Intel Says Its First Discrete Graphics Chips Will Be Available in 2020 (marketwatch.com)

Ryan Shrout, reporting for MarketWatch: Intel CEO Brian Krzanich disclosed during an analyst event last week that it will have its first discrete graphics chips available in 2020. This will mark the beginning of the chip giant's journey toward a portfolio of high-performance graphics products for various markets including gaming, data center and artificial intelligence (AI). Some previous rumors suggested a launch at CES 2019 this coming January might be where Intel makes its graphics reveal, but that timeline was never adopted by the company. It would have been overly aggressive and in no way reasonable with the development process of a new silicon design. In November 2017 Intel brought on board Raja Koduri to lead the graphics and compute initiatives inside the company. Koduri was previously in charge of the graphics division at AMD helping to develop and grow the Radeon brand, and his departure to Intel was thought to have significant impact on the industry.

Comments

i741

[mandark1967]

liquid cooled and running at 50Mhz with an overdrive chip

Re:i741

[kelemvor4]

liquid cooled and running at 50Mhz with an overdrive chip

Or any of the many many others they made after that.

It's pretty obvious that Ryan Shrout just doesn't know what he's writing about.

Re:i741

[ChatHuant]

liquid cooled and running at 50Mhz

Sorry, but you can't really push the 741 far above maybe 10 kHz...

First discrete graphics?

[slew]

What about this one?

https://en.wikipedia.org/wiki/...

Re:First discrete graphics?

[kriston]

I came here to mention this. The article is wrong, though the i740 came about when Intel licensed the technology from the Real3D division of Lockheed Martin. Intel later purchased the intellectual property after Real3D was closed.

Re:First discrete graphics?

[Type44Q]

They're trying to pretend they didn't, since Starfighter was a P.O.S.

Re:First discrete graphics?

[Anonymous Coward]

But should a half-ass attempt count?

Re:First discrete graphics?

[Anonymous Coward]

The article quotes Intel CEO Brian Krzanich, so either the article severely deformed what the CEO said in the analyst event, or the CEO himself is wrong.

Re: First discrete graphics?

[Anonymous Coward]

The what?

Seriously... I remember a time of sanity, when people on /. still had an actual clue about the thieving criminal bullshit that is "copyrightâoe or any kind of imaginary property. When people like you were still told, not to use FUD propaganda terms by the Content Mafia.

Nowadays, kids just grew up into a world bullied into acting like it is normal, and they have no chance from the start. Unless they actually do a reality check.

Kids, ... The only and exclusive purpose of that concept, is to give literal coke-headed* criminals leverage that enables them to steal from people and run a protection racket, by leeching off the work of actual creative people, inventors, and other people who actually work, without actually working for it themselves one freaking bit.
Protecting the work of creative people and rewarding them for it has nothing to do with it AT ALL. Which they proved, every time they tried to lower the amount of money they still had to pass on to those who actually did the work. Remember the writers' strike? Not getting a single cent for digital streaming of the shows was what it was about. Or how musicians got only 3.5% (so peanuts) iof a CD's selling price, and the industry actually had the audacity to try to lower it to 2.5%, back in the early 2000s.)

(Of course the media industry leeches work somtimes: When itâ(TM)s about avoiding work, or their cocaine reserves get low and new propaganda, legislation and attaching-to-the-neck-of-bands needed to be done.)

So please ... Don't harm artists and inventors by spreading that propaganda.

(* I have backed that fact up in court, since I worked in the music, TV and games industry, and just wait for a chance to do it again)

Re:First discrete graphics?

[Anonymous Coward]

Yeah. It was dog slow and had several compatibility problems. Of all the competition at the time (Ati/3Dfx/PowerVR/S3 and others) they managed to be the worse of them all.

No wonder they want to pretend it never happened.

Re:First discrete graphics?

[Solandri]

In modern terminology, the difference between "discrete" and "integrated" graphics is not whether it's a standalone plug-in card. It's the presence of VRAM - high-speed, high-bandwidth RAM dedicated for use by the video card for 3D rendering. Discrete GPUs come with their own VRAM. Integrated GPUs use system RAM (though they're increasingly showing up with their own small buffer of high-speed RAM that acts more like a cache), with a much smaller amount of dedicated RAM for framebuffers.

