Patreon Logo Support us on Patreon to keep GamingOnLinux alive. This ensures all of our main content remains free for everyone. Just good, fresh content! Alternatively, you can donate through PayPal Logo PayPal. You can also buy games using our partner links for GOG and Humble Store.
Latest Comments by YoRHa-2B
DXVK 1.4.3 released helping games with a large number of different shaders
19 Oct 2019 at 3:28 am UTC Likes: 10

Quoting: LinuxwarperDo you think that it will be easier to get closer to Windows performance with VKD3D for DX12 games than with DXVK and DX11 games? Considering the similarities DX12 and Vulkan share and the opposite for DX11 and Vulkan.
Believe it or not, but mapping D3D12 to Vulkan is harder than D3D11. Yes, the APIs are similar in concept, but there are many little details which make it rather painful and introduce overhead in vkd3d. D3D11 might need more code, but it's fairly straight-forward for the most part since its abstractions are higher-level.

That said, in games where D3D12 is significantly better than D3D11, you'll find that vkd3d beats dxvk by a fair margin as well, so it's by no means slow, but there are quite a few nasty issues to sort out.

DXVK 1.4.3 released helping games with a large number of different shaders
18 Oct 2019 at 10:11 pm UTC Likes: 14

Quoting: LunielleIt's not that amazing when you see games like Monster Hunter World achieve just 70% of FPS compared to Windows.. Why do some games have such low performance while others can perform, even better than Windows sometimes? Aside from that DXVK is truly amazing.
Different games do vastly different things, and with this little data it's impossible to tell why it runs poorly on your system (are you CPU-bound? GPU-bound?).

Monster Hunter World in particular is known to be crazy (read: stupid) with its multithreading, which can cause all sorts of performance issues on wine especially if you aren't using fsync. Should it be DXVK's fault, then yeah, Nvidia's D3D11 driver still has a significant edge and that isn't likely to ever change, it's a translation layer after all.

Quoting: Liam DaweConsidering this is [...] running on an unsupported platform, it's amazing.
While there's truth to that, we're really looking at getting as close to parity as possible, and just 70% of Windows performance is pretty bad to be perfectly honest. Especially when most games get ≥85% these days, at least on AMD GPUs.

The Linux port of Shadow of Mordor from Feral Interactive has gained a Vulkan Beta, a massive difference
18 Oct 2019 at 11:51 am UTC

Quoting: appetrosyan
Quoting: TheRiddickIs windows 10 using DX11 or DX12?
Both AFAIK. In DXMD you could choose the API from the options.
Shaodw of Mordor doesn't support D3D12, only 11.

The Linux port of Shadow of Mordor from Feral Interactive has gained a Vulkan Beta, a massive difference
17 Oct 2019 at 2:12 pm UTC Likes: 4

Quoting: dubigrasuSo considering how DXVK improved in the mean time, I'm curious if this is still the case.
Yes it is, that's due to the crazy amount of shaders that the game has. The benchmark shows the same stutter if you run it for the first time, subsequent runs only work fine due to the state cache.

Same goes for SotTR, and quite a few other games, although it's less bad with the new ACO compiler.

Quoting: aejsmithThe AO change is described as lower quality in the article but in practice running it at half res makes virtually no noticeable difference to the quality.
Agreed (although one could argue that it's technically not the same). In any case, was there any particular reason for the change besides AO being unnecessarily expensive? I'm curious about the thought process behind this decision.

The Linux port of Shadow of Mordor from Feral Interactive has gained a Vulkan Beta, a massive difference
17 Oct 2019 at 12:33 pm UTC Likes: 2

Quoting: x_wingAO has some problems with RadeonSI and Unity, could this change also be a workaround? (I know it's RADV in this case, but maybe there is a low level relation)
No, unrelated. The RadeonSI issue is also not really an "AO" issue, it's a DCC bug that just happens to affect the AO implementation of Unity Engine. Reducing the resolution won't fix that either.

The Linux port of Shadow of Mordor from Feral Interactive has gained a Vulkan Beta, a massive difference
17 Oct 2019 at 12:08 pm UTC Likes: 40

Quoting: SpykerWell, if it has no visual impact on the game, using half-precision AO should be considered as an optimisation, right ?
I don't disagree and it's one of those little things that you can get with an official Linux port but not when running the Windows version through Proton. Another example of this is Rise of the Tomb Raider where Feral changed the shadow map fromat from D16 to D32 because D16 used to be slow/broken on RADV (and AMD actually recommends you use D32 over D16 anyway, even if it needs more memory).

FWIW, since someone already attacked me over my comment: The only reason I looked into this in the first place is because DXVK seemed unusually slow in comparison and I obviously wanted to know why, to see if there's maybe something I can improve. I'm not attacking Feral, far from it, I like their work and having officially supported games like this is important.

Shadow of the Tomb Raider Definitive Edition arrives on Linux on November 5th
17 Oct 2019 at 11:09 am UTC Likes: 3

Quoting: EhvisWhat is the reason why that specific feature could be missing from the port?
HBAO+ is part of Nvidia's Gameworks, which means that it's implemented in a closed-source, Windows-only library that they can't just port over. It's missing from RotTR as well.

The Linux port of Shadow of Mordor from Feral Interactive has gained a Vulkan Beta, a massive difference
17 Oct 2019 at 11:04 am UTC Likes: 34

And here's why it beats Proton and, in fact, Windows:


Not complaining though since the game's AO is quite inefficient and reducing resolution doesn't seem to have a visual impact. And in the CPU-bound case, the port probably still beats us, so good job.

Shadow of the Tomb Raider Definitive Edition arrives on Linux on November 5th
17 Oct 2019 at 12:07 am UTC Likes: 6

Quoting: DesumGoing to be interesting to compare Feral's pre-baked Vulkan wrapper to DXVK.
For starters, expect the Feral port to run smooth as butter right from the get-go. DXVK stutters a lot in this game because it has to compile so many shaders at draw time; Feral will just compile everything during the loading screen like they did with Rise. They can do that because they know what the game will do in advance, even if it's just a simple solution similar to DXVK's state cache.

Then, maybe (hopefully) they'll base their port on the D3D12 renderer which is much faster than the D3D11 one. Other than that, Wine itself also has significant overhead in SotTR for some reason, even with Fsync and all the Proton 4.11 improvements.

Feature-wise, expect to lose HBAO+. RTX and DLSS don't work on Proton anyway.

Saves may be incompatible with the Windows version of the game. This is basically the only gripe I have with Feral's games, it's annoying and seems somewhat unnecessary.

Shadow of the Tomb Raider Definitive Edition arrives on Linux on November 5th
15 Oct 2019 at 8:37 pm UTC Likes: 2

Quoting: KimyrielleVP I remember getting burnt by users for that less-than-ideal Witcher II port they did using their translation layer. Not sure if that was enough to discourage them, but yes, I haven't seen much from them lately, either.
Well they've done this and some other racing game from the same company. Both rather mediocre ports of bad games.