Alder Lake: Fedora 35 and the 12900k
Vložit
- čas přidán 3. 11. 2021
- Wendell checks out Alder Lake on Fedora 35 and gives us the rundown on the Linux experience.
Check out the main channel vid here for more: • Alder Lake 12900k and ...
**********************************
Thanks for watching our videos! If you want more, check us out online at the following places:
+ Website: level1techs.com/
+ Forums: forum.level1techs.com/
+ Store: store.level1techs.com/
+ Patreon: / level1
+ L1 Twitter: / level1techs
+ L1 Facebook: / level1techs
+ Wendell Twitter: / tekwendell
+ Ryan Twitter: / pgpryan
+ Krista Twitter: / kreestuh
+ Business Inquiries/Brand Integrations: Queries@level1techs.com
IMPORTANT Any email lacking “level1techs.com” should be ignored and immediately reported to Queries@level1techs.com.
-----------------------------------------------------------------------------------------------------------
Intro and Outro Music By: Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 3.0 License
creativecommons.org/licenses/b... - Věda a technologie
Efficiency cores on a 200+ Watt CPU is… interesting.
LTT just confirmed thats on specific workloads, gaming is much lower
It's not really about power efficiency, it's about die space. Intel could not fit 16 cores on a CPU and sell it for a reasonable price otherwise.
I was feeling positive about big.LITTLE on Intel x86, but now I've seen some reviews, I wonder how much it's about increasing core count without needing dual PSUs and custom loops.
Those "P" cores are insanely power hungry.
It's die area efficient.
I would think it is also market hype to claim high core counts, they did crap like this back in early 2000s when amd cpus had more performance, but clock speed was all the rage and Intel just made high clock but lower performance cpus. and the dumb people bought into it.
@Wendell
Speaking of virtualized workloads.. I have recently discovered a Linux project called ACRN (pronounced: acorn) which is an ultra-leightweight hypervisor tailored to virtualization on edge and IoT devices.
This project exists for several years and literally nobody is talking about it. It takes some inspiration from Xen running an equivalent to its Dom0 for management. The spin of this project is you are practically reading information from your target motherboard, then build a highly customized kernel for it.
Pretty sure this would make a really great topic for you to cover.
Thanks for bringing this up, never knew about it
More info please
Where can I learn more about it 😳🤔🧐
Always loved that LEGO stand. Timeless makeshift utility.
Thanks for making this video :) I am going to try being a linux gamer soon so this video is very helpful for me :) I look forward to the three week followup.
Can't wait for part 2.
Hey Wendell
Since it has been more than a year, could you give us a small update in the comments on how does i9 12900k perform on latest fedora say v38
This is the exact video I needed. While my Debian desktop runs perfectly with the R5 3600, my hedonism really wanted to try the new Alder Lake shinies. But if this much work still needs to be done (especially in terms of scheduling and kernel updates), I might just wait until next gen before any upgrades.
Let's just hope 13 isn't an unlucky number.
I would not suggest to run Debian with Alder lake you would like the newest kernel.
Thanks Wendell
Linux would probably need very little work to get the scheduling to work if they just pulled classic big.LITTLE, but instead they have their HW scheduler magic because they couldn't be arsed to make Windows's scheduler less of a joke...
What is the difference between big.little and alder lake that would prevent this?
Are any of the ARM schedulers non-proprietary? Do they even have one that doesn't zucc?
Smartphone benchmarks tell a different picture.
UHD 770 iGPU performance would have been interesting.
But I guess the FPS increase linearly with frequency, meaning 20% compared to UHD 750.
Great breakdown!
would definitely like to see what single thread performance is like in linux compared to other processors.
It's remarkable how bad cpu's have become at idling over the last few years. Remember when intel actually cared about efficiency, Ivy bridge?
My 3570k idled at 30w on the igpu, and it was overclocked at a fixed frequency!
Edit: The entire system pulled 30w from the wall on an 80 plus white 450w psu.
im still on the 5960x here and i love it.
@@theSFCchannel That's a nice one.
to be fair, an i3-9100f or newer is on par on performance with 3-4th gen i7's. and it draws like 65watts when maxed out..
@@satyarsh665 Did you test it with a killawatt?
Not sure that I follow your complaint; I have overclocked 10700k systems that pull 20w from the wall at idle. Intel desktop CPUs have done excellent at idle for years now, typically pulling all the way back to single digits in watts. AMD CPUs on the other hand do struggle with idle power consumption while excelling under full load.
