Intel (ME) Management Engine

File: riscv.png (10KB, 366x289px) Image search: [Google]

10KB, 366x289px

> is there any alternative...

soon

>>56196338
AMD:
>Early anecdotal reports indicate that AMD's boot guard counterpart will be used on most OEM hardware, disabled only on so-called "enthusiast" CPUs.
https://libreboot.org/faq/#amdpsp
Dunno.

There are alternatives using a few ARM SoCs. Performance of course is not great and graphics acceleration wouldn't work. Eventually, yeah, RISC-V looks promising.

File: bdff5e4df5710735d8a1ecdc5dd9efcf_XL.jpg (143KB, 750x627px) Image search: [Google]

143KB, 750x627px

>>56196935

this Fudzilla slide from 2015 shows that Zen will have the Platform Security Processor (PSP) - AMD's version of Intel's ME cancer

>>56196338
AMD has been just as shitty since 2013

But if you're going to be this paranoid, the only way to truly be safe is to buy your own fab and make your own chips.

>>56197244

or if you can't find $5B+ in your couch cushions, use a CPU with a open design:

>>56196403

>>56196935
Libreboot are tinfoil faggots that unironically follow any bullshit from FSF.

>>56196403
Long way, long way. IMHO, they will only catch on in embedded and microcontrollers. It doesn't what it takes to rise to general purpose user CPUs, it won't even get as popular as bloody Loongson.

File: patterson.jpg (93KB, 708x531px) Image search: [Google]

93KB, 708x531px

>>56197938
>Long way, long way

first hardware is expected next year (from the lowRISC project). yeah you won't be able to play Doom 7 on Winblows 11, but there's a lot of potential for a solid general purpose workloads (e.g. jerking off to traps on /b/)

also... Russia, India and China are into RISC-V hook like and sinker, because they have national security interests to consider and cannot trust backdoored CPUs from Jewtel and AMD

>>56198055
lowrisc will probably produce something resembling the OGD1 (google if you don't remember). The commercial users who are the only who have what it takes to popularize the arch enough, are only interested in using it as a free ucontroller instead of ARC/ARM/MIPS. See Nvidia.

>you won't be able to play Doom 7 on Winblows 11
Nicely memed, but current windows/games is not really even the target to watch here. They need to provide something like Cortex-A7 or A-53 first. And those are still shit - in-order cores et cetera. But at least they can support rich OS like desktop Linux.

>>56198535
>lowRISC project
Checked their slides now, and apparently they use Rocket cores. The custom microcontroller core that Nvidia made for running their embedded management is supposed to be higher performance than Rocket, which tells you something about the desktop-grade grunts of the Rocket. Expect something bellow Cortex-A5 in per-MHz performance and on lower clock IMHO, with immature software ecosystem holding it back as a bonus.

>>56197938
>Libreboot are tinfoil faggots that unironically follow any bullshit from FSF.
Did they rustle your hairy asshole, amdpajeet?

>>56198627
No, they are just retarded. wrt Intel too, actually.

>amdpajeet
Try harder, sonny.

>>56198535
>The commercial users who are the only who have what it takes to popularize the arch enough

it doesn't have to be "popular", it just has to be a viable alternative to the cancer Intel/AMD duopoly (like Linux/BSD are in the OS space)

commercial users may not give a fuck about the security of their customers (us) but they sure as fuck care about their own security

>>56197244
>But if you're going to be this paranoid, the only way to truly be safe is to buy your own fab and make your own chips.
Sorry to burst your bubble but that ain't gonna suffice.
http://www.ieee-security.org/TC/SP2016/papers/0824a018.pdf

>>56198577

no one is making claims that RISC-V is gonna come onto the scene as the performance king

i don't know shit about ISAs, but my guess is that RISC-V is much cleaner and more elegant than x86 and if some manufacturers decide to build big die chips on RISC-V they may outperform Intel/AMD in a few years

>>56196403
>>56197891
That's not gonna solve anything - who the fuck guarantees that the silicone that you'll buy and use matches the hardware design?
Without access to top-grade failure analysis equipment (costs millions), domain expertise & countless man-hours invested you won't be able to tell if the hardware has been backdoored.

>>56199084
>i don't know shit about ISAs
Should've stopped there pal. Intel, AMD, ARM, etc. invest billions of dollars per year in R&D to get better performance per watt & transistor count.
A free & open-source project with very limited funding isn't gonna get anywhere near the performance levels of a commercial core.

>>56199106
>That's not gonna solve anything

> google "define anything"

it solves the problem brought up by op. it does not solve the problem of backdoored silicon.

troll harder

>>56199153

>...R&D to get better performance per watt & transistor count.

at least i know what ISA means

>implying an NSA spy at your silicon fab isn't gonna put in a backdoor anyways
Give it up.

File: eoma-all-2-1_jpg_project-body.jpg (32KB, 749x480px) Image search: [Google]

32KB, 749x480px

>>56196338
There's a project to make a modular PC card thing that you can put in a laptop chassis or other devices. They're applying for FSF Respects Your Freedom certification, meaning no proprietary firmware and no ME. It's a crappy ARM chip, but of course if someone makes a better card later you can just replace it.

https://www.crowdsupply.com/eoma68/micro-desktop

>>56200934
You realize that Intel ME has access to the PCI bus anyways right?

>>56200959
How is that related? You mean the Intel ME on the main CPU in the laptop? No no, it's not a laptop, it's just a display and keyboard and touchpad, the computer is all in the card.

>>56201002
Oh I thought it was like a PCI PC card for laptops, never mind then.

>>56196338
POWER, PowerPC, SPARC, MIPS, etc.

>>56199084
>but my guess is that RISC-V is much cleaner and more elegant than x86

So what if it gives you a 5% boost? The transitional costs of moving to a completely new architecture and bearing the support burden for a wholy new ecosystem while pushing uphill against the incompatible software applications burden is going to kill you swiftly.

BTW, those 5 % are easily absorbed into the disadvantage you will have because your libc, kernel, multimedia software and whatever cpu-sensitite code will not be as optimized as x86 - or even ARM. So you probably won't have any advantage from your nicely clean architecture - maybe only after it is 10-15 years in the mainstream.

With the complexity of todays CPUs, after a certain size, the ISA doesn't matter anymore, only the microarchitecture ( = Haswell, Cortex-A57, Zen, etc) does. The only lasting impact is probably from code size, but in that parameter, x86 wins over RISC-V unless you use the special compressed mode which is speciality for embedded and not viable for mainstream.

>>56201281
>what advantage do you get by moving from x86 to RISC-V?
Lack of Intel ME / AMD-SP. Also the architecture is auditable down to the schematics.

>>56198577
>>56201281
Well, don't get me wrong, I do like the idea and it could be nice and exciting to have such CPUs.

It is just that I am very sceptical about the chances it has to break through - or at least to break through any time soon.

So don't get too excited, don't expect to be able to use it in next 10 years, and maybe be pleasantly surprised.

>>56201345
That is not a significant advantage really. What I was saying has to largely do with economics and practical constraints, while closedness and potential backdoor scare are to a large degree just theoretical and tinfoil matters.

This argument might appeal to some governments, but if such government would act on the sentiment and make its own RISC-V chips, it won't be anything useful or generally available. See the Russian Elbrus thingy.

BTW, if there was significant market for CPUs without PSP/Trustzone and IME built in, you could just convince x86 makers to cover it. It would be easy for them to issue such products.