Intel Says New Software Updates Make Computers 'Immune' to Meltdown and Spectre Vulnerabilities

[RealityM]

I get that most day to day users may not care about this on their desktops, but step back and think about this a minute. You have a potentially 20-30% CPU performance hit on the cloud. That means that in order to achieve the same performance this week as they did last week, cloud computing providers will have to bump their capacity by potentially 20-30%. Along with that comes more power demands which renewable sources may or may not be able to meet...

The required increase in power and capacity is worse than you think.

Let's assume that "pre-patch" capacity = 100

• a 20% reduction to 80 requires a 25% increase to get back to 100
• a 30% reduction to 70 requires a 43% increase to get back to 100
• a 50% reduction to 50 requires a 100% increase to get back to 100 (not likely but easier numbers)

 

[cmaier]

Not nice but has nothing to do with the topic.

Why not? The same guy. Maybe he’s patented the solution to this attack, too.

 

[Mick-Mac]

So I guess I’m showing my age here, but I’m surprised you can do a firmware upgrade on the Intel processor itself. I can certainly see firmware in the computer being updated. It’s a tad scary to think a firmware update by Apple for their computers could yield an Intel Processor that executes code differently to an update issued by Microsoft for Windows machines. I know it’s not likely, but I just didn’t think it possible...

 

[PickUrPoison]

So I guess I’m showing my age here, but I’m surprised you can do a firmware upgrade on the Intel processor itself. I can certainly see firmware in the computer being updated. It’s a tad scary to think a firmware update by Apple for their computers could yield an Intel Processor that executes code differently to an update issued by Microsoft for Windows machines. I know it’s not likely, but I just didn’t think it possible...

Yes, CPUs get microcode updates, but I’m pretty sure the firmware residing on the CPU itself doesn’t actually get changed; rather the update gets loaded as part of the boot process. It doesn’t stick, and gets patched anew each boot cycle. (Processors do get new microcode but it requires a new hardwware stepping afaik.)

 

[cube]

So I guess I’m showing my age here, but I’m surprised you can do a firmware upgrade on the Intel processor itself. I can certainly see firmware in the computer being updated. It’s a tad scary to think a firmware update by Apple for their computers could yield an Intel Processor that executes code differently to an update issued by Microsoft for Windows machines. I know it’s not likely, but I just didn’t think it possible...

It has nothing to do with age, but with being unfamiliar with how Intel CPUs have been implemented since 1995.

 

[cmaier]

Yes, CPUs get microcode updates, but I’m pretty sure the firmware residing on the CPU itself doesn’t actually get changed; rather the update gets loaded as part of the boot process. It doesn’t stick, and gets patched anew each boot cycle. (Processors do get new microcode but it requires a new hardwware stepping afaik.)

The way it typically works is that at boot, the bios can tell the cpu “here’s some replacement microcode for instruction XYZ.” Each time the power is cycled you have to do it again. Of course, not every cpu instruction is microcoded in the first place, so this only gets you so far.

 

[panjandrum]

Well, good news on the gaming front - these fixes not to impact gaming performance in any meaningful way:

https://www.techspot.com/article/1554-meltdown-flaw-cpu-performance-windows/

 

[oldmacs]

I really hope Apple will be patching at least back to El Capitan. Wish they'd patch Lion for the devices they left stranded on Lion. Same goes for iOS. - iOS 9 and 10 should be patched to cover the A5 devices Apple kept around for wayyy too long.

 

[Baymowe335]

Trying to put things into perspective.

More than half a year ago, a team of researchers reported the Meltdown and Spectre design flaws to Intel, Kocher P. et al. reported in their paper that a permanent fix for Spectre was not possible for the current microprocessor design. Intel had roughly 1/2 a year of ample time to come up with "temporary solutions" to their problems - until January 3. 2018, the date the public was informed about the issues. Three days later Intel say abracadabra we have a fix that makes our hardware "immune" to potential attacks.

Smells very fishy to me; Dear Intel, I'm not buying it.

So you’ll just keep your tin foil hat on without any facts to back up your conspiracy theory ?

 

[cmaier]

Trying to put things into perspective.

More than half a year ago, a team of researchers reported the Meltdown and Spectre design flaws to Intel, Kocher P. et al. reported in their paper that a permanent fix for Spectre was not possible for the current microprocessor design. Intel had roughly 1/2 a year of ample time to come up with "temporary solutions" to their problems - until January 3. 2018, the date the public was informed about the issues. Three days later Intel say abracadabra we have a fix that makes our hardware "immune" to potential attacks.

Smells very fishy to me; Dear Intel, I'm not buying it.

Paul Kocher has, in the past, found Attacks and then patented countermeasures. For example:

https://www.google.com/patents/US8879724

(And notice my name is in this document in at least a couple places - My own work came up in a challenge to the novelty of his inventions.).

When he says it’s impossible to fix he either meant that you can’t fix the processor itself (without making a new one with a different design) or maybe he plans to offer his own solution.

 

[O.N.Y.X]

So you’ll just keep your tin foil hat on without any facts to back up your conspiracy theory ?

Likely. The "tin foil hat" I got from the guys who did and published unbiased research on the matter ( https://meltdownattack.com/ ) and a former Intel employee who explains why a rush job won't suffice in this case ( https://mobile.twitter.com/securelyfitz/status/949370010652196864 ).
[doublepost=1515323329][/doublepost]

Paul Kocher has, in the past, found Attacks and then patented countermeasures. For example:

https://www.google.com/patents/US8879724

(And notice my name is in this document in at least a couple places - My own work came up in a challenge to the novelty of his inventions.).

When he says it’s impossible to fix he either meant that you can’t fix the processor itself (without making a new one with a different design) or maybe he plans to offer his own solution.

Just FYI, I'm not praising Kocher in my post. That's why I worded it "Kocher P. et al.". He just happened to be first author on the Spectre paper and co-author of the Meltdown paper. If he patented inventions only to force them upon other companies that appears borderline parasitic but definitely selfish and, to me, quite annoying. On the other hand, as an independent researcher he is allowed to do so, especially if he's the only one to offer a working solution.