[Bug 44315] Buffer maps cause CPU-GPU synchronization (Guild Wars 2, The Witcher 3)

wine-bugs at winehq.org wine-bugs at winehq.org
Mon Sep 24 20:54:28 CDT 2018


https://bugs.winehq.org/show_bug.cgi?id=44315

Firerat <firer4t at googlemail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |firer4t at googlemail.com

--- Comment #44 from Firerat <firer4t at googlemail.com> ---
(In reply to Kai Krakow from comment #41)
> It looks like pba requires inclusion of many (if not all) d3d staging patches

This is not true, PBA requires no staging patches.

I believe the reason PBA was crafted against staging was that some staging was
required in order for Retail WoW to run correctly. Although that is a guess, I
could be wrong...

Historically patching wine-devel was trivial, only really had to deal with the
missing wined3d-csmt/Makefile.in. Since wine-staging 3.5, when the wined3d-csmt
was dropped, applying the PBA patchset to wine-devel is even easier.

for a while I ran wine-devel-pba without issue
I belive Shmerl has firsthand experience with PBA on wine-devel, perhaps they
can offer some insight.

currently I'm looking at making my envvar based patch less hacky, dropping some
junk from it and moving the tunables into wined3d-main.c's  wined3d_dll_init

PBA toggle ( default off )
geo / cb heap size ( default 512 / 128 )
these tunables will be in registry,  or via envvar (envvar winning).

tbh, that is all done, at least mostly.
don't get too excited by it, from a users perspective very little will change 
, they will just have to turn PBA on if they want it instead of off if they
don't.

(In reply to Kai Krakow from comment #41)
>It is also difficult to follow the individual changes that are done with the various wine branches that integrate pba because most of the time these are an undefined collection of various staging and pba patches squashed together into one single, huge patch committed to the repository

yeah, I guess they didn't use `patchinstall.sh --backend=git-am`
slower but so much better

sadly that doesn't help much with wine-staging, instead of adding a patch to a
patchset to alter functionality they edit existing patch!

check git history of programs/winecfg/winecfg.rc , apparently the current
version of csmt toggle was commited Sun Dec 14 20:42:45 2014 +0100
The actual history is missing, like someone went at it with a sledgehammer. I
don't know how often that kind of thing happeneds,  but it means I cannot
really trust the history of wine-staging patches .. and I don't fancy reading
diffs of diffs to figure out who did what when.

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.



More information about the wine-bugs mailing list