People keep on saying it's the first time in history that a x86 hybrid architecture was used but I remember Intel had a mobile CPU that they had 1 normal big core and 4 Atom cores (Lakefied). Just for correctness' sake.
I’m thinking after a little maturity the alder lake 12900k could make a great UnRaid CPU where the e-cores are managing pihole, UNMS, Sonarr, lidarr, radarr, Plex, etc. and P-cores can go to 1 or 2 gaming VMs. Question does become would it still be better to run Threadripper?
I wonder how this type of chip would do with custom CPU schedulers, like those in the TKG kernel. I'm sure those schedulers need updating too but maybe they would do a bit better in some aspect?
Linux on ARM has probably done a lot of the heavy lifting for big+little cores on x86
Ohh thank you I’m using the same motherboard I was thinking I was crazy not researching more about what to use with proxmox and then truenas but I’m thinking maybe I just need to wait for the drivers to catch up I went to as close to next gen as I could and I feel I got it all at a decent price till I saw nvme ssds seemed to drop $30 a few weeks after I bought everything and currently testing with unraid and unraid is really easy but might fall back to proxmox and just wait for everything to catch up
Thanks friend
Nice to see Intel actually innovate! It should be fun to see not the next generation AMD CPUs, but the ones after that. Finally some real competition and evolution in x86 CPU land!!
x86 is not a optimal base to start with. This ISA is antique and has many milestones around the neck which block innovation. As one of many workarounds intel uses internally a RISC core with a wrapped x86 translation layer on it, which has more drawbacks than intel is ready to admit. X86 is only kept by intel because it's a patent protected cash cow to them.
Also the P+E cores idea has been copied by intel from the LITTLE.big concept.
Many people believe that intel's alder lake was a try against AMD. But in truth intel tried to win back apple as a customer and to compete against apple's M1 iterations. Intel failed horribly when taking the important performance per watts factor into account.
AMD zen5+4D will completely destroy then intel 12 CPUs
@@JR-uy2nd If AMD manages to do so then very probably also by increasing the power consumption dramatically. I highly doubt this to be a good idea. We can observe a similar development in the latest GPU fights, where cancerous GPUs consume several hundred watts of power with a big heat output.
Giant heat coolers are one of the worst atrocities in the history of computers.
@@aladdin8623 "x86 is antique" lmao, wait until you find out when ARM got started...
@@dial2616 You seem to err, that i didn't know about ARM's history. I know. But you don't seem to know, that in comparison to x86, there were a lot more changes in arm isa versions an extensions to the point, that you need new software due to incompatibilities to prior arm versions. X86 on the other hand drags along a whole bunch of outdated and antique extensions and commands.
I'm still pulling my hair out over every single scheduler I've tried (CFS, MuQSS, PDS, UPDS, cacule) switches my game threads between CCDs... and not just every now and then but literally every second. The number of cache misses is absolutely incredible because of this. I get ~10% better gaming performance if I just disable my second CCD on my 5950X
Were you able to turn on NUMA on Ryzen? I can't remember. I hope someone hooks up the cluster-aware scheduling soon either way.
You should not cross CCD's boundaries except if you want Numa enabled VM, that's special setup of KVM. Also you have to pass additional kernel boot options in order to mitigate the use of specific cores. There is a good guide o. Arch Linux to do so.
How about you increase the time quants? Maybe it's a good trade.
Enabling automatic (load-based) core-parking on Windows for 8 of the cores helps significantly with this with moderate and light tasks in my experience. It works by putting the other CCX to sleep when the extra processing power isn't needed, thus stopping this costly CCX-hopping. Perhaps something like this could help?
Can you use cgroups to force it to a single CCD?
Just curious if you could someday do an in-depth video on using Fedora Rawhide. Explaining all the pros and mostly cons of using Rawhide. I use Rawhide and love it. But from my perspective, if it crashes, it does not
affect me, as I just reinstall it. I do not use it enough though to say how often it crashes, and primarily just use
Gnome, so do not know if it crashes more frequently on the Fedora alternative DE's. Rawhide does have all the debugging stuff running in the background. I am not familiar with all that kind of stuff.
Choose this as my first video to watch on the new Intel chips. Don't run Linux but just love your faces
@7m0s The L1 cache bandwidth numbers measured by Aida64 are invalid.
Thank you
When you are doing the revisit, please try out pinning for only P cores.
