Regression: Crash due to high memory usage when compiling Eevee shaders on AMD GPUs #99120

New Issue

Will · 2022-06-24T02:27:51+02:00

Will commented

2022-06-24 02:27:51 +02:00

System Information

GPU:	OS:	User:
AMD Radeon RX 6900 XT (sienna_cichlid, LLVM 13.0.1, DRM 3.42, 5.15.41-1-MANJARO) AMD 4.6 (Core Profile) Mesa 22.0.4	Linux-5.15.41-1-MANJARO-x86_64-with-glibc2.35 64 Bits	@celticsamurai
Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.0 Core Profile Context 22.8.2.220819	Windows-10-10.0.22000-SP0 64 Bits	@mano-wii

Blender Version
Broken: 80859a6cb2 is the first bad commit
Worked: 3.1.2

Short description of error
Whenever I attempt to load a particular file, the memory gets so high that it crashes several programs if not the entire computer. With another file, the memory usage soars to ridiculous heights trying load textures. This happens in Manjaro, Arch, and EndeavorOS. As it turns out, this seems to be exclusive to 3.2, as downgrading to 3.1.2 resolved the issue.

Exact steps for others to reproduce the error

Open any of the attached files
Go to Material Preview view

Attached is a .blend file with only nodes that seriously affect memory usage (eevee_memory_bug.blend)
And another containing the original file from where the bug was detected (Earth.blend)

eevee_memory_bug.blend
Earth.blend

**System Information** |GPU:|OS:|User: |---|---|---|--- |AMD Radeon RX 6900 XT (sienna_cichlid, LLVM 13.0.1, DRM 3.42, 5.15.41-1-MANJARO) AMD 4.6 (Core Profile) Mesa 22.0.4|Linux-5.15.41-1-MANJARO-x86_64-with-glibc2.35 64 Bits|@celticsamurai |Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.0 Core Profile Context 22.8.2.220819|Windows-10-10.0.22000-SP0 64 Bits|@mano-wii **Blender Version** Broken: 80859a6cb2 is the first bad commit Worked: 3.1.2 **Short description of error** Whenever I attempt to load a particular file, the memory gets so high that it crashes several programs if not the entire computer. With another file, the memory usage soars to ridiculous heights trying load textures. This happens in Manjaro, Arch, and EndeavorOS. As it turns out, this seems to be exclusive to 3.2, as downgrading to 3.1.2 resolved the issue. **Exact steps for others to reproduce the error** - Open any of the attached files - Go to Material Preview view Attached is a .blend file with only nodes that seriously affect memory usage (eevee_memory_bug.blend) And another containing the original file from where the bug was detected (Earth.blend) [eevee_memory_bug.blend](https://archive.blender.org/developer/F13441568/eevee_memory_bug.blend) [Earth.blend](https://archive.blender.org/developer/F13415865/Earth.blend)

Will commented

2022-06-24 02:27:51 +02:00

Added subscriber: @celticsamurai

Pratik Borhade commented

2022-06-24 05:29:47 +02:00

Added subscriber: @PratikPB2123

Pratik Borhade commented

2022-06-24 05:29:47 +02:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Pratik Borhade commented

2022-06-24 05:29:47 +02:00

Broken: version: 3.2, branch: makepkg (modified), commit date: 2022-03-31

@celticsamurai hi, can you reproduce this in 3.2 stable version?: https://www.blender.org/download/

> Broken: version: 3.2, branch: makepkg (modified), commit date: 2022-03-31 @celticsamurai hi, can you reproduce this in 3.2 stable version?: https://www.blender.org/download/

Will commented

2022-06-24 05:33:54 +02:00

In #99120#1379322, @PratikPB2123 wrote:

Broken: version: 3.2, branch: makepkg (modified), commit date: 2022-03-31

@celticsamurai hi, can you reproduce this in 3.2 stable version?: https://www.blender.org/download/

Would 3.2.0-3 be the stable version? If so, the issue seems to persist.

