[WCCF] Intel Skylake 2015 Platform Details Revealed

[Ajay]

I don't like being the party-pooper when it comes to CPU speculation.

That said, if you're willing to allow a much larger margin of error in your guesses and the fact that engineers are supposed to be rational and that you don't have unlimited time and resources.

I'm not mad at you or engineers (I'd have to be angry at myself) - just as witeken noted, these kinds of PR slides seem to be useful at first but ultimately are confusing. Trying to read the 'tea leaves' in them winds up being a fools errand.

 

[Idontcare]

For Broadwell or Skylake?

For Broadwell, I'd expect pretty much nothing in terms of IPC gain. We didn't see one in Westmere either, remember? For Skylake, that sort of improvement will be a disappointment at the least.

For Skylake over Broadwell.

 

[IntelUser2000]

For Skylake over Broadwell.

Yea, that would be a disappointment. I'd put a absolute minimum bar(at equal frequencies) at 20%, or they'll find themselves quickly reducing Core chips to Atom-levels of pricing. Alternative is to go for a architecture that for example, brings 40% gain but reduces frequency to be ~20%.

It really does not matter if they are at the end of the "IPC era" or that it takes some heroic efforts to get the extra performance. If they can't do it, regardless of the reasons, their market segmentation will have to change dramatically. Too bad, so sad.

Broxton is probably Intel's equivalent of "A57". That 30-50% gain means Core will need to be that much faster to be ahead.

That's not the end. They'll have to find something for Tock after Skylake, because 2 years after Broxton will probably bring significant changes as well.

 

[mikk]

For Broadwell or Skylake?

For Broadwell, I'd expect pretty much nothing in terms of IPC gain. We didn't see one in Westmere either, remember? For Skylake, that sort of improvement will be a disappointment at the least.

We did see 3-4% for IVB over SB.

 

[Homeles]

We did see 3-4% for IVB over SB.

IVB was a tick, though.

 

[shady28]

You can't compare a Gen7 HD4600 to a Gen9 GPU, certainly if you don't even know the amount of EUs.

This entire thread is sourced from an article that shows the i3 / i5 Skylake being G2 and the i7 being GT4e.

Since a GT2 is in fact an HD4600, and since the post I was responding to suggested AMD and Nvidia would be in trouble - I think a comparison of recent AMD APU iGPUs to the HD4600 is quite appropriate since those Intel chips (i3 and low end i5) would be in the same price range as a high end AMD APU.

 

[Sweepr]

This entire thread is sourced from an article that shows the i3 / i5 Skylake being G2 and the i7 being GT4e.

There's no evidence that GT4e will be restricted to Core i7 models yet. Also, it is safe to assume that Skylake GT2 will be faster than any Haswell GT2 part.

 

[Homeles]

There's no evidence that GT4e will be restricted to Core i7 models yet. Also, it is safe to assume that Skylake GT2 will be faster than any Haswell GT2 part.

A lot faster.

 

[IntelUser2000]

We did see 3-4% for IVB over SB.

But Haswell is a Oregon team design, and Broadwell is a Tick of that, same as Westmere was to Nehalem. The former, latter than few extra instructions, had zero IPC gain.

 

[SAAA]

Hi there, long time forum lurker finally registered. Finally we have some Skylake infos! I was awaiting them so much, probably I'm gonna update with that tock from a 7 year rig...

A lot faster.

About this: if the rumors are true wouldn't a 32EUs part with improvments from gen7.5 to gen9 be similar as performance to a 40EUs haswell HD5000-5100 (without -e)? That's a lot of performance for minimum mainstream parts, abeit weaker than most GPUs it would be enough for non-gaming uses, or not?

 

[witeken]

Hi there, long time forum lurker finally registered. Finally we have some Skylake infos! I was awaiting them so much, probably I'm gonna update with that tock from a 7 year rig...



About this: if the rumors are true wouldn't a 32EUs part with improvments from gen7.5 to gen9 be similar as performance to a 40EUs haswell HD5000-5100 (without -e)? That's a lot of performance for minimum mainstream parts, abeit weaker than most GPUs it would be enough for non-gaming uses, or not?