O boy O boy O boy O boy!!
Wendells back!
Im curious on how you updated the uefi. I havent been able to do that for a few years now since all mybsystems run Linux
The bioses have a built.in updater the past 5-6 years+ so if you get into your existing uefi you can bootstrap a new version
Usually just download the BIOS, put it on a FAT partition/FAT USB stick and then from the BIOS just go to the updater and select the file and of you go.
Thank you so much for this video for my needs at the current time 11th Gen is better suited for my needs.
Agree with you on the desktop, dont care or want E Cores, those would be disabled, and then you have no more scheduler problems, but get that juicy high ipc
How does this High Performance / High Efficiency design compare to the mobile Big / Little designs? I would have assumed the Linux schedulers used for these ARM devices would be applicable to Alder Lake, but apparently they don't?
Where can I get one of those big die layouts? Would love to hang something like that on my wall.
Right now there's no scheduler patch in the horizon for 5.16 kernel so as a Linux user I would wait until Intel releases kernel patches that will make the scheduler work with the new Alder Lake CPU correctly.
Let's get those CPUs to 1500W and it can replace a space heater.
Are you cold?
Go recompile your gentoo system.
Still cold?
Better fire up that 3090 and play some games.
But I only get 5 months of heating season, better make it a sauna for next 7. 🥵😵
That's the spirit!
Should I use the z690 hardware raid or use the Linux software zfs? Or combo?
What is a good CPU from intel to use as a server, vm, containers, storage use?
2 videos in the same day, dam
Have you been able to check out if the ECC of the DDR5 works?
Very cool
If you run your system as a dedicated VFIO backend, you could, theoretically, reserve all of the efficiency cores for CLI only Linux base OS operations, and then grant all of your performance cores/threads to a QEMU Windows 10 VM with GPU passthrough for gaming. You could also use the base Linux system as a NAS, accessible to the Windows VM and other devices on the network. Correct me if I'm wrong.
You could, but that would ignore the fact that Windows also have background processes to run of its own. Those should be scheduled to e-cores, if you don't want to leave performance on the table.
@@EtienneMaheu not if I'm running Windows 10 AME. I'll never use Windows 11. It almost spies on users as much as US ISP's.
@@needsLITHIUM in that case, you'll be leaving performance on the table no matter what you do. Linux doesn't need 8 background threads to run a hypervisor so some of those e-cores will remain idle. On the other side, anything but windows 11 cannot differentiate between e and p cores, so assigning even 1 e-core to the vm will cause massive performance and even stability issues due to the different core capabilities and the scheduler might decide to move a high performance load on the e-core.
@@EtienneMaheu I wouldn't do this with an i9 for that reason. You don't need an i9 to play video games that are incompatible with Proton or do pro audio production in a DAW. Also, using those extra e-cores for other things - like running a nested NAS for backups and a firewall, both exclusively for the VM, for example - would handily make use of those cores and not bother the rest of the system by using "too much." Linux native would be fine for most else. I would use separate users as separation for the Desktop and VM host, and configure the core affinity accordingly for each user.
if you have a server, could you pass trough the p-cores to a VM and e-cores for something like docker?
Yes core pinning works fine. I wish it worked better without having to pin though
150W idle? Kinda disappointed in Intel they didn't prepare patches in time. Hopefully it will change fairly soon.
Hell my Debian desktop is running an FX8320 slightly undervolted to manage heat while minimizing fan noise, with DDR3 at 1333 and it seems snappy to me. For the .01% of tasks would really benefit from more CPU meh, or for huge tasks I'll go rent some CPU time.
Have Intel announced a tuned version of their Clear Linux for these new CPUs?
Likely not, but Clear Linux is Intel's test ground, they likely already have the patches for the HW in their Kernel (which is out-of-tree anyway). The question here is.... Will they upstream them? As far as I understand it, baring any of the core kernel team's objections, Intel does tend to do this. Just don't expect everything.
fun fact: intel vmd is for some reason available for my tigerlake laptop, despite only having 1 nvme slot
Why 3 weeks rather than when 5.16 lands for followup, given that has a bunch of Intel patches due for Alderlake (though I don't recall off the top of head how applicable for this stuff)?
EDIT: Apparently there's nothing in the pipe for 5.16 regarding core/thread handling. 😕
@3:02 - I have the exact same errors since I put in my new Samsung NVME M.2 980...let's see what you say about this...
