[Bug 47128] New: Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

Bug ID: 47128
           Summary: Heroes of the Storm: Unreasonable I/O disk usage on
                    leading to stuttering during game gameplay
           Product: Wine
           Version: unspecified
          Hardware: x86-64
                OS: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: -unknown
          Assignee: [e-mail]
          Reporter: [e-mail]
      Distribution: ---

This was experienced from me and a few people I play with on Linux + Wine

Distribution: Ubuntu 19.04
Hardware: 19.04, the machine has an AMD Ryzen 5 1600 CPU, GTX NVIDIA 1060 3GB,
2x4GB dual channel DDR4 HyperX RAM, Crucial MX 500 SSD, MSI b350m mortar
motherboard, nvidia driver 430.09
Wine version: Tested 4.6, and a couple of others leading pretty much the same
results.

What was tested:
-Playing from an ntfs-3g partition on an HDD (Seagate 5200RPM)
-Playing from an ntfs-3g partition on an SSD (Crucial MX500)
-Playing from an ext4 partition on an SSD and HDD (the same as above)

No matter what configuration, graphics driver, DXVK, wine d3d11 or not, using
disk monitoring software like iotop clearly shows unreasonable I/O usage (close
to 100%) when those stutters happen for small reads (less than 1 Megabyte)
leading the game to be mostly unplayable thanks to those, which for example do
not happen on other wine games with the same specs/config even more
graphics-demanding.

This happens during the normal gameplay, not during the loading.

This is obviously worse when the game is run from an HDD, slightly better but
still present on an SSD. Of course the I/O usage is different from the one
Windows has in this game.

I am obviously available for further testing if you tell me what to do to
better diagnose that.

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

pattietreutel  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

Zebediah Figura  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|unspecified                 |4.6
                 CC|                            |[e-mail]

--- Comment #1 from Zebediah Figura  ---
And this does not occur on Windows?

Any clue what files are being read from/written to?

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

--- Comment #2 from Matteo Bruni  ---
Is the bulk of the stuttering limited to the first few seconds of the game,
maybe related to the first time you get some hero on screen? I suspect the
stutter is due to shader compilation / loading of compiled shaders from the
shader cache.

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

--- Comment #3 from Luca  ---
(In reply to Matteo Bruni from comment #2)
> Is the bulk of the stuttering limited to the first few seconds of the game,
> maybe related to the first time you get some hero on screen? I suspect the
> stutter is due to shader compilation / loading of compiled shaders from the
> shader cache.

It's definitely not the first time since in all my testing, I made a custom
game with the same charatchers and map, and those stutter still occurs

(In reply to Zebediah Figura from comment #1)
> And this does not occur on Windows?
>
> Any clue what files are being read from/written to?

iotop was pretty generic, pointing only to the hots .exe wine process

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

--- Comment #4 from Luca  ---
also, I enabled in the DXVK_HUD=compiler options to make it easier for me to
check for shader compilation, those rarely occur in my testing, instead again
the I/O spikes with the stutters are noticeably higher than the normal

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

David Brooks  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #5 from David Brooks  ---
I am having the same problem.

Distribution: Ubuntu 19.04 
Hardware: Intel 4 core 3GHZ, NVIDIA 880M, NVIDIA Driver 418.56, 16GB RAM,
Samsung EXT4 SSD.
Wine Version: Staging 4.8

Is there any data or testing I can provide or do in order to help bring a
resolution to this issue?

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

--- Comment #6 from Matteo Bruni  ---
Created attachment 64926
  --> https://bugs.winehq.org/attachment.cgi?id=64926
NtPowerInformation() patch

Can you test if the attached patch makes a difference?

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

--- Comment #7 from Luca  ---
(In reply to Matteo Bruni from comment #6)
> Created attachment 64926 [details]
> NtPowerInformation() patch
>
> Can you test if the attached patch makes a difference?

I could not test it since the wine version I built is buggy and does not even
start the launcher, and faults the game executable (non-staging wine 4.12)

Now I don't have the time to further testing, I will update this as soon I
build it correctly

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

--- Comment #8 from Luca  ---
(In reply to Matteo Bruni from comment #6)
> Created attachment 64926 [details]
> NtPowerInformation() patch
>
> Can you test if the attached patch makes a difference?

Ok so I managed to build a wine-staging package with that patch applied but it
did not seem to make any difference

Any way I can test for further understanding this problem?

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

--- Comment #9 from Luca  ---
I have an update on this, it's not wine-related.

I reproduced the same environment using DXVK on Windows, and it lags the same,
probably related to some sort of shader compile.

So it's related to the game engine which behaves bad with Vulkan, I have to say
it's the only game I know doing this

[Bug 47128] Heroes of the Storm: Unreasonable I/ O disk usage on leading to stuttering during game gameplay

https://bugs.winehq.org/show_bug.cgi?id=47128

Luca  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |NOTOURBUG

--- Comment #10 from Luca  ---
DXVK problem with the game engine

[Bug 47128] Heroes of the Storm: Unreasonable I/O disk usage on leading to stuttering during game gameplay

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=47128

Mathew Hodson  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #11 from Mathew Hodson  ---
https://source.winehq.org/git/wine.git/commitdiff/b37371bf20be58ee7bd0b63dba...
mentioned this bug.

Does that commit fix this issue or is it a workaround?