Will Skylake be a big leap forward?

sethk

2[H]4U
Joined
May 3, 2005
Messages
2,133
This may sound like a strange question, but ever since Intel started focusing on mobile first, I've been underwhelmed by their progress, and each generation seems only very incrementally (~10%) better than the last, even if it's been a relative while between platform launches. Here are my questions:

  • On mobile, will skylake enable an iPad like experience? Even being generous, will it allow a ~10-11" form factor with ~1.5lbs or less and an "all day" battery life (i.e. > 8 hours in regular use)?
  • On desktop, will we see a 25% improvement over haswell*dollar for dollar? i.e. at the usual $240, $340 price points for the K series i5 and i7 processors, will the skylake equivalents be 25% faster or better?

With no real competition from AMD, I think Intel has become complacent. Eventually the ARM licensees will eat their lunch, or so it seems, but I hope they have some surprises in store, it feels like a long time since they've been able to impress me as much as they did in years past.
*Not even counting Broadwell
 
10% is not much to someone that upgrades their processor every generation, but for most of Intel's customers (enterprise and non-enthusiast consumers) that typically wait 3-5 years between systems, they are getting quite an increase in performance.

Current system: 100 (baseline)
Next gen: 110 (10% increase each gen)
Next gen: 121
Next gen: 133
Next gen: 146
 
It's not much though when you think back to what used to happen in a year. Pentium 133 one year..Pentium 200 the next. 50% performance boost via clock increases alone. Similar with Pentium II 300 to Pentium II 450. It wasn't every year that there was a 50% boost, but it was a lot of them.

And then 1998-2002 was crazy as the competition between AMD and Intel became heated (as did processors, this is when TDPs really went up). I got a K6-2 300 in 1998 and in late 2001 or early 2002 I was rocking an Athlon XP 1800+ (1.53 Ghz). 5x the Mhz in 4 years and more IPC as well as I recall. And the battle between AMD and Intel kept going well into the dual core era.
 
With no real competition from AMD, I think Intel has become complacent. Eventually the ARM licensees will eat their lunch, or so it seems, but I hope they have some surprises in store, it feels like a long time since they've been able to impress me as much as they did in years past.
*Not even counting Broadwell

I think Intel still has an incentive to make faster chips. People won't replace unless there is a reason to. If there's nothing worth upgrading from my 2500k to then I'll just keep running it. This = no $$$ for Intel.
 
The IPC bump will be nice, but I don't expect anything earth-shattering...

The real story will be the Iris Pro graphics and gaming... For low end systems, iGPUs should be able to cover all the basics and then some, running low end 'esports' games fine... For high end gaming systems, with DX12 it should have a HUGE impact on future games being able to use it for post-processing...

Most people aren't anticipating the Iris Pros impact on DX12. Big :)
 
I think Intel still has an incentive to make faster chips. People won't replace unless there is a reason to. If there's nothing worth upgrading from my 2500k to then I'll just keep running it. This = no $$$ for Intel.

I agree. I recently had problems with MoBo and considered upgrading to Skylake (which would cost €600+ for a CPU, DDR4 RAM and MoBo), but since performane difference isnt that great, I just bought a "new" MoBo for €60. I dont see value, in upgrading from one 4core CPU, to another 4core CPU.
 
waiting for Zen to upgrade myself.
Intel havent invented enough,
 
I've been on the 60 series platform for 4.5 years now. I'm ready for an upgrade even if it's not a huge boost. My stuff will eventually wear out so I might as well replace it now while I can still resell the old stuff for a good price.
 
At 1440p (with a GTX970) my 2500k@4G is still holding it's own. The CPU market for gamers hasn't really moved much, if at all.
 
The 6700K is the only processor that really has me interested to upgrade from my 3570K. ~20% higher stock clocks and a slightly larger cache. Not sure if I'm $400-500+ interested though to get a new proc and board. The old setup would replace the board in my MAME cabinet which has an old Athlon X2 which would then become the leftover parts and wouldn't have much resell value.
 
I'm really curious about the benchmarks, like I am every generation. My X58 Xeon rig still feels quick, and I haven't seen enough of a jump in multithreaded performance to justify the expense. Chipset improvements for newer SSD technology, and UEFI for faster boot would be nice, but once I'm in Windows is there even a noticeable difference?
 
I WANT MOAR CORES, MOAR CORES DAMNIT :D unlocked ones.