On the question of "why e-cores?": For a first stab at this setup, this seems good.... BUT, that wattage does seem kind of high for a i5 and we have all seen what big.LITTLE has done in energy constrained environments. I don't think that Intel is done with tweaking the power profile of their hardware from this point out with the next generation, in-fact I think that it will become more important to do more with less wattage when it come to HPC. Their are already ARM efforts to stab at the data centre market, which is the money bin for Intel, AMD, etc. I don't think that they are just going to concede that, not without a brawl in the streets.
What's wrong with using taskset to tell a process which cores it is allowed to run on ?
actually we do care about efficiency in desktops: California (& other West Coast states IIRC) are enabling legislation that limits the amount of power electrical equipment can consume, particularly in idle mode.
Is it possible to buy a die layout like yours?
I can see it. 8 little cores is more than enough to run your ide smoothly. Add in 8 big cores to run your code on? Glorious. I'm basically doing that now by sending code to my home server from my laptop.
I had the opposite reaction. Why always big cores? Smaller cores=more cores for less money. I would love to see a chip with 40 E cores for a home server. The E cores' performance is nothing to sneeze at. Imagine 40 cores during the skylake generation! Something like that could disrupt the market! (threadripper) Even on a hybrid BIG.little design like Alder Lake without intel's hardware thread scheduler working, having all those extra cores to breakout into non-time-sensitive/security segregated tasks while maintaining a high speed desktop could be a boon. I mean how am I supposed to feed three thin clients, run a home NAS/router, seed bittorrent and LBRY and have a superb gaming experience without ALL THE CORES!
There is, on server side market, dual socket, ant it's outstanding.. How low is your cpu consumption since 78 cores out of 80 are sitting iddle 99% of the time!
ARM phones have big.LITTLE do they run on special linux kernels? Or how is this solved on those?
Very different architecture, based on similar design philosophy. The fix for these will probably not be found from arm.
Exactly what I wanted to know ever since Intel announced this mixed model!
Why do you run Fedora and not Opensuse Tumbleweed?
I hear that Intel has their own version of linux for their chips: ClearLinux, or something like that. I wonder if that distro would be better at handling the new chip. I mean, I would hope so.
It does. Phoronix did a test about that with several distros.
engagement challenge for Linux goodness
Kinda surprising that there wouldn't be a good Linux scheduler for heterogeneous core sizes. Android is just a locked down version of Linux for ARM mobile cores, and ARM has had big/little architecture for a while, hasn't it? Or do those Android schedulers not work on x64 because Intel implemented big/little in a bizarre non-compatible way?
As a man who pays for his own power bill... I want efficiency in a desktop XD
Hmmm, what you said about single core performance is not true at all.
If you are not worried about efficiency just because it plugs into the wall... you draw half the power I use with a single machine. That is pretty taxing for the environment.
twice the power*?
The best reviews on CZcams!
This is a special cpu for Windows 11 to show off potential for future performance. Clearly aim at M1 pro and Max. But at a big energy use hit.
So, Linux' schedulers don't deal with these different types of core well, (yet).
No it already deals with it on arm platform.
@@goldnoob6191 well, it doesn't seem to deal with different types on x86 well yet
only because one architecture does it, doesn't mean, that it's portable to others since they may work too differently in this aspect
@@kuhluhOG no of course, I only mean the higher level kernel is ready for that and for sure the driver is not ready and the scheduler is not yet optimized but almost half this work belongs to the driver.
I assume It will be there pretty soon they did a tremendous work for M1 SoC.
But don't be too exited, when it comes to performance the problems in VMs indeed no one should let the kernel manage your cores.. And several other attention should be taken into account, one of them is huge pages, the other one already take place concerning FS segment register or something like that.
And I don't talk to those having precompiled binaries!
@@goldnoob6191 "And I don't talk to those having precompiled binaries!"
you do know that this includes every Linux distro except Linux from Scratch and Gentoo, don't you?
0:30 That's not completely true. I mean there was Lakefield. But does that count?
maybe a dumb question but if linux already runs on arm with its big little cores, why doesn't it work for these intel chips?
The problem is not with the algorithms, but with the data those algorithms need to work well and the tuning. That's different for every architecture.
