Raja Koduri, Executive Vice President & Chief Architect, Leaves Intel

[HairyLobsters]

He has only been there 5 minutes, they released one generation and hes gone? wow.

Also helped with EVO graphics, if that counts.

 

[GerKNG]

Using the, want to be closer to family again

translating from corpo speech to english:
he was fired.

 

[cvaldes]

We need a betting poll and pool to predict where he goes next.

Semiconductor startup in stealth mode would be the odds-on favorite.

 

[bug]

Some more details here.

Raja Koduri, Main Driver Of Intel’s GPU Efforts, To Leave Company | CRN

Raja Koduri tells CRN why he is leaving Intel as the company's chief architect after leading the buildout of its graphics capabilities for more than five years.

www.crn.com

So, between his medical condition and considering retirement, he goes on to start his own company. That doesn't quite add up, but I wish him all the best nonetheless.

 

[mechtech]

Oh dear, it's a shame as GPU market could use a third player.

So could the x86-64 CPU market

 

[Daven]

So could the x86-64 CPU market

Not possible until Intel is close to bankruptcy

 

[trsttte]

Generative AI for gaming? What does that even mean? Is it for image quality? World generation? ... what??

Read the recent press releases from Nvidia, tools to help with game development, creating assets and stuff like that.

The GPU market has many players (Apple, Qualcomm, ARM, Samsung, Matrox, Mediatek, Imagination Technologies, etc) What you mean is a dedicated GPU you can buy as a component in a box so you can build your own PC and play games on it. Unfortunately AMD and Nvidia seem to be the only ones putting much effort in this space.

There's many players, but none of them is competing with AMD and Nvidia in their key markets, and I don't mean just gaming gpus, server accelarators like the AMD MI300 or NVIDIA’s Grace Hopper soc's, I don't see anyone competing with that. Intel is trying to go there with Ponte Vecchio but I don't how much success that's having

 

[Minus Infinity]

Even MLiD posted that Raja was in deep shit with the company over Alchemist back in mid-to-late 2022. The writing has been on the wall for a long time. Will the gpu division continue or will the goals be scaled way back, maybe discrete is dead and Battlemage and Celestial only will come in iGPU form.

 

[Crackong]

We all know what happened AFTER Raja left AMD, AMD gone sky-rocketed.
I guess Intel is taking her chances now.

 

[phanbuey]

Its not that he does a bad job... it's more that he promises and amazing job at a ridiculous timeline and then lies about it when it's clearly not going well to keep the show going.

It's hard to pin it just on him, but this type of thing seems to follow him where he goes, so it's probably the culture he cultivates with his team.

ARC graphics didn't get disbanded for poor performance, ARC got disbanded for lying to management.

 

[Jism]

His targets are just too high. Polaris was aimed to compete with a 1080. It was more like 1060. The Arc was aimed at a 3060, but alot of issues, inexperience drivers and quite higher power draw makes it so / so product. It has some advantages, but does not justify buying a Intel GPU right now. It needs a few revisions, and by that time it could actually be something.

 

[Icon Charlie]

From a business perspective I see things a different way.

I think he got shown the door. Reasons? Well there are plenty. And if you are an grognard like me you knew about this man and his... actions???

heh heh heh

 

[Frick]

I expect Intel to continue with Battlemage and Celestial. What I don't expect much from is the Koduri startup. Generative AI for gaming? What does that even mean? Is it for image quality? World generation? ... what??

Roblox is working on it as well.

Are Roblox’s new AI coding and art tools the future of game development?

New initiative aims to take game development past “the hands of the skilled few.”…

arstechnica.com

"In an example Corazza presented at the conference, a user could ask the system to "make orb turn red and destroy after 0.3 seconds when player touches it." The system then generates a seven-line Lua function that does just that, based on a coder-defined orb object provided earlier in the code.
Another prompt for a function to "create a 3 by 3 grid of orbs around orb" similarly generates a few lines of code to place a small grid of those objects in the game scene."

