How "good" is Intel Graphics? A thought experiment

From the pixels, bits, and shaders to the graphic cards that power them. Discuss the latest from AMD and NVIDIA here.

Moderators: morphine, SecretSquirrel

How "good" is Intel Graphics? A thought experiment

Postposted on Sun May 18, 2014 9:39 pm

Just hypothetically thinking here; but if Intel's IGP technology, the IRIS Pro 5200, were spun off and scaled to size of a typical AMD or NVIDIA gpu, would it be competitive with them?
I dont think, therefore I am not.
WhatMeWorry
Gerbil
 
Posts: 76
Joined: Mon Apr 13, 2009 6:01 pm
Location: Illinois

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Mon May 19, 2014 7:58 am

WhatMeWorry wrote:Just hypothetically thinking here; but if Intel's IGP technology, the IRIS Pro 5200, were spun off and scaled to size of a typical AMD or NVIDIA gpu, would it be competitive with them?


Probably not.
If you look at the IGP stuff, the GCN architecture in Kabini/Kaveri is faster than the GT2 (20EU) and GT3 (40EU) stuff from Intel. Only the GT3e with embedded RAM is faster, and you can't really scale that up the same way.

You might think in terms of raw die area that it's competetive, but you also have to factor that Intel's 22nm process means that they are squeezing in 60% more transistors to the same area that AMD manage with a 28nm process.
<insert large, flashing, epileptic-fit-inducing signature (based on the latest internet-meme) here>
Chrispy_
Gerbil Jedi
Gold subscriber
 
 
Posts: 1777
Joined: Fri Apr 09, 2004 3:49 pm

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Mon May 19, 2014 8:11 am

AMD's integrated graphics benefit from being the same architecture as their GPUs (even past APUs were based on past GPUs) - so they benefit from all the per-game performance optimizations that are built into the driver. Intel's drivers, as far as I know, are meant to be general-purpose, and aren't as well optimized. That might explain at least part of it.
I do not understand what I do. For what I want to do, I do not do. But what I hate, I do.
derFunkenstein
Gerbil God
Gold subscriber
 
 
Posts: 21255
Joined: Fri Feb 21, 2003 9:13 pm
Location: WHAT?

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Mon May 19, 2014 9:11 am

From what I've seen online, the intel IGP (especially Iris Pro) is not only not die-size competitive with discrete GPUs (or in AMD's case, their APU), it's also not competitive in performance per watt compared to mobile discrete parts from nvidia and AMD.

But then we've seen that embedded RAM solutions don't always perform better than a straight GPU solution of the same die size (cf Xbox One).
Wind, Sand and Stars.
Voldenuit
Minister of Gerbil Affairs
 
Posts: 2453
Joined: Sat Sep 03, 2005 11:10 pm

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Mon May 19, 2014 11:10 am

As far as die sizes go, it is really weird looking at Haswell and trying to scale up: Intel doesn't use the entire die are in some versions. That complicates the hypothetical picture and makes any rough estimate, well rough. So here are the dies sizes (source)

Haswell 2C GT3 = 177 mm^2
Haswell 4C GT3 = 264 mm^2
Haswell 4C GT2 = 181 mm^2
Haswell 2C GT2 = 130 mm^2

Going from GT2 to GT3 (adding 20 EUs) adds 41 mm^2 of die space.
Going from dual to quad core adds 37 mm^ 2 of die space.
Uncore in the 2C GT2 configuration would be 52 mm^2.

I'd guesstimate that a 160 EU part would around 400 mm^2 with a wide GDDR5 memory bus. Note that I'm rounding up quite a bit to account for additional ROPs, TMUs and wider memory bus necessary to scale. Cranking up the base clock speeds to 1.3 Ghz base with turbo up to 1.6 Ghz give it another boost. The dedicated GDDR5 memory bus would address the bandwidth issue which would allow us to use GT3e the baseline instead of the vanilla GT3. Assuming perfect linear scaling, we'd be looking at a part roughly 5 times faster than GT3e in terms of compute. Even with these performance bumps, I just don't see it being competitive with the top of the line from AMD or nVidia. Putting such a chip up against the R9 280X or GTX 770 would be interesting and something that Intel could be far more competitive with (the R9 280X is about 5.5 times faster in theoretical compute).