> In #99120#1379322, @PratikPB2123 wrote: >> Broken: version: 3.2, branch: makepkg (modified), commit date: 2022-03-31 > @celticsamurai hi, can you reproduce this in 3.2 stable version?: https://www.blender.org/download/ Would 3.2.0-3 be the stable version? If so, the issue seems to persist.

Pratik Borhade commented

2022-06-24 05:49:26 +02:00

Could you upload crash logs then?: https://docs.blender.org/manual/en/dev/troubleshooting/crash.html#linux

With another file, the memory usage soars to ridiculous heights trying load textures

Loading textures you mean at the time of rendering?

Could you upload crash logs then?: https://docs.blender.org/manual/en/dev/troubleshooting/crash.html#linux > With another file, the memory usage soars to ridiculous heights trying load textures Loading textures you mean at the time of rendering?

Will commented

2022-06-24 05:56:44 +02:00

In #99120#1379334, @PratikPB2123 wrote:
Could you upload crash logs then?: https://docs.blender.org/manual/en/dev/troubleshooting/crash.html#linux

With another file, the memory usage soars to ridiculous heights trying load textures

Loading textures you mean at the time of rendering?

When I'm just posing the scene and I want to see it with the textures, it does this. The main problem is that the memory usage is just far too high and texture loading is taking too long. I even booted Blender through the terminal to see if I could get more details on what was causing the holdup. I would upload crash logs if I had them, but they're now lost, and I don't want to know what will happen to my computer if I attempt to replicate the issue.

> In #99120#1379334, @PratikPB2123 wrote: > Could you upload crash logs then?: https://docs.blender.org/manual/en/dev/troubleshooting/crash.html#linux > >> With another file, the memory usage soars to ridiculous heights trying load textures > Loading textures you mean at the time of rendering? When I'm just posing the scene and I want to see it with the textures, it does this. The main problem is that the memory usage is just far too high and texture loading is taking too long. I even booted Blender through the terminal to see if I could get more details on what was causing the holdup. I would upload crash logs if I had them, but they're now lost, and I don't want to know what will happen to my computer if I attempt to replicate the issue.

Pratik Borhade commented

2022-06-24 05:58:12 +02:00

Changed status from 'Needs User Info' to: 'Needs Triage'

Omar Emara commented

2022-06-24 08:49:06 +02:00

Added subscriber: @OmarEmaraDev

Omar Emara commented

2022-06-24 08:49:06 +02:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Omar Emara commented

2022-06-24 08:49:06 +02:00

The crashes are most likely Blender getting killed by the OOM Killer, so I guess you may not necessary find a crash log. If this happens again, you can look at the end of the output of sudo dmesg to see if there are any messages about Blender getting killed. Alternatively, if you rebooted, you can look at the end of the output for journalctl -b -1.

Can you monitor the GPU memory as well and describe the mechanism at which the memory increase happens? Does GPU memory keep increasing until it is full, then host memory starts increasing until it is full, then the crash happens?
Or does the memory increase merely concern host memory?

Can you reset to factory settings before opening the file to see if a difference in a preference is causing this? File > Defaults > Load Factory Settings.

I don't want to know what will happen to my computer if I attempt to replicate the issue.

The OS is capable of handing such cases gracefully, that's why it kills Blender in the first place, so I wouldn't worry too much about it.

The crashes are most likely Blender getting killed by the OOM Killer, so I guess you may not necessary find a crash log. If this happens again, you can look at the end of the output of `sudo dmesg` to see if there are any messages about Blender getting killed. Alternatively, if you rebooted, you can look at the end of the output for `journalctl -b -1`. Can you monitor the GPU memory as well and describe the mechanism at which the memory increase happens? Does GPU memory keep increasing until it is full, then host memory starts increasing until it is full, then the crash happens? Or does the memory increase merely concern host memory? Can you reset to factory settings before opening the file to see if a difference in a preference is causing this? File > Defaults > Load Factory Settings. > I don't want to know what will happen to my computer if I attempt to replicate the issue. The OS is capable of handing such cases gracefully, that's why it kills Blender in the first place, so I wouldn't worry too much about it.