The video cards from the era you've linked either used system RAM, or only did 2D graphics using a few MB of onboard RAM for the framebuffer. So they are analogous to today's integrated graphics. The need for the GPU to have gobs of its own high-speed VRAM didn't arise until 3D graphics began pushing frames faster than you could transfer needed data across the bus from system RAM to the video card. Most of that VRAM is taken up by textures used for 3D graphics, so only 3D graphics cards have large amounts of it. A framebuffer, found on both 3D and 2D graphics cards, is only 8 MB for 1080p 32-bit color. So there's no need for large amounts of VRAM in an integrated video card.

Back then, we called them a 3D video card vs a 2D video card. That nomenclature was abandoned once even low-end 2D video cards became capable of rudimentary 3D graphics. The distinction then shifted to whether it was a "serious" 3D graphics card with its own dedicated VRAM, or whether it was a 2D video card (commonly integrated into the motherboard) which could do 3D graphics in a pinch by borrowing system RAM to use as VRAM.

Re:First discrete graphics?

[Anonymous Coward]

Many of the Real3D people were hired by Intel and still work there on the Intel integrated graphics chipsets that did originate from the i740. ( I was working there in the graphics group at the time so I know)

Re:First discrete graphics?

[Anonymous Coward]

In electronics you can choose to build a circuit (like a computer graphics card) either from dscrete components or use Integrated Circuits; Intel is known for doing the latter. (how about a trillion separate transistors?)

https://en.wikipedia.org/wiki/Electronic_component

Re:First discrete graphics?

[DontBeAMoran]

When you're doing a half-ass attempt, you can at worst expect half-shit results.

Re:First discrete graphics?

[kriston]

No, the i740 had its own, dedicated VRAM. Therefore, this is not Intel's first discrete graphics chipset.

Re:First discrete graphics?

[Anonymous Coward]

Uh, all graphics cards whether 2D or 3D had their own RAM, and even integrated graphics card (e.g. ATI Rage Pro, or even ISA graphics in old OEM PCs) had their own RAM soldered to the motherboard.

Even if you only had 4MB on a graphics card that did both 2D and 3D, well you had 4MB for everything, framebuffer (double-buffered + Z buffer) and textures.
The original 3dfx Voodoo did only 3D but had its own framebuffer : 2MB framebuffer memory and separate 2MB for textures.
So that was not much but you'd run something like Quake 2, or ports of Playstation 1 games, and you marveled at the 640x480 resolution and filtered textures with perspective correction.

The i740 was unique in that it did what you describe. It had some 4MB RAM on the graphics card, used for the framebuffer. It borrowed system RAM, over AGP 2X, to access textures. It was not a 2D video card but a 2D/3D one.
The problem is that reading textures over AGP 2X was too slow (and might gobble up your CPU's RAM bandwith if you only have a 66 MHz FSB)
Intel developed an AGP 4X version, but that didn't really get released. Instead they used that tech and turned it into the i810 chipset, which used only the system RAM, and it was very successful if infamously slow. It was still a good feature to have a game run at all.

Re: First discrete graphics?

[Anonymous Coward]

When you're doing a half-ass attempt, you can at worst expect half-shit results.

Huh? At worst you can expect full-shit results. No matter if it's a half-ass attempt or the best attempt in the world.

What the fuck are you smoking?

Re:First discrete graphics?

[Anonymous Coward]

From the very wiki article you seem to have skipped over:

  A unique characteristic, which set the AGP version of the card apart from other similar devices on the market, was the use of on-board memory exclusively for the display frame buffer, with all textures being kept in the computer system's main RAM.

As he said, there is a minimal framebuffer on the card, the rest being in system memory. Hence, integrated.

Re:First discrete graphics?

[John Napkintosh]

They even have a more modern attempt based on many cores with an onboard OS: https://en.wikipedia.org/wiki/...

Re: First discrete graphics?

[Anonymous Coward]

Or even the 80860, although it was marketed as a general purpose RISC CPU.

Re:First discrete graphics?

[Anonymous Coward]

I remember it not badly received, it did about well against stuff like Riva 128, Voodoo1. Left the crap like S3 Virge, SiS, Cirrus Logic, Mystique, Rage II etc. behind.
But as often happened it was about outdated on launch day still or whenever those you cited launched.

