Intel Skylake / Kaby Lake

[Sweepr]

Honestly an entry level Skylake-Y Core M convertible running Windows 10 should be way more interesting and productive than a 12'' iPad Pro running iOS with a stylus attached. If the 3DMark 11 result translates to other benchmarks its iGPU should be comfortably faster than Apple's A8X (there's rumours Apple won't refresh the iPad Air line this year while iPad Pro brings a new SoC). Looking at 3DMark 11 (+41%) it could come close to Geforce GT820M performance (in this particular benchmark), not bad at all for fanless devices that can play actual games instead of mobile junk.

An ASUS T300 Chi with Broadwell-Y Core M, 4GB RAM and 1080p IPS screen, 128GB SSD is currently available for $527 @ Amazon. Give me a Skylake-Y Core M + Windows 10 version of this for less than $600 and I'm in ASUS.

 

[TidusZ]

Not sure either. ;-)

But, maybe I'm nostalgic for those great 32nm Intel launches where the focus was on raw power. Desktop power! With a tiny little bullet point about the neutered laptop chips designed only for emergency use when a proper desktop wasn't available.

Sadly, it was during this golden era of overclocking mojo that Apple was building the walls around its poisoned garden of tedium. Soon the world would be embracing an era of personal mobile distraction where not-really-good-enough would be enough to launch a new era of clueless consumerism.

Amen. The only good thing about these new processors is I can save money not buying them

 

[Sweepr]

Linux 4.3 Will Have Many Intel Graphics Improvements, Better For Skylake

Daniel Vetter of Intel's Open-Source Technology Center has sent in many Intel DRM driver changes to be queued up in DRM-Next for the Linux 4.3 kernel.

This drm-intel-next load is quite big given that there's three batches of changes due to Vetter having held off on sending out this pull request for the code to land in DRM-Next.

Skylake is getting all cleaned up for Linux 4.3, which is good since the first of the Skylake processors are launching in early August... However, for stable Linux users, these Skylake changes may be too late as Linux 4.3 will not officially ship until late into the year.

With Linux 4.3, the Skylake support is no longer considered "preliminary" after an ABI issue with planes was corrected. There's also more workarounds for this "Gen9" graphics processor, MOCS programming support for Skylake and Broxton, improvements to the Skylake DPLL code, a fix for a hard-hang when trying to reset the GPU, and various other changes.

The Intel DRM driver for Linux 4.3 is also making greater use of the atomic mode-setting interfaces with more code being ported over, has various fixes for Valley View and Cherry View hardware, there's some frame-buffer compression improvements, improvements to front-buffer tracking, resource streaming for Mesa, and various other changes.

More details on the Intel DRM changes queued up for Linux 4.3 thus far can be found via this pull request. Stay tuned for our first Intel Skylake Linux testing next month as soon as the CPUs start rolling out.

www.phoronix.com/scan.php?page=news_item&px=Linux-4.3-Intel-DRM-Next

Ps: We get that Sandy Bridge, Ivy Bridge and Haswell are still great processors guys, there's no need to justify the fact that you're not upgrading by downplaying Skylake.

 

[Wreckem]

ASUS ROG Z170 Motherboards Lineup Unveiled

http://wccftech.com/asus-rog-z170-m...-extreme-maximus-viii-hero-maximus-viii-gene/

So people were asking where U.2 without m.2 to u.2 on Z170 is. Right there by itself only on the Asus ROG's most expensive board($350?). For that price one should go X99. The smart choice remains the boards wotht m.2 to u.2. So yah it's not going to be common place to not have to use an adapter until the next HEDT platform.

 

[PPB]

Linux 4.3 Will Have Many Intel Graphics Improvements, Better For Skylake



www.phoronix.com/scan.php?page=news_item&px=Linux-4.3-Intel-DRM-Next

Ps: We get that Sandy Bridge, Ivy Bridge and Haswell are still great processors guys, there's no need to justify the fact that you're not upgrading by downplaying Skylake.