Will commented

2022-06-25 02:31:41 +02:00

The GPU memory seems pretty steady; it's the RAM that seems to gradually increase. I should mention that this only seems to happen when loading a particular set of admittedly complex textures that were made using tons of nodes (but again, Blender didn't used to have a problem loading it swiftly). And yeah the memory just keeps going until it's full, at which point it crashes itself as well as several other programs, and sometimes even the whole computer. Something else worth mentioning is that I have to forcefully shut down Blender when the RAM increase is taking place. Now here's what happened when I tried to CTRL+C the program in the terminal. I invariably have to forcefully terminate the program before it gets worse.

^CTraceback (most recent call last):
  File "/home/wayne/.config/blender/3.2/scripts/addons/blenderkit/timer.py", line 27, in timer
    @bpy.app.handlers.persistent
KeyboardInterrupt
Starting daemon server on port 10753
Daemon server started on address http://127.0.0.1:10753, PID: 9493, log file located at: /home/wayne/blenderkit_data/blenderkit-daemon-10753.log
Traceback (most recent call last):
  File "/usr/lib/python3.10/site-packages/urllib3/connection.py", line 174, in _new_conn
    conn = connection.create_connection(
  File "/usr/lib/python3.10/site-packages/urllib3/util/connection.py", line 95, in create_connection
    raise err
  File "/usr/lib/python3.10/site-packages/urllib3/util/connection.py", line 85, in create_connection
    sock.connect(sa)
ConnectionRefusedError: [Errno 111] Connection refused

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/lib/python3.10/site-packages/urllib3/connectionpool.py", line 703, in urlopen
    httplib_response = self._make_request(
  File "/usr/lib/python3.10/site-packages/urllib3/connectionpool.py", line 398, in _make_request
    conn.request(method, url, **httplib_request_kw)
  File "/usr/lib/python3.10/site-packages/urllib3/connection.py", line 239, in request
    super(HTTPConnection, self).request(method, url, body=body, headers=headers)
  File "/usr/lib/python3.10/http/client.py", line 1282, in request
    self._send_request(method, url, body, headers, encode_chunked)
  File "/usr/lib/python3.10/http/client.py", line 1328, in _send_request
    self.endheaders(body, encode_chunked=encode_chunked)
  File "/usr/lib/python3.10/http/client.py", line 1277, in endheaders
    self._send_output(message_body, encode_chunked=encode_chunked)
  File "/usr/lib/python3.10/http/client.py", line 1037, in _send_output
    self.send(msg)
  File "/usr/lib/python3.10/http/client.py", line 975, in send
    self.connect()
  File "/usr/lib/python3.10/site-packages/urllib3/connection.py", line 205, in connect
    conn = self._new_conn()
  File "/usr/lib/python3.10/site-packages/urllib3/connection.py", line 186, in _new_conn
    raise NewConnectionError(
urllib3.exceptions.NewConnectionError: <urllib3.connection.HTTPConnection object at 0x7f9935f45570>: Failed to establish a new connection: [Errno 111] Connection refused

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/lib/python3.10/site-packages/requests/adapters.py", line 440, in send
    resp = conn.urlopen(
  File "/usr/lib/python3.10/site-packages/urllib3/connectionpool.py", line 785, in urlopen
    retries = retries.increment(
  File "/usr/lib/python3.10/site-packages/urllib3/util/retry.py", line 592, in increment
    raise MaxRetryError(_pool, url, error or ResponseError(cause))
urllib3.exceptions.MaxRetryError: HTTPConnectionPool(host='127.0.0.1', port=10753): Max retries exceeded with url: /report (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object at 0x7f9935f45570>: Failed to establish a new connection: [Errno 111] Connection refused'))

