By: Robert Williams (crispysilicon.delete@this.gmail.com), April 8, 2021 7:50 pm
Room: Moderated Discussions
Linus Torvalds (torvalds.delete@this.linux-foundation.org) on April 8, 2021 9:56 am wrote:
> Robert Williams (crispysilicon.delete@this.gmail.com) on April 8, 2021 9:15 am wrote:
> > Linus Torvalds (torvalds.delete@this.linux-foundation.org) on April 8, 2021 8:41 am wrote:
> > > Sure, that was only one particular implementation (I only
> > > ever had one machine that for a while had TSX enabled)
> >
> > Just out of curiosity, what machine was it? Just trying to get a feel for how long ago.
>
> It's certainly some time ago. I had to go back and try to
> look in my email archives. Looks like it was a i7-4770S.
>
> I had a TSX patch that actually seemed to improve performance for a while - or at least not
> make it worse. And then I realized that it was because the non-TSX path was doing unnecessarily
> stupid things, and when I fixed that, the TSX side was no longer competitive.
If you still have your results, might not be a bad idea to take another look come next generation just for comparison? Sapphire Rapids is supposed to add TSX suspend load address tracking according to the manual. Haven't heard anything about it being disabled this time.
How much slower is a question on a few minds I think?
> Robert Williams (crispysilicon.delete@this.gmail.com) on April 8, 2021 9:15 am wrote:
> > Linus Torvalds (torvalds.delete@this.linux-foundation.org) on April 8, 2021 8:41 am wrote:
> > > Sure, that was only one particular implementation (I only
> > > ever had one machine that for a while had TSX enabled)
> >
> > Just out of curiosity, what machine was it? Just trying to get a feel for how long ago.
>
> It's certainly some time ago. I had to go back and try to
> look in my email archives. Looks like it was a i7-4770S.
>
> I had a TSX patch that actually seemed to improve performance for a while - or at least not
> make it worse. And then I realized that it was because the non-TSX path was doing unnecessarily
> stupid things, and when I fixed that, the TSX side was no longer competitive.
If you still have your results, might not be a bad idea to take another look come next generation just for comparison? Sapphire Rapids is supposed to add TSX suspend load address tracking according to the manual. Haven't heard anything about it being disabled this time.
How much slower is a question on a few minds I think?