Monday, December 21st 2020

Intel 11th Gen Core "Rocket Lake" to See CES 2021 Unveil, Availability Later in Q1

Intel's 11th Generation Core "Rocket Lake-S" desktop processor family could be formally unveiled in just a few weeks from now, with HotHardware reporting a 2021 International CES unveiling (virtual event), for both the processors and their companion Intel 500-series chipset motherboards. This would put the unveiling around mid-January for the virtual launch event.

Availability is a different story, with the report predicting that while the Intel 500-series chipset motherboards will be available from mid-January, the processors won't arrive until March. Older reports predicted a market availability of these processors to almost miss Q1, and spill over into Q2 (April-June). Since "Rocket Lake-S" is based on the current LGA1200 package, the Intel 500-series chipset motherboards are expected to support existing 10th Gen Core "Comet Lake-S" processors, along with out-of-the-box support for the 11th Gen.
Source: HotHardware
Add your own comment

18 Comments on Intel 11th Gen Core "Rocket Lake" to See CES 2021 Unveil, Availability Later in Q1

#1
Hyderz
oooh interesting to see what the new cpu brings this time around. i've heard the i9 has 2 less cores than the 10th gen but still achieve higher ipc.
Posted on Reply
#2
fancucker
Intel taking back the gaming crown
Posted on Reply
#4
ratirt
Hyderz
oooh interesting to see what the new cpu brings this time around. i've heard the i9 has 2 less cores than the 10th gen but still achieve higher ipc.
with a first look, nothing new.
Posted on Reply
#5
amit_talkin
Honestly, I have lost the track of Intel lineup since coffee lake. So many "lakes" to get confused .
Posted on Reply
#6
TheLostSwede
Hyderz
oooh interesting to see what the new cpu brings this time around. i've heard the i9 has 2 less cores than the 10th gen but still achieve higher ipc.
Current chips that Intel are circulating amount partners apparently perform like a 9900K in games..
Posted on Reply
#7
ZoneDymo
ok but really, who in their right mind cares?
Posted on Reply
#8
HansRapad
ZoneDymo
ok but really, who in their right mind cares?
OEM, SI, and whoever relying on Intel Superior Supply Chain over AMD

Intel still superior on supply chain level
Posted on Reply
#9
Turmania
For me it is all about reliability and longevity therefore, I would chose Intel to the other competitor whom has yet to prove itself on that front. Same goes for GPU side as well, I would chose Nvdia for its reliabilty. But the reasons for chosing them is getting narrower each passing product launch and the competitor fixing its reliability and longevity deficit.
Posted on Reply
#10
Octopuss
Turmania
For me it is all about reliability and longevity therefore, I would chose Intel to the other competitor whom has yet to prove itself on that front. Same goes for GPU side as well, I would chose Nvdia for its reliabilty. But the reasons for chosing them is getting narrower each passing product launch and the competitor fixing its reliability and longevity deficit.
What drugs have they been giving you?
You're something like an antivaxxer in the IT.
Posted on Reply
#11
chris.london
TheLostSwede
Current chips that Intel are circulating amount partners apparently perform like a 9900K in games..
That is disappointing, but not completely unexpected. (Haswell and Skylake also hardly increased gaming performance.) Source?
Posted on Reply
#12
Ravenmaster
Garbage stop-gap processors. Wait for Alder Lake which will support DDR5 and PCI-E 5.0
Posted on Reply
#13
Chrispy_
I strongly suspect that the IPC gains are in special fixed-function execution blocks and that general x86-64 performance is unchanged.

Intel loves tacking on extra rubbish for one specific application and then flogging the benchmarks that include that application like a dead horse.
amit_talkin
Honestly, I have lost the track of Intel lineup since coffee lake. So many "lakes" to get confused .
They kind of messed up the naming by separating their architecture and manufacturing node last year. Here's a list (made mostly to clarify for my own benefit, not just yours!)

Skylake Architecture (6th-Gen to 11th-Gen):
  • 2015 - Skylake 14nm (new architecture and process)
  • 2016 - Kaby Lake (no real architectural or process changes, but on matured process)
    • 2018 - Amber Lake (rebrand for laptops, no changes)
    • 2018 - Whiskey Lake - (rebrand for laptops with Meltdown vulnerability mitigations)
  • 2017 - Coffee Lake (no per-core architectural changes from Kaby Lake but scaled to 8 cores)
    • 2019 - Cascade Lake (Coffee Lake renamed for servers and HEDT with Spectre/Meltdown fixes)
  • 2019 - Comet Lake (Coffee Lake, scaled up to 10 cores, some Spectre vulnerability mitigations)
Palm Cove Architecture (8th Gen failure):
  • 2018 - Cannon Lake 10nm (stillborn as a result of broken 10nm process. CPU was defective and cancelled)
Sunny Cove Architecture (10th-Gen onwards):
  • 2019 - Ice Lake 10nm+ (Laptop only, Palm Cove architecture tweaked to make it work on the still-broken 10nm process)
  • 2021 - Rocket Lake (Backported to 14nm++ because of 10nm problems) This backport is called Cypress cove but Sunny Cove in all but name)
Willow Cove Architecture (11th-Gen onwards):
  • 2020 - Tiger Lake 10nm++ (Laptop only, basically Ice Lake + Xe graphics and more vulnerability mitigation)
Golden Cove Architecture:
  • 2021-2022 - Alder Lake 10nm+++ (Laptop only, Tiger Lake + Atom in a single package, DDR5 support, and a whole bunch of dubious fixed-function extensions slapped on top of Tiger Lake's basic core layout)
So, TL;DR...
10nm still terribad, Intel still not fully spectre/metldown fixed, Rocket Lake is basically 2018's failed attempt to launch Palm Cove backported to 14nm for desktop. Soon we will have a new, three-year-old design from Intel for our desktops! :D
Posted on Reply
#14
TheLostSwede
chris.london
That is disappointing, but not completely unexpected. (Haswell and Skylake also hardly increased gaming performance.) Source?
Sorry, can't give you one, as it was from a friend in the industry.
It's still not final CPUs or boards/UEFI, but I doubt it'll improve by a lot.
I should say that the person that told me has been an avid Intel user for years and is now considering going with an AMD build when upgrading next.

