Intel CEO confirms first dGPUs in 2020

I have increasingly wondered if this will pan out with Intel buying the discrete graphics arm of AMD and it becoming the Intel discrete graphics arm....

It's a win for Intel in that suddenly patent issues become largely a non-issue, cross licensing is all well and good but it is grabbing the experience and knowhow to leverage it into something that works as a competitive product and being able to further develop new techniques and features to keep competitive long term. It probably is cheaper than trying to just develop their own development and manufacturing arm. They can bring something reasonably competitive to market sooner and likely with less teething troubles.

It's a win for AMD because a) they get huge gobs of cash off Intel to buy that business. b) they probably would retain enough expertise in the APU segment to keep pushing that forward so long as they get written into the sale that they retain rights to all the patents so far awarded & they can invest said huge gobs of cash into making their APUs stronger long term. c) If Intel sails off into the horizon to go combat NVidia then that distracts them from crushing AMD in the CPU market, which it seems AMD are actually more interested in long term... at least from an investment and R&D POV.

It's a win for the consumer because AMDs issue with developing new discrete GPUs never really had anything to do with talent or drive... it seems it had more to do with the fact they were struggling for revenue in all sectors CPU or GPU related at one point and decided to prioritise one over another for investment and R&D. Intel have very big pockets to properly fund the necessary future R&D to be a real competitor to NVidia.

Perhaps I am reading this wrong but if you think about the really great AMD GPU's, they really had their roots in what used to be ATI. It was thought when AMD bought ATI that would mean real investment in discrete graphics would be a thing, but in reality it didn't pan out that way... AMD were never strong enough to be properly competitive in both sectors at the same time.
 
I have increasingly wondered if this will pan out with Intel buying the discrete graphics arm of AMD and it becoming the Intel discrete graphics arm....

It's a win for Intel in that suddenly patent issues become largely a non-issue, cross licensing is all well and good but it is grabbing the experience and knowhow to leverage it into something that works as a competitive product and being able to further develop new techniques and features to keep competitive long term. It probably is cheaper than trying to just develop their own development and manufacturing arm. They can bring something reasonably competitive to market sooner and likely with less teething troubles.

It's a win for AMD because a) they get huge gobs of cash off Intel to buy that business. b) they probably would retain enough expertise in the APU segment to keep pushing that forward so long as they get written into the sale that they retain rights to all the patents so far awarded & they can invest said huge gobs of cash into making their APUs stronger long term. c) If Intel sails off into the horizon to go combat NVidia then that distracts them from crushing AMD in the CPU market, which it seems AMD are actually more interested in long term... at least from an investment and R&D POV.

It's a win for the consumer because AMDs issue with developing new discrete GPUs never really had anything to do with talent or drive... it seems it had more to do with the fact they were struggling for revenue in all sectors CPU or GPU related at one point and decided to prioritise one over another for investment and R&D. Intel have very big pockets to properly fund the necessary future R&D to be a real competitor to NVidia.

Perhaps I am reading this wrong but if you think about the really great AMD GPU's, they really had their roots in what used to be ATI. It was thought when AMD bought ATI that would mean real investment in discrete graphics would be a thing, but in reality it didn't pan out that way... AMD were never strong enough to be properly competitive in both sectors at the same time.

RK had a similar idea once, it didn't go well for him. Or maybe it did if his goal was to become an Intel employee.

If they didn't spin off RTG at it's low when they were desperate for cash they sure as hell won't do it now when they're about to kick Intel harder than they ever did. They are the only company with high performance CPU & GPU which you hear them say often and I'm sure they have every intention of staying so.
 
I have increasingly wondered if this will pan out with Intel buying the discrete graphics arm of AMD and it becoming the Intel discrete graphics arm....

It's a win for Intel in that suddenly patent issues become largely a non-issue, cross licensing is all well and good but it is grabbing the experience and knowhow to leverage it into something that works as a competitive product and being able to further develop new techniques and features to keep competitive long term. It probably is cheaper than trying to just develop their own development and manufacturing arm. They can bring something reasonably competitive to market sooner and likely with less teething troubles.

It's a win for AMD because a) they get huge gobs of cash off Intel to buy that business. b) they probably would retain enough expertise in the APU segment to keep pushing that forward so long as they get written into the sale that they retain rights to all the patents so far awarded & they can invest said huge gobs of cash into making their APUs stronger long term. c) If Intel sails off into the horizon to go combat NVidia then that distracts them from crushing AMD in the CPU market, which it seems AMD are actually more interested in long term... at least from an investment and R&D POV.

It's a win for the consumer because AMDs issue with developing new discrete GPUs never really had anything to do with talent or drive... it seems it had more to do with the fact they were struggling for revenue in all sectors CPU or GPU related at one point and decided to prioritise one over another for investment and R&D. Intel have very big pockets to properly fund the necessary future R&D to be a real competitor to NVidia.

Perhaps I am reading this wrong but if you think about the really great AMD GPU's, they really had their roots in what used to be ATI. It was thought when AMD bought ATI that would mean real investment in discrete graphics would be a thing, but in reality it didn't pan out that way... AMD were never strong enough to be properly competitive in both sectors at the same time.