Pointong facts hurt the blind sheep most likely it seems. People shelling 200 bucks for having an 5% perf increase each year can not be justified by reason. 200 bucks not spent in areas where they would be best spwnt such as gpus, even more storage or RAM, etc is feally bad. Only the most Time constrained, cpu performance dependant prpductivity software users would see a point in that perf to $ gain. But then those people roll by LGa2011 and xeons.

 

[Sweepr]

Pointong facts hurt the blind sheep most likely it seems.

Reported.

People shelling 200 bucks for having an 5% perf increase each year can not be justified by reason. 200 bucks not spent in areas where they would be best spwnt such as gpus, even more storage or RAM, etc is feally bad. Only the most Time constrained, cpu performance dependant prpductivity software users would see a point in that perf to $ gain. But then those people roll by LGa2011 and xeons.

2014 Core i7 4790K was 12% faster than 2013's Core i7 4770K, Core i7 6700K could very well bring another 10% on top of that this year, where does your doom and gloom 5% fit here?

Point is, Devil's Canyon users are not upgrading anyway, who's interested in Skylake is using Core i7 4770K or older IB/SB/Nehalem/Yorkfield chips, very few are 'spending $200 each year'.

Some people simply want Skylake to do bad to justify their recent upgrades and that's quite boring.

 

[PPB]

Reported.



2014 Core i7 4790K was 12% faster than 2013's Core i7 4770K, Core i7 6700K could very well bring another 10% on top of that this year, where does your doom and gloom 5% fit here?

Point is, Devil's Canyon users are not upgrading anyway, who's interested in Skylake is using Core i7 4770K or older IB/SB/Nehalem/Yorkfield chips, very few are 'spending $200 each year'.

Some people simply want Skylake to do bad to justify their recent upgrades and that's quite boring.

12% faster in clockspeed, which doesnt scale linearly to software. And the leaks are showing us marginal gains, even regressions, to a product 2 gens behind uarch wise. So 5% anually is pretty in line from SB foward.


Ah, to the 200 bucks add the mandatory aftermarket heatsink, and the ddr4 coat compared ddr3 this round. It is certainly hard on the pocket to be a intel fanboy these days.

 

[Sweepr]

I'm sure you will see plenty of designs using Skylake in the market soon enough

Crossing my fingers for a Retina MacBook Air...

Apple users are particularly interested in Thunderbolt 3.

Meanwhile gamers will be happy to hear that Intel is finally moving forward on external graphics via Thunderbolt, and after more than a few false starts, external GPUs now have the company’s blessing and support. While Thunderbolt has in theory always been able of supporting external graphics (it’s just a PCIe bus), the biggest hold-up has always been handling what to do about GPU hot-plugging and the so-called “surprise removal” scenario. Intel tells us that they have since solved that problem, and are now able to move forward with external graphics. The company is initially partnering with AMD on this endeavor – though nothing excludes NVIDIA in the long-run – with concepts being floated for both a full power external Thunderbolt card chassis, and a smaller “graphics dock” which contains a smaller, cooler (but still more powerful than an iGPU) mobile discrete GPU.

5K 60Hz displays, external dGPUs anyone?
Macbook Air and Macbook Pro deserve a proper Skylake refresh later this year.

 

[Sweepr]

12% faster in clockspeed, which doesnt scale linearly to software.

11.8% faster than Core i7 4770K according to (AMD's favourite) Hardware.fr:
www.hardware.fr/articles/924-19/indices-performance.html

And the leaks are showing us marginal gains, even regressions, to a product 2 gens behind uarch wise. So 5% anually is pretty in line from SB foward.

Depends on the leak and particular software you pick (just look take a look at the OP), better wait for actual reviews.
Your 5% is false and misleading, Core i7 4770K, Core i7 4790K and (likely) Core i7 6700K outclassed their predecessors by more than that.

Ah, to the 200 bucks add the mandatory aftermarket heatsink, and the ddr4 coat compared ddr3 this round. It is certainly hard on the pocket to be a intel fanboy these days.

Not really, you can always buy a discounted Haswell if the price difference is very significant. Both are better options than investing in AM3+ crap in 2015.