Looks like some more benchmarks have leaked, which pretty much verifies what I said above.
videocardz.com/newz/intel-core-i9-11900-rocket-lake-s-engineering-sample-tested-on-b560-motherboard
Posted on Reply
#15
efikkan
Chrispy_
I strongly suspect that the IPC gains are in special fixed-function execution blocks and that general x86-64 performance is unchanged.
Intel loves tacking on extra rubbish for one specific application and then flogging the benchmarks that include that application like a dead horse.
While I agree application specific instructions are gimmicks and crap that should be avoided, but no such instructions will yield any IPC gain, and would require the software using it to be rewritten using low-level code to utilize it, but you should have known all this if you knew what you were talking about.

If Cypress Cove is essentially a full Sunny Cove, we should expect a similar performance gains. I've not yet seen any solid information pointing to improvements beyond Sunny Cove, but it's not impossible, considering how late it was designed. Sunny Cove is a major improvement over Skylake, with large improvements in the front-end and the caches, plus load/store throughput, address calculation and integer multiplication. Overall, these improvements add up to a performance bump in non-vector workloads comparable to Sandy Bridge -> Skylake, so this is no minor improvement.
Posted on Reply
#16
Chrispy_
efikkan
If Cypress Cove is essentially a full Sunny Cove, we should expect a similar performance gains. I've not yet seen any solid information pointing to improvements beyond Sunny Cove, but it's not impossible, considering how late it was designed. Sunny Cove is a major improvement over Skylake, with large improvements in the front-end and the caches, plus load/store throughput, address calculation and integer multiplication. Overall, these improvements add up to a performance bump in non-vector workloads comparable to Sandy Bridge -> Skylake, so this is no minor improvement.
That's why Rocket lake is in a different section, as it's Cove architecture rather than Skylake. The IPC gains should be comparable to Ice Lake vs Whiskey Lake, which were actually quite big (~18% in a clock-for-clock test by Anandtech) but the problem with Ice Lake was poor clocks on the broken 10nm+ process, ofsetting the IPC gains and disguising the architectural improvement. Backported to 14nm++ those ~18% gains could really help Intel catch up with Zen3, provided they don't melt the socket in doing so!

Contrary to what you say, general x86-64 IPC does increase, even ignoring the application-specific FF hardware, because the Cove architecture has more cache per core, improved prefetch and TLB updates that are far better than those of Skylake and not dissimilar to the cache and prediction changes from Zen1 to Zen2. There are no special instructions or low-level code rewrites to take advantage of those things.
Posted on Reply
#17
ARF
Turmania
For me it is all about reliability and longevity therefore, I would chose Intel to the other competitor whom has yet to prove itself on that front. Same goes for GPU side as well, I would chose Nvdia for its reliabilty. But the reasons for chosing them is getting narrower each passing product launch and the competitor fixing its reliability and longevity deficit.
How reliability with Intel CPUs when there are so severe security exploits and vulnerabilities?!
How longevity when AMD has 16 cores while Intel is just about to downgrade to half of that?!!
Octopuss
What drugs have they been giving you?
You're something like an antivaxxer in the IT.
Trolmetol Forte® ;)
HansRapad
OEM, SI, and whoever relying on Intel Superior Supply Chain over AMD

Intel still superior on supply chain level
Corruption and anti-competitive practices which result in billions of $ fines and no trust in Intel in the people with more ethics and morality than greed and envy.
Posted on Reply
#18
efikkan
Chrispy_
That's why Rocket lake is in a different section, as it's Cove architecture rather than Skylake.
I wouldn't put too much emphasis on the naming. Sunny Cove is implemented in Ice Lake, Willow Cove(minor change) in Tiger Lake, then a new architecture in Golden Cove implemented as Sapphire Rapids and Alder Lake. These are all just marketing names, they "mean" nothing.
Chrispy_
The IPC gains should be comparable to Ice Lake vs Whiskey Lake, which were actually quite big (~18% in a clock-for-clock test by Anandtech) but the problem with Ice Lake was poor clocks on the broken 10nm+ process, ofsetting the IPC gains and disguising the architectural improvement. Backported to 14nm++ those ~18% gains could really help Intel catch up with Zen3, provided they don't melt the socket in doing so!

Contrary to what you say, general x86-64 IPC does increase, even ignoring the application-specific FF hardware, because the Cove architecture has more cache per core, improved prefetch and TLB updates that are far better than those of Skylake and not dissimilar to the cache and prediction changes from Zen1 to Zen2. There are no special instructions or low-level code rewrites to take advantage of those things.
I believe you must have misread me. I said the application specific hardware (you called it fixed function) doesn't yield any IPC gains, but not that there aren't any IPC gains. If Cypress Cove is essentially Sunny Cove, I would expect ~18% IPC gain (which is why I compared it to SB -> Skylake), plus the benefits of AVX-512, and any other goodies or gimmicks they might have put in there :)
Posted on Reply
Add your own comment