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/wayne/.config/blender/3.2/scripts/addons/blenderkit/timer.py", line 49, in timer
    results = daemon_lib.get_reports(app_id)
  File "/home/wayne/.config/blender/3.2/scripts/addons/blenderkit/daemon_lib.py", line 53, in get_reports
    resp = session.get(url, json=data)
  File "/usr/lib/python3.10/site-packages/requests/sessions.py", line 542, in get
    return self.request('GET', url, **kwargs)
  File "/usr/lib/python3.10/site-packages/requests/sessions.py", line 529, in request
    resp = self.send(prep, **send_kwargs)
  File "/usr/lib/python3.10/site-packages/requests/sessions.py", line 645, in send
    r = adapter.send(request, **kwargs)
  File "/usr/lib/python3.10/site-packages/requests/adapters.py", line 519, in send
    raise ConnectionError(e, request=request)
requests.exceptions.ConnectionError: HTTPConnectionPool(host='127.0.0.1', port=10753): Max retries exceeded with url: /report (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object at 0x7f9935f45570>: Failed to establish a new connection: [Errno 111] Connection refused'))
Terminated

The GPU memory seems pretty steady; it's the RAM that seems to gradually increase. I should mention that this only seems to happen when loading a particular set of admittedly complex textures that were made using tons of nodes (but again, Blender didn't used to have a problem loading it swiftly). And yeah the memory just keeps going until it's full, at which point it crashes itself as well as several other programs, and sometimes even the whole computer. Something else worth mentioning is that I have to forcefully shut down Blender when the RAM increase is taking place. Now here's what happened when I tried to CTRL+C the program in the terminal. I invariably have to forcefully terminate the program before it gets worse. ``` ^CTraceback (most recent call last): File "/home/wayne/.config/blender/3.2/scripts/addons/blenderkit/timer.py", line 27, in timer @bpy.app.handlers.persistent KeyboardInterrupt Starting daemon server on port 10753 Daemon server started on address http://127.0.0.1:10753, PID: 9493, log file located at: /home/wayne/blenderkit_data/blenderkit-daemon-10753.log Traceback (most recent call last): File "/usr/lib/python3.10/site-packages/urllib3/connection.py", line 174, in _new_conn conn = connection.create_connection( File "/usr/lib/python3.10/site-packages/urllib3/util/connection.py", line 95, in create_connection raise err File "/usr/lib/python3.10/site-packages/urllib3/util/connection.py", line 85, in create_connection sock.connect(sa) ConnectionRefusedError: [Errno 111] Connection refused During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/usr/lib/python3.10/site-packages/urllib3/connectionpool.py", line 703, in urlopen httplib_response = self._make_request( File "/usr/lib/python3.10/site-packages/urllib3/connectionpool.py", line 398, in _make_request conn.request(method, url, **httplib_request_kw) File "/usr/lib/python3.10/site-packages/urllib3/connection.py", line 239, in request super(HTTPConnection, self).request(method, url, body=body, headers=headers) File "/usr/lib/python3.10/http/client.py", line 1282, in request self._send_request(method, url, body, headers, encode_chunked) File "/usr/lib/python3.10/http/client.py", line 1328, in _send_request self.endheaders(body, encode_chunked=encode_chunked) File "/usr/lib/python3.10/http/client.py", line 1277, in endheaders self._send_output(message_body, encode_chunked=encode_chunked) File "/usr/lib/python3.10/http/client.py", line 1037, in _send_output self.send(msg) File "/usr/lib/python3.10/http/client.py", line 975, in send self.connect() File "/usr/lib/python3.10/site-packages/urllib3/connection.py", line 205, in connect conn = self._new_conn() File "/usr/lib/python3.10/site-packages/urllib3/connection.py", line 186, in _new_conn raise NewConnectionError( urllib3.exceptions.NewConnectionError: <urllib3.connection.HTTPConnection object at 0x7f9935f45570>: Failed to establish a new connection: [Errno 111] Connection refused During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/usr/lib/python3.10/site-packages/requests/adapters.py", line 440, in send resp = conn.urlopen( File "/usr/lib/python3.10/site-packages/urllib3/connectionpool.py", line 785, in urlopen retries = retries.increment( File "/usr/lib/python3.10/site-packages/urllib3/util/retry.py", line 592, in increment raise MaxRetryError(_pool, url, error or ResponseError(cause)) urllib3.exceptions.MaxRetryError: HTTPConnectionPool(host='127.0.0.1', port=10753): Max retries exceeded with url: /report (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object at 0x7f9935f45570>: Failed to establish a new connection: [Errno 111] Connection refused')) During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/home/wayne/.config/blender/3.2/scripts/addons/blenderkit/timer.py", line 49, in timer results = daemon_lib.get_reports(app_id) File "/home/wayne/.config/blender/3.2/scripts/addons/blenderkit/daemon_lib.py", line 53, in get_reports resp = session.get(url, json=data) File "/usr/lib/python3.10/site-packages/requests/sessions.py", line 542, in get return self.request('GET', url, **kwargs) File "/usr/lib/python3.10/site-packages/requests/sessions.py", line 529, in request resp = self.send(prep, **send_kwargs) File "/usr/lib/python3.10/site-packages/requests/sessions.py", line 645, in send r = adapter.send(request, **kwargs) File "/usr/lib/python3.10/site-packages/requests/adapters.py", line 519, in send raise ConnectionError(e, request=request) requests.exceptions.ConnectionError: HTTPConnectionPool(host='127.0.0.1', port=10753): Max retries exceeded with url: /report (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object at 0x7f9935f45570>: Failed to establish a new connection: [Errno 111] Connection refused')) Terminated ``` ![image.png](https://archive.blender.org/developer/F13227096/image.png)

