Intel has announced new developments at its Architecture Day 2018:
Sunny Cove, built on 10nm, will come to market in 2019 and offer increased single-threaded performance, new instructions, and 'improved scalability'. Intel went into more detail about the Sunny Cove microarchitecture, which is in the next part of this article. To avoid doubt, Sunny Cove will have AVX-512. We believe that these cores, when paired with Gen11 graphics, will be called Ice Lake.Willow Cove looks like it will be a 2020 core design, most likely also on 10nm. Intel lists the highlights here as a cache redesign (which might mean L1/L2 adjustments), new transistor optimizations (manufacturing based), and additional security features, likely referring to further enhancements from new classes of side-channel attacks. Golden Cove rounds out the trio, and is firmly in that 2021 segment in the graph. Process node here is a question mark, but we're likely to see it on 10nm and or 7nm. Golden Cove is where Intel adds another slice of the serious pie onto its plate, with an increase in single threaded performance, a focus on AI performance, and potential networking and AI additions to the core design. Security features also look like they get a boost.
Sunny Cove, built on 10nm, will come to market in 2019 and offer increased single-threaded performance, new instructions, and 'improved scalability'. Intel went into more detail about the Sunny Cove microarchitecture, which is in the next part of this article. To avoid doubt, Sunny Cove will have AVX-512. We believe that these cores, when paired with Gen11 graphics, will be called Ice Lake.
Willow Cove looks like it will be a 2020 core design, most likely also on 10nm. Intel lists the highlights here as a cache redesign (which might mean L1/L2 adjustments), new transistor optimizations (manufacturing based), and additional security features, likely referring to further enhancements from new classes of side-channel attacks. Golden Cove rounds out the trio, and is firmly in that 2021 segment in the graph. Process node here is a question mark, but we're likely to see it on 10nm and or 7nm. Golden Cove is where Intel adds another slice of the serious pie onto its plate, with an increase in single threaded performance, a focus on AI performance, and potential networking and AI additions to the core design. Security features also look like they get a boost.
Intel says that GT2 Gen11 integrated graphics with 64 execution units will reach 1 teraflops of performance. It compared the graphics solution to previous-generation GT2 graphics with 24 execution units, but did not mention Iris Plus Graphics GT3e, which already reached around 800-900 gigaflops with 48 execution units. The GPU will support Adaptive Sync, which is the standardized version of AMD's FreeSync, enabling variable refresh rates over DisplayPort and reducing screen tearing.
Intel's upcoming discrete graphics cards, planned for release around 2020, will be branded Xe. Xe will cover configurations from integrated and entry-level cards all the way up to datacenter-oriented products.
Like AMD, Intel will also organize cores into "chiplets". But it also announced FOVEROS, a 3D packaging technology that will allow it to mix chips from different process nodes, stack DRAM on top of components, etc. A related development is Intel's demonstration of "hybrid x86" CPUs. Like ARM's big.LITTLE and DynamIQ heterogeneous computing architectures, Intel can combine its large "Core" with smaller Atom cores. In fact, it created a 12mm×12mm×1mm SoC (compare to a dime coin which has a radius of 17.91mm and thickness of 1.35mm) with a single "Sunny Cove" core, four Atom cores, Gen11 graphics, and just 2 mW of standby power draw.
Why no stats? It just says "single threaded improvement", "new instructions", "improved scalability". That reads more like a wishlist.
It's not a wishlist, it's what Intel has been announcing every year now for at least five years, since they ran out of ideas on what to do in their CPUs. A lick of paint here, a minor tweak there, some graphics crap that no-one cares about because Intel can't do graphics to save themselves, and new instructions including IOMAAYXPRT, WSSXPPSSFFTY, and MPLLXNGRT that some guy in Intel's sales department thought up on the toilet this morning, or possibly later in the day, around 4:20. Shit, I've got a ten-year-old Intel quad-core CPU for which the only real differentiating factor from today's CPUs is that it has AES instructions. Which in my case, and most people's cases, are entirely superfluous. There's nothing to announce because they've run out of ideas for what to do.
All the development efforts have been in the [User] Management Engine.
Ah, of course! Those buffer overflows and leaks don't just write themselves.
AMD and nVidia may be more expensive (or not) but I'll stick with them. For all the same reasons I don't run Intel CPU's or Intel network devices. There is nothing about Intel that I like, or that I can respect.
Anti-Semite! Now eat our chips, Goy, they are best-served with dips like you.
Intel will not be cheaper if this is true: https://www.techpowerup.com/img/iQWJhazDsExN7biT.jpg [techpowerup.com]
Keep in mind they were showing off a 28 core 5 GHz cpu that required like 1 kW of industrial cooling for ~$10k not long ago...
This is the newer leak: https://wccftech.com/amd-ryzen-3000-specs-prices-leaked-upto-16-cores-5-1ghz-on-am4/ [wccftech.com] which came from AdoredTV on YouTube.
Notice the Ryzen 9 3850X, slated for May 2019? If correct, that would be AMD's "50th anniversary chip" (they were founded on May 1, 1969).
Sorry, this didn't have the (I assume) expected impact on me. What is exceptional about this vs the rest of the craziness in these leaks?