Re:First discrete graphics?

[Anonymous Coward]

pretty sure agp didn't start until around tnt2/geforce 1 (with sdr!)

Re:First discrete graphics?

[DigiShaman]

That was normal for AGP. In fact, it's called the "AGP aperture". In BIOS, you could allocate 4MB to 256MB of system RAM to the card. So it's not like Intel was being cheap with their implementation of an AGP video card. It's just that the GPU was...well...crap compared to the rest of the competition.

Re: First discrete graphics?

[Anonymous Coward]

Stop it!
The fact that they relied on AGP for making pretend video memory for texture doesn't make it "integrated". Call it "lousy execution of a card", " worst 3D video card ever made" or "Intel's ugliest 3D child" if you will... But integrated it isn't.

The 740 had a successor. The 810 if I remember right. This one was integrated.

Re: First discrete graphics?

[DontBeAMoran]

half-ass vs full-ass

Re:First discrete graphics?

[Anonymous Coward]

I worked on this. Intel bought it from Real3D and then acquired the company, including Tom Piazza, the senior principle engineer behind Intel's graphics line for almost 2 decades. Got a problem with intel graphics, blame Tom. Kidding, he's a cool guy. Not. Kidding. He is. Not. Hah. LOL.

Re: First discrete graphics?

[Anonymous Coward]

Or this one
https://en.m.wikipedia.org/wiki/Larrabee_(microarchitecture)

Re:First discrete graphics?

[Jeremy Erwin]

Here's what John Carmack had to say about the i740..

Good throughput, good fillrate, good quality, good features.
A very competent chip. I wish intel great success with the 740. I think that it firmly establishes the baseline that other companies (especially the ones that didn't even make this list) will be forced to come up to.
Voodoo rendering quality, better than voodoo1 performance, good 3D on a desktop integration, and all textures come from AGP memory so there is no texture swapping at all.
Lack of 24 bit rendering is the only negative of any kind I can think of.
Their current MCD OpenGL on NT runs quake 2 pretty well. I have seen their ICD driver on '95 running quake 2, and it seems to be progressing well. The chip has the potential to outperform voodoo 1 across the board, but 3DFX has more highly tuned drivers right now, giving it a performance edge. I expect intel will get the performance up before releasing the ICD.
It is worth mentioning that of all the drivers we have tested, intel's MCD was the only driver that did absolutely everything flawlessly. I hope that their ICD has a similar level of quality (it's a MUCH bigger job).
An 8mb i740 will be a very good setup for 3D development work.

IIRC, the "integrated vs discrete" distinction came about because intel's integrated chipsets were so very slow, didn't support the ATI/NVIDIA extensions to OpenGL, and the developers didn't want to spend time optimizing non ATI/nVidia codepaths.

Re: First discrete graphics?

[Anonymous Coward]

Homophobe!

Re:First discrete graphics?

[radarskiy]

The key is "the *growing* use of textures". (emphasis mine)

Intel loves deep pipelines and caching. The latter plays to their self-image as a manufacturing company first and a design company second; and the former looks good in the presence of the former and cooperative workloads. In this case, the workloads changed faster than their GPU design.

Re:First discrete graphics?

[samwichse]

"Intel also sold the i740 to 3rd party companies, and some PCI versions of the accelerator also were made. They used an AGP-to-PCI bridge chip and had more on-board memory for storing textures locally on the card, and were actually faster than their AGP counterparts in some performance tests."

Seems totally discrete to me.

Probably most of the uses of this new chip will be similar... sold to OEMs to integrate as they will.

Re:First discrete graphics?

[Anonymous Coward]

Let's see, if this iteration gets to stores, or is it canceled too by some genius bean counter.

nvidia on suicide watch

[Anonymous Coward]

fuck you geforce i dont want to update driver right now

Intel High Performance

[Anonymous Coward]

By High Performance, they mean it will be comparable to a GPU made 2 years ago, at 2X the price, at 1/2 the performance, with 3X the power consumption, 1/4 the driver support , 0% documentation, and of course, only Windows support.

Re: Intel High Performance

[Anonymous Coward]

You must be talking about one of those other platforms; Intel have been instrumental in revolutionising Linux to a modern Desktop, with great support.