If the current trend continues, and Gen9 GT2 will have 30 or 32EUs, that's already a 1.6x increase. And then 2 major architectural updates. Gen9 is going to be a lot faster than the 25% you need for 32EUs to be equivalent to the 40 Gen7 EUs you predict.

 

[SlickR12345]

Lets be real here, Intel's new Igpu are not going to be good for gaming, like you can probably game on 1024x resolutions and a mix of medium and high settings on newer games, but if you want stable gaming at higher resolutions you need a discrete GPU, that said they will create competition for the lower end graphic cards and if AMD and Nvidia want to stay in the business there they will need to lower prices, which is great for consumers.

As I've said the past 2 generations of GPU's have been very expensive for pretty much minimal gains, once you could buy a 8800GT or GTX 460 1GB for $150 after few months of being on the market, after that the prices skyrocketed with a GTX 760 being a $250 card right now and that is if you get it on the cheap.

If you buy from local stores it usually goes for $280.

So the difference between GTX 460 1GB at $150-170 to GTX 760 at $250-280 is gigantic.

The GTX 760 should technically be from $150 to $180 depending on the amount of memory!

 

[SAAA]

This is also the reason I hope they don't focus too much on the IGP side, using 14nm scaling and increasing resources like Apple/IBM did with their cores should give a bit more than 10% on the CPU side, especially considered there are no 6 cores planned in mainstream parts.
Otherwise they have to follow Kaveri and AMD route of HSA, I really don't like having half of the die space unused and these GFlops should be useful for many parallel workloads.

 

[Homeles]

Whoops, double post.

 

[Homeles]

Lets be real here, Intel's new Igpu are not going to be good for gaming, like you can probably game on 1024x resolutions and a mix of medium and high settings on newer games, but if you want stable gaming at higher resolutions you need a discrete GPU, that said they will create competition for the lower end graphic cards and if AMD and Nvidia want to stay in the business there they will need to lower prices, which is great for consumers.

As I've said the past 2 generations of GPU's have been very expensive for pretty much minimal gains, once you could buy a 8800GT or GTX 460 1GB for $150 after few months of being on the market, after that the prices skyrocketed with a GTX 760 being a $250 card right now and that is if you get it on the cheap.

If you buy from local stores it usually goes for $280.

So the difference between GTX 460 1GB at $150-170 to GTX 760 at $250-280 is gigantic.

The GTX 760 should technically be from $150 to $180 depending on the amount of memory!

What AMD and Nvidia are charging for discrete GPUs isn't exactly relevant to integrated graphics. Also, in typical enthusiast fashion, you're acting like large form factor desktops are the only platform that exist.

Even HD 4000 can game well enough at 1366x768, which is a common laptop resolution. Allow 3 generations of growth (Ivy -> Skylake), at approximately 1.4x each generation... and we're at nearly 3x the performance of HD 4000. That's enough for 1920x1080. And 3x could very well be a conservative estimate.

About this: if the rumors are true wouldn't a 32EUs part with improvments from gen7.5 to gen9 be similar as performance to a 40EUs haswell HD5000-5100 (without -e)? That's a lot of performance for minimum mainstream parts, abeit weaker than most GPUs it would be enough for non-gaming uses, or not?

We know next to nothing about Gen9, and Skylake in general. Gen 8 is supposed to be the largest architectural overhaul Intel has had in nearly a decade. Details on future platforms are sorely lacking in general.

 

[blackened23]

"iGPU doesn't game well". Really, slick? Well compared to a desktop dGPU, no joke. Are you kidding me. Why you're even comparing it to desktop performance is beyond me. These iGPUs were originally created for ultrabook and portable form factors and the vast majority of those buying such machines - macbooks, ultrabooks, etc, most are not gamers. This mindset of 100% of every user being a gamer is absurd. The iGPU was not created for such, and for basic tasks it is just fine. The notion that everyone is only using a giant ATX case with an LGA desktop is also absurd. iGPU was not created for that form factor in the first place.

However, iGPU has been advancing fairly rapidly and the Iris Pro is actually very impressive considering it can fit inside of a macbook pro and actually provides good gaming performance. No, it isn't a dGPU. Thank you captain obvious for pointing that out.

