Linus Torvalds: Discrete GPUs are going away

[Rakehellion]

Its what i keep saying. dGPUs are on deathrow.

Wait, so now people want GPUs you can never upgrade? I thought this was a major piece of ammo the Mac-haters used.

 

[ShintaiDK]

Wait, so now people want GPUs you can never upgrade? I thought this was a major piece of ammo the Mac-haters used.

Plenty of things in your PC you cant upgrade. Specially if its a laptop.

 

[KingFatty]

I want to spend less. If I could buy one motherboard and chip, that gave me performance I needed while saving money, I'd do that. But so far, nothing has satisfied me, so on the last laptop I bought, it had a discrete video card, and my desktop has discrete video cards on top of the integrated.

But I have accepted integrated sound cards and network cards. Long long ago, I swore I would always get a soundblaster sound card. But that ended once the integrated sound cards got good enough.

I could see a time in the future when integrated video cards get good enough. I mean maybe they'll be like quadruple 780ti performance, so that could satisfy me

 

[Reaper_Tech]

YEa hes right in a way but were still years off yet. But eventually discrete GPUs will probably die out.

 

[Techhog]

Five years sounds completely plausible for killing discrete GPU's in mobile devices. Hell... the Iris Pro 5200 on my Macbook Pro are already good enough to handle Civilization V and a few other games without any issue.

Oh the Windows desktop, I'd give it a decade. For the Linux desktop, we're basically already there... most Linux users aren't gamers, and most of the 3D gaming drivers for Linux suck.

While I agree that it's plausible, this is a bad example due to the fact that I highly doubt that whether or not you could play Civ V would have affected your decision to buy that Macbook.

 

[ultimatebob]

While I agree that it's plausible, this is a bad example due to the fact that I highly doubt that whether or not you could play Civ V would have affected your decision to buy that Macbook.

Nope, it didn't have a damn thing to do with my decision... it was just an unexpected bonus

After watching my flimsy plastic Dell XPS 15 fall apart in 2 years even though I rarely traveled with it, I wanted something built better.

I was actually thinking of getting a Thinkpad T series instead, but the Macbook was only about $150 more than a Lenovo with similar specs at the time. I figured that it was worth that much just to avoid having to use Windows 8.

 

[3DVagabond]

It is hard to believe iGPs will ever catch up to dGPUs when the first ones have only sub 100W thermal envelopes shared with CPUs while the latter have up to 500W of juice to stretch their legs.

They don't have to catch up. They only have to get "good enough".

 

[frozentundra123456]

They don't have to catch up. They only have to get "good enough".

Define "good enough". It means different things to different people. Plus games are becoming increasingly demanding and resolutions are increasing. So "good enough" is a moving target. As I said earlier, I see "good enough" as more meaningful in mobile, where power, heat and space are limited.

I see "good enough" as much less relevant in the desktop. Why would I want "good enough" when I can have "much better"?

 

[ShintaiDK]

Define "good enough". It means different things to different people. Plus games are becoming increasingly demanding and resolutions are increasing. So "good enough" is a moving target. As I said earlier, I see "good enough" as more meaningful in mobile, where power, heat and space are limited.

I see "good enough" as much less relevant in the desktop. Why would I want "good enough" when I can have "much better"?

Good enough is essentially when the ROI on dGPUs are unable to pay for any new development.

Since we can pretty much assume the mobile space is a complete loss now for the dGPU. Then we can wonder what volume is needed to substain a desktop dGPU. The question is if one of the 2 gonna give up. or they both will drive it faster to its death.

Already seems we gonna have 5 years (Or longer) with 28nm dGPUs.

 

[3DVagabond]

Define "good enough". It means different things to different people. Plus games are becoming increasingly demanding and resolutions are increasing. So "good enough" is a moving target. As I said earlier, I see "good enough" as more meaningful in mobile, where power, heat and space are limited.

I see "good enough" as much less relevant in the desktop. Why would I want "good enough" when I can have "much better"?

Good enough is when the masses are satisfied. 1%'ers aren't going to keep the dGPU industry profitable.

 

[NTMBK]

Good enough is when the masses are satisfied. 1%'ers aren't going to keep the dGPU industry profitable.

For a good example of this, just look at the plasma TV market. The technology is much better than LCD, but the costs are much higher and most people just don't care enough about the difference in quality. Result? Manufacturers are shutting down plasma TV production.

 

[3DVagabond]

For a good example of this, just look at the plasma TV market. The technology is much better than LCD, but the costs are much higher and most people just don't care enough about the difference in quality. Result? Manufacturers are shutting down plasma TV production.

LCD is so cheap too that the plasmas aren't profitable when competing on price with them. Add to that the additional cost of shipping them safely as they are more fragile.

 

[ShintaiDK]

Plasma also lost heavily outside the US due to power consumption.

 

[AtenRa]

Good enough is essentially when the ROI on dGPUs are unable to pay for any new development.