Although for now I'll take a 10ghz 5960x vs 2x 5960x
 
waiting for Zen to upgrade myself.

At this pace, with AMD bleeding money and disappointing like never before (AMD hasn't been this bad in terms of positioning in their long, money-bleeding history), there is likely never going to be Zen. Chances are, Lisa Su got in, saw how much the PC industry was stagnant, and how much the SD (Smart Device) industry was growing like crazy and thought "What the hell are we doing?"
 
I think Intel still has an incentive to make faster chips. People won't replace unless there is a reason to. If there's nothing worth upgrading from my 2500k to then I'll just keep running it. This = no $$$ for Intel.
The thing is though that there a lot more people out there that are running older setups than yours. So while you yourself may not need to upgrade, the five other people with older Intel or AMD setups are going to need to upgrade. In other words, in terms of population, there's always going to be more people who need to upgrade than those who don't need to upgrade. Not to mention that trying to aim for massive improvements every release means greater risk and a higher chance of eventually hitting a wall in just how much performance they can squeeze out of a platform. Which means more R&D money and time spent to develop a whole new platform. Which is of course relatively risky. Then there's the issue of current software not exactly being programmed well to take full advantage of the performance we're getting now.

Think of the car market: People tend to keep their cars for 5+ years yet car manufacturers are still pumping out millions of new cars every year.
 
At this pace, with AMD bleeding money and disappointing like never before (AMD hasn't been this bad in terms of positioning in their long, money-bleeding history), there is likely never going to be Zen. Chances are, Lisa Su got in, saw how much the PC industry was stagnant, and how much the SD (Smart Device) industry was growing like crazy and thought "What the hell are we doing?"

They need Zen in a huge way. Their only saving grace the past few gens was the server market and fell behind that too. Zen right now would put them in competition with Intel and they could very easily squeeze in some design wins again.

Even if they do shift away from the desktop market they should still aim for the server/pro markets, too lucrative, especially right now with the big data craze.
 
Does anyone know how much total system memory the consumer Skylakes will support? 64GB? 128GB?
 
10% is not much to someone that upgrades their processor every generation, but for most of Intel's customers (enterprise and non-enthusiast consumers) that typically wait 3-5 years between systems, they are getting quite an increase in performance.

Current system: 100 (baseline)
Next gen: 110 (10% increase each gen)
Next gen: 121
Next gen: 133
Next gen: 146


Then, you bring in overclocking. If my 2600K can do 4.4 easily, but the new one can do 'only' 4.0 (not scientific numbers), then those gains are lost. So, for the enthusiast crowd, we need a better processor that can overclock well.

For business work? Every 3 years we upgrade. The difference between 3 years ago and now - CPU's aren't that much better. It's the move from standard HDD to SSD that brought most of the speed increases.
 
I think you will see a 7 to 10% bump in IPC and MAYBE a tiny bump in OC clocks. Basically the same progression we have seen out of Intel for a while now.
 
I really hope that skylake will OC as high as my 2500k. I want a new build, I want to go smaller (mITX) and I want more RAM. usb 3.1, PCIe 3.0 and m.2, wireless ac
 
Then, you bring in overclocking. If my 2600K can do 4.4 easily, but the new one can do 'only' 4.0 (not scientific numbers), then those gains are lost. So, for the enthusiast crowd, we need a better processor that can overclock well.

For business work? Every 3 years we upgrade. The difference between 3 years ago and now - CPU's aren't that much better. It's the move from standard HDD to SSD that brought most of the speed increases.

The same is true for the average consumer. The real sloths in the industry are software developers. How many programs that businesses and consumers use every day are still single-threaded only, manage memory poorly, have bloated superfluous code from ten years ago, and would be bogging PCs down even worse if it weren't for SSDs masking the complete lack of performance improvement?
 
The IPC bump will be nice, but I don't expect anything earth-shattering...

The real story will be the Iris Pro graphics and gaming... For low end systems, iGPUs should be able to cover all the basics and then some, running low end 'esports' games fine... For high end gaming systems, with DX12 it should have a HUGE impact on future games being able to use it for post-processing...

Most people aren't anticipating the Iris Pros impact on DX12. Big :)

I haven't really read anything about this.. I (like many other enthusiasts) have been writing off the iGPU performance since with a discrete card it wouldn't mean anything. If I read your post correctly, the iGPU will still be used even with a discrete card ("high end gaming systems")?