Intel have literally gone from zero to hero. Linux is the best platform out there by Intel's treatment of Linux. Other graphic chip companies have been less supportive, and have been shown Linus's middle finger.

Re:Intel High Performance

[Anonymous Coward]

By High Performance, they mean it will be comparable to a GPU made 2 years ago, at 2X the price, at 1/2 the performance, with 3X the power consumption, 1/4 the driver support , 0% documentation, and of course, only Windows support.

Well, if history serves as a guide, at least Intel will probably have open source drivers. Most likely only half the perf of their closed source buggy Windows drivers, but hey, at least they will be open source...

Discrete MOBILE GPU probably.

[Anonymous Coward]

Because while they can lie about the power consumption (see: Atom) they can't lie about the lack of processing power.

Ok...

[beheaderaswp]

Since I use GPUs a lot for non gaming applications this is interesting.

Normally I'd not be interested because it's Intel who will have to play catch up. But with Raja involved this might actually have life.

Wait and see....

Re:Ok...

[DigiShaman]

You might be interesting to watch the review of an old Intel prototype GPU card, and the history of it from Linux Tech Tips. It truly is interesting and a review of history done justice. I know my PC hardware, but wow, this was very interesting and a bit of trivia I wasn't aware of. Highly recommend watching it.

https://www.youtube.com/watch?...

What I got out of that review was that Intel's card was less ASIC and more programmable. It was more of a cross between a CPU and GPU, but with lots of cores. Unfortunately, it never performed as well as a DirectX specific card, and was very power hungry. It was however far more flexible and adaptable to whatever you can program it to do. For it's time, it probably would have been the Bit Coin miner to have at the time.

Ryan Shrout's brother

[bestweasel]

Off topic maybe but just wondering if he has a brother called Tristan.

So same year as the new XBOX?

[Kenja]

Coincidence? Yeah, probably...

As Intel could never compete...

[gweihir]

I hope this means that GFX will get mature and the endless cycle of "faster" will come to an end. There is a lot of evidence of massive slowdown at this time already, only a few years after CPUs. Finally having mature tech here would be endlessly beneficial.

Re:As Intel could never compete...

[Anonymous Coward]

Gotta drive a pair of 8K displays at 90Hz before graphics are done.

Re: As Intel could never compete...

[Type44Q]

Fuck 90; 240hz.

Re: As Intel could never compete...

[Anonymous Coward]

Dual 8Ks will be mostly in peripheral vision, so yes, definitely need the faster refresh rate.

I'd love something to drive down

[rsilvergun]

video chip prices. It's been 2 years and a 1060 is still selling over MSRP.

Diversification

[Bruce Perens]

Intel is making graphics chips, and IHOP is making hamburgers.

Re:Diversification

[Anonymous Coward]

Well at least their both engaging in speculation.

Re:Diversification

[jwhyche]

I had a IHOP hamburger a few days ago. It was pretty good. Not sure I can say the same about intel graphics chips.

Re:Diversification

[Bruce Perens]

Back in the desktop era, PC manufacturers didn't want Intel to have good graphics. Graphics card upgrades had much higher margins than the base PC. So, you can consider that Intel graphics have been deliberately handicapped.

I've been in this long enough to remember when the big graphics company was Silicon Graphics, and before that Evans and Sutherland. Intel just needs to hire good people.

Let's hope this try goes better than Intel's last attempt at making better graphics chips (which seemed to fizzle out), and better than things like Xscale which Intel discontinued before the market was really ready for it. Qualcomm would probably not be at the top of wireless handset chips now now, if Intel had kept Xscale going.

Re:Diversification

[Anonymous Coward]

It's somewhat ironic if IHOPs hamburgers are pretty good since their pancakes are kind of crap.

Re:Diversification

[Anonymous Coward]

And Spybook is making billions.

Re:Diversification

[rogoshen1]

let's also hope they don't intentionally cripple thunderbolt to bolster their graphics cards.

An nvidia card over a TB3 is still light years better than anything intel has produced -- and will likely stay that way for several years.

Re:Diversification

[Snotnose]

IHOP is cooking their hamburgers on Intel chips.

