By: Foo_ (foo.delete@this.nomail.com), March 31, 2021 9:32 am
Room: Moderated Discussions
Andrey (andrey.semashev.delete@this.gmail.com) on March 31, 2021 7:45 am wrote:
>
> I didn't benchmark it but Google finds a few results:
>
> http://pages.cs.wisc.edu/~rajwar/papers/SC13_TSX.pdf
> https://dl.acm.org/doi/10.1145/2503210.2503232
Those two papers seem to be the same :-) (same authors, same abstract)
> This paper is behind a paywall, but the abstract cites 1.41x and
> 1.31x improvements for HPC and network intensive applications:
Well, what they call "network intensive" seems to be a userspace TCP stack. Notwithstanding the fact that high-performance TCP stacks are generally in the kernel, how insightful their measurements are depends a lot on how optimized the particular TCP stack is. I frankly have no idea.
>
> I didn't benchmark it but Google finds a few results:
>
> http://pages.cs.wisc.edu/~rajwar/papers/SC13_TSX.pdf
> https://dl.acm.org/doi/10.1145/2503210.2503232
Those two papers seem to be the same :-) (same authors, same abstract)
> This paper is behind a paywall, but the abstract cites 1.41x and
> 1.31x improvements for HPC and network intensive applications:
Well, what they call "network intensive" seems to be a userspace TCP stack. Notwithstanding the fact that high-performance TCP stacks are generally in the kernel, how insightful their measurements are depends a lot on how optimized the particular TCP stack is. I frankly have no idea.