https://www.pcgamer.com/rocket-lake-proves-intel-is-broken/

Intel's manufacturing is broken and the newly announced 14nm Rocket Lake CPUs prove it
By Jeremy Laird November 18, 2020

A sudden gust of wind blows the edges of Intel smoke screen off the new 14nm Intel Rocket Lake.

The next six months are make or break time as Intel hurtles towards existential crisis.

Tales of Intel’s woes have become so routine of late that the very existence of Rocket Lake and what it implies has been largely overlooked. Rocket Lake says very, very bad things about the viability of Intel’s entire business model. And that, in turn, makes it very significant for the PC as a whole.

All of a sudden, Rocket Lake is now a 14nm backport, one that was never planned nor designed that way previously.

Rocket Lake, of course, was Intel’s next 10nm desktop CPU architecture.  As of this week Rocket Lake is suddenly a 14nm backport of Intel’s 10nm Sunny Cove CPU core architecture, as seen in 10th Gen Ice Lake notebook chips. Rocket Lake won’t be released until next year, which means Intel will be launching a new CPU design in 2021 on the ancient 14nm node.  (Intel’s original plan was to move to 10nm in 2016)

Yes, really.   14nm.   Again 14nm ......

Intel has tried to sell the whole ‘back porting’ thing as a positive, a sort of groovy and inclusive approach to CPU manufacturing. “Hey guys, relax. We’re flexible, we can port from node to node. It’s freestyle. It’s all good,” Intel seems to be saying.

The reality is that there’s really no such thing as a node-agnostic CPU architecture. It’s going to cost a huge amount of money to port those 10nm Sunny Cove cores, PCI Express 4.0 I/O and Xe-based graphics to 14nm for Rocket Lake.   The die sizes will have to expand, the core counts will have to go down and it is questionable if XE-based graphics are going to work right on a back ported 14nm lithography.

So, it’s not groovy or flexible. It’s a move made out of desperation because Intel’s 10nm production node still isn’t good enough for the prime time. Let’s repeat that. Rocket Lake will be launched in 2021 at 14nm because Intel’s 10nm still won’t be good enough for a desktop CPU launch.

If this is true, then you can count on Intel HAVING to pay TSMC to build their processors for them in 2022-23, because Intel will be 3-4 full lithography nodes behind the competition by that point in time.




14nm to 10nm to 7nm to 5nm with 3nm being in trial runs at Apple and AMD by the time 14nm Intel Rocket Lake ships in any real volume --- I would count 4-5 full lithography nodes behind at that point, personally.

Next, 14nm Rocket Lake will be just as physically big as an old style Intel processor --- Rocket Lake may actually have to lose a core or two compared to its original press releases for these now recognized size and thermal overheating reasons.

https://www.tomshardware.com/news/intel-11th-generation-rocket-lake-cpu-sisof...

Unnamed Intel Rocket Lake Eight-Core CPU Fails To Defeat Ryzen 3 3300X which only has 4 cores



A fresh round of Rocket Lake benchmarks (via Tum_Apisak) has set the rumor mill into action once again. The highest Rocket Lake configuration so far has been eight cores, so this nameless 11th Generation processor may be the Rocket Lake flagship chip or a variant of it.

If you haven't been following Intel's products, Rocket Lake marks the end of the Skylake era. Exercising Intel's latest Cypress Cove cores, Rocket Lake was poised to bring a real performance upgrade for users, unlike the previous iterations of Skylake that only offered minor improvements to existing architecture to justify the introduction of a new product. Intel has deferred to say how much performance we should see exactly, but boasted about IPC (instruction per cycle) enhancements in the double figures.

SiSoftware listed this particular Rocket Lake-S sample with an eight-core, 16-thread configuration with 16MB of L3 cache. This description certainly matches that of the core specifications of a presumed octa-core Core i9 SKU.

The Rocket Lake part reportedly runs with a 1.8 GHz base clock and 4.4 GHz boost clock. The processor's clock speeds are pretty far away from the values that we previously saw for another unidentified Rocket Lake processor, which flaunted a 3.41 GHz base clock and 4.98 GHz boost clock. In all likelihood, the SiSoftware chip could be an early engineering sample with a low 65W or 35W TDP. On the flipside, SiSoftware may be misinterpreting the clock speeds, which happens a lot when it comes to unreleased hardware.

Intel 11th Generation Rocket Lake Benchmarks

Rocket Lake (8 Cores)   431.10 Mpix/s      7.55 GB/s      28.32 kOPTS      21.79 GFLOPS      325.18 Mpix/s      4.04 kPT
Ryzen 3 3300X            474.62 Mpix/s      13.59 GB/s      42.92 kOPTS      28.08 GFLOPS      274.02 Mpix/s      5.32 kPT
Core i5-9600K             473.90 Mpix/s      10.85 GB/s      44.52 kOPTS      30.50 GFLOPS      354.65 Mpix/s      4.16 kPT

Without a model name to pin the 8 core Rocket Lake processor to, it's pretty hard to deduct which processor it's replacing. Furthermore, we don't have all the details of the platform that was used in the tests, so we suggest you take these results with a pinch of salt for now.

