• 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 125 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Some of you need to google “fail fast” and a few other terms.

I have worked with startups that have pushed out 1.0 releases of software (and in some cases: hardware) and they were worse than the Intel release. What is important is how Intel follows up over the next 6-12 months. If they are able to drastically improve the quality of their drivers and the hardware, they will succeed. If not? Game over.

I 100% moving DX9 to emulation btw. Intel’s GPUs handle it like garbage, but other GPUs will handle it better. Don’t use Intel’s attempt to ditch DX9 as a benchmark.
 
eek2121 said:
Some of you need to google “fail fast” and a few other terms.

I have worked with startups that have pushed out 1.0 releases of software (and in some cases: hardware) and they were worse than the Intel release. What is important is how Intel follows up over the next 6-12 months. If they are able to drastically improve the quality of their drivers and the hardware, they will succeed. If not? Game over.

I 100% moving DX9 to emulation btw. Intel’s GPUs handle it like garbage, but other GPUs will handle it better. Don’t use Intel’s attempt to ditch DX9 as a benchmark.
Click to expand...

You 100% what?

You love it, I know you do! xD
 
Aapje said:
This seems absurdly optimistic. I'd be surprised if they get to that level at the end of next year.
Click to expand...

I think you missed the point of the post. Whether they do it this year or the next, they are improving quickly. Don't forget that they'd be displacing incumbents in an industry that's one of the most difficult to penetrate. It shouldn't have been expected that they would replace AMD or Nvidia that quickly. They didn't get there that quick either.

guidryp said:
According to who?
Click to expand...

This is from @Exist50 that have proven correct with information revealed before.

Also, it makes sense with the rumors and reports about BK when he was CEO. Dadi Perlmutter and Mooly Eden was either forced or coerced into leaving the company. Remember they got promotions after Pentium M and Core 2 Duo platforms and are from Haifa.
 
eek2121 said:
Some of you need to google “fail fast” and a few other terms.

I have worked with startups that have pushed out 1.0 releases of software (and in some cases: hardware) and they were worse than the Intel release.
Click to expand...
It's plenty late for a "fail fast" approach though, Intel missed several excellent windows of opportunity where that would have worked way better than in the current environment. And I think you are cutting them too much slack there, neither Intel is a startup, nor is Intel new to graphics. The basics could have been built in the past 2+ decades instead pretending to be a fresh naive startup.

IntelUser2000 said:
Don't forget that they'd be displacing incumbents in an industry that's one of the most difficult to penetrate.
Click to expand...
It only got more difficult though. Before Intel could have profited from lack of supply plus obscene pricing of the competition. Now the main competition will be second hand hardware flooding back on the market.
 
moinmoin said:
It only got more difficult though. Before Intel could have profited from lack of supply plus obscene pricing of the competition. Now the main competition will be second hand hardware flooding back on the market.
Click to expand...

It's an issue for sure there's no denying that. They should have done it earlier but they did not and that's it. Should have, could have does not matter. They are doing it and as long as they keep doing it and not cancel it they will get there.

On a sidenote, cultural and political revolutions also bring technological revolutions. Getting out of that cycle requires someone that's exceptional, since everyone lives in the same universe after all. So it makes perfect sense to me they got GPUs out at a time when everything is ramped up to 11. 2019 was too quiet considering what happened before.

Intel, out of all companies getting a GPU out is a big departure that lines up with all these big changes around the world. After the relative peace, we're getting another fierce competition and the survivor of that in both the CPU and GPU markets will be the ones to be left to be duopolies for the next decade.
 
eek2121 said:
Some of you need to google “fail fast” and a few other terms.
Click to expand...

This doesn't make any sense, since fail-fast in the business sense is about doing bold experiments to test the feasibility of a product, but Intel clearly had the goal of producing a viable product that was merely not that competitive, but could gain decent market share with aggressive pricing.