 

[frozentundra123456]

Pointong facts hurt the blind sheep most likely it seems. People shelling 200 bucks for having an 5% perf increase each year can not be justified by reason. 200 bucks not spent in areas where they would be best spwnt such as gpus, even more storage or RAM, etc is feally bad. Only the most Time constrained, cpu performance dependant prpductivity software users would see a point in that perf to $ gain. But then those people roll by LGa2011 and xeons.

People spend money going to sporting events, eating out, vacations, buying more expensive cars than they need, buying fancy houses, all kind of hobbies. If someone wants to spend 1000.00 on a new computer, what is that to you? And there are more than desktops to upgrade to as well. Obviously the performance gains are not worth it to you, which is fine. My computer is perfectly fine for my uses too, but I dont feel it is my place to tell others how to spend their money.

 

[IntelUser2000]

Again, Broadwell-Y Core M's performance varies greatly depending on the device.
ASUS T300 Chi for example scores 2.64 @ CB 11.5 64-bit running a Core M-5Y71.

Asus's own marketing materials said 2.3 CB 11.5 64-bit. I've seen benchmarks of the Chi, and while in CB it performs pretty well depending on the sites, if you look at in comparison to the "normal" U chips and other benchmarks, its in line with 2.3 points. Cinebench was a pretty good baseline benchmark for a while but with Core M it threw that away.

The fact that only a single device out of all of them performs "OK" is quite an evidence that you are playing lottery when making Core M systems. That said, kudos to Asus for making a dud not a dud.

 

[IntelUser2000]

Depends on the leak and particular software you pick (just look take a look at the OP), better wait for actual reviews.
Your 5% is false and misleading, Core i7 4770K, Core i7 4790K and (likely) Core i7 6700K outclassed their predecessors by more than that.

Based on reading AT and Toms benches, from 2600K to 3770K to 4770K, OC results were actually slightly worse on 3770K, it was similar with 4770K, and better on 4790K.

IPC: 2600K to 3770K, 5%
3770K to 4770K, 7-8%
4770K to 4790K, 11%

Early 2011, to mid 2014 is 3.5 years. Annual gain is about 7%. Notice too that with the 4790K, Intel factory overclocked it, while the maximum air cooled limit was barely higher than 4770K. With counting overclocking, the changes from 4770K to 4790K might be 5-6%. That would make annual gain 5.4%.

 

[Sweepr]

Asus's own marketing materials said 2.3 CB 11.5 64-bit.

NotebookCheck got 2.64 points. I consider ASUS T300 Chi and Apple's 12'' Macbook the best Broadwell-Y Core M implementations.

Based on reading AT and Toms benches, from 2600K to 3770K to 4770K, OC results were actually slightly worse on 3770K, it was similar with 4770K, and better on 4790K.

IPC: 2600K to 3770K, 5%
3770K to 4770K, 7-8%
4770K to 4790K, 11%

880 to 2600K -> 20.1%*
2600K to 3770K -> 8.8%*
3770K to 4770K -> 9.7%*
4770K to 4790K -> 11.8%*
Skylake-S 65W -> 11% better SPECint than 84W Haswell Refresh

*Hardware.fr average application performance.
I'm talking about stock performance. Sorry but changing the goalpost to include OCing won't turn the 5% a year (stock) lie to true.

 

[WaitingForNehalem]

So people were asking where U.2 without m.2 to u.2 on Z170 is. Right there by itself only on the Asus ROG's most expensive board($350?). For that price one should go X99. The smart choice remains the boards wotht m.2 to u.2. So yah it's not going to be common place to not have to use an adapter until the next HEDT platform.

Yeah, that is dumb. Looks like I will be getting the PCI-E version of the Intel 750 then.

 

[Dresdenboy]

"Surface 3 seems snappy enough to me"
- Seems? Really?

Don't mix UI performance with some heave calc algo performance. The iPhones' success was build on the efficient interweaving of GPU accelerated UI operations based on user input. It always seemed faster (more fluidly responding) than other phones with faster cores (before A6 and A7).

