5 ms·

It's probably a bit of both. Intel started implementing speculative execution decades ago but the first exploits like Specter/Meltdown weren't published until a

It's probably a bit of both. Intel started implementing speculative execution decades ago but the first exploits like Specter/Meltdown weren't published until a few years ago so that alone is strong evidence that there might be ancient attack vectors that we haven't even considered yet. On the other hand, interest in this topic among the public has definitely grown since branding departments started getting their hands on the exploits and Bloomberg published that sensationalist Micro story so it's a self reinforcing cycle.

WrtCdEvrydy 7y ago
Interestingly, there was a 1998 paper about speculative execution that basically outlined all of Spectre/Meltdown.
- akga 7y ago
  Sounds interesting. Do you have a link to it?
  - Kurakuan 7y ago
    I'm not sure if there was another paper in '98, but this one in 1995 may be the one being referenced: https://web.archive.org/web/20180506083456/https://pdfs.semanticscholar.org/2209/42809262c17b6631c0f6536c91aaf7756857.pdf https://web.archive.org/web/20180506083456/https://pdfs.sema...
- cpach 7y ago
  And 'cperciva found some weird stuff long before Spectre. He reported to Intel and their reply was basically “meh”.
  - cperciva 7y ago
    Intel is a very different company in 2020 than they were in 2005.
    - cpach 7y ago
      Alright! What are the most significant changes they have made?
      - cperciva 7y ago
        They've gone from "try to get someone fired to get them to shut up" to "offer millions of dollars of bounties to researchers". It's a completely different attitude.
    - saagarjha 7y ago
      Are they better, or worse?
kevin_thibedeau 7y ago
> sensationalist Micro story Considering it was published with the intent to "move markets" I would consider it a fraudulent Micro story.
ngneer 7y ago
Right, but side channels in Intel products were known. It is unreasonable to expect the literature to account for every single aspect of your design. Had the side channel threat been taken more seriously at the time, it would presumably have led to more robust designs.

WrtCdEvrydy 7y ago

Interestingly, there was a 1998 paper about speculative execution that basically outlined all of Spectre/Meltdown.

akga 7y ago
Sounds interesting. Do you have a link to it?
- Kurakuan 7y ago
  I'm not sure if there was another paper in '98, but this one in 1995 may be the one being referenced: https://web.archive.org/web/20180506083456/https://pdfs.semanticscholar.org/2209/42809262c17b6631c0f6536c91aaf7756857.pdf https://web.archive.org/web/20180506083456/https://pdfs.sema...
cpach 7y ago
And 'cperciva found some weird stuff long before Spectre. He reported to Intel and their reply was basically “meh”.
- cperciva 7y ago
  Intel is a very different company in 2020 than they were in 2005.
  - cpach 7y ago
    Alright! What are the most significant changes they have made?
    - cperciva 7y ago
      They've gone from "try to get someone fired to get them to shut up" to "offer millions of dollars of bounties to researchers". It's a completely different attitude.
  - saagarjha 7y ago
    Are they better, or worse?

kevin_thibedeau 7y ago

> sensationalist Micro story Considering it was published with the intent to "move markets" I would consider it a fraudulent Micro story.

ngneer 7y ago

Right, but side channels in Intel products were known. It is unreasonable to expect the literature to account for every single aspect of your design. Had the side channel threat been taken more seriously at the time, it would presumably have led to more robust designs.