Could you provide info/sauce?
 
There's been articles all around the web about it. The big benefit will be to laptop and mobile systems though desktops get some perks.

The iGPU can handle simple gaming and desktop stuff with no real issues. It also has the advantage of being quite efficient on power use compared to discrete GPUs. For mobile with discrete GPU, they can include an onboard and improve battery life. For desktops, they can allow you to effectively powerdown the GPU and use the iGPU until the GPU horsepower is actually needed.

DX12 access to "metal" also in general can allow the iGPU to be used for certain operations as opposed to sitting idle.
 
I'm thinking it is finally time to upgrade my i7 920 @ 4ghz, mostly because I'm just bored. Hoping that Skylake will be a worthwhile investment.
 
DX12 access to "metal" also in general can allow the iGPU to be used for certain operations as opposed to sitting idle.

Don't bet on that, that would be very complicated to program for, at least in a gaming situation.
 
Then, you bring in overclocking. If my 2600K can do 4.4 easily, but the new one can do 'only' 4.0 (not scientific numbers), then those gains are lost. So, for the enthusiast crowd, we need a better processor that can overclock well.

For business work? Every 3 years we upgrade. The difference between 3 years ago and now - CPU's aren't that much better. It's the move from standard HDD to SSD that brought most of the speed increases.

That would be correct...to an extent.

An i5-2500K, for example, would be 33% faster when OC'd to 4.4 GHz vs stock clocks.

An i5-4670K/4690K OC'd to 4.0 GHz would be 47% faster than a stock 2500K.

For those that lost the CPU lottery and got stuck with a subpar example of a Haswell that can only OC to 4.0 GHz are still getting about the same performance of the segment equivalent SB that can OC to 4.8 GHz. Any Haswell that can OC higher than 4.3 GHz provides an IPC win vs just about any SB, if even only by a little.

2500K
IPC (3.3 GHZ) = 100 (baseline)
4.0 GHz = 121
4.4 GHz = 133
4.8 GHz = 145
5.0 GHz = 152
5.2 GHz = 158
5.4 GHz = 164

3570K
IPC (clocked at same 3.3 GHz) = 110 (10% increase over SB)
4.0 GHz = 133
4.4 GHz = 147

4670K/4690K
IPC (clocked at same 3.3 GHz) = 121 (10% increase over IB)
4.0 GHz = 147
4.2 GHz = 154
4.3 GHz = 158
4.4 GHz = 161
4.5 GHz = 165
4.6 GHz = 169
4.7 GHz = 172
4.8 GHz = 176

6600K
IPC (clocked at same 3.3 GHz) = 133 (10% increase over HWL)
4.0 GHz = 161
4.5 GHz = 181
5.0 GHz = 202
 
With no real competition from AMD, I think Intel has become complacent. Eventually the ARM licensees will eat their lunch, or so it seems, but I hope they have some surprises in store, it feels like a long time since they've been able to impress me as much as they did in years past.

Competition with ARM and AMD stepping aside is why Intel has been all-in on mobile chipsets for the last half a decade. Things have moved so fast at the bottom end that the Atom X7 is now close to eating the Core M's lunch at a fraction of the price. One more tick-tock and they'll be indistinguishable unless something radical happens. ARM has lit a fire under Intel in a way AMD could never dream of.

Desktop performance is quickly becoming more a function of GPU than CPU, and Intel decided long ago to let Nvidia and AMD battle that space out. There's no space left for them to make the sort of impact on desktops and servers that they can have on mobile. If you're expecting to "be impressed," you have your eyes on the wrong prize.
 
I haven't really read anything about this.. I (like many other enthusiasts) have been writing off the iGPU performance since with a discrete card it wouldn't mean anything. If I read your post correctly, the iGPU will still be used even with a discrete card ("high end gaming systems")?

Could you provide info/sauce?

DirectX 12... Multi Adapter Technology...

http://wccftech.com/directx-12-mult...-coherently-demo-shows-big-performance-gains/

http://wccftech.com/amd-sheds-more-light-on-explicit-multiadapter-in-directx-12-in-new-slides/

iGPUs will be able to do post-processing effects like lighting, AA, and more... Games built in DirectX 12/Vulkan should be natively built to utilize this, I assume.

