Why can't Intel compete with AMD in GPU Sector?

[frozentundra123456]

Would it not have been cool had Intel bought ATI instead of AMD. I am sure they would have negotiated a better deal and anyway they could have absorbed the cost easily. But just think about dgpus on 14 or even 22 nm.

 

[mrmt]

Would it not have been cool had Intel bought ATI instead of AMD. I am sure they would have negotiated a better deal and anyway they could have absorbed the cost easily. But just think about dgpus on 14 or even 22 nm.

8 years after the acquisition AMD never ever could make up for the 6 billion they spent with ATI. Hell, not even Nvidia could make up this number in the latest 8 years, let alone the price premium someone would have to pay to get Nvidia and not ATI. I'd say that Intel did right in not acquiring neither ATI nor Nvidia.

 

[BSim500]

As far as dGPU's "dying off" figures are concerned, a lot of people are simply switching to 2-3 year upgrade cycles (especially after a generation of rebranded cards). Similar slowdown is seen on CPU's with people holding onto Sandy/Ivy Bridge's for 3-4 years. Others are simply increasing the number of 2nd/3rd devices, which increases iGPU relative share without actually reducing dGPU share in absolute terms (ie, buying an Intel Atom based tablet doesn't = throwing your desktop in the bin).

 

[witeken]

In short, not worth.....

Edit: 2013, Intel pulled in 52 billion in revenue. AMD pulled in 5 billion in Revenue. Focus all their money for a market that isn't even 10% of their revenue? No thanks.....

Intel is already focusing many years on the GPU, but they don't compromise the CPU while doing so.

 

[III-V]

It's not about having more capability to execute ...

It's a question of TRUST! A lot of consumers aren't willing to give Intel the time of day when it comes to a discrete GPU.

How do you propose that Intel can gain a solid foothold in the discrete graphics market when they have no following ?

Because if Intel bests Nvidia or AMD (unlikely, but bear with me), then word will get out and people will gladly go over to Intel. As Intel does close in on AMD and Nvidia, and grabs market share like they have been, it will become increasingly important for them to focus on providing great drivers, and I think this problem has already been resolved to a great extent. Frankly, their drivers are rather good these days, however I question their ability to quickly push through things like hotfixes.

Plus, Intel's IGP / GPU drivers are legendary... for poor quality and support. (No Haswell IGP drivers for XP, for example.)

Good. Screw XP, and everyone still using it. Intel's drivers have actually improved tremendously, starting at least from Sandy Bridge onward.

So your whole argument is that Intel can improve & AMD can't.?

Way to completely distort what he said.

Intel has a LOT of money but they cannot make a video card? how come? apart from AMD always beating them in iGPU, they intel also has no video card to this day.

As mrmt pointed out, it wouldn't be profitable. I do expect the dGPU market to grow a bit with the upcoming 4K era, but there isn't really room for a third player, unless Intel wanted to blow their cash reserves on sticking it to Nvidia. I don't see why they would want to do that, though.

As for the IGP thing, it takes time to catch up, when you previously had no interest in doing so. Expect the gap between Intel and AMD's IGPs to close significantly, just as it's been closing since Clarkdale.

 

[monstercameron]

That's why I can't wait for mantle and its ilk. Take the onus of updates from drivers onto game devs

 

[III-V]

That's why I can't wait for mantle and its ilk. Take the onus of updates from drivers onto game devs

What is Mantle's worth, exactly? It will become irrelevant with DX12. It is a waste of time and money for AMD.

 

[rtsurfer]

Way to completely distort what he said.

So...

They don't have ambitions to compete in the discrete market. When we talk about integrated igpus I expect that even Intels desktop GT2 for Skylake is able to compete with the fastest desktop Kaveri. Their Gen7 is 2.5 years old technology with flaws. Haswell is based on this. For desktop Skylake-S will bring the first real GPU update since 2012 Ivy Bridge.

For an APU it's pretty new. You can't compare a CPU+integrated GPU and a whole new platform with a dedicated standalone GPU because validation times are much longer. There is a reason why AMDs APU is lagging behind their standalone GPUs. I doubt Carrizo will bring a big improvement over Kaveri btw. Desktop existence is still unclear as well.

