New Vulnerability hits Intel processors - Lazy FP State Restore

[Armageus]

Disagree with a few things in that article in respect to Hyper-threading - it often keeps things running smoothly with modern day programs/environment when a lot is going on even in cases where the overall performance increase is negligible and personally I see a fair gain from HT in a good spread of tasks.

I saw that too - certainly for day-to-day tasks HT on an I3 for example is an overall improvement in responsive multi-tasking.

I'm not sure AMD's SMT is immune to variations of this one though.

Not just their SMT implementation either - sounds like Bulldozer's shared architecture might be susceptible to similar flaws.

 

[jigger]

Disagree with a few things in that article

Of course you do. I wouldn't worry though as people have been paid to try and uncover amdflaws.com

 

[Rroff]

Of course you do. I wouldn't worry though as people have been paid to try and uncover amdflaws.com

Would be nice if you actually bothered to read the context of what I was disagreeing with...

I saw that too - certainly for day-to-day tasks HT on an I3 for example is an overall improvement in responsive multi-tasking.

Yeah i3 is a good example - on those tablet/netbook hybrids that use it you really see a difference in the responsiveness of the OS when doing much at all with 2 core versus 2 core 4 threads.

 

[David Bisset]

Disagree with a few things in that article in respect to Hyper-threading - it often keeps things running smoothly with modern day programs/environment when a lot is going on even in cases where the overall performance increase is negligible and personally I see a fair gain from HT in a good spread of tasks.

I'm not sure AMD's SMT is immune to variations of this one though.

Yeah me too - found the claims that turning it off not being a performance step back very peculiar, and the article author supporting that even more odd (the original thread about it points out it's a crude & expensive fix, and more than enough performance comparisons have been done to show it's not true)

 

[hominid]

Another day, another Intel exploit!

 

[drunkenmaster]

Something the OpenBSD guy stated seemed to imply that he believe Intel were very lacks in refusing to add additional checks and blamed a shortcut by Intel on why hyperthreading creates the vulnerability. In other words it sounds like another situation in which Intel used a short cut rather than a more secure method that other companies use to gain a little performance at the expense of security. This seems to be a bit of a recurring theme in that all the checks Intel skipped have turned out to be major security flaws and for the most part AMD refused to skip these steps/checks, it cost them performance but also left their chips dramatically more secure than Intel's.

The more of these that pop up the more server is going to shift towards AMD in the next couple of years. The idea that hyperthreading isn't very useful for information when Intel push HT as a major feature including in their server chips. They area already going to suffer massively on core count, when people start offering these with HT disabled for security and halving the thread count to 28 cores when AMD will launch 96 thread chips next year.... ouch.

 

[PaulCa]

Some of these vulnerabilities are with the architecture and affect both Intel and AMD CPUs. Spectre for example.

 

[humbug]

Some of these vulnerabilities are with the architecture and affect both Intel and AMD CPUs. Spectre for example.

Of course.

AMD's was patch through the OS ages ago.

 

[humbug]

Something the OpenBSD guy stated seemed to imply that he believe Intel were very lacks in refusing to add additional checks and blamed a shortcut by Intel on why hyperthreading creates the vulnerability. In other words it sounds like another situation in which Intel used a short cut rather than a more secure method that other companies use to gain a little performance at the expense of security. This seems to be a bit of a recurring theme in that all the checks Intel skipped have turned out to be major security flaws and for the most part AMD refused to skip these steps/checks, it cost them performance but also left their chips dramatically more secure than Intel's.

The more of these that pop up the more server is going to shift towards AMD in the next couple of years. The idea that hyperthreading isn't very useful for information when Intel push HT as a major feature including in their server chips. They area already going to suffer massively on core count, when people start offering these with HT disabled for security and halving the thread count to 28 cores when AMD will launch 96 thread chips next year.... ouch.

Its been confirmed next year AMD will a 64 core 128 thread monster, Rome. HEDT this year will be 32 core 64 thread, that's more cores than Intel's biggest server chip.

12 - 16 core Mainstream next year

Intel no longer have IPC, all they have left is clock speed, IMO next year they ain't even going to have that.

 

[Smffy]

What does Intel's bad times have to do with Trump etc? Or do you mean Trump's administration will be forced to take it seriously due to all the problems?

There are probably a couple of Mexicans working for Intel so naturally the best thing to do is persecute the entire company