Im wondering why Linux has not an optimized scheduler for little.big architectures yet, because all those android devices have such configurations now for years
I can’t read those bar graphs on my phone.
finally some body looking at 12th gen and Linux will Intel big/small make Linux faster? Anyway I still waiting on Zen4 APU's until then I can stay on 2200G which for a Linux desktop run well. My windows rig is unsupported by Windows 11 and I won't to focus more on Fedora Linux my daily driver. Windows is only used for DJ gear and Gaming. All my coding and work is on Linux now and I am not going to switch back as Windows sucks. Which is why I feel these CPU are trash for me.
kinda sad but expected for a new architecture
Homogeneous architecture? IBM XT/AT/PS2/compatibles were definitively a heterogenous compute platform on disaggregate processing system in package before Intel gulped it all up. mb
And this is a Linux site? mb
I have a job for u Wendell, write and contribute (no wonder there is a problem) the Linux game transcode and supporting drivers. mb
Meanwhile, no Thread Director support even on 6.2.
16 PCIe lanes from the CPU is kinda meh, Intel really should go with at least 20 so the GPU and NVMe are directly connected.
Nvme #1 is on CPU, has 4xpcie4
@@charlese2833 So it has 20 PCIe lanes from CPU, Ok
Intel, Please 2P cores and 8E cores in an i3
what would be the advantage of that Combination ?
I would probably prefer 4p+2e.
@@tanmaypanadi1414 Pretty much anything other than gaming. Also, what you're talking about would be larger die than the suggested 2P+8E, might as well have 2P+12E for that die space
Intel has stated that 4 E cores offer more performance for the die space compared to P cores, but a 40E core 12900K with a PL2 of 120w wouldnt be that useful for gaming because only a few games today can use 16 of those 40 cores, i've only come across 1-2 games that can use all 32 threads of my 3950x. 40E cores would be faster in blender, and pretty much everything else, but not alot of gamers want that limitation on games.
Which is why i ask for an i3 with 2+8 because most gamers dont use i3, like they've already shown this will be available for mobile, just please let it be an option on the desktop. For enterprise desktops i'd almost want 0P+16E
Now that we're talking about it, i would like a 12830E with a 2+32 or 0+40 configuration for my virtualization/container server.
@@tanmaypanadi1414 E cores are great for multitasking(which is different from multi-threading) multi tasking, with many programs at once often results in cache misses, requring the moving of data from different tiers of cache, or from system memory. With many E cores instead of a few P cores, there will be less need to load, and unload programs from each core, meaning that the processor spends less time at idle waiting for data to move around. This is probably the sole reason four E cores offer more performance than one P cores. If the P core didnt have to swap programs in and out of cache, it would probably be faster, but because it is wasting time promoting and demoting things in its cache, the E cores are able to spend more time working on the 4 tasks at hand
Apart from the i5 Alder Lake chip overall I feel very underwhelmed by this release. Marginal performance increases achieved by massively increasing power consumption which doesn't bode well for laptop performance.
*Responsibility
Speak for yourself, I do care about efficiency on desktop. If I want to heat a room, I turn up the radiator instead of buying Intel CPUs.
this cpu was named after the lake i almost drowned in
Thanks for sharing. Glad you didn't go beyond almost.
@@gordongoodman8342 ty lol
Not to worry just buy now and btel will have everything workig with the next gen dropin CPUs
this may be awesome for laptops, but it seems like such a marketing stunt to call the i9 a 16 core when it is really an 8+8. To my mind I'd rather have a true 16 core like a Ryzen 5950x. The same can apply all the way down the line. I guess at the i5 12600k level you are getting the 6/12 big + 4 small cores. So you get the small cores for "free" compared to the Ryzen 5600x 6/12. They can keep your discord running, etc.
Unless the efficient cores can do something amazing well that the performance cores can't do, why not just have all performance? Power is not an issue in the desktop.
Why not make the i9 12900k a 12 core/24 thread? Would you not rather have 10-12 performance cores, rather than 8? Forget the efficiency cores when you can plug in.
This all seems designed to put a bigger number on a box just like the Mhz wars. Intel have decided to put more less powerful cores in the package, but pretend they are the same in the marketing. They are way smaller, but way slower. Just to get a bigger number in the advertisements.
It's got to be such a pain for the schedulers as well.
Phone manufacturers have been doing this for years
@@marcogenovesi8570 very true and obviously that is an environment where power usage and heat matters a great deal.