Re:Diversification

[Anonymous Coward]

OEMs loved using a low end graphics card, like TNT Vanta, Rage 128 variant, Radeon 7000, 9200SE, FX5200 (64bit instead of 128bit), X300.

This sold, and is just like what still happens with laptops today (I disabled the nvidia on mine, to save a bit RAM and *network bandwith* not running the nvidia driver and tools)

XBOX is AMD maybe PS5?

[Joe_Dragon]

XBOX is AMD maybe PS5?

Re:XBOX is AMD maybe PS5?

[EvilSS]

I don't see either moving away from AMD for Intel unless Intel gives the chips away at a loss. Even then...

Wiat a min... I thought Intel was done...

[bobbied]

Didn't we have a story last week about how Intel was on death's door because they couldn't get their yield on the new chips high enough? Wasn't AMD ready to pounce? Now this?

Re:Wiat a min... I thought Intel was done...

[epine]

Didn't we have a story last week about how Intel was on death's door because they couldn't get their yield on the new chips high enough?

10 nanometer

Currently Intel's 10 nm process is denser than TSMC's 7 nm process and available in limited quantities, but volume production is delayed until 2019. However, TSMC's 7 nm process [in name only] is planned to be soon available in high volume shipments or mass produced devices.

After you've been the 800-lb gorilla for four decades, death's door is merely running abreast, because gorilla's don't historically adapt well to sustained sprints.

But in this case, everyone on the track must soon round the corner onto the EUV obstacle course.

This 10 nm design rule is considered likely to be realized by multiple patterning, given the difficulty of implementing EUV lithography.

TSMC is surely feeling their oats these days, but tackling EUV from the vanguard position has to practically scare them pantsless, if they've got any sense; they'll probably concede half a step to the once-tireless vanguard gorilla.

I imagine this will only be temporary, though. EUV from the vanguard might just be Intel's last gorilla glory. But it buys them time, so they're definitely not 800-lb shaggy deadman running abreast, just yet.

Re:Wiat a min... I thought Intel was done...

[Tough Love]

Good luck to Intel with their GPUs. Intel won't be getting me back from the forseeable future. Threadripper 2 for me this fall. Btw, it is said that losing Raja Koduri will actually speed up AMDs GPU evolution, because he changed direction too often.

Re:Wiat a min... I thought Intel was done...

[AHuxley]

Selling a lot of old CPUs on a card as a new expensive GPU product line will help with that.

Discretion...

[Anonymous Coward]

Will come in very handy for watching porn at work.

Re:Discretion...

[Anonymous Coward]

Personally, I prefer the warmth of analog graphics cards.

Remember "FUD"

[Anonymous Coward]

Wow. Intel has fallen back on its old tricks. Announce a product that is going to be "so aweseome" way out in the future. I'm telling you, this news from Intel is going to make me put my graphics card purchases on hold! Gotta get me some of that "Intel Inside"!!

Re: Remember "FUD"

[Anonymous Coward]

Anyone with experience buying GPUs will not be tempted to wait for an Intel Gen-1 GPU.

What this really means is that nVidia and AMD might skip the next few incremental updates and crush Intel out the gate.

The best I can hope for out of this...

[Anonymous Coward]

The best I can hope for out of this is that intel will do ok, adopt freesync, and force Nvidia to get with the program and drop the stupid proprietary gsync.

Re:As reliable as a pentium 60!

[AvitarX]

That doesn't look like floating point to me.

Will ot be free for all Intel has stolen from us?

[Anonymous Coward]

No?

Fuck Intel.

Make it bad at mining please

[xack]

A GPU good at graphics but bad at mining is needed so people can start gaming and designing again. I hope Intel cripples mining capabilities in their card.

Re:Make it bad at mining please

[Anonymous Coward]

Gaming and mining involve the same kind of math. Your suggestion is impossible. Government (over)regulation of cryptocurrency will be what does it in, not graphics card manufacturers cutting their own throats by intentionally gimping their products.

Competition

[Only Time Will Tell]

It will be interesting to see what this does to the duopoly enjoyed right now between Nvidia and AMD. It would be nice to see some price pressure on the market and new tech coming out. We've been in this two-horse race pretty much since the late 90s/early 00s.

Organizational Culture & Priorities

[Anonymous Coward]

OK, so they hired Raja Koduri. That is good.