Robert Guetzkow commented

2022-06-25 11:01:16 +02:00

Added subscriber: @rjg

Robert Guetzkow commented

2022-06-25 11:01:16 +02:00

@celticsamurai Could you please check if the issue still occurs when you disable the BlenderKit add-on?

Will commented

2022-06-25 18:54:55 +02:00

In #99120#1380198, @rjg wrote:
@celticsamurai Could you please check if the issue still occurs when you disable the BlenderKit add-on?

While these textures did come from the media library, disabling it didn't seem to help.

> In #99120#1380198, @rjg wrote: > @celticsamurai Could you please check if the issue still occurs when you disable the BlenderKit add-on? While these textures did come from the media library, disabling it didn't seem to help.

Omar Emara commented

2022-06-27 08:30:36 +02:00

Changed status from 'Needs User Info' to: 'Needs Triage'

Omar Emara commented

2022-07-04 15:04:54 +02:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Omar Emara commented

2022-07-04 15:04:54 +02:00

Not sure if this will help, but does opening blender with blender --debug-memory yield any useful messages?

Not sure if this will help, but does opening blender with `blender --debug-memory` yield any useful messages?

Will commented

2022-07-04 19:50:22 +02:00

In #99120#1384769, @OmarEmaraDev wrote:
Not sure if this will help, but does opening blender with blender --debug-memory yield any useful messages?

Nothing that I can see, but maybe there's something there that you can see?

Switching to fully guarded memory allocator.
Read prefs: /home/wayne/.config/blender/3.2/config/userpref.blend
xps_tools updater verbose is enabled
xps_tools Updater: Read in JSON settings from file
Traceback (most recent call last):
  File "/usr/share/blender/3.2/scripts/modules/addon_utils.py", line 335, in enable
    mod = __import__(module_name)
  File "/home/wayne/.config/blender/3.2/scripts/addons/rprblender/__init__.py", line 39, in <module>
    from .engine.engine import Engine
  File "/home/wayne/.config/blender/3.2/scripts/addons/rprblender/engine/__init__.py", line 54, in <module>
    pyrpr.init(rpr_lib_dir, logging.Log(tag='core'), config.pyrpr_log_calls)
  File "/home/wayne/.config/blender/3.2/scripts/addons/rprblender/pyrpr.py", line 110, in init
    import __rpr
  File "/home/wayne/.config/blender/3.2/scripts/addons/rprblender/__rpr.py", line 2, in <module>
    import _cffi_backend
ModuleNotFoundError: No module named '_cffi_backend'
Error: LZ4 module didn't load. Unsupported Python version!