Broadwell's Iris Pro apparently destroys AMDs APU and Intel HD Graphics, and Skylake is going to have a more powerful version of that.

Current gen games may not notice a huge difference, but anything next-gen/DX12 should see a huge boost... and the consoles will ensure that happens quick (engine makers are already developing and game makers will want the best looking and fastest stuff as always).
 
Last edited:
With no real competition from AMD, I think Intel has become complacent. Eventually the ARM licensees will eat their lunch, or so it seems, but I hope they have some surprises in store, it feels like a long time since they've been able to impress me as much as they did in years past.
Intel has been improving IPC for legacy software each generation pretty steadily. x86 ISA extensions are becoming more important, as are newer compilers to optimize for newer processors. If spending more on R&D than most of the rest of the industry combined is "complacent", there's not really much to say. :p That's just not reasonable expectations of what can be done on tick-tock cycles.

ARM has strengths and weaknesses. For handheld and most small tablets, it's hard to beat the low cost, low power and diversity of ARM SoCs available for every nook. Intel is having a hard time in those traditional ARM strongholds, regardless of how much Atom has improved, lowered cost or its suitability for many of those devices. At the same time, ARM is finding it near impossible to get into desktops, laptops and servers to any significant degree, strongholds for x86. In both cases, even in some stretched definition of equal or good enough, that's not good enough to supplant an established ecosystem.
 
DirectX 12... Multi Adapter Technology...

http://wccftech.com/directx-12-mult...-coherently-demo-shows-big-performance-gains/

http://wccftech.com/amd-sheds-more-light-on-explicit-multiadapter-in-directx-12-in-new-slides/

iGPUs will be able to do post-processing effects like lighting, AA, and more... Games built in DirectX 12/Vulkan should be natively built to utilize this, I assume.

Broadwell's Iris Pro apparently destroys AMDs APU and Intel HD Graphics, and Skylake is going to have a more powerful version of that.

Current gen games may not notice a huge difference, but anything next-gen/DX12 should see a huge boost... and the consoles will ensure that happens quick (engine makers are already developing and game makers will want the best looking and fastest stuff as always).

WOW. Thanks very much for linking these articles. All of a sudden it looks like the huge iGPU gains in Skylake are going to meaningfully impact game performance, even with a dGPU in the system. This is huge news! I wonder why people aren't talking about this more?
 
WOW. Thanks very much for linking these articles. All of a sudden it looks like the huge iGPU gains in Skylake are going to meaningfully impact game performance, even with a dGPU in the system. This is huge news! I wonder why people aren't talking about this more?

Well, it isn't a proven victory yet. Just like stacked VRAM for multi-GPU setups, only time will tell. I'm excited and axious to see results, but I'm not going to set myself up for disappointment in case it just doesn't bring anything meaningful.
 
I'm sitting on a Yorkfield Q9550 that proved to be a poor overclocker (or, quite likely, I just didn't know what I was doing coming from a Core 2). Do you suppose it's holding back my GTX 670?;)

Seriously, though, I have just gotten off the unemployment line and moved from the top of a short salary ladder to the bottom of a much taller one while still getting a raise. I'm feeling the itch but I don't know where to start. Haswell? Devil's Canyon? Wait for Skylake, since my games are well behind the curve? I recently finished Skyrim and I'm working through Fallout--NV without issues, but I added FarCry3 to my Steam library during a big sale and I'm pretty sure it will kill my rig if I don't upgrade before playing that.
 
If you primarily game where CPU isn't as critical as GPU, then I'd entertain the notion of a Devil's Canyon Haswell refresh if you want to jump now.

Wait for Skylake if you want the latest NVME support (PCIe 3.0 4x iirc), which the new Z100 series chipset will bring if I remeber right...along with whatever other new features. If IGP offloading for certain rendering functions in Vulkan and DX12 ring true, that may be another reason to wait for Skylake since it will have a stronger IGP.
 
I'm thinking it is finally time to upgrade my i7 920 @ 4ghz, mostly because I'm just bored. Hoping that Skylake will be a worthwhile investment.

I'm in the exact same boat as you. A little while back I considered dropping in a 6 core Westmere. I ended up changing my mind and used the money instead to buy an all in one liquid cooling solution, and just squeezing a higher OC out of my 920.

I'm hoping that the new socket for Skylake-e will be compatible with this liquid cooler and thus I'll be able to achieve a decent OC.
 