I suggested a 4790K (besides other stuff) to a video editing friend. And he is disappointed, as it's not faster (even incl. switching to SSDs, 16 GB etc.) than his older i5 with half the passmark score he looked at.

Video "editing" benchmarks just test final rendering or encoding, not the actual work.

 

[cytg111]

"Seems? Really?"

- Point was "Seems" is about as unscientific as it gets, it is really saying a whole lot of nothing. Precisely nothing.

 

[Eug]

I'd be shocked if the mythical iPad Pro got Core M. More likely it it would get an ARM A9X.

Core M will likely exist only in the MacBook Retina on the Apple side.

Again, Broadwell-Y Core M's performance varies greatly depending on the device.
ASUS T300 Chi for example scores 2.64 @ CB 11.5 64-bit running a Core M-5Y71. The same chip inside Lenovo ThinkPad Helix 2 scores 2.04.
Based on this it's hard to make a meaningful prediction about Skylake-Y's Core M but I'm hoping it delivers more consistent performance across different devices on top of 'up to 17% & 41% faster' (CPU & iGPU) performance bump. Considering it's the same proccess, looks like an impressive achievement.

The i5-4200U you mentioned earlier scores 2.5 in Cinebench 11.5.

Personally, I'm not so concerned about yearly performance improvements but cumulative improvements, as I keep my computers a long time.

I'm coming from a 13" 2.26 GHz Core 2 Duo MacBook Pro, which has Penryn P8400. That thing has held up well with an SSD in it, and supports the latest OS X El Capitan. Cinebench 11.5 is only around 1.35 though. Not sure how much Passmark is usable as a bench these days, but it gets 1475 compared to 3278 with i5-4200U and 3059 for Core M 5Y71.

I'm a little confused as to the specs though, since Apple's Core M MacBook Retina upgrade is for a 1.3 GHz - 2.9 GHz machine, whereas the 5Y71 is a 1.2 GHz - 2.9 GHz chip. Apple does have a 1.2 GHz SKU, but it only goes up to 2.6 GHz, which sounds like 5Y51.

Regardless, even though that MacBook has held up pretty well, Skylake Core M will be a MASSIVE upgrade for me, esp in the context of the MacBook Retina:

Core 2 Duo P8400 --> Core M Skylake = over twice the CPU performance
nVidia GeForce 9400M --> Skylake = dunno the comparative performance
OK non-Retina screen --> Awesome Retina screen
4.5 lbs --> 2.0 lbs.
USB 2 --> USB 3.x
802.11n --> 802.11ac
Thunderbolt (3)
Huge increase in battery life (partially because my MBP is so old now, but it never had great battery life to begin with).
Bluetooth 4.1
HEVC H.265 in hardware

And of course, fanless.

As alluded to in other posts, CPU performance isn't everything, since so much other stuff is important for the "feel" for performance, make things "seem" fast. All the stuff that Skylake brings to the table is impressive though. This is a big deal IMO. I'm glad I didn't bite with Broadwell.

As for importance of CPU performance for me though, it will actually matter. I am starting to notice the C2D machine slow down a bit even just surfing, so getting a 100%+ CPU performance increase is certainly welcome. More important to me though is the extended battery life and huge weight loss. I guess I'm a weakling, but these days 4.5 lbs just seems way too heavy to me.

 

[jpiniero]

Apple is almost certainly using the cTDP up mode on the Macbook.

 

[cytg111]

Also, its fine and all with the IGP focus, 40% improvement BDW->SKL, if leaks has merit, however tell me the last time you did a heavy duty graphics task that fitted into the "race to finish or throttle" paradigm? Never? Point is, these improvements are worth diddly squat if they are not sustainable. For the mobile part, IMO, a better move would be to regress performance but be able to sustain it.

 

[Sweepr]

I'd be shocked if the mythical iPad Pro got Core M. More likely it it would get an ARM A9X.

Core M will likely exist only in the MacBook Retina on the Apple side.

That's the most likely scenario.
Though it would be incredible if they released a Skylake-Y iPad Pro capable of dual-booting iOS and Mac OS X. This, coupled with their own stylus and an optional keyboard, Apple's own Surface Pro killer device.

