. If everything Intel does is to benefit only themselves and their products and has no value to anything else what are they achieving, same goes for AMD and Nvidia.

 

No matter how much Intel submits to open source software I still see significant efforts in closed ecosystems like OmniPath interconnect rather than contributing to the many existing open standards consortiums made up of all the other industry players sans Intel and Nvidia. Intel very much has an open door policy, but puts a bear trap in the door way. "Come in, we're happy to hear what you have to say". *slam* *leg trapped*.

Yes and no.

Prime examples are Intel sound and WiFi on Linux, where the former had next to no support from Intel until relatively recently, the latter has had spectacular support for years on end.

 

However both have been using, still are, and from the looks of things, are just going to rely on even more, closed source firmware.

The Intel ME fiasco is another good example, even better than the others.

 

So it is not just locking people into proprietary systems (though also that, just to a lesser degree), but more about moving things as much as possible, to systems where they technically follow standards, but in reality are trying to to do so via proxy, adapters, etc.

[leadeater]

26 minutes ago, moriel5 said:

So it is not just locking people into proprietary systems (though also that, just to a lesser degree), but more about moving things as much as possible, to systems where they technically follow standards, but in reality are trying to to do so via proxy, adapters, etc.

I'd put that under the bear trap in the door way situation.