VKD3D-Proton v2.10 brings various game fixes, DirectStorage additions and lots more

Also included is in this release is Enhanced Barriers, but there's no games shipping with it so it's not well tested yet.

OK, I don't even know what this is. But I think the mere fact that VKD3D-Proton is supporting things before there are even any games that use them is a pretty good sign. Chasing taillights my ass.


Last edited by Purple Library Guy on 11 September 2023 at 5:45 pm UTC

"Tired of your players figuring out ways to get past your carefully arranged invisible walls and waist-high barricades, and sequence breaking, maybe even – GASP – speed running your game? Try our NEW Enhanced Barriers™ technology! Keep your players in their places at all times, and never again see an embarrassing 5 minute world record run for your artistic vision which should take 87.5 painstaking hours to complete!"

I don't actually know what it does either, but that's what came to mind. 🤣

A bit unclear here - do we know if these are also (at least as they pertain to games like Starfield) in
GE-Proton8-14? It says they updated to the latest VKD3D-git and of course GitHub just says 'last week' without an actual timestamp (really?).

Presumably the features have been in Git for a while before they tagged this release, and the performance seems... okay, so I'm going to assume 'yes', but I'm wondering if this might be a good reason to flip back to Proton Experimental for Starfield.


Last edited by Kithop on 11 September 2023 at 7:19 pm UTC

It says they updated to the latest VKD3D-git and of course GitHub just says 'last week' without an actual timestamp (really?).

Hah, nice to see someone else that doesn't like those dumbed down "timestamps". In trying to be more human readable, they make it harder. Unless you're watching commits and see "3 minutes ago" or something, "yesterday" or "3 weeks ago" doesn't cut it. It's especially annoying when trying to bisect something. (I don't do git bisect, I make educated guesses on commits to nail it down, browsing commits in sequence with the web interface)

However, on the web interfaces, if you hover, it should pop up some alt text with at least a date and time (e.g. "3 months ago" will show it in the form of "Jun 5, 2023, 2:33 pm EDT" (based on my time zone). Both github and gitlab's web interfaces do that now. Even git.kernel.org dumbs it down like that now in some places.

It still irritates me, I'd rather just see something like "2023-09-11 16:28:21 -0700"

P.S. To answer your question, NO, Proton-GE does not have the latest git commits for vkd3d-proton and dxvk. Even current master is going to pull dxvk and vkd3d-proton from sept. 1 commits.

Proton Experimental has updated them 4 days ago.

Proton Experimental Bleeding Edge 5 hours ago


Last edited by Grogan on 11 September 2023 at 8:15 pm UTC

A bit unclear here - do we know if these are also (at least as they pertain to games like Starfield) in
GE-Proton8-14? It says they updated to the latest VKD3D-git and of course GitHub just says 'last week' without an actual timestamp (really?).

Presumably the features have been in Git for a while before they tagged this release, and the performance seems... okay, so I'm going to assume 'yes', but I'm wondering if this might be a good reason to flip back to Proton Experimental for Starfield.

If you are really curious: Check the commit on the vkd3d-proton submodule. Find the commit in the log for the tagged vkd3d-proton release to see the date: Aug 31, 2023.

(All commits after this didn't make it into the GE release obviously.)

It seems strange to me that driver developers have to make their drivers work with games instead of game developers making their games work with drivers.

It seems strange to me that driver developers have to make their drivers work with games instead of game developers making their games work with drivers.

There’s a large amount of duct tape, popsicle sticks, paper clips and bubble gum holding most software together, unfortunately.

It seems strange to me that driver developers have to make their drivers work with games instead of game developers making their games work with drivers.

Kinda been the case since the start when ya think about it.Quake wasn't exactly made to run on a GPU, but a lot of GPUs were made to run Quake. Too oversimplified, but the jist is 3d games came firat and GPUs came in later to run them but better.
When the cutting edge hardware and software development is best as a joint venture.

I hesitate to call Starfield cutting edge though. I mean I saw swf files inside some of the game data archives... Meaning it is likely in some way using Macromedia Flash, or some such derivative.

It seems strange to me that driver developers have to make their drivers work with games instead of game developers making their games work with drivers.

That's just how it is. Part of those GPU vendor partnerships generally include quite a lot of back-and-forth for optimising and just plain making the game works.

The fact of the matter is that game devs are mostly there to make games, not tinker around figuring out how the driver works.

It says they updated to the latest VKD3D-git and of course GitHub just says 'last week' without an actual timestamp (really?).

Hah, nice to see someone else that doesn't like those dumbed down "timestamps". In trying to be more human readable, they make it harder. Unless you're watching commits and see "3 minutes ago" or something, "yesterday" or "3 weeks ago" doesn't cut it. It's especially annoying when trying to bisect something. (I don't do git bisect, I make educated guesses on commits to nail it down, browsing commits in sequence with the web interface)

However, on the web interfaces, if you hover, it should pop up some alt text with at least a date and time (e.g. "3 months ago" will show it in the form of "Jun 5, 2023, 2:33 pm EDT" (based on my time zone). Both github and gitlab's web interfaces do that now. Even git.kernel.org dumbs it down like that now in some places.

It still irritates me, I'd rather just see something like "2023-09-11 16:28:21 -0700"

P.S. To answer your question, NO, Proton-GE does not have the latest git commits for vkd3d-proton and dxvk. Even current master is going to pull dxvk and vkd3d-proton from sept. 1 commits.

Proton Experimental has updated them 4 days ago.

Proton Experimental Bleeding Edge 5 hours ago

How come Proton Experimental Bleeding-Edge is 4 days behind the standard Proton Experimental? Am I missing something here shouldn't the BE version be the one getting it earlier.

Edit: Oh I think I miss read you post and see how you meant Bleeding Edge included everything up to the 5 hours ago and not just added it :D


Last edited by Obscure on 12 September 2023 at 1:10 pm UTC

How come Proton Experimental Bleeding-Edge is 4 days behind the standard Proton Experimental? Am I missing something here shouldn't the BE version be the one getting it earlier.

Edit: Oh I think I miss read you post and see how you meant Bleeding Edge included everything up to the 5 hours ago and not just added it :D

The "5 hours ago" was just for the subprojects dxvk and vkd3d-proton. They were synced 5 hours ago at the time (Proton bleeding edge had the most current of them vs. Experimental and Glorious Eggroll's Proton-GE that were set up to pull 4 day and 10 day old subproject commits respectively)

When you see something like:

"This branch is 53 commits ahead, 3 commits behind whatever"

It just means something was committed to that branch and not the one you're looking at, for whatever reason (perhaps it's N/A). It doesn't necessarily mean the older branch "whatever" has something newer.


Last edited by Grogan on 12 September 2023 at 5:37 pm UTC

Does anyone else also get weird block-effects in a lot of Starfield's particle-effects, like the engine-trails and such? - I have nothing but stock drivers on Pop!_OS and Steam with Proton Experimental. And my best guess it's because of an incompatibility or bad translation or something.