Massive security hole in CPU's incoming?Official Meltdown/Spectre Discussion Thread

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

dark zero

Platinum Member
Jun 2, 2015
2,571
107
106
Ouch... This is far worse than I expected... Is making the processor performance goes to even before the Sandy Bridge era....
 

FIVR

Diamond Member
Jun 1, 2016
3,753
908
106
He'd be looking at jail if your implication is accurate.
Maybe a few years ago but in the current political climate there is almost no chance he will be prosecuted. Equifax CFO sold millions of dollars in stock and options after they found out about the hack and he is sitting pretty right now. I think BK with his well-known Trump connections will be fine.
 

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
19,421
1,761
126
I bet someone at the NSA is going
"DOH! we got caught trying to be big brother Trump... lets pretend like we dont know anything."

:X

but on another note...

If we do get a performance hit that big, Intel better be doing recalls... the chip is no good, infact its worse then last gen, and would make no sense to keep it over last gen.
 

Rifter

Lifer
Oct 9, 1999
11,522
750
126
I bet someone at the NSA is going
"DOH! we got caught trying to be big brother Trump... lets pretend like we dont know anything."

:X

but on another note...

If we do get a performance hit that big, Intel better be doing recalls... the chip is no good, infact its worse then last gen, and would make no sense to keep it over last gen.
Yeah if this is true Epyc sales and market share really are going to be Epyc.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,054
2,856
136
Several months ago I think there was a rumor was Intel managers were telling validation teams to hurry up. It's ridiculous because validation can't be sped up. Errors are caught over time, not at your will.

If that was true, then they would be paying the price for it. This bug, the minor bug Skylake-SP has(LSD), Atom C2000, Puma series modems, I woudn't be surprised if they were ALL affected by it. How unfortunate.
 

Dayman1225

Golden Member
Aug 14, 2017
1,075
767
146
Several months ago I think there was a rumor was Intel managers were telling validation teams to hurry up. It's ridiculous because validation can't be sped up. Errors are caught over time, not at your will.

If that was true, then they would be paying the price for it. This bug, the minor bug Skylake-SP has(LSD), Atom C2000, Puma series modems, I woudn't be surprised if they were ALL affected by it. How unfortunate.
you mean this? https://danluu.com/cpu-bugs/
 
  • Like
Reactions: turtile

StinkyPinky

Diamond Member
Jul 6, 2002
6,659
608
126
Um....those are massive performance hits. We're not talking 1% here. Intel should recall every single CPU impacted if this is the case.
 

Dayman1225

Golden Member
Aug 14, 2017
1,075
767
146
Um....those are massive performance hits. We're not talking 1% here. Intel should recall every single CPU impacted if this is the case.
So you want them to recall every CPU they've sold for the past 10 years? Yeah, um not happening....
 

mnewsham

Lifer
Oct 2, 2010
14,535
426
136
Um....those are massive performance hits. We're not talking 1% here. Intel should recall every single CPU impacted if this is the case.
Lol, you're talking about probably 85% of computers sold in the last 8 years. That's hundreds of millions of computers. No recall is happening.
 

VirtualLarry

No Lifer
Aug 25, 2001
53,207
7,661
126
I remember when Linux had to re-write their page-table / -fault handling, when Intel came out with their Core2 line-up, because it further defined the behavior of how that all works, more specifically, than P4 and earlier, in which case "looser" code worked.
 

StinkyPinky

Diamond Member
Jul 6, 2002
6,659
608
126
Clearly I'm talking about a recall under warranty. It's a faulty product, and in many countries that is a breach of warranty. Intel would be facing massive fines.
 
  • Like
Reactions: DarthKyrie

Jan Olšan

Senior member
Jan 12, 2017
266
258
106
I fear that if we made recall/replacement mandatory standard (which might not be possible with older CPUs or logic issues that can't be fixed just with some respin), it would make the business way too risky for the vendors.

Errata are not something you can completely rule out, they just happen, these chips are way too complex. Something like this can always happen and it would be unfortunate if a really massive recall actually killed essentially irreplaceable companies (it's mostly luck - probably - that it happened to Intel and not AMD, or VIA).

AMD TLB bug in Barcelona B2 didn't result in a recall either (although it was smaller in extent and not directly a security issue).

Newly released current Gen Coffee Lake is not effected by this bug correct?
Its architecture is old, I'd expect it is affected, Cannon Lake possibly too. If it was already fixed in it, that would likely mean Intel knew about the issue but didn't inform public/users to keep selling stuff.
 

DarkKnightDude

Senior member
Mar 10, 2011
981
44
91
Great way to start off the year...

What would the gaming impact be? I doubt most games really make all that many syscalls.
 

Justinbaileyman

Golden Member
Aug 17, 2013
1,960
242
106
Well I just got my Rig on launch day and just cant afford a complete overhaul to switch back to AMD Ryzen. Guess I will just have to sit through BS this time around. I am hoping Video Encoding and Gaming isn't hit to hard with this bug.
 
  • Like
Reactions: psolord

Eddward

Member
Apr 10, 2012
56
19
81
It may be "old" but Intel has been fixing errata in the Skylake Core, from Skylake S through Coffee Lake S
True, Kaby and Coffee are basically newer revisions of Skylake core. Anyway Coffee retail chips are being produced since mid summer, so I doubt are not affected. And even if not then I dont understand the post above with benchmarks where is clearly performance hit on the Coffee Lake chips.
 
Thread starter Similar threads Forum Replies Date
G CPUs and Overclocking 38

ASK THE COMMUNITY