    - Loading Rokoko Studio Live for Blender...
    - Loaded Rokoko Studio Live for Blender successfully!

register_class(...):
Warning: 'BonePhys_UL_BoxList' doesn't have upper case alpha-numeric prefix
register_class(...):
Warning: 'EditPanel' does not contain '_PT_' with prefix and suffix
blenderkit addon: Verbose is enabled
blenderkit addon: Read in JSON settings from file
Starting daemon server on port 10753
Daemon server started on address http://127.0.0.1:10753, PID: 7571, log file located at: /home/wayne/blenderkit_data/blenderkit-daemon-10753.log
mmd_tools registed 132 classes
Warning: property 'need_active' not found in item 'FILE_OT_execute'
blenderkit updater: Running background check for update
blenderkit addon: Starting background checking thread
blenderkit addon: Checking for update now in background
blenderkit addon: Checking for update function
blenderkit addon: Read in JSON settings from file
blenderkit addon: Determined it's not yet time to check for updates
blenderkit addon: Aborting check for updated, check interval not reached
blenderkit addon: Finished check update, doing callback
blenderkit addon: Running background update callback
blenderkit addon: BG thread: Finished check update, no callback

> In #99120#1384769, @OmarEmaraDev wrote: > Not sure if this will help, but does opening blender with `blender --debug-memory` yield any useful messages? Nothing that I can see, but maybe there's something there that you can see? ``` Switching to fully guarded memory allocator. Read prefs: /home/wayne/.config/blender/3.2/config/userpref.blend xps_tools updater verbose is enabled xps_tools Updater: Read in JSON settings from file Traceback (most recent call last): File "/usr/share/blender/3.2/scripts/modules/addon_utils.py", line 335, in enable mod = __import__(module_name) File "/home/wayne/.config/blender/3.2/scripts/addons/rprblender/__init__.py", line 39, in <module> from .engine.engine import Engine File "/home/wayne/.config/blender/3.2/scripts/addons/rprblender/engine/__init__.py", line 54, in <module> pyrpr.init(rpr_lib_dir, logging.Log(tag='core'), config.pyrpr_log_calls) File "/home/wayne/.config/blender/3.2/scripts/addons/rprblender/pyrpr.py", line 110, in init import __rpr File "/home/wayne/.config/blender/3.2/scripts/addons/rprblender/__rpr.py", line 2, in <module> import _cffi_backend ModuleNotFoundError: No module named '_cffi_backend' Error: LZ4 module didn't load. Unsupported Python version! - Loading Rokoko Studio Live for Blender... - Loaded Rokoko Studio Live for Blender successfully! register_class(...): Warning: 'BonePhys_UL_BoxList' doesn't have upper case alpha-numeric prefix register_class(...): Warning: 'EditPanel' does not contain '_PT_' with prefix and suffix blenderkit addon: Verbose is enabled blenderkit addon: Read in JSON settings from file Starting daemon server on port 10753 Daemon server started on address http://127.0.0.1:10753, PID: 7571, log file located at: /home/wayne/blenderkit_data/blenderkit-daemon-10753.log mmd_tools registed 132 classes Warning: property 'need_active' not found in item 'FILE_OT_execute' blenderkit updater: Running background check for update blenderkit addon: Starting background checking thread blenderkit addon: Checking for update now in background blenderkit addon: Checking for update function blenderkit addon: Read in JSON settings from file blenderkit addon: Determined it's not yet time to check for updates blenderkit addon: Aborting check for updated, check interval not reached blenderkit addon: Finished check update, doing callback blenderkit addon: Running background update callback blenderkit addon: BG thread: Finished check update, no callback ```

Omar Emara commented

2022-07-05 08:15:59 +02:00

Changed status from 'Needs User Info' to: 'Needs Triage'

Omar Emara commented

2022-07-05 08:15:59 +02:00

There are no useful messages unfortunately as far as I can see either.

Philipp Oeser commented

2022-07-05 11:50:37 +02:00

Added subscriber: @lichtwerk

Philipp Oeser commented

2022-07-05 11:50:37 +02:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Philipp Oeser commented

2022-07-05 11:50:37 +02:00

Loading any file reproduces this, at least in my Linux distros.

