Intel Talks "Battlemage" Xe2-LPG and Xe2-HPG Graphics Architectures

erek

[H]F Junkie
Joined
Dec 19, 2005
Messages
8,842
Anyone else excited?

"Had everything gone to plan, particularly cost-effective availability of a 7 nm-class foundry node, "Battlemage" was supposed to take off in 2022. Instead, the the group crawled with a first-gen "Alchemist" launch in 2022, by which time NVIDIA and AMD had advanced their architectures (to "Ampere" and RDNA2, respectively). "Now as we go forward in our roadmap, we realized this is a very, very expensive - the QA process and the segmentation. The Thinking was we needed to differentiate our IP and customize it per each segment," said Tom Peterson, an Intel Fellow from the former AXG. "[…] We are going to just have one thing and it goes everywhere unmodified. That's more the strategy we are looking at going forward. And that's because, that's really the only way to get IP reused to really work," he added."

8cEaHoiclGbLryk7.jpg


Source: https://www.techpowerup.com/304375/intel-talks-battlemage-xe2-lpg-and-xe2-hpg-graphics-architectures
 

Zarathustra[H]

Extremely [H]
Joined
Oct 29, 2000
Messages
36,070
I'm glad they are still pushing, but damn, someone should fire their marketing department.

These names for the Arc series GPU's are moronic and infantile.
 
Joined
Mar 19, 2011
Messages
727
I'm glad they are still pushing, but damn, someone should fire their marketing department.

These names for the Arc series GPU's are moronic and infantile.
Because something like Navi is much better?

navi.jpg


I'm being slightly facetious, but in reality let's be honest, most of the internal names are infintile anyways.
 

cjcox

2[H]4U
Joined
Jun 7, 2004
Messages
2,496
Obviously people have choices, but I think you can make a pretty good claim still that Nvidia GPU and Intel CPU is the majority case out there. And there's something somewhat unatural about an Nvidia GPU and AMD CPU pairing (though it is certainly done). So, with Nvidia on its current "insanity" rage on size, power and of course, price. A true bit of competition should impact Nvidia the worst. Right now, however, "the cheap alternative" that way out delivers Intel on performance is from the AMD side. So... we'll see.

Intel really could have been the "be all, end all" if they had released during the crypto crazy price days.... now... it's sort of a "guess" as to whether Intel can pull off something that people will "want" (that won't frustrate them).

And of course, it's Intel.... not exactly the Mr. Sta-Puff we want.
 

sc5mu93

[H]ard|Gawd
Joined
Jul 11, 2018
Messages
1,129
Because something like Navi is much better?

View attachment 546342

I'm being slightly facetious, but in reality let's be honest, most of the internal names are infintile anyways.
i think what is annoying about the whole AMD naming, is they just got stuck.
First Polaris, then Vega, then Navi, then Navi2, then Navi3. there are a crapton of other stars to name this stuff with...
 

trikat

Weaksauce
Joined
Mar 5, 2006
Messages
115
Honestly it's tough rooting for Intel when they played dirty in the past. And are getting free money subsidies for fabs. But competition is good.
 

Lakados

Supreme [H]ardness
Joined
Feb 3, 2014
Messages
7,306
Honestly it's tough rooting for Intel when they played dirty in the past. And are getting free money subsidies for fabs. But competition is good.
TSMC is getting the same free money, and AMD has done some amazing stuff but I want to say TSMC has had as much to do with AMDs success as AMD has.

Like if AMD was still with Global, or they used Samsung instead I don’t see them being anywhere close to where they are now.
 

toast0

2[H]4U
Joined
Jan 26, 2010
Messages
2,308
We are going to just have one thing and it goes everywhere unmodified. That's more the strategy we are looking at going forward. And that's because, that's really the only way to get IP reused to really work

Tada! They are learning. Look at Zen; one chiplet for mid grade, (one chiplet that binned poorly for low end), two chiplets for high end, four or eight chiplets for server. All the chiplets are mostly the same, so you get a ton of skus from one mask, and you can figure out the product mix after fabrication. Perfect for a underdog without their own fabs, which is where Intel is for GPUs. If you run the fabs, you can afford to have a bunch of different dies for each segment, but that's not where Intel is here.
 

PC-Only

n00b
Joined
Dec 1, 2021
Messages
39
I'm excited to see what Intel can make, but far more excited for the potential competition in the market. Video cards may have gone back to "normal" pricing, but they're still damn expensive and often difficult to obtain.
Not sure how this can be considered "normal" pricing - even with the quotes. It is vastly inflated right now, even after accounting for inflation. Just less inflated than it was.
 

Lakados

Supreme [H]ardness
Joined
Feb 3, 2014
Messages
7,306
He was talking about Intel being the underdog.
Underdog isn't the right analogy, those are lean, agile, scrappy, everything Intel isn't. AMD has come into its own and they are by all accounts a heavyweight champion, Intel right now is like the old punched-out fighter looking for their comeback bout.
This would be the part of the movie where Intel finds the scrappy kid and starts training them to take on the current champion (AMD) after realizing that they are just too old for the fight, but the writers give AMD a shitty personality so we immediately dislike them and overlook the fact that Intel used to be just like them but is too old and tired to be able to behave that way any more.
 

Meeho

Supreme [H]ardness
Joined
Aug 16, 2010
Messages
5,777
The Thinking was we needed to differentiate our IP and customize it per each segment," said Tom Peterson, an Intel Fellow from the former AXG. "[…] We are going to just have one thing and it goes everywhere unmodified. That's more the strategy we are looking at going forward.
That's not a strategy, that's a Hail Mary.
 
Top