Intel published a developer guide confirming details of its upcoming Alder Lake processors.
Desktop "Alder Lake-S" processors will include up to 8 "Golden Cove" performance cores (P-cores), 8 "Gracemont" (Atom) efficiency cores (E-cores), and 32 graphics execution units (Gen 12.2 EUs). A smaller die will include only up to 6 P-cores and no E-cores, to be used in lower-end products such as a 6-core Intel Core i5-12400 or a quad-core i3.
Mobile "Alder Lake-P" processors will include up to 6 P-cores, 8 E-cores, and 96 graphics EUs. A smaller "ultra mobile" die will include up to 2 P-cores and 8 E-cores.
AVX-512 is physically present on Golden Cove cores, but disabled in Alder Lake.
The guide mainly focuses on software implementations for hybrid CPUs. It provides various optimization strategies for Alder Lake, including lack of optimization, a "Good Scenario", and the "Best Scenario". According to the document, lack of optimization will not mean that the CPU will be unable to distribute workloads for hybrid CPUs, which should be handled by ThreadDirector anyway, but some may be distributed to the wrong types of cores, should the scheduling algorithm not recognize the task.
In the "Good Scenario," Intel assumes that the application will be aware of the hybrid architecture. The primary tasks should target Performance cores, whereas non-essential and background threads with lower priority should target Effcieent cores.
The "Best Scenario" goes into further detail about which workloads specifically should target Efficient cores: Shader Compilation, Audio Mixing, Asset Streaming, Decompression, Any other non-critical work.
Intel's Thread Director combines a microcontroller with software-based scheduling:
Intel's Thread Director controller puts an embedded microcontroller inside the processor such that it can monitor what each thread is doing and what it needs out of its performance metrics. It will look at the ratio of loads, stores, branches, average memory access times, patterns, and types of instructions. It then provides suggested hints back to the Windows 11 OS scheduler about what the thread is doing, whether it is important or not, and it is up to the OS scheduler to combine that with other information about the system as to where that thread should go. Ultimately the OS is both topologically aware and now workload aware to a much higher degree.
Inside the microcontroller as part of Thread Director, it monitors which instructions are power hungry, such as AVX-VNNI (for machine learning) or other AVX2 commands that often draw high power, and put a big flag on those for the OS for prioritization. It also looks at other threads in the system and if a thread needs to be demoted, either due to not having enough free P-cores or for power/thermal reasons, it will give hints to the OS as to which thread is best to move. Intel states that it can profile a thread in as little as 30 microseconds, whereas a traditional OS scheduler may take 100s of milliseconds to make the same conclusion (or the wrong one).
On top of this, Intel says that Thread Director can also optimize for frequency. If a thread is limited in a way other than frequency, it can detect this and reduce frequency, voltage, and power. This will help the mobile processors, and when asked Intel stated that it can change frequency now in microseconds rather than milliseconds.
[...] On the question of Linux, Intel only went as far to say that Windows 11 was the priority, and they're working upstreaming a variety of features in the Linux kernel but it will take time. An Intel spokesperson said more details closer to product launch, however these things will take a while, perhaps months and years, to get to a state that could be feature-parity equivalent with Windows 11.
See also: Intel 12th gen Alder Lake-P and Alder Lake-M mobile SKUs to enter production between Q4 2021 and Q1 2022; Up to 14 cores, Xe GT3, PCie Gen5, and DDR5 on the anvil
Linux 5.16 To Add Intel Encrypted PXP, Alder Lake S Declared Stable & Ready
Alder Lake Support Added To Intel's TCC Driver In Linux 5.15
Scheduler Changes For Linux 5.15 - Still No Sign Of Any Intel Thread Director Optimizations
Previously: Windows 11 Bashes Some AMD Procs; Boosts Some Intel Core i7 Alder Lake
Related Stories
Upcoming Intel processors will support scalable AVX-512 instructions, which one former Intel employee calls a "hidden gem":
Imagine if we could use vector processing on something other than just floating point problems. Today, GPUs and CPUs work tirelessly to accelerate algorithms based on floating point (FP) numbers. Algorithms can definitely benefit from basing their mathematics on bits and integers (bytes, words) if we could just accelerate them too. FPGAs can do this, but the hardware and software costs remain very high. GPUs aren't designed to operate on non-FP data. Intel AVX introduced some support, and now Intel AVX-512 is bringing a great deal of flexibility to processors. I will share why I'm convinced that the "AVX512VL" capability in particular is a hidden gem that will let AVX-512 be much more useful for compilers and developers alike.
Fortunately for software developers, Intel has done a poor job keeping the "secret" that AVX-512 is coming to Intel's recently announced Xeon Scalable processor line very soon. Amazon Web Services has publically touted AVX-512 on Skylake as coming soon!
