Intel’s 6th-gen Core processors are set to launch later this year, and while the new “Skylake” processors will be 14nm chips just like the current 5th-gen “Broadwell” processors, Intel is introducing a new microarchitecture which should offer reduced power consumption, better performance, and a big boost in graphics performance.
In fact, Skylake chips could offer up to 50 percent better gaming graphics than Broadwell chips.
According to a series of leaked slides published by Fanless Tech, here’s a bit of what we can expect:
- 10 to 20 percent better performance-per-watt for single-threaded CPU tasks
- Up to 50 percent better Intel HD graphics
- Support for hardware-accelerated HEVC encoding
- Integrated support for 4K cameras or up to 4 HD cameras
- Up to 30 percent longer battery life (If a device got up to 8.5 hours of run time with a Broadwell chip, the same hardware should last for over 11 hours with a Skylake chip).
Intel says the new chips will feature enhanced audio processing technology, improved touch sensor technology, support for Cat6 LTE, WiDi 6.0, Bluetooth 4.1, eMMC 5.0, SDXC 3.0, and more.
Not every chip in the Skylake lineup is created equal. THer will be S-series desktop chips, H-series processors for powerful laptops, U-series chips for lower-power models, and Y-Series “Core M” processors for low-power, possibly fanless notebooks, tablets, and other devices.
The Intel Core M Skylake chips should be quite a bit more powerful than their Core M Broadwell predecessors. Intel is promising up to 17 percent more CPU power, up to 41 percent faster graphics, and up to 1.4 hours of additional battery life (when playing full HD video on a system with a 35 Whr battery).
You can find more details at Fanless Tech.
Guess we have to wait till Zen comes out with their HBM FX APU’s.
These are good for games that were out before 2010.
Still mid to high end dGPU from 2010 are still better than today’s iGPU/APU for gaming any day.
You sure this is legit? Broadwell is spelt wrong and I’m pretty sure it supports PCI-E 3 unless they downgraded over Haswell.
If it was from some two bit Chinese chip manufacturer you’d kinda expect errors but Intel don’t usually make these kind of mistakes.
the last L just got moved to the next line. Intel’s presentation decks from the past have always been pretty shoddy.
The PCIE Support is likely referring to the U and Y series… Things like eMMC support are only for the more mobile product range that you’d see put into some tablets, etc. and relies more on the native support of the platform rather than what the motherboard chipset can support….
Basically, the minimum supported specs are now apparently better…
Intel cannot tick tock for now but there still is room for improvement 😉
Well, the tick tock cycle basically just described the two step advancement between every FAB advancement and every architectural advancement… So, basically, they can’t tick tock but they can still tock tock :-p
Like Broadwell advanced the Intel GPU to Gen 8 and now Skylake advances it again to a Gen 9 GPU… There just won’t be a FAB, tick, for while… but things like 3D transistor designs can still let them pack more in the same space, etc. as well as improving existing designs to be more efficient…
Fingers crossed for 100% HEVC hardware decoding.
Check the second slide…
It doesnt really say anything. It suggests that Broadwell and Skylake both have “HW accelerated HEVC decode”, it doesnt distinguish them. It kindof implies that they have the same thing.
No, it shows that it adds Hardware Accelerated Encoding and that suggests they significant improved support as now it can support full video editing in HEVC and not just playing…
It also states Faster and Ultra HD quality media transcode, which further suggests they improved support…
I don’t know. I’m interpreting that differently than you are.
When they say they add encoding support, I don’t think that has anything to do with decoding.
Transcoding doesnt really sounds like it has anything to do with playback, or decoding.
Perhaps, though to provide transcoding support for even lowest end variants it usually means they improved efficiency… and mind this also involves a more advance version of their Quick Sync, which usually means all hardware acceleration performance is improved at least a little bit…
But assuming the improvements may still be marginal then just wait until Kaby Lake comes out in the later half of 2016 with a Gen 10 GPU… It definitely already lists full HEVC 10-bit support, [email protected], etc. and by then we can be more sure of the driver support… While skipping over Skylake should ensure an even bigger upgrade benefit when you do upgrade…
Along with other factors like DX12 support may actually start to be relevant by then, more actual 4K content to watch, etc…
Man, if the whole 50% better GPU is true, AMD APU are screw. The only advantage they have is gone.
Well, yes and no… as AMD’s APU push extra things like GPU processing, etc. and not just traditional CPU and GPU performance but otherwise, yes…
It took two rapid GPU advances, first with Broadwell to a Gen 8 GPU and then again with Skylake with a Gen 9 GPU but it looks like Intel has finally turned things around for their iGPU…
Though, it remains to be seen if they can provide proper driver support… which is the other thing they had to really improve upon… and whether they can maintain the lead once they have it… Mind that this has yet to be compared to AMD’s releases that are due out later next year, such as the Zen, etc…
I can agree that Intel has turned things around in terms of performance, but Intel still needs to turn things around in price. When the eDRAM iGPUs have a $100 price premium over the non eDRAM iGPUs that is a problem.
Broadwell GT3: 47-48 execution units, up to 128 MB eDRAM
Skylake GT3: 48 execution units with or without 64 MB eDRAM
Skylake GT4e: 72 execution units with 64 MB or 128 MB eDRAM
72 is 50% more than 48… so you will only get a 50% speed increase if you buy the most expensive Skylake iGPU and compare that to Broadwell GT3 iGPUs.
True, but no one should assume “up to” means every single model will be in that range as there is always other variables like power range, emphasis on power efficiency versus performance variables, price range, etc. and Intel makes multiple tier products with different optimizations…
Also, up to 50% doesn’t mean every type of performance variable is equal… Just the max that could be provided under best scores for any particular benchmark… So we also still have to see what are the strengths and weaknesses of the new architecture and compare that to the old one too…
Anyway, here’s another slide giving more detail on those lower tier offerings…
Mind, this isn’t just a GPU scaling like they did with Ivy Bridge to Haswell but an actual architectural advancement as they go from a Gen 8 (Broadwell) to a Gen 9 (Skylake) GPU… So it’s not just a matter of how many EU’s are configured…
Well if a Skylake chip with 16 EUs can significantly outperform a Broadwell chip with 16 EUs, same with 48 and so on, then those are the benchmarks I’ll be looking for. And maybe I’ll try to buy a laptop with the full GT4e 72 EUs.
Comments are closed.