What isn't so good is the Intel culture and priority list, which has never valued high performance graphics. Every Intel graphics solution, every single one, has sucked balls. It never stopped them from marketing the bejeezus out of them, which only served to lower Intel's graphics cred even further.

What is worse, for me and for many others, integrated Intel graphics have come to be a general marker of a low-end, poorly designed and configured system. If it relies upon Intel graphics then it is a garbage system. No, don't give me the Buts. I don't care about Iris Pro. I don't want to hear about how "this time it is different". I've fallen for the hype too many times in the past and I'm not falling for it again. Intel graphics are crap, and they are a sure sign of a cheap-o computer that won't work well for me or others.

I wish this Raja Koduri fella well, and it would be great if he turned this particular ship around. Even talking in terms of discrete graphics is an improvement, since dGPUs don't have that "Sure Sign Of Garbage" stigma attached to them. However I'm willing to bet this is more of a Ray Ozzie/Groove type of move. A new hire of an outsider who is supposed to Change Everything, but winds up changing nothing and leaves quietly.

Re: Organizational Culture & Priorities

[Anonymous Coward]

Linus Tech Tips did a nice video about Intel graphics after Linus bought an old Intel development graphics card off eBay.

Re: Organizational Culture & Priorities

[Anonymous Coward]

This one?

Re: Organizational Culture & Priorities

[Anonymous Coward]

That's it!

Discrete graphics?

[DontBeAMoran]

So it's a GPU that won't tell anyone about the kind of porn you watch?

Re:Discrete graphics?

[TeknoHog]

Either that, or it's huge and power-hungry because of all those separately packaged transistors. (Like Honest Bender's Dating Service, some things can be both discrete and discreet.)

Re:Discrete graphics?

[thegarbz]

So NVIDIA doesn't have discrete graphics cards? https://www.neowin.net/news/nv...

Re:Discrete graphics?

[Anonymous Coward]

That's 'discreet'.

Subambient

[Anonymous Coward]

So, Intel presents more BS-ware. Is this one too clocked at "5 GHz"? Sorry Intel, your sub-ambient, water-cooled bullshit doesn't fly outside Taiwan.

Can't they master

[Anonymous Coward]

one subject first? Like building a CPU that is secure, a graphics chip that doesn't error, drivers that work, and wireless chipsets that are not broken?

Diversification is a good strategy when you are *good* at what you do, but Intel isn't good at what they do.

Time to branch out

[TsuruchiBrian]

Now that Intel has finally asserted and solidified it's superiority and dominance over all it's competitors in the CPU market, it only makes sense to branch out into other markets.

Intel probably doesn't have to try that hard anymore. Their lead is so big that they can probably just continue to profit indefinitely even without any real innovation on their part.

That said, I'm hearing that Intel has managed to get a 28-core chip running at 5 GHz on all cores. Their advancements in tablecloth technology has made this possible with standard watercooling solutions.

Re:Time to branch out

[Anonymous Coward]

LOL. Can have some of that shit you're smoking? You're completely off your rockers.

Support

[Anonymous Coward]

Let's remind Intel of Karma. Where's the GMA500 (Intel integrated graphics) linux driver?

Way, way behind.

[Bing Tsher E]

Back in the day, Motorola had the 6845 in their 6800 processor family. I think even Zilog has a CRT controller of some kind. Intel has waited until now? Really?!?

If unlike nVidia, they provide open source drivers

[ffkom]

... and unlike AMD, they provide _stable_ open source drivers, then I'm all ears. At this point in time, the Intel iGPU drivers are the only ones I can trust to run 365/24 in a Linux system.

Wanting that sweet, sweet cryptocurrency cash...

[Anonymous Coward]

Goodbye Bitcoin, Ethereum, and many, many scam-laden derivatives. We barely knew thee. Intel's entering the ICO market a bit late in the game methinks just as cryptocurrency is making a final exit. Various electricity providers are starting to crack down, thereby implicitly regulating (and eliminating) the currency by regulating the most basic requirement to mine it.

It will do ray tracing for free

[AHuxley]

That will be the limits of the free software spread over many Intel CPU's sold as a new GPU.
Want more? The app creator will have to tell the all the CPU's on the Intel GPU what to do for their app.