This assumption that 100% of users are gamers (especially with macbooks/ultrabooks) when most of them are more than likely students or people that actually do real work on their portables, the gaming assumption is absurd. The "everyone is a desktop user" notion is also absurd. Obviously if you want 1080p gaming on a desktop you're buying a dGPU. Another shout out to captain obvious for this fact. But if you're of the intended market for iGPU - ultrabooks - if you want to game you can buy an ultrabook with a mobile dGPU. Pretty neat huh. However, iGPU has advanced very rapidly. If Iris Pro is any indication, Broadwell and Skylake will be milestones in terms of integrated graphics. Should be interesting to say the least. No they won't match a GTX 780ti. I'll just go ahead and once again point out the obvious. If the iGPU doesn't match your needs for your ultrabook, you have options. Get a mobile dGPU. The Kepler and Maxwell based mobile dGPUs will provide greater performance. If you want that, buy it. If you don't want iGPU, don't use it.

Regardless, there will be a day when iGPU will converge with mobile dGPU. Intel's iGPU development is on a far faster pace than AMD's with their mobile APUs. I am talking ultrabook form factors here, not desktop. Comparing HD4000 to Iris Pro, it is a huge jump - let's just put this in perspective here. Iris Pro is RIGHT NOW being used in ultrabooks. PORTABLE ultrabooks. The 7850k is a LGA DESKTOP CPU. If a mobile iGPU in a portable box like the Brix Pro or a macbook can match an LGA desktop APU designed for the desktop, what are the obvious conclusions about how well the next iGPU will perform? It will be even better. And it will be a mobile designed iGPU. Again, a mobile iGPU will not match the best dGPUs. Or even mid range dGPUs. That much is obvious. But there will be a day when 1080p gaming on a mobile ultrabook will be a thing. Even with high details. Not sure when, but the pace of jumps with iGPU are fairly massive IMO. But if that isn't enough for you, you have options. Mobile dGPU, playing on your desktop, etc etc etc.

http://wccftech.com/amd-kaveri-apu-a10-7850k-intel-i5-4570r-iris-pro-graphics-showdown/

 

[SlickR12345]

"iGPU doesn't game well". Really, slick? Well compared to a desktop dGPU, no joke. Are you kidding me. Why you're even comparing it to desktop performance is beyond me. These iGPUs were originally created for ultrabook and portable form factors and the vast majority of those buying such machines - macbooks, ultrabooks, etc, most are not gamers. This mindset of 100% of every user being a gamer is absurd. The iGPU was not created for such, and for basic tasks it is just fine. The notion that everyone is only using a giant ATX case with an LGA desktop is also absurd. iGPU was not created for that form factor in the first place.

However, iGPU has been advancing fairly rapidly and the Iris Pro is actually very impressive considering it can fit inside of a macbook pro and actually provides good gaming performance. No, it isn't a dGPU. Thank you captain obvious for pointing that out.

This assumption that 100% of users are gamers (especially with macbooks/ultrabooks) when most of them are more than likely students or people that actually do real work on their portables, the gaming assumption is absurd. The "everyone is a desktop user" notion is also absurd. Obviously if you want 1080p gaming on a desktop you're buying a dGPU. Another shout out to captain obvious for this fact. But if you're of the intended market for iGPU - ultrabooks - if you want to game you can buy an ultrabook with a mobile dGPU. Pretty neat huh. However, iGPU has advanced very rapidly. If Iris Pro is any indication, Broadwell and Skylake will be milestones in terms of integrated graphics. Should be interesting to say the least. No they won't match a GTX 780ti. I'll just go ahead and once again point out the obvious. If the iGPU doesn't match your needs for your ultrabook, you have options. Get a mobile dGPU. The Kepler and Maxwell based mobile dGPUs will provide greater performance. If you want that, buy it. If you don't want iGPU, don't use it.