It seems more likely to me than slapping the old "Black Edition" label on it from the Bulldozer days. It's also similar to how Intel recently released an i7-8086K commemorative chip. AMD could be copying that idea and the timing is just right.
The YouTuber went into greater detail about why he thought his leak was accurate. You can watch it yourself if you want. Or just wait until CES in January.
The most important point, that the top Ryzen chips will have 12 and 16 cores, has been floating around for a long time now. That was reinforced when AMD raised Threadripper 2 up to 16-32 cores despite not increasing cores per CCX on the "12nm" node. And now we know that the Epyc chips will have 64 cores using 8 chiplets. The 16-core Ryzen is a done deal.
> The 16-core Ryzen is a done deal.
Sadly for their profits, what most of the market needs is covered by a quad-A73.
I'm very glad for the progress (Go AMD!), by my current quad i5 is about 5 years old, and unless it kicks the bucket, it looks like it should be enough until Ryzen 4.
Chicken, meet egg. Or vice versa. With core counts reaching 16 for mainstream CPUs, someone will come up with software that can take advantage of it. And don't say "bloat will do it!" because you actually have to try to waste those resources.
I wish you luck on stretching to Ryzen 4 or 5. Maybe you can hold out until the "5nm" node for some extra gains. If you get an APU or GPU, make sure it comes with AV1 hardware decoding.
If you get an APU or GPU, make sure it comes with AV1 hardware decoding.
This reminds me of the hardware h.264 in my GPU. I don't know if there are any programs that support it even now. I wouldn't be surprised if it took years before any software supports the AV1 hardware decoding.
I'd say first in line is your web browser.
This might be the most informative response I've received. I bookmarked that!
I think you've got it in reverse. Intel's monopoly stagnated the PC market, so people stopped coming up with cool new stuff to do with them.
Most of the market needs a computer to check email/facebook/twitter, order on Amazon, make/watch youtube/porn, and do some basic office stuff.A quad A73 is plenty for all of the above. Intel delivering more power would not have changed that much, just enabled faster spying.
I got a free Titan card for my PC, yet without playing the latest AAA games, I don't seem to be hurting the processor, yet I am pretty satisfied with the scene complexity compared to ten years ago.I'm sure my future Ryzen 3 oe 4 will blow my socks off.
A quad A73 is plenty for all of the above. Intel delivering more power would not have changed that much, just enabled faster spying.
It isn't just delivering more power, it is also doing the same while consuming less power, or much cheaper.
But in general, the high core count is a qualitative difference from the gradually increasing single thread speed on 2-6 cores intel has been offering. We don't know yet who will be able to take advantage of this.
Same here. One correction though... AMD is not more expensive than Intel. Not by $/core, not by $/performance. And that will become very apparent when they start releasing "7nm" desktop chips next year. But this list of Intel stuff still warrants a look anyway.
To give credit where it's due, Intel dominates in low-Watt x86. Their sub-8W Celerons, Pentiums, and Atoms are pretty ubiquitous in Chromebooks and whatever the Windows equivalent is while there are basically no AMD chips in those (there are some ARM chips). I like those machines since they are light, fanless, and have great battery life. They also do pretty well with the 15 Watt laptop chips, i.e. the ones found in "ultrabooks". And then even when we factor in AMD's expected IPC and clock improvements in 2019, Intel might still be ahead in single-threaded performance across the board.
So when I see the "hybrid x86" demo chip, that's pretty interesting. Neither Intel or AMD has given much thought to putting cores of different sizes and performance/efficiency on the same die. That's the big.LITTLE/DynamIQ approach that is found in hundreds of millions of smartphone chips. It will be interesting to see if these kinds of designs benefit laptop and tablet users. Also, my understanding is that Intel could mix cores from different process nodes. So they could put one big "10nm" core alongside several "14nm+++++" cores.
This 3D packaging approach they are experimenting with is interesting but is far from what may be available in a decade or two, where 3D placement of layers of cores and memory could allow computers to be thousands of times faster [darpa.mil].
If you're interested in AMD, then you should be looking at this leak* [wccftech.com] and keeping your eyes peeled on CES 2019. AMD is likely to increase "7nm" Ryzen core counts up to 16, from 8. That's right, Ryzen with 16 cores, not just Threadripper. Combine that with about a 20-25% performance increase from IPC improvements and clock speeds, and you can see why Intel is getting antsy and showing off its goods.
Apparently this new info about "12nm" mobile APUs [wccftech.com] just came out. I'm not going to touch it since I want to see them double core counts to 8, from 4, which they might do on the "7nm" node in 2020. And I still want AV1 support [wikipedia.org] in any new hardware.
*SPOILER* (click to show)
*SPOILER* (click to hide)
*The YouTuber who released the leak has walked back some details, such as AMD's use of "14nm" I/O dies to connect 8-core chiplets shown in the diagrams. That was a guess on his part, and his sources say that it will just use Infinity Fabric.
I saw "Xe" with the 'e' smaller than the 'X', and my next thought was of Blackwater Security, a private military and law enforcement training company that changed its name to "Xe Services" [wikipedia.org] in 2009 and then reestablished as "Academi" a year later.
Does Intel really have any credibility left for preannouncements?
We at least are pretty sure that the thing they demoed exists.
Some of this stuff is basically a copy of what AMD did so it's not very surprising or unlikely.