The results from the SiSoftware submissions aren't very inspiring though. The octa-core Rocket Lake processor seemingly put out worse scores than the Ryzen 3 3300X quad-core processor and Core i5-9600K hexa-core chip.

Not performing as well as the existing Intel Core i5 processors and being clearly inferior to Ryzen 3 has been confirmed now in another set of samples reported by a second big name review house.    

This sort of results could mean "delayed due to something or another" and then a quick death for the Rocket Lake line of products .......

Intel will either fix it or kill it ...... fast .......    Bob Swan will not spend any more money on this one, he is too good of a bean picker to make that mistake.

AI and some benchmarking & clever marketing lies clustered around a trick filled product introduction won't fix this one like it did for Tiger Lake as this the bloody wound is bleeding too fast and is simply too too big and ugly to slap a bandage on it.

DO YOU CHOOSE TO DRINK THE HIDDEN ACCELERATOR  KOOL-AID ???

AI and "hidden support softwares" are not present in all applications, so the AI stuff does not work all the time and it should not be given up to the public as "CPU" improvements in advertising or in benchmark test results.  

Class action suits are in motion even now for the Tiger Lake buyers who got defrauded by this Intel Marketing ploy.

Where and how you spend your money is 100% how you vote in this issue.

Apple controls their software universe, and if you need it they REQUIRE that it gets put into all software and drivers automatically.   Apple also names their neat little tricks and states their use openly.

AMD just builds faster and better processors ---- you sign up for your own tricks if you think you need any.

Intel keeps secrets and lies to everybody, that is just the new Intel and how it rolls.  

You pay more for Intel so we hope the Intel Marketing tricks amuse you enough to give you recompense for the extra 30% more that the little Intel Inside logo costs you.

Otherwise, go join a class action suit to get some of your money back.


====================================================


Just type this into your browser and pick the one that fits your circumstances best --- likely you can participate in more than one as Intel has provided multiple issues to sue them over ......

current intel class actions                   on the Google Chrome search engine

https://www.google.com/search?q=current+intel+class+actions&oq=current+intel+...

https://www.youtube.com/watch?v=iZBIeM2zE-I&t=198s

Linus at Linus Tech Tips covers some of this and offers a few sets of numbers and comparative predictions about Rocket Lake vs AMD.

He also asks that Lisa Su to PLEASE quit kicking the dead Intel donkey in the balls.

Please stop kicking it, it is finally dead and it is not going to move again.


    ....... does this mean when the dead Intel donkey eventually appears to get back up it is actually a zombie donkey ?????






==================================================



For those who deem the effort to look worth to be worth the effort, the new Cinebench 23 has ranked all current processors by Multi-core results while still offering the best of the installed cores single best core speed numbers for reference.  Note please, current modern games all use multi-core processing now so any BS about using that one single "best" core to rate chipsets has become just some more Intel Marketing BS.

This INCLUDES the Apple M1 and will include all ARM chipsets when they show up, so it Cinebench 23 is a good reference benchmark because it can correctly rank the different brands and types of processors --- and it avoids many of Intel's magic minute, hyperclocking and "turbo boost" games as it does make sure the chipset warms up to its steady state and the processor is system BIOS throttled to its settled performance level.   Cinebench 23 reviews entire systems including the cooling system (listing models and brands) as well as just giving out the averaged results for a given processor.

Reminder, honest companies test completed production units and rank them using Cinebench 23 as this test impartially reports how good your entire thermal system really is.   There is a lot to like about Cinebench 23 because it can sort out so much of the Intel Marketing BS easily and with no fuss ......


https://www.cgdirector.com/cinebench-r23-scores-updated-results/                (click to the left to see pretty graph of this data)

