INTEL CPU FLAW LEAKS CRYPTO KEYS TLBEED

And here we go again! Grab a seat grab some popcorn. I wonder how big the impact industry wide is going to be this time. Meltdown is the gift that just keeps giving. I wonder how much more bugs we're going to see this year. It's only june.

You won't believe how glad I am that I still have a couple of years of college left. This must be one insane year for people working in this field. It's just one shitfest after another all year long now.

Attached: nkqstlpdet801.jpg (1018x768, 87K)

2018 is even worse for intel, how's 2019 gonna be?

imdb.com/title/tt1476431/

>Still have my Sandy Bridge Toughbook
>Still waiting to replace it
>Anxiety slightly higher every day
I really wish OEMs would stop voluntarily participating in intel's monopoly.

>7nm Zen2 competing with 14nm++ Skylake re-re-re-re-release
It's going to be a fun year.

NONONONONONO

Attached: 1503848584264.jpg (1000x1000, 253K)

>You won't believe how glad I am that I still have a couple of years of college left. This must be one insane year for people working in this field. It's just one shitfest after another all year long now.
God damn I've been thinking about that a lot for high up system admins. There are a LOT of tough calls that have to be made in the last year regarding how to mitigated this shit on shared farms. I'm sure some of them feel like eating a bullet every time there is a new announcement... which seems to be happening like every month recently.

we haxx0r googl when?

Just another speculation.
Same shit happened with spectre.
Intel is vulnerable and we proved it.
Arm is vulnerable and we proved it.
Amd might be but we won't prove it.
Same shit, different vulnerability.

The taste of the last flop hasn't gone yet.
AMD has a severe security flaw in Zen.
What's wrong?
ASMedia forgot again to lock their private keys for their usb peripherals.

THE GHOST OF KRZANICH STILL HAUNTS INTEL
JUST FUUUUUUUCK MY SHIT UUUUUUUP OOOOOOOOOOO

AMD flatout lied about not being vulnerable to Spectre v2. I don't really trust them and I don't like the complete lack of pentesting that happens with their chips for some odd reason. Did you know that 10h/12h (Phenom II/Athlon II/OG Bulldozer) chips have a bit that can be set to entirely disable speculative execution that was introduced a good decade ago? It's like they (((knew))).

Kek

Attached: 1527749837990.png (300x250, 8K)

same, what a shame.

Feeling pretty happy about owning my 1090t and 2700x.

Still waiting on variant 3b and 4 mitigation via firmware. Intel still hasn't released it, despite it being production ready, per their own documents.

And now I have to disable hyperthreading and whatever the fuck AMD's equivalent is on my machines.

Fuck.

Server is no different from computer, just with more abstraction

Your 2700x is likely affected, although the security researchers kept that an open question.

"""Flaw"""

Attached: Naamloos-2.jpg (600x600, 40K)

b-b-but mommy su.. muh kikelake.... muh muh

Stop being Dutch.

>captcha is dutch street signs
>even google knows

i guess i have to wait 5 years before a secure cpu is released

Got a source for any of that, buddy?

Attached: 1525353398975.png (1066x600, 429K)

>the researchers claim to have a 98 percent success rate in recovering a 256-bit EdDSA encryption key after a single signing operation.
lmao
the absolute state of intel

Attached: 1529779667206.png (594x514, 140K)

>>the researchers claim to have a 98 percent success rate in recovering a 256-bit EdDSA encryption key after a single signing operation.
>lmao
>the absolute state of intel

>>>>>>>>>>>>>>>>>>>>>>>>>>>the researchers claim to have a 98 percent success rate in recovering a 256-bit EdDSA encryption key after a single signing operation.

Attached: 1280x720-pwI.jpg (1280x720, 140K)

>Theo de Raadt has stated that, while the attack could theoretically be prevented by preventing tasks with different security contexts from sharing physical cores

That isn't even a fix You're approximately one step from essentially having disabled hyper-threading at that point

What the fuck is wrong with these people

Attached: Hell.jpg (650x400, 102K)

Look at this Pajeet just lie.

>not having disabled it in BIOS already
What are you mentally slow? Enjoy getting hacked.

Attached: flat,800x800,070,f.u12.jpg (625x800, 84K)

>32 thread CentOS server just became a 16 core 16 thread server

Hold me, Jow Forums.

Attached: 1529037458844.png (645x773, 12K)

Buy AMD, it's as easy as 123

Attached: images (2).jpg (443x332, 25K)

Wonder how many times (((they))) used this side channel attack.

>mfw I don't own a chip with hyperthreading
i5 master race reporting in

Is this another one of those
>YOUR MACHINE IS FUCKED AND IN PERIL IF SOMEONE IS ABLE TO PLUG A USB DEVICE INTO YOUR COMPUTER
things?

These things don't mean shit without physical access, and most of them require physical access to your hardware. In such a case you're fucked either way. Unless you own a Mac because no one could even remove your SSD because the whole machine is the SSD.

Attached: macbook_pro_touch_bar_ssd[1].jpg (800x573, 96K)

It's a
>Weird shit runs sandboxed in your web browser and breaks out to swipe your keepass database master password as soon as you enter it
Kinda deal

Attached: 1527229832097.jpg (750x824, 269K)

Better get Ryzen or Power9.

Attached: 62A8EF02-08FB-4DD2-8781-5E9B6CC09D40.jpg (819x1200, 96K)

I bet by 2020 Intel will essentially hit Bulldozer levels of performance.

Chromium's been working on some isolation stuff recently.

