4 ms·

A lot of jumping to conclusions in this thread, but no, Intel did not block undervolting across the board. As the linked post explains, they implemented an opt

A lot of jumping to conclusions in this thread, but no, Intel did not block undervolting across the board. As the linked post explains, they implemented an option for vendors to block runtime undervolting from the BIOS/motherboard side. This was in response to a published CVE that used undervolting as a way of inducing and exploiting instability in the system: https://plundervolt.com/ https://plundervolt.com/ If you buy an enthusiast motherboard with overclocking and voltage adjustment controls, it's extremely unlikely that the motherboard vendor would choose to block undervolting for you. Being able to adjust and tweak things is the whole point of premium motherboards. It would be illogical for them to choose to block a popular feature. On the other hand, if you're using an OEM motherboard from some vendor that takes security seriously, I wouldn't be surprised to see undervolting blocked for security reasons. This seems logical, because not many people are buying OEM Dell workstations so they can go in and tweak every last detail in the BIOS. IMO, Intel did the right thing here. They responded to a CVE by providing an option for vendors to enable that addresses the issue. Meanwhile, enthusiasts can continue doing what they want. Seems like much ado about nothing. There's even a fantastic timeline and explanation from an Intel community member as the top comment on Reddit: https://old.reddit.com/r/intel/comments/10slejm/intel_blocks_undervolting_the_whole_story/j73ge9a/ https://old.reddit.com/r/intel/comments/10slejm/intel_blocks...

saurik 4y ago
The security issue you linked to was not something that provided the owner of the computer security but instead provides security for third-parties against the owner of the computer and their meddling notions of privacy or transparency. It is totally fine to make an argument for why they made a rational decision here, and you can even argue it has something to do with benefits to the user--though I am unsure I've ever heard of, say, a movie company banning computers from HQ video over stuff like this--but using the term "security" is disingenuous as like, I read your comment and seriously upvoted it as "sounds reasonable" until I decided to go and actually read about this fascinating exploit... and then I was all "oh :/".
- FreakLegion 4y ago
  You're referring to Plundervolt? At a glance it could be used to extract private keys from e.g. FIDO2 platform authenticators. I don't know how widespread SGX use is at this point, but it already shows up in common Windows Hello flows, 1Password, some crypto wallets, and will probably be fairly common in passkey implementations. Based on that I'd say it's fair to disable undervolting for security reasons.

saurik 4y ago

The security issue you linked to was not something that provided the owner of the computer security but instead provides security for third-parties against the owner of the computer and their meddling notions of privacy or transparency. It is totally fine to make an argument for why they made a rational decision here, and you can even argue it has something to do with benefits to the user--though I am unsure I've ever heard of, say, a movie company banning computers from HQ video over stuff like this--but using the term "security" is disingenuous as like, I read your comment and seriously upvoted it as "sounds reasonable" until I decided to go and actually read about this fascinating exploit... and then I was all "oh :/".

FreakLegion 4y ago
You're referring to Plundervolt? At a glance it could be used to extract private keys from e.g. FIDO2 platform authenticators. I don't know how widespread SGX use is at this point, but it already shows up in common Windows Hello flows, 1Password, some crypto wallets, and will probably be fairly common in passkey implementations. Based on that I'd say it's fair to disable undervolting for security reasons.