I'm not even going to attempt to tackled the concept of power consumption. Haswell is a very power optimized design and their 84W desktop allocation is shared between the CPU and GPU. The dynamic nature make its it very difficult to determine just how much of that power consumption is due to just the GPU. Even then, they do have a power envelope of 300W to work in the PCI-e spec which is 3.5 times more. I fathom that my 160 EU part above would fit into the 300W spec but I don't know if it'd go below 225W for a 6 + 6 pin configuration under load. On the flip side, I'd argue that Intel would have better idle power consumption than either AMD or nVidia.

Then again, Intel has fab space to use and the next Xeon Phi is estimated to be around 700 mm^2 before the eDRAM is factored in. Scaling the number of EU's to 300 along with other units would be around 700 mm^2 by my guestimation. Toss in some of Intel's fast eDRAM and having 500 GB/s of bandwidth is feasible. Such a design would be performance competitive with the best from AMD and nVidia but also radically more expensive. Large dies like that are not cheap, inherently have poor yields plus the eDRAM only adds more to the cost. So ultimately if Intel really, really wanted to, I believe that they could take the performance crown but it would be a Pyrrhic victory.
Dual Opteron 6376, 128 GB DDR3, Asus KGPE-D16, Radeon 6970
Mac Pro Dual Xeon E5645, 48 GB DDR3, GTX 770
Core i7 3930K@4.2 Ghz, 32 GB DDR3, GA-X79-UP5-Wifi
Core i7 2600K@4.4 Ghz, 16 GB DDR3, Radeon 6870, GA-X68XP-UD4
the
Gerbil
Gold subscriber
 
 
Posts: 57
Joined: Tue Jun 29, 2010 2:26 am

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Mon May 19, 2014 11:32 am

I dunno. Feel like I'd rather see a really big GPU from AMD or Nvidia manufactured on Intel's process tech.
jihadjoe
Gerbil Team Leader
 
Posts: 248
Joined: Mon Dec 06, 2010 11:34 am

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Tue May 20, 2014 4:38 pm

jihadjoe wrote:I dunno. Feel like I'd rather see a really big GPU from AMD or Nvidia manufactured on Intel's process tech.


Who wouldn't! It'd be like skipping an entire GPU generation with what NVIDIA or AMD could design and then squeeze into Intel's 22nm tech.

Which is a funny thought, Intel has been producing 22nm chips for 3rd parties for over a year now. I would venture that in a few more years if the foundry business will be at the point Intel would be more eager to fund expensive plants / nodes by opening up to producing NVIDIA's big chips.
Kougar
Gerbil Team Leader
 
Posts: 245
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Tue May 20, 2014 4:54 pm

The way I see it is this- Intel has all of the IP that they'd need in order to tweak their current graphics technology into a state that could then be properly scaled up to compete with AMD and Nvidia's top offerings, should they choose to do so.

But I don't see the incentive for them to do that. They don't really have the incentive to fab their competitors CPUs or GPUs either- Intel is likely not very interested in making 'toys', and their 'tools' such as the Xeon Phi compete directly with the higher-end GPUs for the compute market.
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Maximum Gerbil
 
Posts: 4968
Joined: Fri Oct 15, 2004 10:41 pm

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Tue May 20, 2014 5:57 pm

Airmantharp wrote:The way I see it is this- Intel has all of the IP that they'd need in order to tweak their current graphics technology into a state that could then be properly scaled up to compete with AMD and Nvidia's top offerings, should they choose to do so.


Intel already tried that though. Larrabee went pretty poorly and was never able to even reach performance parity with current GPUs of the time despite the promises to offer several times more performance. Intel had to redesign the whole thing into a general compute card, but the graphics results of their Larrabee failure went into Intel's HD Graphics initiative. So making it "big" again into another Larrabee I don't see working at all for them.

Airmantharp wrote:But I don't see the incentive for them to do that. They don't really have the incentive to fab their competitors CPUs or GPUs either- Intel is likely not very interested in making 'toys', and their 'tools' such as the Xeon Phi compete directly with the higher-end GPUs for the compute market.