DirectX 12... Multi Adapter Technology...

http://wccftech.com/directx-12-mult...-coherently-demo-shows-big-performance-gains/

http://wccftech.com/amd-sheds-more-light-on-explicit-multiadapter-in-directx-12-in-new-slides/

iGPUs will be able to do post-processing effects like lighting, AA, and more... Games built in DirectX 12/Vulkan should be natively built to utilize this, I assume.

Broadwell's Iris Pro apparently destroys AMDs APU and Intel HD Graphics, and Skylake is going to have a more powerful version of that.

Current gen games may not notice a huge difference, but anything next-gen/DX12 should see a huge boost... and the consoles will ensure that happens quick (engine makers are already developing and game makers will want the best looking and fastest stuff as always).

This technology will increase FPS 10% at the cost of a good 50% extra processing lag, look at the example charts for when those frames actually get sent to the display. It's a terrible tradeoff.
 
I don't need more speed, I need more power efficiency, and less heat. Seriously... with my M.2, even my OS loads in less than 10 seconds from power on.... and all my games load in less than a minute, so there's no clean requirement for something 25% (or whatever) faster. There is an advantage to being able to passively cool my CPU though.
 
This technology will increase FPS 10% at the cost of a good 50% extra processing lag, look at the example charts for when those frames actually get sent to the display. It's a terrible tradeoff.

50% extra processing lag? Where do you get that from?

The FPS increase shown is only 10% and that is said to only utilize 70% of the iGPU. This is also in April/May, likely on a Haswell with Intel HD Graphics. Broadwell's Iris Pro is more than 2x the iGPU that Intel HD Graphics is, and Skylakes Iris Pro is even better than Broadwells... Imagine a potential 20%+ FPS gain with iGPU on/off (20% on a Titan-X build, not a 290/970 build, which would likely see larger gains)...

because of multi-adapter, I will probably go with Skylake over Haswell-E or Skylake-E for the next rig, as I think games may benefit more from the iGPU than from the extra cores... Better single thread on DX 9-11 games, and multi-adapter for DX12 games.

I just need a skylake ITX motherboard with U.2!
 
The IPC bump will be nice, but I don't expect anything earth-shattering...

The real story will be the Iris Pro graphics and gaming... For low end systems, iGPUs should be able to cover all the basics and then some, running low end 'esports' games fine... For high end gaming systems, with DX12 it should have a HUGE impact on future games being able to use it for post-processing...

Most people aren't anticipating the Iris Pros impact on DX12. Big :)

That's where I see the difference, too. The typical budget box with integrated graphics is about to take a giant leap forward in capability. Combined with small form factors, it blurs the lines between computers and consoles even further.

At 1440p (with a GTX970) my 2500k@4G is still holding it's own. The CPU market for gamers hasn't really moved much, if at all.

Then, you bring in overclocking. If my 2600K can do 4.4 easily, but the new one can do 'only' 4.0 (not scientific numbers), then those gains are lost. So, for the enthusiast crowd, we need a better processor that can overclock well.

For business work? Every 3 years we upgrade. The difference between 3 years ago and now - CPU's aren't that much better. It's the move from standard HDD to SSD that brought most of the speed increases.

That's exactly where I'm at. My 2500k, overclocked, is going nowhere. If I want to upgrade, there's way more bang for the buck to be had in upgrading the GPU. I'll see FPS gains and likely use less power.

And it's the same with my work laptops, too. The old probook had an i5-460m, and the new one (about a year old now) has an i5-3230m. As far as the CPU between the two, the difference is negligible. Getting an SSD made a huge difference, though. If not for that I could hardly tell the difference between the two.
 
AMD needs to get off their asses and starting bringing the heat to intel. I'm still on the 2500K but i really want to upgrade to something worthwhile, but with no real competition, intel has no reason to make anything astounding performance jump wise. mid-range intel cpus should have been at at least 6 cores standard already and enthusiast grade cpus should have been OCing to 5Ghz with no effort by now. Seems like power consumption has been the only thing that intel has been really chasing.

Seems like only GPU's and SSD's have brought any significant performance gains in the last 6 years.

(This was not meant to be some childish AMD bash, intel has gotten lazy as shit as well)

/rant :)
 
That i7-6700T looks pretty nice. At 35W I could remove my H100i from the loop and just air cool my system.

Or not, not socket compatible.
 
Last edited:
Back
Top