He says that he expects Intel to deliver groundbreaking IGP performance Skylake & Carrizo will not bring any major improvements.

So is it distorting everything he said what I posted this

So your whole argument is that Intel can improve & AMD can't.?

Or is it just me interpreting his post.?

What is Mantle's worth, exactly? It will become irrelevant with DX12. It is a waste of time and money for AMD.

So did Nvidia also waste their time when they brought Gsync to the market as it will become irrelevant with AMD bring Adaptive Sync in collaboration with the VESA standard.

 

[ThatBuzzkiller]

Because if Intel bests Nvidia or AMD (unlikely, but bear with me), then word will get out and people will gladly go over to Intel. As Intel does close in on AMD and Nvidia, and grabs market share like they have been, it will become increasingly important for them to focus on providing great drivers, and I think this problem has already been resolved to a great extent. Frankly, their drivers are rather good these days, however I question their ability to quickly push through things like hotfixes.
Good. Screw XP, and everyone still using it. Intel's drivers have actually improved tremendously, starting at least from Sandy Bridge onward.

I agree that their drivers seem fine for the most part but is it really worth the effort for Intel trying to break in the discrete graphics market to gain hold when they already have the most promising GPGPU solution and will likely dominate laptops with their IGPs ?

Another problem that Intel has to face is dealing with lower profit margins too for this endeavor to work ...

 

[tential]

Intel is already focusing many years on the GPU, but they don't compromise the CPU while doing so.

You know EXACTLY what I mean. Intel's goal will never be to dethrone AMD in the GPU sector but to give "good enough" performance without compromising their GPU.

 

[roob]

why would they want to enter a shitty market? just look at nvidia's quarterly numbers: 1b revenue and a paltry 200m in gross income. big deal.

profanity is not allowed in the technical forums
Markfw900

 

[tential]

why would they want to enter a ****** market? just look at nvidia's quarterly numbers: 1b revenue and a paltry 200m in gross income. big deal.

2% of intel's Revenue. Seems like a HUGE market intel is just clamoring to get into.

http://ycharts.com/companies/NVDA/r_and_d_expense

Intel spends ~$3 Billion a quarter in R&D and Nvidia spends ~$350 million a quarter.

Assuming Intel would have to spend equal to Nvidia to compete, they'd spend $300 million a quarter ($1.2 billion a year) to enter into a market that has $1B in revenue for the best competitor?

Seems like a great market to enter into.

 

[ShintaiDK]

2% of intel's Revenue. Seems like a HUGE market intel is just clamoring to get into.

http://ycharts.com/companies/NVDA/r_and_d_expense

Intel spends ~$3 Billion a quarter in R&D and Nvidia spends ~$350 million a quarter.

Assuming Intel would have to spend equal to Nvidia to compete, they'd spend $300 million a quarter ($1.2 billion a year) to enter into a market that has $1B in revenue for the best competitor?

Seems like a great market to enter into.

Yep. Not to mention the result may end up as 3x AMD. 3 companies with too low marketshare to make money.

 

[rtsurfer]

Yep. Not to mention the result may end up as 3x AMD. 3 companies with too low marketshare to make money.

Since Intel isn't reinventing the wheel here.
Aren't there chances of them stepping into more of Nvidia & ATi's GPU patents.

Then they will have to fork out Billions of Dollars in licensing fees like they paid to Nvidia sometime back.

That could significantly increase the cost of their APU endeavor.

 

[Enigmoid]

2% of intel's Revenue. Seems like a HUGE market intel is just clamoring to get into.

http://ycharts.com/companies/NVDA/r_and_d_expense

Intel spends ~$3 Billion a quarter in R&D and Nvidia spends ~$350 million a quarter.

Assuming Intel would have to spend equal to Nvidia to compete, they'd spend $300 million a quarter ($1.2 billion a year) to enter into a market that has $1B in revenue for the best competitor?

Seems like a great market to enter into.

To be fair they are spending >50% of that R and D money anyway for their IGPs. Only real difference would be the scaling of the architecture and production costs, as well as competent drivers.