Fail-fast only really make sense for innovative products, where you need market exposure, but the viability of GPUs is evident, so there's no need for bold experiments to learn that. The problems that Intel has been experiencing can all easily be determined by in-house testing, so there's no need for market exposure at this stage.

In fact, Intel has been acting that way as well, withholding these cards from (most) of the market because they could easily determine themselves that the drivers are/were a total disaster.
 
IntelUser2000 said:
I think you missed the point of the post. Whether they do it this year or the next, they are improving quickly. Don't forget that they'd be displacing incumbents in an industry that's one of the most difficult to penetrate. It shouldn't have been expected that they would replace AMD or Nvidia that quickly. They didn't get there that quick either.
Click to expand...

You are shifting the goal posts. Earlier you made a specific prediction that the drivers would be at AMD level at the end of the year.

Now you walk that back to 'improving quickly,' which is debatable, since if they were really improving quickly, the drivers would be in a much better state from when they first had access to working boards.

Also, you seem to extrapolate their current pace of improvement, but you cannot simply assume that this pace can be maintained. Generally, it becomes harder and harder to catch up, the closer you get. It's like accelerating in a car. You may have a car that goes 0 to 60 in 3 seconds, but that doesn't mean that you can reach the speed of light if you press the pedal for 3100 hours. Your acceleration will be slower and slower, until you end up at a certain maximum speed, way slower than the speed of light.
 
eek2121 said:
Some of you need to google “fail fast” and a few other terms.

I have worked with startups that have pushed out 1.0 releases of software (and in some cases: hardware) and they were worse than the Intel release. What is important is how Intel follows up over the next 6-12 months. If they are able to drastically improve the quality of their drivers and the hardware, they will succeed. If not? Game over.

I 100% moving DX9 to emulation btw. Intel’s GPUs handle it like garbage, but other GPUs will handle it better. Don’t use Intel’s attempt to ditch DX9 as a benchmark.
Click to expand...

They are about five years late for 'fail fast'.

This not what Intel did, at all. This was all the result of poor management, and apparently a rogue marketing department.
 
Aapje said:
You are shifting the goal posts. Earlier you made a specific prediction that the drivers would be at AMD level at the end of the year.
Click to expand...

Not at all. I made a mistake thinking end of the year was longer than it really was, but that wasn't the point and it was merely an example. You basically have two major camps - one that thinks Intel would just take over, and another that thinks they'll never get there. Actually that could be viewed as the shift in mindset for everyone from six months ago to a month ago. Even if it takes 5 years to do so, and they only need to get close enough before taking serious share.
 
IntelUser2000 said:
Even if it takes 5 years to do so, and they only need to get close enough before taking serious share.
Click to expand...

That's based on the assumption that they are willing to give it 5 years and that taking serious market share with a third best card is a given. You take those things for granted, but they are far from self-evident.
 
Aapje said:
That's based on the assumption that they are willing to give it 5 years and that taking serious market share with a third best card is a given. You take those things for granted, but they are far from self-evident.
Click to expand...
They use this architecture in their CPUs and they need working drivers for integrated graphics too so they can not back down now.
 
moinmoin said:
I really hope Intel is still aiming higher than their standard for iGPU drivers so far...
Click to expand...
Well, I guess they realised that their driver were horrid. Maybe it used to work decently since they were so slow that massive driver overhead wasn't an issue. Then they realised that their best GPU would run like GT 1030 if they do not start from scratch.
 
Man, leave it to Intel to screw the pooch. They're finally launching entry level cards when a) supply has caught up with demand, b) nVidia/AMD are about to launch new generation, and c) ETH is finally going PoS which will cause even further price reductions. Had they launched their cards a year ago they may have carved out a foothold in the GPU market, now however...
 
Tup3x said:
They use this architecture in their CPUs and they need working drivers for integrated graphics too so they can not back down now.
Click to expand...

Discrete GPU has completely different memory management problems to worry about. Things like the shader compiler will be common, but there will be a lot of unique code for discrete.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top