 

[GreiverBlade]

"He intends to form a brand new startup operation that will focus on AI-generative software for computer games."
in short, he will ask ChatGPT for ideas/concepts on how to do that for games?

i wonder how big/exaggerated his promise will be, this time.

 

[Dirt Chip]

I fully expect Intel to cancel ARC in the next year and concentrate on iGPU as well as Ponte Vecchio. I’m guessing Intel will squeeze out a few more dollars using the Alchemist refresh, breakeven then call it a day.

I think they will keep it for good, certainly in the next 5 years.
Guess we'll see

 

[ZoneDymo]

Even MLiD posted that Raja was in deep shit with the company over Alchemist back in mid-to-late 2022. The writing has been on the wall for a long time. Will the gpu division continue or will the goals be scaled way back, maybe discrete is dead and Battlemage and Celestial only will come in iGPU form.

Imagine unironically referencing MLiD....dear lord.

 

[R-T-B]

Hi,
Using the, want to be closer to family again

I mean that's not at all what anyone claimed, but yeah.

 

[Chomiq]

He did his part, he moves on. It's not the 50's or 60's anymore. People don't stay at the same job for the rest of their lives.

 

[BoboOOZ]

This is just his usual job cycle.
No point taking it to the realms of "ARC is doomed" or "Intel is on bankrupt course" kind of shenanigans.
He done his part, now moving to the next interesting thing. That's about it.

Please tell me it is sarcasm

 

[qlum]

I don't know if he was good or bad for intel, but his departure could signal intel wanting to pull out from it's gpus especially since its not doing great as a company.

 

[sephiroth117]

Oh dear, it's a shame as GPU market could use a third player.

Intel is working on it, with or without Raja

 

[wNotyarD]

Please tell me it is sarcasm

Actually, there isn't much room for sarcasm in that statement.
Looking at AMD, his cycle involved Hawaii, Tonga, Fiji, Polaris and Vega until he got the boot. All of them missed their performance targets big time, but they did set the foundation for current AMD graphics. Say whatever you want about raytracing, but current AMD options are completely competitive.
If Intel execs have any brains, they shouldn't kill Arc graphics. But for that to succeed it would take at least two more generations working over Raja's designs until they can rework it without his bias.

 

[Daven]

Intel is working on it, with or without Raja

The article and this whole thread is calling into question whether or not ‘Intel is working on it.’ It is unclear to many given Intel recent moves such as planned layoffs, product cancellations and refreshes.

As I posted earlier, I believe Arc will be cancelled after the Alchemist refresh.

 

[BoboOOZ]

Actually, there isn't much room for sarcasm in that statement.

Well, it's obvious that he was put in a golden cupboard at Intel before going to some startup, that's not a good way to go out. That's a nice way to fire somebody, but that's about it.

Looking at AMD, his cycle involved Hawaii, Tonga, Fiji, Polaris and Vega until he got the boot. All of them missed their performance targets big time, but they did set the foundation for current AMD graphics. Say whatever you want about raytracing, but current AMD options are completely competitive.

RDNA is his conception, and yes it's competitive. What people reproach Raja while at AMD is the stupid marketing hype part, remember "poor Volta"? I haven't heard anybody say that RDNA is a bad architecture.

If Intel execs have any brains, they shouldn't kill Arc graphics. But for that to succeed it would take at least two more generations working over Raja's designs until they can rework it without his bias.

Yes, Intel should keep ARC if they afford it. But if it came out on time, it would've sold great at the MSRP, because any potato GPU sold well in that period. Coming out late, it's losing a lot of money for Intel and, unfortunately, they can't really afford these losses.
So yeah, there was plenty of room for sarcasm in that post, a missed opportunity.

 

[stimpy88]

In fairness, he couldn't design a GPU. The whole project he was head of was a shit-show, from silicon to drivers.