Why would AMD sell its graphic business at a time when gpus are seeing huge expansion into areas other than gaming?

Probably one of AMD biggest mistakes in the last decade was selling off its mobile gpu business. I bet it hurts AMD everytime they see snapdragon SOC sales figures. Furthermore, selling off its gpu business means AMD is basically giving its semi custom business to intel.
 
Why would AMD sell its graphic business at a time when gpus are seeing huge expansion into areas other than gaming?

Probably one of AMD biggest mistakes in the last decade was selling off its mobile gpu business. I bet it hurts AMD everytime they see snapdragon SOC sales figures. Furthermore, selling off its gpu business means AMD is basically giving its semi custom business to intel.
Agree on the rest, but it's not guaranteed AMD would have been Qualcomms vendor of choice for the graphics IP on all the Snapdragons
 
Agree on the rest, but it's not guaranteed AMD would have been Qualcomms vendor of choice for the graphics IP on all the Snapdragons

I doubt Qualcomm bought AMD’s mobile gpu without a non compete clause that locked them out of the smartphone market.

The fact that AMD has made no real attempt at that market hints that such clause exists.

Adreno is proof they had capable mobile tech and is the reason they can’t compete in the space.
 
I doubt Qualcomm bought AMD’s mobile gpu without a non compete clause that locked them out of the smartphone market.

The fact that AMD has made no real attempt at that market hints that such clause exists.

Adreno is proof they had capable mobile tech and is the reason they can’t compete in the space.
I meant that even though Qualcomm chose AMD Z430 (Adreno 200) for the initial Snapdragons, they might have hopped over to PowerVR or ARM later on if AMD hadn't sold their mobile graphics division to them
 
I really hope they can give good products. I don't doubt they will have a good hardware, but for them it's all about the drivers. AMD is out of the loop in the high end for gaming (either it's too late or just non existant. We'll see what they do with Navi), I hope Intel can go fight nVidia here.
 
It's amazing the amount of hype that is surrounding intel entering the discrete GPU market.

All they have released so far is a teaser video and some date two years in the future.

The teaser video shows us what appears to be a single-slot card of some kind, though the largely featureless and likely-rendered image doesn't offer much more to go on than that.

https://techreport.com/news/33995/intel-teases-discrete-graphics-card-on-new-twitter-account

If you look at the responses to the above article you will see many who think that intel will not only compete but excel.

I for one remember Larrabee and how it was going to crush Nvidia and put them out of business.
https://en.wikipedia.org/wiki/Larrabee_(microarchitecture)

An Inconvenient Truth: Intel Larrabee story revealed
http://vrworld.com/2009/10/12/an-inconvenient-truth-intel-larrabee-story-revealed

Nvidia and AMD will not be standing still over the next two years. And Nvidia has been quoting that GPU's take about three years to develope. If intel produces a discrete GPU in two years time I believe that it will not be competitive with Nvidia's high end.
 
Last edited:
Intels primary goal at GPUs was ray tracing. I think in 2 years they will bring up ray tracing GPUs.

If this is true they will have not much trouble with ray tracing patents, because ray tracing just started now.

Maybe it was a mistake that Nvidia start the ray tracing race early. Now Intel knows there goals and can react. For NVidia it should be better when they started ray tracing after Intel presented there new gpu.
 
Intels primary goal at GPUs was ray tracing. I think in 2 years they will bring up ray tracing GPUs.

If this is true they will have not much trouble with ray tracing patents, because ray tracing just started now.

Maybe it was a mistake that Nvidia start the ray tracing race early. Now Intel knows there goals and can react. For NVidia it should be better when they started ray tracing after Intel presented there new gpu.

Three years is the time frame to develop brand new GPU architectures so that would push intel's gpu out to 2021.

As for Nvidia releasing ray tracing in a GPU now I say it is better to be a leader than a follower.
 
I'm pretty sure Intel is already halfway through development if not more by now. They wouldn't make a big public announcement before even beginning work on the thing. Likely, they only announced it now because they may soon be working more openly with multiple players in the industry, so they though they'd rather have people know through them than through a leak. Building hype early is important for a new product so devs want to start experimenting with the pre-release kits so there already are optimized software out by the time the card is released to the public.
 
I'm pretty sure Intel is already halfway through development if not more by now. They wouldn't make a big public announcement before even beginning work on the thing. Likely, they only announced it now because they may soon be working more openly with multiple players in the industry, so they though they'd rather have people know through them than through a leak. Building hype early is important for a new product so devs want to start experimenting with the pre-release kits so there already are optimized software out by the time the card is released to the public.

If that was the case then it would have to be a 14nm part and that would just be a massive fail......
2021 would seem like the year of the 10nm intel dgpu , assuming they can get 10nm out the door and into volume during 2020.
 
So you really think they went out and said "We will build a dGPU" went out and made a snazzy promo video for it, announced it in a conference, created some hype, and then after that and was said and done they called up their engineers and said "ok guys, so now you have to start thinking about how we are gonna make this."
 
Last edited:
Back
Top