Regardless, there will be a day when iGPU will converge with mobile dGPU. Intel's iGPU development is on a far faster pace than AMD's with their mobile APUs. I am talking ultrabook form factors here, not desktop. Comparing HD4000 to Iris Pro, it is a huge jump - let's just put this in perspective here. Iris Pro is RIGHT NOW being used in ultrabooks. PORTABLE ultrabooks. The 7850k is a LGA DESKTOP CPU. If a mobile iGPU in a portable box like the Brix Pro or a macbook can match an LGA desktop APU designed for the desktop, what are the obvious conclusions about how well the next iGPU will perform? It will be even better. And it will be a mobile designed iGPU. Again, a mobile iGPU will not match the best dGPUs. Or even mid range dGPUs. That much is obvious. But there will be a day when 1080p gaming on a mobile ultrabook will be a thing. Even with high details. Not sure when, but the pace of jumps with iGPU are fairly massive IMO. But if that isn't enough for you, you have options. Mobile dGPU, playing on your desktop, etc etc etc.

http://wccftech.com/amd-kaveri-apu-a10-7850k-intel-i5-4570r-iris-pro-graphics-showdown/

Everything you wrote is wrong.

If igpu's don't compete with descrete gpu's then they don't take ANY marketshare, so nvidia/amd don't care, that is obviously not true even for someone living under rock.

igpu's can't perform though, they are bad, they are good for 1024x at medium and high mix of settings on older games. In 2 years time games will use even more resources and be way harder on graphics, you can't compare a game that released in 2012 on how its going to perform on an igpu that comes out in 2016.

 

[Enigmoid]

Everything you wrote is wrong.

If igpu's don't compete with descrete gpu's then they don't take ANY marketshare, so nvidia/amd don't care, that is obviously not true even for someone living under rock.

igpu's can't perform though, they are bad, they are good for 1024x at medium and high mix of settings on older games. In 2 years time games will use even more resources and be way harder on graphics, you can't compare a game that released in 2012 on how its going to perform on an igpu that comes out in 2016.

He is not saying that they don't compete but rather that they are not designed to compete at the high, mid, or even upper low end of the dgpu market.

The igp will always be constrained far more than a standalone gpu, forced to share die area and thermals with the CPU and lacking goodies like bandwidth.

However, on the low power mobile market igps have the potential to take the market by storm. Less motherboard traces and no PCI-e means lower power as does sharing system RAM.

This progress argument has been used time and time again yet never have igp's ever been capable of the (relative) performance offered today by Haswell and Kaveri.

 

[Homeles]

However, on the low power mobile market igps have the potential to take the market by storm. Less motherboard traces and no PCI-e means lower power as does sharing system RAM.

Really, they already have. I must say that Nvidia's Maxwell definitely bought themselves more time, though, and hopefully AMD will have a competitive answer as well.

 

[Ajay]

The igp will always be constrained far more than a standalone gpu, forced to share die area and thermals with the CPU and lacking goodies like bandwidth.

Though dGPUs will get there first, eventually, CPUs will have HBM, which will really improve iGPU performance and kill off low end, and possibly mid-range dGPUs (depending on how the economics and timing works out). So long term, low end dGPUs will go into the dustbin of history. The million dollar question is when does this transition happen?

 

[Homeles]

Though dGPUs will get there first, eventually, CPUs will have HBM, which will really improve iGPU performance and kill off low end, and possibly mid-range dGPUs (depending on how the economics and timing works out). So long term, low end dGPUs will go into the dustbin of history. The million dollar question is when does this transition happen?

If IntelUser2000 is to believed, it'll be sometime after Skylake for Intel. Cannonlake might be a great introduction for it, especially if Intel continues on their trend of having graphics "tocks" coincide with their CPU "ticks."

I am fairly confident that AMD has been working hard to be the first to leverage stacked DRAM for their graphics products. Apparently, they were supposed to have it as early as their 28nm GPU Tahiti. That didn't happen of course, but they certainly aren't far away, and will undoubtedly have a product utilizing HBM out within a year.

I don't know when we'll see it on an APU, though. There's a lot of cost reduction that needs to occur before it can make it to that market without killing AMD's margins.

We'll definitely see it by the end of the decade, though. Perhaps as early as 2016.

 

[AtenRa]

Though dGPUs will get there first, eventually, CPUs will have HBM, which will really improve iGPU performance and kill off low end, and possibly mid-range dGPUs (depending on how the economics and timing works out). So long term, low end dGPUs will go into the dustbin of history. The million dollar question is when does this transition happen?

