/tech/ - Technology★

it must be pretty fucking bad.

They can talk a load of shit.

It's impossible to enforce that.

Oooooooh, It feels so good to have custom BIOS without any microcock whatsoever.

>>959262

>>>/g/ is that way if you want to masturbate loudly and act a dumbass, pal. All CPUs have microcode these days. Now gtfo.

>>959268

>I run x86 guise look at me!

(you)

>>959268

You literally can exlude cpu microcode before compiling coreboot and libreboot you retarded nigger.

>>959268

yes but you can keep using the version that is on the cpu if you dont want the shitty updates

>>959293

The CPU still has microcode, all you're doing is not providing a newer version of the microcode. That's what already happens by doing nothing you doublenigger.

>>959271

I was referring to mainstream desktop computing. You're right though. There are a lot of different CPU architectures.

>>959298

>security updates

>shitty

>>>/g/

Intel has reversed the decision after Debian and Gentoo refused to host it.

Redhat also ignored the license change and posted benchmarks anyway, for VMs the performance hit can be as big as 30%.

>>959568

>30%

Absolutely Depreciated.

>>959243

This is what I wonder. How can they enforce it? Do they have some kind of beacon that phones home in the case someone runs benchmarking tools etc.

>>959568

More information:

http://archive.fo/Dx1I9

https://www.tomshardware.com/news/intel-cpu-microcode-benchmark-mitigation,37684.html

The brand new licensing agreement:

Redistribution and use in binary form, without modification, are permitted, provided that the following conditions are met:
Redistributions must reproduce the above copyright notice and the following disclaimer in the documentation and/or other materials provided with the distribution.
Neither the name of Intel Corporation nor the names of its suppliers may be used to endorse or promote products derived from this software without specific prior written permission.
No reverse engineering, decompilation, or disassembly of this software is permitted.
“Binary form” includes any format that is commonly used for electronic conveyance that is a reversible, bit-exact translation of binary representation to ASCII or ISO text, for example “uuencode.”

>>959240 (OP)

I hope novideo makes their own CPU capable of 30fps raytracing at 1080p for just $1200 so we have a future where Intel has some economic pressure to stop hiring pajeets.

So what new backdoors will the revised firmware contain? I’ve a strong suspicion that caching and hyper threading have to be substantially revised to be safe and only OpenBSD is being proactive with security, as usual.

>>959592

>I hope novideo makes their own CPU

t. someone who has never had the displeasure of doing literally anything with the Tegra SoCs

>>959592

>nvidia CPUs

>full of minor vulnerabilities that get coincidentally discovered right before the release of the next generation of chips

>microcode update "mitigations" make certain features run on x87 in order to convince the goyim to buy a new, faster CPU

>on top of false advertising regarding L3 cache size and others

>Limited to Windows 10, refuses to run AMD drivers

I can't wait.

>>959568

>30%

That was literally how much they led AMD by

>>959635

I wonder if Intel is doing like MySQL used to and win benchmarks by knowingly having not implemented proper safety.

>>959242

Most workloads show a 1/3rd drop in performance, databases will show 2/3rds loss in performance.

A modern, mitigated Intel processor will end up being about the same speed as a 2006-2008 era unit. It's going to be a death blow.

Buy AMD if you missed out when it hit $7 and $10 and $15 and $20 and $25, it will be at $100 soon.

>>959592

They wanted to, but the licensing hell makes it basically impossible to legally implement modern x86 if you're not Intel or AMD. So they're stuck with ARM and therefore mostly embedded systems.

>>959627

Faggot, I'm writing this on a Tegra laptop.

> Another issue is whether the customer should install the fix at all. Many computer users don’t allow outside or unprivileged users to run on their CPUs the way a cloud or hosting company does. For them, these side-channel and timing attacks are mostly irrelevant, and the slowdown incurred by installing the fix is unnecessary.

This guy doesn't even know that the whole "Web 2.0" meme rests on allowing untrusted users to run untrusted code on the user's machine in the form of Javascript.

>>959268

>what is ARM

the absolute state of /tech/

>>959701

He knows, he's just a paid liar.

>>959701

A lot of the CPU exploits can't be done via javascript. Luckily activex is dead.

>>959691

>Faggot, I'm writing this on a Tegra laptop.

Shit, I forgot that a few laptops got made with TK1s. I was more referring to the TX1/TX2 SoCs/SoMs.

>have to use a special version of Ubuntu modified by Nvidia which is hardly ever updated

>if you want to use anything different the GPU portion of the silicon basically becomes worthless

>shit I/O despite people asking for better PCIe/SATA/USB/Ethernet/etc since the TK1

>the ethernet controller is actually behind the USB3 controller (which is only initialized during Linux boot) so get fucked if you want to network boot it

>you wanted more memory bandwidth? too bad

>>959701

>This guy doesn't even know that the whole "Web 2.0" meme rests on allowing untrusted users to run untrusted code on the user's machine in the form of Javascript.

Javascript is yesterdays news old man, all the cool and hip Web 3.0 kids are adopting WebAssembly.

>>959240 (OP)

> WebAssembly

< A little slow on my "web technoloies"

< Google WebAssembly

< tl;dr: NuJava run by the browser.

>>959767

Meant to reply to >>959760

>>959767

It runs closer to the CPU than JS though which is the worst part.

>>959417

>buys CPUs that are capable of having security vulnerabilities

>a literal arithmetic device

>with security vulnerabilities

>get your daily blackbox CPU level patch goy its for your security XDDD

>>>/techpol/

>>960555

Goddamn! They're fucked.

>>960241

That board is deader than dead but less so is:

>>>/poltech/

>>960555

Since you are too lazy to post the source

https://www.phoronix.com/scan.php?page=article&item=l1tf-foreshadow-xeon&num=1

One thing to note is that for bare metal systems you probably don't need to use the full mitigation but for things like cloud providers where they are running lots of untrusted VMs they need to use the full mitigation. Enterprise customers are Intels bread and butter and they are probably fucking pissed at the massive decrease in performance they have to take in order for their services to be secure.

>>959767

It's actually not bad, it's just a subset of LLVM IR based on an imaginary le32 architecture to make it portable. It frees the web from having to use javascript. We're finally starting to return to the thin client web that we were supposed to have 20 years ago.

>>959580

If you bench mark it they won't buy advertising with you any more or offer you early access to products for day one reviews like your rivals have.

What is the fucking total at right now. Has someone added up all these bugs and their fix impact? I feel like it's either 50% or over at this point with all the fixes combined in the last few years. What a fucking mess.

>>959243

my toughts exactly, no matter what the license says, i own the hardware, i get to do whatever i want with it, intel doesn't lend me my cpu no matter how hard it wants me to believe that

fuck. why i didn't listen to the AMD people. I will always buy AMD from now on