Since we can pretty much assume the mobile space is a complete loss now for the dGPU. Then we can wonder what volume is needed to substain a desktop dGPU. The question is if one of the 2 gonna give up. or they both will drive it faster to its death.

Already seems we gonna have 5 years (Or longer) with 28nm dGPUs.

And how did you come to that conclusion ??

 

[3DVagabond]

Plasma also lost heavily outside the US due to power consumption.

Yes. That would be another issue.

 

[ShintaiDK]

And how did you come to that conclusion ??

Because not many laptop models are sold with a dGPU anymore. And if they do, its more or less only nVidia.

 

[AtenRa]

Because not many laptop models are sold with a dGPU anymore. And if they do, its more or less only nVidia.

85 Laptops with NVIDIA 800 series Mobile GPUs.

Impressive that 76 of them are with GT860/870 and 880 HighEnd GPUs.

 

[ShintaiDK]

85 Laptops with NVIDIA 800 series Mobile GPUs.

Impressive that 76 of them are with GT860/870 and 880 HighEnd GPUs.

Your link shows 66 for some reason. Out of how many laptops?
http://www.newegg.com/All-Laptops-No...Category/ID-32

Not even to talk about the low volume of those actually selling with a dGPU. And you may note the majority of the laptops you link are MSI(35). MSI isnt exactly know to be the biggest laptop supplier.

Your link contains 1 Dell (Out of 458), 4 Acer(Out of 133) and 4 Lenovo(Out of 416).

Thanks for proving my point.

 

[Revolution]

I'm ready to buy an iGPU @$200 which can give avg. 60fps BF4 @ultra setting.
Did I missed something ?

 

[Insert_Nickname]

Good enough is when the masses are satisfied. 1%'ers aren't going to keep the dGPU industry profitable.

Hate to break it to you, but the "masses" where already satisfied with the GMA900. 10 years ago. Any DX11 level IGP with video acceleration is "good enough" for 98% of the market.

Actually most users don't even know what they're missing with a better GPU for gaming. I've even seen people gaming on Brazos and pre-SB Intel IGPs. They didn't even know you could get better graphics performance then that...

 

[Techhog]

Already seems we gonna have 5 years (Or longer) with 28nm dGPUs.

I laugh harder every time you say this. Actually, I laughed hardest this time because I didn't think this was what you meant. This is the most absurd thing I've ever read on this forum! You could have at least said that "after 20nm, dGPUs won't shrink anymore" so that you would have had some sort of leg to stand on.

If you're going to make an argument, please use common sense. We KNOW that 20nm GPUs are coming next year. That's barely even debatable. Focus on facts and rumors, not on your deep desire for an Intel monopoly on everything in x86 PCs.

Ugh, I can't stop facepalming over this statement. I just... wow. WOW.

Yes, it's likely that iGPs will take over someday. However, it's not right around the corner like some of you want it to be. Making up stuff like this won't change anything.

EDIT: Maybe I misunderstood you, and you meant 5 years total instead of 5 more years... which is still wrong and and completely baseless, but not as silly I guess.

 

[3DVagabond]

Hate to break it to you, but the "masses" where already satisfied with the GMA900. 10 years ago. Any DX11 level IGP with video acceleration is "good enough" for 98% of the market.

Actually most users don't even know what they're missing with a better GPU for gaming. I've even seen people gaming on Brazos and pre-SB Intel IGPs. They didn't even know you could get better graphics performance then that...

Well, it's amazing they sell any graphics cards then, isn't it? Actually I wasn't referring to those masses. The masses that want something better than a console.

 

[Techhog]

Well, it's amazing they sell any graphics cards then, isn't it? Actually I wasn't referring to those masses. The masses that want something better than a console.

They'll want something better than a console at a similar price to a console. That would mean GT5 on i3, and that's a good 5-7 years away. By then, there will be new consoles on the horizon.

AMD may get there in 3-4 years, but they don't have the market share to cause the type of shift you're looking for. Additionally, consoles will be cheaper by then.

 

[3DVagabond]

They'll want something better than a console at a similar price to a console. That would mean GT5 on i3, and that's a good 5-7 years away. By then, there will be new consoles on the horizon.

AMD may get there in 3-4 years, but they don't have the market share to cause the type of shift you're looking for. Additionally, consoles will be cheaper by then.

Consider that consoles are already using AMD APU's and then consider how long it might take AMD to get that performance on PC.

 

[Techhog]

Consider that consoles are already using AMD APU's and then consider how long it might take AMD to get that performance on PC.

I addressed that. They'll hit XBox One in 1-2 years, of course, but that's ignoring any other factors and just straight-up assuming that same hardware = same real-world performance = better sales for AMD APUs. It's ignoring any differences between PCs and consoles, the possibility of badly optimized PC games, the growth of 4K, tablet adoption, and console price cuts.