If I look at those die shots it seems that for the 8 efficiency cores on the 12900K you could fit in another 2 performance cores. I don't think that this makes sense as these efficiency cores are actually quite fast and I would guess they are especially fast at running the "easier" instructions and also it may not be a big deal if they have to wait for some data from main memory. (but I don't know that much about the architecture)... Also you can't really run a piece of silicon at stupidly high power output as it overheats eventually from the internal thermal resistance alone. And the 12900K at 250W is already drawing a lot of power and running pretty hot.
@@jellowiggler I meant the "using all cores in the name" like calling "octacore" a phone that is in fact a big.little with 4 cores big and 4 cores small, or even 2 big 2 medium and 4 small
Yeah the answer is pretty obvious. The E cores are really meant to be space efficient. 8+8 alder lake outperforms a hypothetical 10+0 which would use the same die area. Instead of 12+0 you could have 8+16 (ie raptor lake) which extends the gap further.
2 E cores roughly match a P core, but you can fit 4 in the same space.
Microsoft and Intel are in cahoots. That's for sure.
I thought Linux would have a huge advantage, cause of big.LITTLE, why is it not used?
i have a 12900k and i have NEVER been able to get a reliable install of any linux on it. i want gentoo but it is constantly crashing during builds. i put mint on there and its fine for... 15 to 20 minutes then it just STOPS. no cursor movement no led's blinking, no nothing, just DERAD.
never over clocked, never been reliable.
This is what I've been curious about! Thanks! LOL Not gonna lie, not really surprised here. It's "good" to have confirmation. It is disappointing though. I had really been hoping that Intel would be working more with Linux kernel devs to get ahead of the usual Linux support curve. I mean we all knew Microsoft was going to flub this, and will probably continue with a half-baked implementation for a long time. It's M$ after all. We know Linux is smarter and capable of great things ... if you don't mind the wait. I just had hoped, for once, the wait wasn't going to be quite so long.
Power usage doesn't matter? You try 100+F degrees in summer with a system burning 250W for the CPU and then whatever the 3090s crank along at. No thanks. And as the world in general is trending that way....hope you have good AC and enough cash to use it.
this CPU consumes as much power as 20 M1 chips
neat trick, but i just want plenty of uniform cores, thank you very much.
This isn't exactly new tech if you look at ARM's processors. When done correctly, you get two benefits:
1.) Running smaller tasks on these cores frees up your bigger ones for heavy workloads. No need for context switching and such.
2.) Smaller cores have great potential for reducing power consumption. Especially for notebooks this is a great use case.
excellent point re: virtualization. Intel are losing me on Alder Lake. I think I will stick to Rocket Lake (or earlier) or AMD processors that don't require a "Thread Director". Maybe it will all come good eventually, but I don't want to be beta testing Intel's latest brainwave for two or three generations before all the kinks are ironed out. Thanks for the great review!
Amd will snap their fingers, ryzen drops by $100 and easy.
Amd stock already up 27% in a month, too bad i cashed out at $105
AMD is inevitable
My big question is does the 12900k have the raw processing power to run a wide empire on a Modded Stellaris game in a huge galaxy?
There are no consumer ready quantum computers yet. :)
It's unlikely a linux user would prefer Intel RST over ZFS.
They can fit 40 E-cores into a 12900k die
Not the first time, Lakefield was heterogenous.
Lol Apple must really regret ditching Intel.
No? Lol
Stick to AMD, that's the future!
This Intel 12900k with efficiency cores is a dirty hack, and from now everybody has to suffer. [First impression]
200w is this a joke?
Who really wants the added headaches that this will cause? I think I'd prefer the simplicity of uniform cores.
Ah. Level1Linux, the red-hair headed child is back.
we do care about power usage on the desktop. electricity is not free. a few of these in a room and you need to turn the ac up. like cooking in the kitchen the stove heat ups the room. winter might be nice fore the extra heat. intel in too heavy. needs to get smaller like 0.05 nm and use no more than 45 watts on its xeon processor with 64 cores, and a terabyte of memory on the chip. and that would be xeon with graphic cores, called xeon EX for home/game/content creators pro consumers.
what is the hold up intel, lazy? hiring only kids that went to the rubbing elbows schools, get off the stick, or pot... hire some new creative don't know it can't be done folks that will make it work. good grief.
you should have never loss apple. whole divisions should be fired for that. hell i'm almost willing to vote to close the doors and turn off the lights on that loss. you've got 6 not seven months to demo a processor that uses a quarter of the power of apple's silicon and 10 times the performance in actual workloads, or start getting your collective resumes together. maybe apple will give you a job.