I should mention that this only seems to happen when loading a particular set of admittedly complex textures

So any file? Or just the complex ones? Could you share one .blend of the complex ones?

The logs also still show Addons in action, does this also happen when starting blender with the following?

blender --factory-startup

> Loading any file reproduces this, at least in my Linux distros. > I should mention that this only seems to happen when loading a particular set of admittedly complex textures So any file? Or just the complex ones? Could you share one .blend of the complex ones? The logs also still show Addons in action, does this also happen when starting blender with the following? ``` blender --factory-startup ```

Will commented

2022-07-07 01:30:50 +02:00

In #99120#1385651, @lichtwerk wrote:

Loading any file reproduces this, at least in my Linux distros.

I should mention that this only seems to happen when loading a particular set of admittedly complex textures

So any file? Or just the complex ones? Could you share one .blend of the complex ones?

The logs also still show Addons in action, does this also happen when starting blender with the following?
blender --factory-startup

From the media library, the Procedural Earth texture is one that Blender does not seem to like.

> In #99120#1385651, @lichtwerk wrote: >> Loading any file reproduces this, at least in my Linux distros. > >> I should mention that this only seems to happen when loading a particular set of admittedly complex textures > > So any file? Or just the complex ones? Could you share one .blend of the complex ones? > > The logs also still show Addons in action, does this also happen when starting blender with the following? > > ``` > blender --factory-startup > ``` From the media library, the Procedural Earth texture is one that Blender does not seem to like.

Will commented

2022-07-10 08:16:44 +02:00

Okay. Slightly new development. When running this in Windows, the RAM usage spikes to 40GB, then crashes.

Will commented

2022-07-19 06:02:02 +02:00

Changed status from 'Needs User Info' to: 'Needs Triage'

Pratik Borhade commented

2022-07-21 05:20:54 +02:00

Might be related #97591 (Cycles HIP error with image textures on Linux and RDNA1)

Omar Emara commented

2022-08-19 14:06:58 +02:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Omar Emara commented

2022-08-19 14:06:58 +02:00

Is this the Procedural Earth texture file you are talking about? https://www.blenderkit.com/get-blenderkit/7cf8399d-d8b6-42c1-a41f-7390c6f35117/
If permitted, can you attach it here?

Is this the Procedural Earth texture file you are talking about? https://www.blenderkit.com/get-blenderkit/7cf8399d-d8b6-42c1-a41f-7390c6f35117/ If permitted, can you attach it here?

Robert Guetzkow commented

2022-08-24 11:36:31 +02:00

@celticsamurai Could you please provide the file / project you were referring to?

Will commented

2022-08-25 05:17:33 +02:00

In #99120#1408001, @rjg wrote:
@celticsamurai Could you please provide the file / project you were referring to?

All righty. Here it is.Earth.blend

> In #99120#1408001, @rjg wrote: > @celticsamurai Could you please provide the file / project you were referring to? All righty. Here it is.[Earth.blend](https://archive.blender.org/developer/F13415865/Earth.blend)

Robert Guetzkow commented

2022-08-25 11:27:20 +02:00

Changed status from 'Needs User Info' to: 'Needs Triage'

Philipp Oeser commented

2022-08-25 11:47:15 +02:00

Can confirm the Procedural Earth shader is really heavy on my machine as well (viweport navigation is really laggy).
But not getting crashes or extraordinary memory usage (also not on the GPU)

Can confirm the `Procedural Earth` shader is really heavy on my machine as well (viweport navigation is really laggy). But not getting crashes or extraordinary memory usage (also not on the GPU)

Germano Cavalcante commented

2022-08-31 14:44:37 +02:00

Added subscriber: @mano-wii

Germano Cavalcante commented

2022-08-31 14:44:37 +02:00

I can confirm excessive RAM consumption by the AMD driver by using the provided file and viewing it in Material Preview mode.

