Re: macOS metal rendering engine in mac port

[Alan Third <alan@idiocy.org>]

On Sun, May 23, 2021 at 10:06:35AM -0700, Aaron Jensen wrote:
> On Sun, May 23, 2021 at 9:13 AM Aaron Jensen <aaronjensen@gmail.com> wrote:
> >
> > I've been using it and haven't seen any flickers. I'll continue using
> > it and report back. I probably would have seen one by now.
> 
> So one important note, I'm selling my 5k2k giant monitor so it's boxed
> up (if anyone is interested, email me :) ). I'm on a 24" 4k monitor
> now running at 1920x1080 scaled, whereas before I was at 3008x1269.
> 
> So, this morning when I run the scroll-up-benchmark it does what the
> macport did. That is, it looks like it hangs for a second, shows maybe
> 1 or two scrolls and then jumps to the bottom.

Told you I'd back-port that fix. ;)

I've seen similar, but only when I start the benchmark then click on
another window. My thinking was that macOS was seeing I'm no longer
using it and put it into some low-performance mode. But clicking back
in the window doesn't fix it. I've no idea what's going on and I've
only seen it happen when using the benchmark so I've not investigated.

> I get roughly 2.3s w/ line numbers and 1.16 without.
> 
> On the new surface-stuff branch, I get 4.7s w/ line numbers and 2.98s w/o.

I've made surface-stuff as much like the (non-metal) mac port as I
possibly can. I'm not seeing any difference. I've put an NSLog at the
start of keyDown and another at the end of updateLayer, and the time
difference is pretty consistently about 3ms, so I don't think the NS
port's IO is slow. That leaves the time between the key being hit and
the NS port registering it, which I don't think we can do anything
about, or the time between us passing the IOSurface to the system and
it actually displaying, which should be identical to the Mac port,
unless it's using some sneaky setting I've yet to discover.

If it's still laggy, I've got absolutely no idea.

-- 
Alan Third