AMD Jabs Intel Good Over Spectre/Meltdown Problems

DooKey

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
12,891
AMD has a nice little sign set up in their booth over at Cloudfest in Germany. In a nut shell they take a nice little shot at Intel over their Spectre/Meltdown woes. You have to see it to believe it and I'm sure you'll love it as much as I did. Check out the twitter post that brought this to our attention.
 
You guys might be right about the spectre/meltdown angle. It's what came to mind when I saw the booth.
 
The only place I have really seen a significant slowdown from S&M (giggle) in XEON's is in AVX and AVX 2 instruction code. I can hardly detect it on VMWare ESX at all.
 
hold-my-beer-shit-just-got-real.jpg
 
Shots fired!
I'm sure they'll just threaten/bribe OEMs to victory while drinking expensive imported Scotch, but it's nice having AMD back
 
I think it's funny. I made a very similar comment in a previous Spectre/Meltdown thread.
 
This Anantech Conference Call With David Kanter and CTS-Labs is as good a clarity as you will get from the situation until AMD goes public on this.

https://www.anandtech.com/show/12536/our-interesting-call-with-cts-labs

Answer: COULD BE real. They sound competent enough to know what they're talking about vulnerability-wise, even if they don't know their ass-end around a chip design (their terrible estimates of fix time justifying the 1-day delay).

But even if it's real, I don't foresee this being as back-breaking as the Intel vulnerabilities. You have to have admin access to the system first.
 
Last edited:
defaultluser said:
But even if it's real, I don't foresee this being as back-breaking as the Intel vulnerabilities. You have to have admin access to the system first.
Click to expand...

That is what makes it completely laughable. There are many paths to further compromise a machine you have admin access too.... of course if you already have admin access, you sort of don't have to anymore. By that logic we can call basic commands like ls, rm, cp ect security issues because if someone has admin access they can use them as intended. lmao

Things that are useful to the nefarious minded are things that grant higher level access... not things that work as they are supposed to. Yes if you have admin level access to a system you can make it do all sorts of things, heck you can even just make it copy the contents of all its drives. That doesn't make the copy command a vulnerability. lol
 
ChadD said:
That is what makes it completely laughable. There are many paths to further compromise a machine you have admin access too.... of course if you already have admin access, you sort of don't have to anymore. By that logic we can call basic commands like ls, rm, cp ect security issues because if someone has admin access they can use them as intended. lmao

Things that are useful to the nefarious minded are things that grant higher level access... not things that work as they are supposed to. Yes if you have admin level access to a system you can make it do all sorts of things, heck you can even just make it copy the contents of all its drives. That doesn't make the copy command a vulnerability. lol
Click to expand...

Um, things that allow persistence of access are very important in security. There are after all lots of ways over time to get access to a system and with a persistent mechanism, all you need to do is get access once. At TLAs, a rather significant portion of their security exploit research is about persistence not just access because a large portion of access happened via social engineering. AKA, if the issues prove to be true, they are rather significant.
 
I'm happy AMD is doing well CPU wise. Competition is good. Hopefully it will last longer than the bright star that was the Venice over a decade ago and for only a brief moment in time.
 
Last edited:
aaronspink said:
Um, things that allow persistence of access are very important in security. There are after all lots of ways over time to get access to a system and with a persistent mechanism, all you need to do is get access once. At TLAs, a rather significant portion of their security exploit research is about persistence not just access because a large portion of access happened via social engineering. AKA, if the issues prove to be true, they are rather significant.
Click to expand...

but again if you have admin access you can make yourself a new admin account. You can give yourself permission to things...
 
Exavior said:
but again if you have admin access you can make yourself a new admin account. You can give yourself permission to things...
Click to expand...

That isn't persistence. One the break in is discovered, you wipe the machine with a new OS Image. Now you have to basically throw out the machine.
 
aaronspink said:
That isn't persistence. One the break in is discovered, you wipe the machine with a new OS Image. Now you have to basically throw out the machine.
Click to expand...

They haven't said they have made it persistent it's in theory possible which is in theory possible on Intel ME too(I think it's more in depth done on intel ME)
to make these hacks you have to specialize your attack so damn well to the hardware that no-one will bother so no-one have bothered on Intel either apart from hacking CL of Z170.
There isn't anything out there that is really more secure which is the point.

If you want to make something unbreakable you make it unpatchable like the good ol 90's hardware but then your absolutely fkin screwed if the hardware has a fault.
 
Rvenger said:
Looks like Linus just killed that guys credibility across the boards. I would love to go drinking with Torvolds, I bet he's a blast.
Click to expand...

Lets be honest, the guy has had a rep across multiple boards for years. I first came across him on semi-accurate. Still amusing.
 
Axiomatic said:
The only place I have really seen a significant slowdown from S&M (giggle) in XEON's is in AVX and AVX 2 instruction code. I can hardly detect it on VMWare ESX at all.
Click to expand...
http://aida64.helpmax.net/en/benchmark-guide/fpu-vp8/
I updated my i7 6700k a few days ago and this was the only benchmark that showed significant losses. Everything else dropped - some noticeable and some not, but this would be quite obvious to anyone that uses this codec or code that hits the CPU in similar ways.
TW2Vxtt.png
 
You must log in or register to reply here.
Back
Top