The symbols are not available and the shader is very complex so it is still a little difficult to identify the cause:

 	[External Code]	
>	blender.exe!GHOST_ContextWGL::activateDrawingContext() Line 109	C++

 	blender.exe!wm_window_set_drawable(wmWindowManager * wm, wmWindow * win, bool activate) Line 1013	C
 	blender.exe!wm_window_reset_drawable() Line 1057	C
 	blender.exe!DRW_opengl_context_disable_ex(bool restore) Line 3181	C
 	blender.exe!wm_draw_window_offscreen(bContext * C, wmWindow * win, bool stereo) Line 945	C
 	blender.exe!wm_draw_window(bContext * C, wmWindow * win) Line 1114	C
 	blender.exe!wm_draw_update(bContext * C) Line 1372	C
 	blender.exe!WM_main(bContext * C) Line 647	C
 	blender.exe!main(int argc, const unsigned char * * UNUSED_argv_c) Line 551	C
 	[External Code]

The problem doesn't occur in 3.1, so it's a regression.
I'll look for the causing commit and try to simplify the file before confirming.

System Information
Operating system: Windows-10-10.0.22000-SP0 64 Bits
Graphics card: Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.0 Core Profile Context 22.8.2.220819

I can confirm excessive RAM consumption by the AMD driver by using the provided file and viewing it in Material Preview mode. The symbols are not available and the shader is very complex so it is still a little difficult to identify the cause: ``` [External Code] > blender.exe!GHOST_ContextWGL::activateDrawingContext() Line 109 C++ blender.exe!wm_window_set_drawable(wmWindowManager * wm, wmWindow * win, bool activate) Line 1013 C blender.exe!wm_window_reset_drawable() Line 1057 C blender.exe!DRW_opengl_context_disable_ex(bool restore) Line 3181 C blender.exe!wm_draw_window_offscreen(bContext * C, wmWindow * win, bool stereo) Line 945 C blender.exe!wm_draw_window(bContext * C, wmWindow * win) Line 1114 C blender.exe!wm_draw_update(bContext * C) Line 1372 C blender.exe!WM_main(bContext * C) Line 647 C blender.exe!main(int argc, const unsigned char * * UNUSED_argv_c) Line 551 C [External Code] ``` The problem doesn't occur in 3.1, so it's a regression. I'll look for the causing commit and try to simplify the file before confirming. --- **System Information** Operating system: Windows-10-10.0.22000-SP0 64 Bits Graphics card: Radeon (TM) RX 480 Graphics ATI Technologies Inc. 4.5.0 Core Profile Context 22.8.2.220819

Germano Cavalcante changed title from ~~Crashing in 3.2~~ to Regression: Crash due to high memory usage when compiling Eevee shaders on AMD GPUs

2022-08-31 18:16:13 +02:00

Germano Cavalcante commented

2022-08-31 18:16:13 +02:00

Changed status from 'Needs Triage' to: 'Confirmed'

Jeroen Bakker commented

2022-11-11 09:01:28 +01:00

Added subscriber: @Jeroen-Bakker

Jeroen Bakker commented

2022-11-11 09:01:28 +01:00

On NVIDIA this doesn't happen, but on AMD it does. this is related to the new way how bump shaders are generated.
The AMD compiler does extensive unrolling which make the compiler use much more memory in the new method.

The change is required in order to develop Eevee-next, but leads to issues when using bump.
This ticket has different symptoms, but same root cause as #98989 (Regression: Eevee bad performance when using procedural bump map.). Close it as duplicate for now.

On NVIDIA this doesn't happen, but on AMD it does. this is related to the new way how bump shaders are generated. The AMD compiler does extensive unrolling which make the compiler use much more memory in the new method. The change is required in order to develop Eevee-next, but leads to issues when using bump. This ticket has different symptoms, but same root cause as #98989 (Regression: Eevee bad performance when using procedural bump map.). Close it as duplicate for now.

Jeroen Bakker commented

2022-11-11 09:03:38 +01:00

Closed as duplicate of #98989

Jeroen Bakker closed this issue

2022-11-11 09:03:38 +01:00

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Regression: Crash due to high memory usage when compiling Eevee shaders on AMD GPUs #99120