>they projected like 10% slowdown

u-u-uh AMD might be v-vulnerable too but we have no idea, this statement alone should be enough to label them as guilty r-right

No one bothers hacking those because not mainstream enough but rest assured they have massive security holes too.

>>they projected like 10% slowdown

Attached: 1522775358675.gif (400x225, 1.74M)

Turning HT off essentially.

>The fix is just to turn off HT in 95% of use cases where it can bleed.

OpenBSD weren't fucking around. There's no way to have your cake and eat it, too.

Attached: 1528649200786.png (257x251, 105K)

>tfw not even SMT is safe.

just fuck my shit up senpai.

Just get sys-firmware/intel-microcode-20180616 for the microcode update, my dude.

So with Meltdown, Spectre, and now disabling HT... We're looking at a 30% performance hit for people who bought Intel, right?

It's all very selective performance degradation but more like I/O took a hit, now it's FP/INT time.
This will affect gaming. I'd like to see the Microsoft/Windows response to this.

What a shitshow

Attached: 1526957034344.gif (1366x768, 1.07M)

If you have an i5, mignt be a 15 % to 25%, but if you have a i3 or i7, get your ass ready for almost 50% - 60% lost performance. YAY INTEL, you did it, you have turned almost all your cpus into Bulldozers!!!!!

Attached: fuuuck.gif (360x362, 519K)

I would not be surprised if AMD hired some 1337 paj33t hax0rs to find all these spl01ts

Cloud is one of those things that have nice uses and horrible uses, and people just like the horrible uses more.

Hello where are proofs

You'd think Fedora would be pushing this down my throat more quickly than *toos.

computer is no different from light-switch, just with more abstraction.

You are red hats beta tester. Not an alpha tester.

n-no! amd cpus are perfect!!! reddit said so!!!

AAAAAAAAAAAAAAAAAAAAAHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH

What if Intel is deliberately leaving speed holes in their designs to make cloud computing so undesirable that it will make businesses leave the cloud and buy servers in the metal?

>being this mad and desperate for a (You)

i'll throw (You) a bone

This TLBleed is nothing compared to MMIOBleed.

Why is no one talking about this?

It affects multi threaded computing which is literally everything post 90s...
Disabling hyperthreading will result in 20-30% reduction in performance on average CPUs.
It will have a worse impact the more threads you contain per core.
This doesn't just affect servers damnit!

>meltdown
>spectre 1
>spectre 2
>spectre 3
>spectre 4
>spectre-NG (next generation) includes 8 variants
>LazyFP
>TLBleed
>IME

(((INTEL INSIDE)))

There has been a huge push to ARM 64 by most big tech companies for years. It is around the corner.

Power is back on the menu, boys. It's the only good alternative.

This.
OpenSPARC
POWER
ARM64
RISC-V
NanoMIPS (just announced a cpl months ago)

Things are looking up lads.

OYYYYY VEEYYYYY
DELID IT

Attached: intel oyvey.jpg (691x771, 112K)

muh Jow Forumscoins

AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

Attached: intel pink.png (807x745, 205K)

This isn't news.

Attacks like this have been theoretically possible since the P4.

FOUR NOTHING WRONG WITH ME

Attached: 1527629778452.jpg (679x758, 54K)

Terry devis was right all this time
Three is no reason to use ring3, MMU, virtual addresses. Those are useless anyway

DELE

You forgot the magical self-destruct system built in atom chips that got reported on right when it's successor came out, you just need to upgrade you filthy goyim!

Attached: 1ELkNkI[1].png (1098x1047, 763K)

hello viceroy long time to see

hows things on jdif?

Does Intel most shit processors ever i have seen?

This.

>It's like they
their engineers were actually competent?

AMD's engineers were always competent. Bulldozer was a management fuck up.

x86 considered deprecated

Attached: images.jpg (150x150, 4K)

Yes yes, all you need now is to triple your IPC and you're good to go.

how do they keep getting away with it

morally disgusting business
proven liars
shit CPUs
faked benchmarks
anti-competitive
anti-consumer

we need a final solution to the intel problem

Attached: 1529550874709.jpg (1116x1011, 116K)

a light switch is no different from cuttint and re-soldering a wire, just with more abstraction

Here is the solution.

Make the major manufacturers build more AMD laptops. Intel is only still alive because of enterprise and fucking normie tops. Enterprise they are already losing because of meltdown and Epyc.

Kill their laptop hegemony kill the beast.

Soldering is no different from using a live wire to heat up metals, just with more abstraction.

>Soldering is no different from using a live wire to heat up metals, just with more abstraction.
No its not you fucking retard that is welding, soldering places a thermally conductive but not electrically conductive insulator between the heating element and soldering surface

Ur mom gay

today*

>As it is, this attack requires malware or a malicious user to have access to the system. While easier avenues for attacks exist—and there is no known instance of this being exploited in the wild—it is concerning for public cloud users, as other guest instances on the same hardware could attempt to use this to exfiltrate data from threads running in other cores.

Problem for shared servers.

> is likely affected
>no one proved it but hey trust me

the absolute state of intlels

>what ia JS?

muh threads

Sparc had 64 threads per chip ( 8 cores x 8 thread each) in like 2007.
en.wikipedia.org/wiki/UltraSPARC_T2

LEON is nice too, 32bit sparc used for radiation resistant hardware, it's under GNU/GPL.

you mean less.

>when casuals pretend to be tech savvy

>It affects multi threaded computing which is literally everything post 90s...

heh

>cringephotocollected.raw