Friday, June 24th 2022
Intel "Raptor Lake" Core i9 Sample Powers Up, 8P+16E Configuration Confirmed
An engineering sample of a 13th Intel Core "Raptor Lake" Core i9 processor hit the web, courtesy of wxnod on Twitter, which confirms its 8P+16E core-configuration in a CPU-Z screenshot. Based on the same LGA1700 package as "Alder Lake," and backwards compatible with Intel 600-series chipset motherboards, besides new 700-series ones, "Raptor Lake" combines eight "Raptor Cove" performance cores (P-cores), with sixteen "Gracemont" efficiency cores (E-cores).
"Raptor Cove" features a generational IPC increase over the "Golden Cove" P-cores powering "Alder Lake," while the "Gracemont" E-cores, although identical to those on "Alder Lake," are expected to benefit from the doubling in L2 cache per cluster, from 2 MB to 4 MB. The ISA as detected by CPU-Z appears to be identical to that of "Alder Lake." The processor is a monolithic silicon chip built on the Intel 7 (10 nm Enhanced SuperFin) silicon fabrication process.
Sources:
wxnod (Twitter), VideoCardz
"Raptor Cove" features a generational IPC increase over the "Golden Cove" P-cores powering "Alder Lake," while the "Gracemont" E-cores, although identical to those on "Alder Lake," are expected to benefit from the doubling in L2 cache per cluster, from 2 MB to 4 MB. The ISA as detected by CPU-Z appears to be identical to that of "Alder Lake." The processor is a monolithic silicon chip built on the Intel 7 (10 nm Enhanced SuperFin) silicon fabrication process.
104 Comments on Intel "Raptor Lake" Core i9 Sample Powers Up, 8P+16E Configuration Confirmed
Every review I've read of alder lake mobile has it stomping the competition, and doing so while on a larger less efficient node, barring the m1 pro (which is a completely different ISA on a smaller node).
Intel Core i7-12700H Review: Alder Lake on the Go | TechSpotAMD is going the exact same route BTW. Zen 5 is going to be 8C Zen 5 and 16C Zen 4D (high density e cores). So you know that design has something going for it. Intel without e cores wouldn't even be in contention at the high end on their current node, and neither would Zen 4 at 10nm.
A) required to use Windows 11 for best performance. Windows 11 is fine but Microsoft's draconian requirements and shoving their anti-consumer practices down everyone's throat means I'm going to be trying Linux, probably SteamOS before I try Windows 11 for anything. I'm starting to feel old, because I used to always switch the latest version of WIndows as soon as I could, I even used and had good success with Vista, but the highly questionable decisions to enforce TPM and require a microsoft + internet access to install the damn OS is just too much, F microsoft and Windows 11. Yes, I feel strongly about it and I'm probably not in the majority but for me Intel is guilty by association with Windows 10's inferior core scheduler.
B) E cores on a desktop that's always plugged into the wall are not really that useful over another P core. P cores can still go into very low-power mode anyway so I highly doubt anyone is even saving that much energy in most desktop use cases.
The pc restarted, it skipped the network settings and I was able to create a local account.
Succeeded:)
Pure, scalable software rendering like V-Ray that doesn't rely on bandwidth or AVX512 gimmickery are still comfortably the domain of Zen3 and I think Intel's problem is that they're power limited. They have the architecture to beat AMD right now but to do so in something like V-Ray would need so many Watts delivered to the socket that it would just turn to hot slag. Hopefully more E-cores and Intel 4nm combine to push the performance envelope forward.
the problem w/ the 12900k is just that its pushed way above its efficiency window like, if you were to chop it back to like 150w you'd still be at something like 80% of its performance, so yeah id imagine more E-cores being good for MT stuff
Anyway I'm all good for now :D
AVX-512 died for this, and that's something which actually does have a large performance impact on a program that I run regularly.
There was a Win10 update a couple of months ago that brength the exact functionality of Win11 to Windows 10.
This is not to say the hybrid approach is perfect but it has given Intel a massive performance improvement from the crud that was 14nm in the 11th gen where they have taken back the lead especially in productivity workloads and a lot of us do use these workloads across encoding, rendering etc oh and yes I game as well . AMD on the other hand have there own highly efficient solution with the advantage of a long lived platform that works extremely well. The key is that we have a choice, a proper choice now that AMD socked it to Intel and made them compete. Not sure what the issue is on e-cores but more than impressed on my workloads with the 12700K and I now have more performance than I even I expected which only bodes well for the longevity of this CPU before I upgrade in 5 odd years to whatever is best at the time..
Raptor Lake with 8 P-Cores of the new Golden Cove type and 16 Gracemont e-cores with a higher cache at the top end will be interesting but it will be the halo product, the 13400, 13600 and 13700 will be the interesting ones and we will soon see how well that pans out but it is Meteor Lake that really matters as it is on the new N4 process node with a 40% lower power consumption and low and behold they are going the chiplet route ala AMD (though it will be the new 3D Foveros platform that will make or break Meteor Lake)...Looking forward to see if they can pull it off...
Another perspective, though, is that the ideal product would marry both concepts: chiplet with interconnect, and several core designs next to each other. In both situations though we're looking at something that can clearly use refinement. But the strides that AMD is making on their chiplet end of the line, are much bigger per generation and over the course of Ryzen versus the samey-timed Intel releases.
It remains to be seen what truly is better. One thing is absolutely certain: the REASON these E-cores extract an advantage right now, is because these CPUs are barely ever pushed to their limits. You quite simply can't, you either run into its retarded boost limit Wattage cap, or you run into a cap you set for yourself after having met 241W once, or you run into heavily reduced clocks. Intel made a great CPU for low intensity consumer segment, but a shite CPU otherwise. This extends to gaming: gaming is a pretty light load, not parallelized at all, and barely loads cores fully, if ever. Not a huge surprise that ADL works for it and the E cores help a bit on background load.
Another aspect in this race is often forgotten: yields and production cost. I reckon AMD has a MUCH higher margin on Ryzen chips than Intel will ever get on their ADL line and followups. Fact is, they're still building monolithic chips and they make them per segment. Ryzen is just a slice of EPYC. And an underlying aspect in that, is that power consumption is going to matter. If you can't keep shrinking, you must go wider. I think the next GPU generation will drive that fact home for us. 450W on top end ADA...