It is timely to examine the new AVX-512 capabilities and their ability to impact beyond the more regular HPC needs for floating point only workloads. The hidden gem in all this, which enables shifting to AVX-512 more easily, is the "VL" (vector length) extensions which allow AVX-512 instructions to behave like SSE or AVX/AVX2 instructions when that suits us. This is a clever and powerful addition to enable its adoption in a wider assortment of software more quickly. The VL extensions mean that programmers (and compilers) do not need to shift immediately from 256-bits (AVX/AVX2) to 512-bits to use the new bit/byte/word manipulations. This transitional benefit is useful not only for an interim, but also for applications which find 256-bits more natural (perhaps a small, but important, subset of problems).
Will it be enough to stave off "Epyc"?
AMD: Windows 11 May Cause Performance Dips of Up to 15% on Ryzen CPUs
AMD: Windows 11 May Cause Performance Dips Of Up To 15% On Ryzen CPUs:
Wccftech reported a few days ago about known issues appearing after users have installed Microsoft's newest Windows 11 operating system—Oracle VirtualBox software and Cốc Cốc browser compatibility issues, as well as Intel networking issues. Today, AMD reported issues that their Windows 11 compatible AMD processors were having with performance while running certain applications after the new OS installation.
AMD urges users to stick with Windows 10 as a workaround, hotfix coming
[...] Known changes to performance affected areas such as
- Measured and functional L3 cache latency may increase by ~3X.
- UEFI CPPC2 ("preferred core") may not preferentially schedule threads on a processor's fastest core.
Intel 12th Gen Core Alder Lake for Desktops: Top SKUs Only, Coming November 4th
The first things we'll go into are the new CPUs that Intel is announcing today: the overclockable models of Intel 12th Gen Core. As with previous launches, we have Core i9, Core i7, and Core i5, with the key highlights including new support for DDR5, PCIe Gen 5, new overclocking features, and a change in how Intel is promoting its Thermal Design Power (TDP).[*]
Each processor has a number of performance cores (P-cores) and efficiency cores (E-cores). The P-cores have SMT, whereas the E-cores do not, so we're dealing with non-standard numbers of total threads. Inside the system, the P-core threads, E-core threads, and SMT threads are categorized for performance and efficiency, which we'll get to later in the article. But with a new hybrid design also comes with new ways to showcase frequencies, and each set of cores will have its own base frequency and turbo frequency. The way power is marketed and used has also changed, designed to be clearer.
All processors will come with 16 lanes of PCIe 5.0 from the processor, and an additional 4 lanes of PCIe 4.0 for storage. Memory support is listed as both DDR5-4800 and DDR4-3200, although systems will only support one or the other, for a maximum of 128 GB. The K processors also feature 32 EUs of Intel's Xe-LP graphics, designated as UHD Graphics 770. Prices will start at $264 for the base Core i5 model, up to $589 for the top Core i9 model.
[*] This is the table:
(Score: 0) by Anonymous Coward on Sunday October 17 2021, @03:10AM (4 children)
so many loser "software engineers".
(Score: -1, Spam) by Anonymous Coward on Sunday October 17 2021, @03:12AM (3 children)
It's those fucking Jew-scum rats. Good for Jews. Bad for everybody else. Poison the Goyim you shitbags.
(Score: -1, Troll) by Anonymous Coward on Sunday October 17 2021, @06:19PM (2 children)
I am not the OP, but this is clearly a poor troll and not spam, so why is this mod still allowed to abuse the spam mod? Any mention of the Jew, whether reasonably assessed as flamebait or troll, is always modded spam by this Jew agent. Shutter their account. They have no integrity.
(Score: 2) by martyb on Sunday October 17 2021, @08:22PM
Spam, like beauty, may lie in the eyes of the beholder.
Going by the moderation guidelines which includes (among others) this definition for spam:
in *my* eyes, it fits the definition of spam.
Other admins may view things differently. I see the post as being spam and let the mod stand. They are free to override the mod if they see things differently.
Wit is intellect, dancing.
(Score: 0) by Anonymous Coward on Monday October 18 2021, @02:11AM
Says the "totally not OP" defender of racist crap. Sadly your brain is not up to the task as your use of integrity clearly shows. Stop being a racist dick and maybe people will not mod your trash as spam.
(Score: 2, Funny) by Anonymous Coward on Sunday October 17 2021, @03:35AM
someone vomitted a bunch of cores on me.
Can we have a dedicated Microsoft Telemetry core?
Oh wait, that is TPM...
nevermind.
(Score: 3, Touché) by Anonymous Coward on Sunday October 17 2021, @04:09AM (9 children)
Take your time, Intel, I don't want your second hand quality anyway. Thanks, Lisa, much appreciated.
(Score: -1, Offtopic) by Anonymous Coward on Sunday October 17 2021, @04:48AM
FUck you shit hound
(Score: 4, Interesting) by Runaway1956 on Sunday October 17 2021, @05:13AM (7 children)
Linux may well have a leg up on mixed core infrastructure. All those phones with different chips? They run unix-like OSs. What OS do those phones NOT run? Windoze. I think Linux will take this new Intel thing in stride.
(Score: 1, Insightful) by Anonymous Coward on Sunday October 17 2021, @06:12AM (6 children)
When they say "mobile" they mean 'battery-powered', like a laptop or a Chrome OS tablet.
Because no one, and I mean no one, is successfully marketing an x86 smartphone. That ship sailed.
