Hacker News Reader

Comment by cbozeman | Hacker News Reader

Turns out the leadership of Intel is stupider than I thought.

It's almost impossible to believe that they actually thought they'd be able to go up against two GPU makers each with 25+ years of experience.

This was always going to be a long uphill battle. Always. I figured it would take at least 10 years for Intel to become competitive, but turns out they tucked tail and ran.

Mediocre!

View on HN

Arc is insanely good for a first attempt at a discrete GPU. It has mid tier performance for mid tier price, and drivers are still getting tons of work from intel that makes them better and better almost every release. It's actually crazy that they managed to do so well (mid-tier is still amazingly powerful, all things considered). If they stick with it, there's no reason to believe they won't be able to compete at the high end, too. I honestly expected a massive trash fire when I first heard of Arc, just because of the countless edge cases and driver/API hell that even Nvidia and AMD seem to run into continuously.

its not Intel's first foray into cancelled discreet GPUs. they've been on and off with this for at least 13 years

They've been trying with discreet gpus waaay longer than that. The 82786 is from the mid-80s.

I'm trying to understand their issue though, are they locked out of doing time tested solutions because of patents or something? My real question is are they just incompetent or are they not able to use known solutions.

They aren't incompetent or ignorant (well...on a macro scale). They just didn't succeed in a market that's a niche in a subset of the market they're after.

Intel makes a ton of money selling graphics silicon. They just haven't quite figured out the "plays the most demanding video games on max mode with a side of ML/AI" market yet.

Predictions of Intel's demise are greatly premature.

3060 wasn't mid tier performance when high-end Arc was released anymore.

Intel has been making GPUs since roughly 2002, 845 chipset series. (There were earlier ones, but I'm

Intel740, 1998.

I was on the team that wrote the DirectX 3 and 5 drivers for the 740 (I wrote the vxds in assembly using MASM 6.11). It wasn't a GPU in today's terms (huge parallel vector engines), it was just a basic 3d pipeline: triangle pipe (with Gouraud lighting, 1st version didn't even have phong/specular) followed by rasterizer, with a 2D engine. You sent it triangles with aRGB color, and it interpolated and rasterized them. Dat's it. Total dud compared to Voodoo 3DFx or ATI Rage3D. But when we integrated it into the 810 chipset it became the main reason why 3D gaming was held back for a decade: it was the top selling graphics engine but it sucked so every game dev had to design for the lowest common denominator. It was never intended to be good. But it sure did result in the promotion of a lot of engineers to principal. (raises hand)

You sent it triangles with aRGB color, and it interpolated and rasterized them. Dat's it. Total dud compared to Voodoo 3DFx or ATI Rage3D

I don't get it - that was what all gfx chips were before shaders came along. That's not unlike how the PS2's GS chip worked too. Texturing and triangles was everything. What feature did the Voodoo have that the 740 didn't?

3dfx/ATI were faster.

Ah - I see. Yeah, raw performance was (and is) really an enabling feature with gfx chips.

Intel 82786, 1986

Wasn't that mostly just PowerVR IP? Or was that mostly the SoC/Atom line.

They only used PowerVR GPUs for a single Atom generation (I think it was the second).

That's Apple who uses PowerVR IP.

Yes, just not good ones

Didn't they buy Chips and Technologies many years ago to improve their GPU offerings and then dumped the idea? Intel does GPUs in fits and starts but never sticks with it long term. It's always a pet project.

C&T wasn't really performance-oriented either; and as far as I know, they were 2D-only. The i740's non-3D parts have a lot of similarity to C&T's latest accelerators.

True. I think intel at the time got spooked by 3rd party MB mfgs (acer, asus, etc) who were integrating video in their mobos but intel wasn’t and they didn’t want to lose market share or maybe they wanted people to at least have a passable basic VGA without needing an add on card.

But for whatever business reason they felt mission accomplished and dropped it. Then over the years they’ve bought others for differing business reasons.

But the common thread is they never stuck to it and never dedicated resources to make it flourish and succeed on its own merit.

Gelsinger did a fantastic job ruining VMware before jumping over to Intel to do the same.

He projects himself as a savior if you've been around him. I'm not sure if it's his deep religious beliefs, but the guy doesn't understand what the market wants. He may be a great engineer, but he's horrible at the helm. Unfortunately, with someone like him, he's got credibility in areas that open doors for him that shouldn't be.

After going through their Larrabee efforts, and OpenGL drivers that used to lie about what was actually supported in hardware, I am not surprised how things turned out.

Reply on news.ycombinator.com