 

[mrmt]

Intel spends ~$3 Billion a quarter in R&D and Nvidia spends ~$350 million a quarter.

Assuming Intel would have to spend equal to Nvidia to compete, they'd spend $300 million a quarter ($1.2 billion a year) to enter into a market that has $1B in revenue for the best competitor?

This spending you are seeing in Nvidia is not all related to graphics, a lot of it is directed to its Tegra business, another sizable chunk is for developing their professional drivers....

 

[ehume]

. . . And then when you consider someone buying a i7 why would they even use the iGPU in a desktop? iris pro seems so useless for desktops.

Maybe all the people who use their computers for solving problems, and don't need much in the way of graphics. None of the machines I use at work have iGPU's. I'll give you that they are not i7's. But I used a low-end GPU with my i7 860, and I'm using the built-in iGPU's on both my 4770k and my 4790k machines.

dGPU's are for gamers. Or maybe if you have a hi-res monitor and you watch movies? I dunno -- maybe the i7's iGPU can even handle that.

But there are many who aren't willing to pay for the dGPU. Intel follows the money.

 

[Blitzvogel]

I think it'll be a long time before we see an end to dGPUs. The amount of capability/mm² and the ability to upgrade that single component versus an APU represents better customization capabilities that both gamers and graphics focused businesses will want. Plenty of people would rather see an APU's graphics die area put to better use as more CPU cores, or not there altogether to reduce cost.

 

[papapapapapapapababy]

please no... if intel pulled of a "conroe like gpu" , then > years and years of total stagnation ( and gpu erratas!) i rather have another player in town , better yet a new gaming os with no locked proprietary dx 367 garbage

 

[jpiniero]

I think it'll be a long time before we see an end to dGPUs. The amount of capability/mm² and the ability to upgrade that single component versus an APU represents better customization capabilities that both gamers and graphics focused businesses will want. Plenty of people would rather see an APU's graphics die area put to better use as more CPU cores, or not there altogether to reduce cost.

Intel is increasing the graphics die % basically to kill the low and medium end dGPUs knowing that will severely hurt AMD and nVidia's dGPU profits. Whether they can make it work is the question.

 

[bullzz]

HBM is not out yet and eDRAM has just started coming out. these two will really disrupt dGPU market. like in laptop market how we saw iGPU get to a "good enough" performance, I expect to see iGPU eat up low to mid dGPU market for desktop.
where dGPUs really make money is HPC market and Intel has Phi for it
can Intel compete with AMD in GPU? may be. may be not. but it does not make sense for business

 

[Morbus]

(No Haswell IGP drivers for XP, for example.)

Haswell was released six months before the end of XP's extended support.

It's no surprise they don't support XP. And it's a good thing too.

 

[Enigmoid]

Haswell was released six months before the end of XP's extended support.

It's no surprise they don't support XP. And it's a good thing too.

Given the quality of intel's gpu drivers its probably a good thing they didn't put resources there.

 

[monstercameron]

What is Mantle's worth, exactly? It will become irrelevant with DX12. It is a waste of time and money for AMD.

It's ilk meaning oglnext and d3d12

 

[Anon_lawyer]

The simple answer is attention span. Intel is a huge company with vast R&D resources, but they also have a vast array of products to support and expand already. At a certain level the key constraint is how many projects can you staff with A-team players, and how many high priority projects can the C-level executives manage at once. Ever hear the expression that if you have 10 priorities you have no priorities? That's true for big companies too. Comparing the size of Intel's major existing business lines it is pretty clear that the dGPU market isn't worth re prioritizing a lot of top people.

You see this attitude at lots of big companies, notably Microsoft and more recently Google. Steve Ballmer was always looking for really big potential businesses, figuring that Microsoft was already so big that a new business had to be huge to be worthwhile. Larry Page keeps pushing this moon-shoot idea, that a big world-changing technology is the only thing worth it to Google. Both are partly just recognition of the reality that you can only do so many things well at once.

Quite often this is the real answer to the question 'how can [giant widely admired company X] screw up [new product Y] so badly?!?'