The incentive is the cost of building foundries and the extremely expensive (and rising) R&D costs of every new process node generation. That's the sole reason Intel is already producing chips for third parties, and in say 5 more years I bet they may actually consider it. Hasn't the volume of CPU chips been decreasing with the shrinking desktop + laptop markets? Another reason Intel is so gung-ho to get chips into tablets, they need to keep their fabs busy producing something.
Kougar
Gerbil Team Leader
 
Posts: 245
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Tue May 20, 2014 9:22 pm

Larrabee wasn't a graphics card- it was an attempt to run graphics on what is now the Xeon Phi, which is an array of x86 cores. What we're talking about is using Intel's actual graphics IP to build a full-size discrete GPU.

As for foundry costs etc- Intel will have to become TSMC/UMC/TI/etc. if they want that business, but they very likely don't. There's far less margin in making competitor's products better. They're better off making great Xeon Phi's than making great Teslas.
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Maximum Gerbil
 
Posts: 4968
Joined: Fri Oct 15, 2004 10:41 pm

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Thu May 22, 2014 12:32 pm

But Larrabee did have some graphics hardware baked into it for the sole purpose of making it a graphics processor, it wasn't just "pure" x86 cores with the GPU stuff done in pure software. Similarly I don't think Intel's current "EU"s would scale well if used to make a large discrete GPU core, but that's the fun part about hypothesizing I guess. :)

Airmantharp wrote:As for foundry costs etc- Intel will have to become TSMC/UMC/TI/etc. if they want that business, but they very likely don't. There's far less margin in making competitor's products better. They're better off making great Xeon Phi's than making great Teslas.


Don't get me wrong, I completely agree with ya on this point. But say 10 years down the road unless some new alternative to silicon-based semiconductors pops up, the cost/benefit analysis may get to the point they might start considering it.
Kougar
Gerbil Team Leader
 
Posts: 245
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Thu May 22, 2014 12:48 pm

All I'm saying is that if they get to the point that their fabbing other companies' designs, it means that they've patently failed with their own designs :).
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Maximum Gerbil
 
Posts: 4968
Joined: Fri Oct 15, 2004 10:41 pm

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Sat May 24, 2014 8:43 am

Airmantharp wrote:All I'm saying is that if they get to the point that their fabbing other companies' designs, it means that they've patently failed with their own designs :).


Fair point. But, do you think Intel would consider making chips for Apple possibly? Like, my thinking was that Intel makes chips for everything. Processors, Phi, iGPs, chipsets, networking and IO controllers, RAID, even NAND. They're trying to compete in everything from big iron down to smartphone SoCs, so technically most companies could be an indirect competitor. Fun to theorize about anyway.
Kougar
Gerbil Team Leader
 
Posts: 245
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: How "good" is Intel Graphics? A thought experiment

Postposted on Sat May 24, 2014 1:03 pm

Kougar wrote:Fair point. But, do you think Intel would consider making chips for Apple possibly? Like, my thinking was that Intel makes chips for everything. Processors, Phi, iGPs, chipsets, networking and IO controllers, RAID, even NAND. They're trying to compete in everything from big iron down to smartphone SoCs, so technically most companies could be an indirect competitor. Fun to theorize about anyway.


Possibly- Apple is already a close partner when it comes to mobile CPUs, and they helped Intel launch the Ultrabook platform by providing proof of concept. As far as fabbing Apple's mobile SoCs, such a partnership might be very attractive to Apple in that they'd no longer be sourcing their mobile SoCs from Samsung, and that Intel's processes might be superior enough to make the R&D for a transition worthwhile.
Canon 6D||[24-105/4L IS USM|100/2.8L Macro IS USM|70-300/4-5.6 IS USM|40/2.8 STM|50/1.4 USM|85/1.8 USM|Samyang/Bower 14/2.8 Full-Manual Rectilinear Wide-angle|
Canon EOS-M|11-22/4-5.6 IS STM|22/2 STM|EF-M 18-55/3.5-5.6 IS STM|
For sale!|24/2.8 IS USM
|
Airmantharp
Maximum Gerbil
 
Posts: 4968
Joined: Fri Oct 15, 2004 10:41 pm


Return to Graphics

Who is online

Users browsing this forum: Bing [Bot] and 5 guests