Examining Intel's Ice Lake Processors: Taking a Bite of the Sunny Cove Microarchitecture
by Dr. Ian Cutress on July 30, 2019 9:30 AM EST- Posted in
- CPUs
- Intel
- 10nm
- Microarchitecture
- Ice Lake
- Project Athena
- Sunny Cove
- Gen11
Using Power More Efficiently: Dynamic Tuning 2.0
A common thread in modern microprocessor design is being able to use the power budget available. There have been many articles devoted to how to define power budgets, thermal budgets, and what the mysterious ‘TDP’ (thermal design power) actually means in relation to power consumption. Intel broadly uses TDP and power consumption simultaneously, along with a few other values, such as power limits 1 and 2 (PL1 and PL2), which apply to sustained power draw and peak power draw respectively. Most Intel processors up until this point will allow a processor to turbo, up to a peak power draw of PL2 for a fixed time, before enforcing a PL1 sustained power draw. This is all very OEM dependent as well. However, for Ice Lake, this changes a bit.
For Ice Lake, Intel has a new feature called Dynamic Tuning 2.0, which implements a layer of machine learning on top of the standard turbo mode. The idea behind DT2.0 is that the processor can predict the type of workload that is incoming, say transcode, and adjust the power budget intelligently to give a longer turbo experience.
Technically the concepts of PL1 and PL2 don’t magically disappear under this new regime – the processor ends up going below max turbo because the algorithm predicts that the user won’t need it, and this saves up ‘power budget’ in order to enable the turbo to work for longer.
This is a topic that Intel will hopefully go into more detail. We do know that it requires collaboration at the OS level, but how these algorithms are trained would be a useful trove of information. It is unclear whether Intel will allow this feature to be enabled/disabled at the user level, for testing purposes, but it should be noted that unless it is by default ‘on’ for OEM systems, we might end up with some systems enabling it while others do not.
Facebook
Twitter
RSS
107 Comments
View All Comments
notashill - Tuesday, July 30, 2019 - link
Are those RAM/SSD targets really "greater than" 8GB/256GB or is it supposed to be "greater than or equal to"?Either way I would love to see an end to companies having >$1000 machines with pathetically low RAM/storage and then charging 500%+ markups to upgrade them to something decent. Like Microsoft's $1200 to go from 4/128 to 16/512.
mkozakewich - Wednesday, July 31, 2019 - link
I can't believe Microsoft has been using 4 GB as their base amount for the last six years. At some point it becomes insulting.repoman27 - Tuesday, July 30, 2019 - link
“Intel uses the ‘U’ designation for anything from 9W to 28W TDP, which covers all the mobile ultra-premium devices.”No they don’t. 9W are Y Series, 15 and 28W are U Series. This is all clearly stated in Intel’s publicly available product briefs for 10th Generation Core processors.
HStewart - Tuesday, July 30, 2019 - link
I be curious for more information on the Y processors - what the performance difference between Y and U. But it looks like these Ice Lake chips are designed for Ultraportable machines and not designed to replace to higher end ones - even like my Dell XPS 15 2in1 - I am really curious about that replacement - it's GPU is probably short lived possibly in updated higher voltage Ice Lake with Gen 11 graphics or new version with Xe graphics. I also have a Dell XPS 13 2in1 with Y processor - I am actually bringing it to meeting today - it is lightweight and does not really need that much power for what I using it for. I think it will be very interesting to compare this new XPS 13 2in1 and the existing XPS 15 2in1 - yes 15 2in1 has faster processor - but it not Ice Lake and that could make a huge difference.Hixbot - Tuesday, July 30, 2019 - link
4.2% annual IPC growth doesn't sound great but it is better than anything we've seen since SandyBridge.Targon - Tuesday, July 30, 2019 - link
And that should make people question the claims about performance increases. Mind you, how much performance has been lost on Intel chips due to the security issues? Intel may be comparing theoretical performance increases, without disclosing the fact that first through 9th generation have actually lost performance since launch due to security mitigations.So, +18% IPC, but -20% due to security mitigations for issues that are still there. Has Intel actually fixed the problems with the memory controller and other problems that allow for Meltdown and the other problems, rather than mitigating around the problem? If a problem has existed since first generation Core processors that turns out to be a design flaw, that also shows that the fundamental core design hasn't changed all THAT much.
rahvin - Wednesday, July 31, 2019 - link
Meltdown and some of the first spectre mitigations are going to be fixed in the hardware. Later spectre variants are probably only fixed in microcode and software.Where that line is drawn is going to be determined by when they froze the physical design for tapeout.
eek2121 - Wednesday, July 31, 2019 - link
I'm not knocking Intel on the IPC growth. If they had an 18% increase, great for them! However, mobile Intel CPUs of any variant (U, HK, Y, etc.) are much slower than their desktop counterparts. My Core i7 2600k absolutely destroys the 6700HK in my laptop. Laptops in general are designed to be low power devices, so performance is never the number one factor in designing a laptop, even on the high end. The only exception to this is the so called 'desktop replacements' that weigh a ton, have desktop class hardware, and basically need to be plugged in to not have the battery die after an hour.That's also the reason I take this announcement with a grain of salt. 18% on mobile is one thing. 18% on the desktop is something else. As I've mentioned to people here and elsewhere, the smaller the process, the harder it is to maintain high clock speeds. Also, from reading certain documentation, it seems that part of that 18% is counting the addition of AVX-512. I could be mistaken though.
erple2 - Wednesday, July 31, 2019 - link
Wow, really? That has not been my experience at all. My 6700hq has generally been (usually significantly) better performing than my 2600k for the vast majority of tasks I've thrown at it.jospoortvliet - Monday, August 5, 2019 - link
Any task that requires sustained compute will of course suffer on thr lower power budget on mobile. But tasks which require short bursts of activity will do better thanks to vastly improved turbo since the 2600k. So depending on what you do your impression might very well be accurate.