Xe-HPG is the Performance Gaming Graphics Architecture to Look Out for from Intel

[btarunr]

Intel appears to have every intention of addressing the performance gaming segment with its Xe graphics architecture. According to information leaked to the web by VideoCardz, Xe-HPG (high performance gaming?) represents a product vertical dedicated to the gaming segment. Among the other verticals are Xe-HPC (high performance compute). The Xe-HPG graphics architecture is being developed for a 2021 market launch. It will feature all the client-segment staples, including a conventional PCI-Express interface, and GDDR6 memory instead of HBM. Intel may also eye DirectX 12 Ultimate logo compliance. Intel's Xe discrete GPU and scalar processor development is already de-coupled with Intel's foundry business development, and so the company could contract external foundries to manufacture these chips.

As for specs, it is learned that each Xe-HP "tile" (a silicon die sub-unit that adds up in MCMs for higher tiers of Xe scalar processors), features 512 execution units (EUs). Compare this to the Xe-LP iGPU solution found in the upcoming "Tiger Lake" processor, which has 96. Intel has been able to design scalar processors with up to four tiles, adding up to 2,048 EUs. It remains to be seen if each tile on the scalar processors also include the raster hardware needed for the silicon to function as a GPU. The number of tiles on Xe-HPG are not known, but it reportedly features GDDR6 memory, and so the tile could be a variation of the Xe-HP. Intel SVP and technology head Raja Koduri is expected to detail the near-future of Intel architectures at a virtual event later today, and Xe-HPG is expected to come up.



View at TechPowerUp Main Site

 

[Jack1n]

I am definitely willing to give Intel a chance (given the reviews are positive). I think they have everything required to make this succeed and I cannot really see any negative aspect to this.

 

[Verpal]

I really hope Intel can make MCM work with consumer level XE GPU, this is likely the only way for Intel to get into the ultra competitive GPU market.

Lets hope this isn't another 10nm debacle.

 

[Flanker]

Would be interesting to see Intel make something truly competitive, but I have to see it to believe it

 

[ebivan]

Well first off all if they want to want to compete with Navi2 and RTX 3000, they should learn that launching months after the established players is not a good idea. Nobody will wait for the newcomer. Also, if they want to sell GPUs to consumers its not a good idea to launch after the the holiday season...

 

[XiGMAKiD]

If they enter the market I hope it's a serious effort and for a long term basis not just dumping whatever they got in hand just to make a quick buck

 

[Legacy-ZA]

They better have ironclad support for their products, with drivers and warranties, that is the only way I would be confident enough to even try one of their cards.

 

[ZoneDymo]

bring it on

 

[Crackong]

And the leather jacket is cooking something (again) from his oven and ready to show us on 1st of September....

 

[The Quim Reaper]

Well, they are being designed (mostly) by the team that was let go from AMD for the Radeon 7 debacle, so I wouldn't hold out much hope for these,TBH.

The only thing that might negate that fact is the virtually bottomless pit of money that Intel can throw at it, compared to what AMD allocated to their GPU division, but we'll see.

 

[Vayra86]

They better have ironclad support for their products, with drivers and warranties, that is the only way I would be confident enough to even try one of their cards.

That... and then there is the economical aspect of a powerful consumer GPU. All of that support isn't free. Making GPUs comes with a long-term investment in all sorts of things gaming related. So far... not seeing it. Some one-off technologies and developments (such as INT scaling), which are great, but don't have any impact whatsoever in the landscape of todays games.

Then there is the cost aspect. Can Intel really compete in the midrange? Because if they cannot, they will not have the volume and customer base to keep a strong support base for a vast number of games going. You can't explain that to shareholders. I wonder how long their breath is going to be in that regard. They won't have that midrange volume the first gen, or the second, or likely even the third. They might need as much as ten years.

Nvidia didn't get big on leather jackets and flashy marketing. They got big because they nailed their support and added services in a way that its there if you want it, and if its missing, it will probably get released fast or first. Their Game Ready driver offensive is another perfect split of great marketing with actual support. The result? We trust them. Trust level for Intel is probably sub zero.

As so often the problem isn't directly hardware. I'm sure Intel can daisy chain enough IGPs to get somewhere.