I'm a little confused as to the specs though, since Apple's Core M MacBook Retina upgrade is for a 1.3 GHz - 2.9 GHz machine, whereas the 5Y71 is a 1.2 GHz - 2.9 GHz chip. Apple does have a 1.2 GHz SKU, but it only goes up to 2.6 GHz, which sounds like 5Y51.

Looks like a special Apple-only model, wouldn't be the first time (expect the same from Skylake-Y).

Regardless, even though that MacBook has held up pretty well, Skylake Core M will be a MASSIVE upgrade for me:

Core 2 Duo P8400 --> Core M Skylake = over twice the CPU performance
nVidia GeForce 9400M --> Skylake = dunno the comparative performance
OK non-Retina screen --> Awesome Retina screen
4.5 lbs --> 2.0 lbs.
USB 2 --> USB 3.x
802.11n --> 802.11ac
Thunderbolt (3)
Huge increase in battery life (partially because my MBP is so old now, but it never had great battery life to begin with).
Bluetooth 4.1
HEVC H.265 in hardware

And of course, fanless.

Indeed, great upgrade. Not only the performance uplift from Skylake-Y but also much lighter, more portable, extra connectivity, etc. And, like most Apple devices, expect second generation Macbook 12'' to carry an important list of improvements.

In 3DMark 11 the new Gen 9 GT2 iGPU could perform close to some low-end discrete dGPUs like the Geforce GT820M, which is a lot faster than your current Geforce 9400M.

 

[Eug]

That's the most likely scenario.
Though it would be incredible if they released a Skylake-Y iPad Pro capable of dual-booting iOS and Mac OS X. This, coupled with their own stylus and an optional keyboard, Apple's own Surface Pro killer device.

I've wondered about this same thing, but I'm not sure I see the point. Personally, I'm not a big fan of dual-booting and neither are average consumers.

Are they really going to do multiple architecture thing yet again?

Also, its fine and all with the IGP focus, 40% improvement BDW->SKL, if leaks has merit, however tell me the last time you did a heavy duty graphics task that fitted into the "race to finish or throttle" paradigm? Never? Point is, these improvements are worth diddly squat if they are not sustainable. For the mobile part, IMO, a better move would be to regress performance but be able to sustain it.

How much stress does Retina with a bunch of windows (including on external displays) put on the GPU? Even if not all that much stress, could a faster part not handle lower workloads better without throttling?

Apple is almost certainly using the cTDP up mode on the Macbook.

I was trying to look that up, but didn't find it. What does that mean? Assume I don't know anything.

 

[StrangerGuy]

Yeah, that is dumb. Looks like I will be getting the PCI-E version of the Intel 750 then.

It's funny how OEMs still think in 2015 that some niche I/O standard will get any sort of traction if they never appear as a baseline feature. Remember eSATA?

 

[jpiniero]

I was trying to look that up, but didn't find it. What does that mean? Assume I don't know anything.

Intel says the 5Y71 runs at 1.4 base if you get the TDP up to 6 W instead of the normal "4.5 W". Maybe Apple is using some sort of intermediate mode so it ends up being 1.3 base.

 

[WaitingForNehalem]

It's funny how OEMs still think in 2015 that some niche I/O standard will get any sort of traction if they never appear as a baseline feature. Remember eSATA?

Yep. The M.2 form factor isn't really ideal for desktops and has overheating issues. SATA Express is limited to 2 PCI-Express lanes which isn't a huge deal but limits its potential. U.2 is clunky and nowhere to be found on boards under $350. That leaves good old PCI-Express as the only viable alternative at the moment.

 

[ShintaiDK]

Yep. The M.2 form factor isn't really ideal for desktops and has overheating issues. SATA Express is limited to 2 PCI-Express lanes which isn't a huge deal but limits its potential. U.2 is clunky and nowhere to be found on boards under $350. That leaves good old PCI-Express as the only viable alternative at the moment.

Is there any other M.2 SSDs than Samsungs x4 that overheat?