The answer is NEVER.

Say you get a Kaveri or IRIS PRO today, you have the fastest iGPU on the market. In 2 years you want to upgrade only the GPU. At the time, a low end ($100) GPU will be 2x times or more faster than your iGPU.

Also, $100 dGPUs will always provide more performance than any $100 APU. Even if Kaveri had GDDR-5, it would not even reach HD7750 level of performance because it would need to raise the iGPU frequency to 900-1000MHz. That would make the TDP to raise to more than 95W.

I believe you see why low end GPUs will always be available in the market. The only thing that will happen (maybe) is that AMD and NVIDIA would need to introduce lower end dGPUs more frequently and with top of the line manufacturing processes than what they are doing now.

One more thing, If AMD can make the hybrid CrossFire to work as it must, they could also benefit by selling both an APU and Lower End dGPU (like R7 240 or R7 250) for added performance and higher performance/price ratio than the competition.

 

[blackened23]

Everything you wrote is wrong.

If igpu's don't compete with descrete gpu's then they don't take ANY marketshare, so nvidia/amd don't care, that is obviously not true even for someone living under rock.

igpu's can't perform though, they are bad, they are good for 1024x at medium and high mix of settings on older games. In 2 years time games will use even more resources and be way harder on graphics, you can't compare a game that released in 2012 on how its going to perform on an igpu that comes out in 2016.

Maybe you should sharpen your reading comprehension skills. The iGPU is designed for ultrabooks and super small form factors (such as the Brix Pro), AIOs, and portable ultrabooks. They were never meant to replace desktop GPUs. At some point, iGPUs will converge will mobile dGPUs because the pace of progress with iGPU is far faster than that of mobile dGPUs. But if you feel you can throw a GTX 780ti into a Brix Pro within the same size constraints as the Iris Pro die size contained within, well okay if you say so. You're just oblivious to the bigger picture for which the iGPU is designed. It was designed at the suggestion of Apple for their mobile macbooks, that is why intel invested into the iGPU originally. And they've made significant strides with their mobile graphics performance, in fact they have the fastest mobile graphics performance at this time. The fact that Iris Pro as a mobile form factor is trading blows with a desktop level 7850k? Mobile chip versus desktop chip, roughly same performance, trading blows. What does that imply in terms of how well intel is progressing in graphics performance?

But if you live in the desktop only world where AIOs, SFF, ultrabooks, macbooks, and mac mini/brix pro devices don't exist, yeah, a dGPU is obviously better. Thanks for reminding us again, no one said otherwise. But good luck fitting your dGPU into a mac mini or Brix Pro inside of that size constraint. That's why intel designed the iGPU. They didn't design it for your 3 foot tall ATX case, even though some SKUs have it for free. It was designed for mobile at the suggestion of Apple, and apple was willing to pay the big bucks for integrated graphics that did not compromise CPU performance or performance per watt.

This is completely beside the point that most people buying macbooks and ultrabooks aren't gamers at all. It's neat that you view everything and anything through the lens of gaming. I assure you, the bigger market for mobile macbooks and ultrabooks aren't 100% gamers that expect to play crysis 3 at 1080p/60 fps. And they aren't 100% desktop users.

 

[AtenRa]

At some point, iGPUs will converge will mobile dGPUs because the pace of progress with iGPU is far faster than that of mobile dGPUs.

Not true, since both use the same manufacturing process and same architecture they are progressing at the same speed at best.

Edit: Also, any 90mm2+ 28nm dGPU is faster than 22nm IRIS PRO. Same will continue to hold true in future products.

 

[witeken]

Not true, since both use the same manufacturing process and same architecture they are progressing at the same speed at best.

dGPUs are 1+ process node behind Intel. If Intel catches up with ~Gen8, they can get more performance out of a given die area.

Edit: Also, any 90mm2+ 28nm dGPU is faster than 22nm IRIS PRO. Same will continue to hold true in future products.

...which is why this won't hold true in future products. Also, products like TK1 show that lower TDPs have higher efficiency.

BTW, he was talking about mobile GPUs.