 

[ebivan]

"We trust them"
you serious? You trust a billion dollar company? You trust any billion dollar company? You understand that all of those care only for their shareholders, stock value and manager salaries. They shit on customers! Not only Nvidia, all of them. They dont care for you, they want your money, everything else is BS!

Sorry, but I have not forgotten how the tried to cheat customers with the 970. Even if the effect was mostly unnoticeable, they still tried to cheat their customers!
They didn't get big with support. They got big with Riva TNT and soon after with Hardware T+L and also with the aquisition of other good companies and their technologies.

 

[Anymal]

Raja, that is not what Me Too movement is about!

 

[Tom Yum]

I was around when Intel was going to dominate video cards with the i740, because it was the first to use AGP and it had high end 'military tech' ....I am sure Intel has the potential to produce a high performance video card, the issue will be driver development and support. They have a terrible track record with optimising drivers for games (look at any review of Intel iGPU's, they are consistently inconsistent in game performance, but always good with benchmarks). They also have a track history of abandoning anything they don't immediately dominate.

 

[JohnSuperXD]

It will take Intel at least one or two product cycles to get into good enough shape for their GPU products. Even though they have a lot of money, still it will take a bit of time to optimize their architecture. According to some of the murmurs, it's not looking good at the moment and we have yet to have a good look at their product. I will wait and see and it's good to see someone entering the GPU market, but if this ever happens, what is Nvidia going to do?

 

[Steevo]

So, still vaporware.....

Maybe they will team up with unlimited detail and Tachyum CPUs to make the best vapors on the internet.

 

[medi01]

1) It's Raja again
2) "Will be manufactured at TSMC" => tell me how Intel could have planned it upfront
3) If Intel didn't plan #2 upfront, tell me how they could have a ready to sell product next year
4) #1

 

[Turmania]

There are some guys in the industrumy whose reputation exceeds what they are all about, Raja is one of them.

 

[dont whant to set it"']

3rd vote selection entry : wishful thinking .

 

[chstamos]

I'd be delighted to see intel addressing the performance market, but they'll first have to address the bottom rung entry level barely-gaming-capable discrete gpu market. They haven't shown anything approaching Geforce 1050 or Radeon 560 performance yet, and their -granted, prerelease and developer focused- Xe DG1 was pretty much an unimpressive take on dung-level 1030 performance.. We were supposed to be getting some kind of discrete GPU in 2020, and instead we're expecting a run of the mill intel graphics incremental igpu with the xe badge (be still my heart).

Hope they do better. But performance market seems to be setting the bar quite high for a company that's struggling with entry level, for the time being.

Maybe we're expecting too much, maybe they set expectations too high themselves with the various press blitzes.

 

[bug]

First of all, it's not a monopoly if you have two players in the market.

And second, it seems Intel has decided these are the parts they'll let others build for them. So, as much as I want a third player*, it seems at least for the time being, desktop is not Intel's priority.

*I don't need a competitor to 3080 or 3080Ti, I'm good with competition at the mid-range.

 

[midnightoil]

At this point, I think Intel have more of a shot of releasing competitive consumer segment GPUs than they do CPUs, as long as they're fabbed at Samsung.

Zen3 / 4 / 5 are going to slaughter the Intel CPU product stack.

 

[Deleted member 177333]

Yep will be following this closely. I've been ready to upgrade GPUs for almost two years now. Any company that can come in and offer a viable alternative to NVidia's overpriced GPUs I absolutely welcome.

 

[geon2k2]

First of all, it's not a monopoly if you have two players in the market.

Yes, its called oligopoly, almost the same.

Adding some launch prices:
780 - 500$
980 - 550$
1080 - 600$
2080 - 800$

Why do you think this happens, because the market works well?

Also it depends how you look, maybe from US perspective there are 2 competitors, 3 soon, from EU perspective everything is US based. So, is there an US monopoly?

 

[ebivan]

First of all, it's not a monopoly if you have two players in the market.

Yeah, you can basically choose between sleeping on a rock or a hard place. Like you can choose between Republican oder Democrat, between Pepsi and Coke, Between Android an iOS, between PC and Mac...
You fell me? It never hurts to have more than one alternative.