(Score: 3, Informative) by takyon on Sunday October 17 2021, @07:19AM
Well, he's correct that Linux is running on hundreds of millions of big.LITTLE ARM SoCs.
Even Windows should already have been able to do it, see the Microsoft SQ1 [notebookcheck.net].
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Runaway1956 on Sunday October 17 2021, @07:27AM (4 children)
Is ChromeOS a unix-like, or a Windoze-like? Is it more akin to Linux, or to Windows? Same questions regarding macOS/iOS and Android. Linux and it's brethren are powering virtually all battery operated devices today. Do you really think the new offering from Intel will slow any of them down? There might be a couple hiccups here and there, but they've all been doing this for years.
(Score: 2) by maxwell demon on Sunday October 17 2021, @08:14AM (2 children)
Did you accidentally reply to the wrong comment? Because the comment you replied to was about Intel processors not being used on mobile devices. I don't see how the question whether an OS is Windows-like or Linux-like is relevant for that. Indeed, the question
can be answered right from the post you replied to, and the answer is very clear: No, the poster quite obviously doesn't think so.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2) by Runaway1956 on Sunday October 17 2021, @08:57AM
Yes, I replied to the comment I intended to comment to.
Lemme try this once more:
Unix-like operating systems have been utilizing multiple processors simultaneously for a long time now. They know how it's done. The fact that Intel is adding something to the x86 hardware world won't change much. The instructions are different? So Linux et al changes their programming to match. Intel is doing something new and different for Intel. They aren't doing anything that hasn't been done before. Linux, BSD, ChromeOS programmers will simply examine how Intel implements this "new" architecture, and program for it.
It could be summarized as, "We've done this before, here, we are just doing it again on a different playing field."
(Score: 3, Insightful) by takyon on Sunday October 17 2021, @09:16AM
Linux (Android) is the king of big.LITTLE (ARM phones, tablets, etc.), so it shouldn't be so hard to adapt to x86 versions of big.LITTLE.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Sunday October 17 2021, @08:54AM
Chromebook is a budget laptop. But like Microsoft's Surface Go with Windows 11, it includes an Android app container. Because some people like the tablet form factor on their desktop OS. *
but your earlier comment was about phones and there's no evidence anyone is putting these chips into phones, where ARM has seen off all challengers.
* if you're curious to play with it, you can boot the open source version of Chrome OS off a USB stick https://www.neverware.com/freedownload#home-edition-install [neverware.com]
(Score: 4, Informative) by takyon on Sunday October 17 2021, @09:13AM (1 child)
Intel's Alder Lake CPUs May Not Work With Older Games [tomshardware.com]
Inb4 Steam Deck has better DRM support than Alder Lake.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 3, Insightful) by coolgopher on Sunday October 17 2021, @11:15AM
DRM = Defective by design.
(Score: 0) by Anonymous Coward on Sunday October 17 2021, @03:34PM (1 child)
the only reason for this is to save energy.
i hope we can eject android with its "cpu (like a postage stamps please buy new stamp at checkout counter for future communication requirments), is valid for 3years" from the mobile space.
let's be honest, nothing requires ARM in the gigantic volumes but googles android...
.
having a "out of band" sensor inside the cpu, measuring and reporting cpu loads s prolly the beginning of a real smart cpu.
as it is now, we can check via OS what the cpu is doing but lets not kid ourselfs, this report itself requires cpu resources. if it can be had "out of band" the os becomes smart and can see "ahead" that it's driving towards "the cliff of saturation" or "forever wait".
.
is it possible to start each task/thread on the little cores and only escalate/move to the heavy cores later? or is a task/thread bound to the core it was started on?
(Score: 3, Informative) by takyon on Sunday October 17 2021, @03:58PM
They can migrate, quickly.
Actually, the AVX situation is probably due to the need for parity between the two core types, so that any thread can run on any core:
https://en.wikipedia.org/wiki/Alder_Lake_(microprocessor)#CPU [wikipedia.org]
x86 still has a long way to go before it could possibly match ARM efficiency, and the top Alder Lake is rumored to use a lot of energy if allowed to:
Intel Core i9-12900K tested: toasty 93C under load, 250W of power used [tweaktown.com]
The 3.9 GHz boost clock for Atom cores, while about 1.4 GHz lower than the performance cores, is higher than any other Intel Atom product ever, AFAIK.
What is true is that the Atom cores have a performance-per-watt advantage, as well as a performance-per-area advantage. Intel can put 4 Atom cores in the place of 1 Golden Cove core.
The next generation, Raptor Lake, will literally double down on this strategy by keeping 8 big cores, and doubling to 16 small cores. It's rumored that Intel will quadruple down with "Arrow Lake-S" [digitaltrends.com] (desktop), again keeping 8 big cores, but doubling again to 32 small cores, for a total of 40 cores, 48 threads.
Google is not the only major ARM user, just look at Apple [notebookcheck.net].
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Sunday October 17 2021, @06:24PM
https://www.youtube.com/watch?v=MtBF6MxquWI [youtube.com]
I don't care about your hardware until you care about my security and freedom.