• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

News Intel GPUs - we've given up on B770, where's Celestial already

Page 54 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
It's interesting how with Gen 9(Skylake) they gave up SIMD+SIMT capability and just went with SIMT to get a reduction in EU size and gain efficiency, but it came back with Xe.

@Dayman1225 A cynical point of view is that they'd have also canned HPG if it wasn't for the fact the consumer GPU prices are at record highs despite the possibility that Xe isn't all that competitive.

HPC is pretty much Ponte Vecchio and it's for one deal. HP sounds like a general compute alternative to Nvidia/AMD and it wasn't competitive and/or customers aren't willing to switch to it.
 
Seems a bit premature to be saying that. You have no idea how well it's going to mine.

It is under the new Intel division - Arc brand. It's GPU is made by TSMC and is 7nm! Do very well! I going buy the lowest bottom graphics card since I do not need one just to help Intel along! LOL!
 
It is under the new Intel division - Arc brand. It's GPU is made by TSMC and is 7nm! Do very well! I going buy the lowest bottom graphics card since I do not need one just to help Intel along! LOL!

I think what he's saying is that it may sell out before it hits store shelves.
 
I mean they have shipped over 70 million Tiger lake units on that same process and over 1 million Icelake SP dies on the original 10nm process - it’s likely that Xe-HP overlapped a lot with PVC in terms of performance and features and/or customers no longer thought it was necessary - so it was turned into a SW dev vehicle
Minor nitpick: 10nm+ (ice lake), 10sf (tiger lake), and 10esf (Intel 7, alder lake) are 3 separate processes. You would want them manufactured on 10esf/Intel 7. The issue is that they can’t kill off 10sf or 10nm+ because they still sell those chips. Intel isn’t going to spin up new products on those older nodes because they will eventually be able to move everything to 10esf. Until that happens and Intel also gets Intel 4 ramped, they are using TSMC to avoid capacity issues.

I suspect ice lake will see an early eol date, then they will convert/add that capacity to 10esf. However, most of that extra capacity will be for sapphire rapids.
 
That's weird. Didn't they just announce a Q2 launch? Hey if they can get them out a month or two early then more power to em. Let's see what they've got!
 
Intel said Q1, but things have been oddly silent so far. With a little over 4 months left, you would think we would see a leaked bench or something.

I don’t need another GPU, but assuming I can, I plan to pre-order one for testing.
 
Intel said Q1, but things have been oddly silent so far. With a little over 4 months left, you would think we would see a leaked bench or something.

I don’t need another GPU, but assuming I can, I plan to pre-order one for testing.

I have become a little concerned because intel keeps stating 3070-ish performance.
Seems like nVidia is ready for a possible 3x performance increase. Basically nVidia can pull out just enough performance to easily double intels best which would suck however having more cards & a third player is still welcome in my opinion.
 
I have become a little concerned because intel keeps stating 3070-ish performance.
Seems like nVidia is ready for a possible 3x performance increase. Basically nVidia can pull out just enough performance to easily double intels best which would suck however having more cards & a third player is still welcome in my opinion.
On the other hand, who needs 3x the performance of a 3070 at 500W? If Intel is cheap and reasonably efficient it may be a good option despite Nvidia selling $3000 cards at the high end.
 
To be fair in the current GPU market it would be very easy to look "cheap" even with good margins, given sufficient supply. Perfect time for Intel to enter really, what are they waiting for?

Finished product.

It may be the ideal time, but contrary to what people believe on the internet, you can't just get product out whenever it's convenient. Technology products have multi-year timelines that really aren't possible to significantly accelerate.
 
Intel could have one of three approaches to build a discrete GPU from scratch. The first and most obvious one would be to scale up its current gen 9.5 architecture. The trouble is, that Intel's SIMD parallelism is more transistor-heavy than even NVIDIA....
Is this some kind of bot post? That makes zero sense unless you live in 2016.
 
Back
Top