CPU Name      Cores      Ghz      Single Score      Multi Score
AMD Threadripper 3990X      64      2.9      1262   75671
AMD Epyc 7702P                      64      2.0    993     48959
AMD Threadripper 3970X      32      3.7   1308   46874
AMD Threadripper 3960X      24      3.8      1307   34932
AMD Threadripper 2990WX      32      3.0      1005   29651
AMD Ryzen 9 5950X              16      3.4      1684   28782
Intel i9 9980XE                      18      3.0      1114   27093
AMD Ryzen 9 3950X              16      3.5      1406   26375
Intel i9 10980XE                      18      3.0      1063   25490
AMD Ryzen 9 5900X              12      3.7       1670  22046
AMD Threadripper 1950X      16      3.4       1027  19635
AMD Threadripper 2950X      16      3.5       1135  18797
AMD Ryzen 9 3900X              12      3.8       1312  18682
Intel i9 10900K                     10      3.7       1415  18034
Intel i9 9960X                         16      3.1       1075  17953
AMD Threadripper 1920X       12      3.5       1054  15038
AMD Ryzen 7 5800X                 8      3.8       1596  14812
Intel i9 9920X                               12      3.5       1067  14793
Intel i9 9900X                               10      3.5        1182  13994
AMD Ryzen 7 3800X                 8      3.9        1346  13848
Intel i7 10700K                         8      3.8        1345  13302
Intel i9 9900K                           8      3.6        1343  12470
AMD Ryzen 7 3700X                 8      3.6        1345   12195
AMD Ryzen 5 5600X                 6      3.7        1593   11201
AMD Ryzen 7 2700X                   8      3.7     1102   10140
AMD Ryzen 5 3600X                 6      3.8        1323     9526
Intel i7 9700K                           8      3.6     1285     9428
AMD Ryzen 5 3600                 6      3.6        1245     9073
AMD Threadripper 1900X         8      3.8      1005    8979
AMD Ryzen 5 2600X                 6      3.6         1094    7523
AMD Ryzen 3 3300X                 4      3.8         1299    6787
Intel i5 9600K                            6    3.7         1196    6596    
AMD Ryzen 3 3100                 4      3.6         1105    5423

https://www.tomshardware.com/news/amd-patent-shows-CPU-FPGA-integration

Stock based mergers can move very quickly as no funds are involved so government approvals are minimal.   AMD and Xylinx are cooking up their first co-joined processors and are pre-announcing them through hard PATENT applications.

AMD has been doing wonders in the CPU industry, with its well-received and high performing Zen 2 and Zen 3 based processors, but now it seems AMD wants to improve performance not just through faster cores, but through the use of FPGAs. Just a few days ago, AMD filed a patent for integrating FPGAs into a CPU, which would allow the processor to run custom instruction sets to extend the its capabilities. As a side note, this patent was made just a few months after AMD's acquisition of Xilinx, a company dedicated to making FPGAs.

FPGAs, or Field Programmable Gate Arrays, are simple yet powerful devices that can run specific instruction sets very quickly. This is different from a standard x86 CPU core that's designed to run a near-infinite variety of instruction sequences, albeit sometimes slowly. If there's a specific task (graphics, physics, encryption, etc.) that's used regularly, it might be beneficial to create a custom instruction on an FPGA that will process the code much more quickly. Plus, FPGAs aren't limited to a single instruction set; they can be re-programmed to run another instruction set if necessary.

This seems to be what AMD is going for, and AMD's implementation would allow the FPGA unit to share registers with the CPU itself. Simply put, this allows the CPU to very quickly offload instructions to the FPGA unit when necessary. We don't know what specific tasks AMD is looking at, but presumably anything currently using dedicated FPGAs or AI accelerators could see instant support.

We also don't know where this FPGA (or FPGAs) would be located. If we're talking about a Zen 2 or Zen 3 based design, the FPGA could be installed on its own separate die (chiplet) connected via the infinity fabric. Alternatively, it could be integrated directly onto the CPU chiplet, sharing a die layer with the cores. This would be the most optimal setup as far as performance goes, but it would require new compute chiplets to be designed.

Aiming for a 2-3 year implementation time frame, AMD intends to preempt anything Intel can come up with by having the ability to CHANGE their chiplets on the fly.

Be afraid Intel, be very afraid .......    

https://www.pcgamer.com/intels-11th-gen-desktop-cpus-could-be-split-between-t...

Buyer beware,  rumor has it that not all of Intel's next-generation desktop CPUs will be based on the newly announced architecture --- yeah that one, the one that is supposed to deliver a double-digit gain in IPC (instructions per clock) performance and all sorts of other goodies.
They will, however, all be confusingly labeled as 11th Gen Core parts.

According to Expreview, however, Rocket Lake-S will be reserved for Intel's 11th Gen Core i9, Core i7, and Core i5 SKUs. Meanwhile, Core i3 SKUs will be a refresh of Comet Lake-S (as will upcoming Pentium and Celeron models), and will not support PCIe 4.0.

Comet Lake Refresh CPUs will also lack Xe graphics. Instead, some models will be paired with Intel's UHD Graphics 630 solution with 24 execution units, while others will lean on the slower UHD Graphics 610 solution with 12 execution units.



Smells like another BS game filled product introduction series from our buddies at Intel ........  
Lots and lots of acetone wiped retreads of old processors being called "Gen 11" just to confuse ordinary folks who are buying PCs just by reading the boxes.

"Hey Charlie, which one did you get, the good one or last year's retread with the slow processor and the last year's non-XE graphics ???"

