News Intel GPUs - Falcon Shores cancelled

[gdansk]

MLID said today that Intel stopped producing B580 already.

It's believable but since he said it I'll disbelieve it.

 

[511]

MLID said today that Intel stopped producing B580 already.

I will always believe opposite of whatever he said 🤣

 

[511]

XeSS 2 Released

Release XeSS SDK 2.0.1 · intel/xess

Introducing XeSS Frame Generation (XeSS-FG) Introducing Xe Low Latency (XeLL) XeSS Super Resolution (XeSS-SR): DirectX 12: Backwards compatible API with XeSS 1.0 & 1.1 & 1.2 & 1.3 Introduced Vulka...

github.com

 

[511]

Looking Ahead at Intel’s Xe3 GPU Architecture

Examining software changes for hints on what Intel's next GPU architecture may bring

chipsandcheese.com

Xe3 Expected Changes

 

[igor_kavinski]

Announcing DirectX Raytracing 1.2, PIX, Neural Rendering and more at GDC 2025! - DirectX Developer Blog

Today in the DirectX State of the Union at GDC 2025, we proudly showcased the next evolution in graphics with the announcement of DirectX Raytracing (DXR) 1.2. This update promises groundbreaking performance improvements and breathtaking visual fidelity, marking another milestone in our mission...

devblogs.microsoft.com

Neural texture compression for B580!

Intel Co-Presents Cooperative Vectors with Microsoft at Game Developer Conference 2025

This technical deep dive was written by Jonathan Dupuy, and is a joint work with Anis Benyoub, Laurent Belcour, Mariusz Merecki, and Thomas Chambon as part of their research efforts at Intel Labs. Highlights: Microsoft will showcase Cooperative Vectors on DirectX during the Advanced...

community.intel.com

Thanks to the new DirectX Cooperative Vectors API, we can directly use AI hardware and speed up the computation by up to a factor of 10 on Intel® Arc™ (B series) GPUs.

EXCITING!!!

 

[vanplayer]

G31, dead.

https://twitter.com/x/status/1905055296839700601

it's died earlier than my older post.

Big die of Battlemage is not looking good. Latest ES was poorly done, it's not ready. Mid 2025 is still possible but don't be surprised if it is canceled.

 

[511]

G31, dead.

https://twitter.com/x/status/1905055296839700601

it's died earlier than my older post.

It died in Q3 but was revived in December many people don't know that

 

[poke01]

It died in Q3 but was revived in December many people don't know that

This don’t sound like we’re talking about GPUs but about instead about dungeons and dragons

 

[511]

This don’t sound like we’re talking about GPUs but about instead about dungeons and dragons

Dungeons and Dragons takes inspiration from Intel.
Intel creates dragon in their dungeons but also kills them on their whim.

 

[DavidC1]

Now a new CEO is here, I will reiterate that Intel should NOT abandon their client dGPU efforts.

All the progress they are making with serving customers and rapidly improving drivers? I guarantee if they do iGPU-only it'll disappear again, back to being a mediocre, HD Audio like irrelevance. I heard nonsense from Intel too many times that they'll "focus on GPU", first on X3000 in 2006, then again with the successor, then again with Sandy Bridge, then again with Haswell, and with Iris Xe parts "now we're really going to focus on GPU".

The actual focus? Didn't happen until their real dGPU product, the ARC dGPU. The rest didn't matter. Because simply until 2 years ago, they had zero incentive.

If they announce they'll focus on "AI" only, which means server GPU, then the same result will happen.

Cheap, Fast, Good, and you can only choose two? Well, only the client market requires the most careful balance between all three. Server GPUs don't care about the first, therefore competitors that have client experience will eventually undercut you, and over time be faster as well, because they get learnings from serving an order of magnitude larger customer base.

 

[Win2012R2]

Now a new CEO is here, I will reiterate that Intel should NOT abandon their client dGPU efforts.

They have no choice - consumer GPU market can barely sustain 2nd player, nevermind 3rd. They should license AMD's iGPUs and be done with it.

 

[511]

They have no choice - consumer GPU market can barely sustain 2nd player, nevermind 3rd. They should license AMD's iGPUs and be done with it.

No need to license AMD iGPU IP when you have invested so much into yours it's not like their arch is bad it's the half baked drivers and software that is the issue.

 

[Win2012R2]

No need to license AMD iGPU IP when you have invested so much into yours it's not like their arch is bad it's the half baked drivers and software that is the issue.

It's sunk money, total failure and AMD needs more scale for their Radeon group so licensing iGPU to Intel is no brainer. It's the only rational way out that is best for both Intel and AMD.

 

[511]

It's sunk money, total failure and AMD needs more scale for their Radeon group so licensing iGPU to Intel is no brainer. It's the only rational way out that is best for both Intel and AMD.

It's not sunk money though Intel Xe2 arch is good Xe1 was a letdown but I can't say that about Xe2 and Xe3 is supposedly even bigger change than Xe1 -> Xe2.

 

[KompuKare]

I think it's easy to see their obvious driver problems and conclude that is their only weakness. And without competent drivers they have to sell low.

However, the last time I looked into this their perf/area and perf/transistor was really poor. For dGPUs perf/transistors is absolutely crucial.

 

[Hans Gruber]

Let's think about this for a minute. Intel is using TSMC for their GPU's. That's good but they will have 18A online by the end of the year. So Intel would have a market advantage using their own silicon moving forward if they were to choose to go that route. Battlemage has been a huge leap forward for Intel. The lack of availability has been a problem. Intel is positioned to take a bite out of Nvidia's 90% market share if they stay the course moving forward.