No thankyou Intel

[Anonymous Coward]

Intel's support of its graphics chips has been fleeting and this is worse because of evolving graphics standards. What you buy from Intel today may be a piece of junk in a years time.

Stick to nVidia and AMD. They care about their long-term reputation. Intel does not.

Q: Is there a way to install a newer Intel graphics driver? According to the Intel download website, the latest graphics driver for the GPU in an Intel Core i5-2400 is version 15.28.24.64.4229, released on 6/5/2015. However, I'd like to run an application which requires... MORE

A: No, there's no way to get a new driver. Intel dropped support for all GPU chipsets below the HD 4000 series with the release of Windows 10. The HD 4000 got a limited release at the time, everything else was abandoned.

https://superuser.com/questions/1243255/is-there-a-way-to-install-a-newer-intel-graphics-driver
https://superuser.com/questions/536129/intel-3150-officially-wont-support-opengl-2-0-for-windows-so-how-come-i-read-it

Intel have done this on their software too eg. Intel® XDK. Here today. Next big thing. Dumped tomorrow.

Re:No thankyou Intel

[Anonymous Coward]

Stick to nVidia and AMD. They care about their long-term reputation.

Yup, nVidia sure cares about their long-term reputation, which is why they intentionally gimp cards more than a year out of date with driver updates and force telemetry through a shitty manager program. AMD absolutely cares about their long-term reputation, which is why they take years to make stable drivers for every GPU they release and suck power like a nuclear-powered vacuum cleaner.

I don't even like Intel, but anything that has potential to actually stimulate some competition in the market is worth hearing out.

Why does everyone discount Intel on this?

[Anonymous Coward]

Intel has fairly up-to-date standards support in their iGPU line. It is not perfect or best but it is there. They have all the necessary building pieces even now, they just need to scale it up. As far as die space or power consumption goes they do appear to be fairly on par with AMD and Nvidia.

The common Intel's GT2 iGPU currently has 192 shader units. AMD's (deservedly) much praised Vega 11 has 704 at 2-3 times the die space. At this size, fixed-function and supporting parts of GPU take up a lot of die space compared to calculation units. Performance gap roughly matches expectations based on relative size difference.

Historically, Intel has been making attempts ar GPUs at various sectors or niches. i740 was not a bad attempt. They did get to the point of it being usable with Larrabee, including software. Intel's drivers have been bad but usable. They are not doing this out of the blue.

With this announcement in mind, partnership with AMD to get embedded GPUs attached to their CPUs also makes a lot of sense. 8x0xG series is the test vehicle for both EMIB as well as power management. When they have their own GPU ready it will be a drop-in replacement (that EMIB houses plain PCI-e).

We as users should definitely welcome new competition in GPU sector, AMD and Nvidia have become complacent in their duopoly.

Re:If unlike nVidia, they provide open source driv

[samwichse]

I don't know man, I've got a system with a Radeon HD 3450 in it (crappy base discrete card Dell put in everything a while back) and it is absolutely rock stable, the uptime is... well, I did a kernel update last week so it's a week, but this has never crashed once except some weird condition where accessing a file on a mounted SMB caused a GPF once and left the system in a weird state.

The older Radeon cards are extremely well supported and stable, although I sure wouldn't want to game on one...

cheap GPU ?

[micahraleigh]

I'm a dev who knows nothing about HW.

I came to this article thinking, "Ooh, maybe I don't have to pay an extra $400 for a GPU on a new computer!"

Is that wishful thinking?

Does this announcement make any handwavey motions in that directions or am I waay off course.

Disagree!

[Anonymous Coward]

Ah, I have to disagree most strongly with this. At least the idea of "mature technology benefits us" part.

As long as the programming interfaces and the general design principles don't change radically, the fast cycling and large generational performance improvements have been a great benefit. And guess what? That's exactly what we have in the marketplace.

Seriously, what is the benefit to a slowdown in the rate of GPU performance improvements? For most people, all the way from personal devices to enterprise servers to HPC, that is an unambiguous loss of opportunity. There's no benefit whatsoever.

Re:If unlike nVidia, they provide open source driv

[ffkom]

I could live with the lesser 3d power of old ATI cards - but not without 4k 60Hz displays, which only the newer ones support.