Hey, they never actually said ---- I definitely ordered the good one but now Intel says you can't test or benchmark their chipsets so I guess I'll never really know for sure what I got.
I should've have ordered an AMD instead.



The first ARM designed 8 core PC chipset shipped today, in China.  This motherboard is running a Chinese Linux version as its only option.   Rumor is that chip was produced in a state owned production facility using some end of life TSMC lithography equipment, and it was run at the original 7nm.   Same set up has already produced and shipped a 24 core ARM design for the Chinese Military.


 

This week at CES

Intel  

No specific facts, no specific announcements, no specific claims of industry leadership.

Intel claims that they have the best most cohesive set of hardware and software to give the best solutions to business users.

They imply they have the "best hardware performance", but actually will only say they have the "best most cohesive set of hardware and software".  

They never say how their stuff works to give you this"most cohesive software/hardware combination" --- but what they showed the computer analysts in private was some hidden AI speeding up a carefully selected set of benchmark tasks.

So, the Intel's secret press presentation is just more Intel BS Marketing, more misleading smoke and mirrors that carefully dances short of any real easily caught lies.   Intel is wary of the class action lawyers as Intel has lost a lot of money in class action legal actions last year and now have EU and USA regulators looking hard at their advertising practices.

Remember, nothing that Intel "promises to show you" is one bit more than AMD can really actually do for you, for totally real, right now.


https://liliputing.com/2021/01/intel-launches-11th-gen-core-h35-chips-for-gam...

Brad Linder drew a few extra bits of information from the Intel Keynote address and even more from the private "invitation only" Intel show-and-tell afterwards.  

Intel says these new 11th-gen Core H35 series processors are 10nm processors designed for notebooks with discrete graphics, but they also included the company’s latest integrated graphics technology for up to twice the graphics performance of a 10th-gen Core H “Comet Lake” processor.

The new chips also offer up to 15-percent better CPU performance than a Comet Lake-H chip, and more than 40-percent CPU performance than a 15-watt Tiger Lake-U processor like those that have powered most Intel-powered thin and light laptops to ship in recent months.

So far there are 12 new members of the 11th-gen Core H family, topping out at the Core i7-11375H quad-core, eight-thread processor with support for single-core speeds up to 5 GHz.

Intel has provided specs for a few of the new chips:

Name      Cores / Threads      TDP      Max Turbo 1-core / 2-core / 4-core   Cache      Memory
Core i7-11375H   4 / 8      28 – 35W   5 GHz / 4.8 GHz / 4.3 GHz      12MB DDR4-3200  LPDDR4x-4266
Core i7-11370H  4 / 8      12-28W      4.8 GHz / 4.8 GHz / 4.3 GHz      12MB  DDR4-3200 LPDDR4x-4266
Core i5-11300H   4 / 8      12-28W      4.4 GHz / 4.4 GHz / 4 GHz      8MB    DDR4-3200 LPDDR4x-4266


But later this year Intel says it will bring desktop-class performance to an 11th-gen mobile H-series processor code-named Tiger Lake H. It’ll be an octa-core processor capable of hitting 5G Hz speeds on multiple cores simultaneously and it’ll support 20 lane PCIe for up to 40 Gbps throughput.

What’s the difference between Tiger Lake-H35 and Tiger Lake-H? At least 10 watts of power. It’s likely that the chips launching later this year will run at up to 45 watts.   (thermally suicidal high power draws for a standard laptop chassis)

Other features for Tiger Lake-H35 include support for WiFi 6 and 6E, PCIe Gen 4 x4, and DDR4-3200 or LPDDR4x-4266 memory.


==================================================


OK, it is a full day later and Intel has had to go back to its fans and EXPLAIN to everybody that VPRO was the big announcement that they made for CES this time around.

VPRO means a very mild hardware bump (yes, we heard that part) that is "effectively magnified" by background software activities (yes, we heard about the AI working in the background to cheat on the benchmark performance of your chipset).

But this is not CES level "progress", it is simply some more BS Intel Marketing tricks again.  

Intel, you SHOULD have multiple government efforts (USA and EU regulators at multiple levels) looking hard at you for defrauding your customer base if you have to go back to tell  that customer base that some very mild BS Intel Marketing cheating was the sum of your progress for this year.

"But later this year Intel says it will bring desktop-class performance to an 11th-gen mobile H-series processor code-named Tiger Lake H. It’ll be an octa-core processor capable of hitting 5G Hz speeds on multiple cores simultaneously and it’ll support 20 lane PCIe for up to 40 Gbps throughput."   And this is a vague bogus sounding future roadmap claim, not reporting hard progress of any kind.  You have been spewing these sorts of roadmap claims all along and they simply get replaced by more roadmap claims as you can never actually get around to doing them.  

So this sort of stuff from Intel gets routinely ignored by everybody now-a-days when you spew it at us.