19 Commits

Author SHA1 Message Date
Vojtech Bubnik
b2138dd348 OSX specific: Setting the QoS level to the highest level for TBB
worker threads: QOS_CLASS_USER_INTERACTIVE.
The one level lower QOS_CLASS_USER_INITIATED makes our tester Filip
unhappy, because when slicing tree supports Filip switches to a browser
on another display wating for the slicer to finish, while OSX moves
the slicing threads to high efficiency low coal burning cores.
2023-02-16 16:31:29 +01:00
Vojtech Bubnik
8472287601 Implementing a test whether the current thread is the main (UI) thread
and using it on AppConfig::save() to assert if save is called
from a worker thread.
The old assert was using thread names, which did not always work
on Windows.
Fixes #7839 #9178 #9370 #9420
2023-02-07 16:17:26 +01:00
Vojtech Bubnik
9bc69efde6 Configuration of TBB work pool threads executed from the background
slicing process:
1) "C" locales are always enforced.
2) OSX Quality of Service level is set to make sure the slicing
  runs on fat cores on Apple Silicon if some fat cores are available.
2022-09-27 18:12:27 +02:00
Vojtech Bubnik
7e77048593 Removed shiny profiler. The Shiny profiler was only working on MSVC
and nowadays the sampling profiler inside Visual Studio is better.
2022-08-24 16:32:01 +02:00
Lukáš Hejl
a47446574e Sets locales before any thread start participating in the GCode processing pipeline.
Locales should be set once per any participating threads in tbb::parallel_pipeline.
It should fix the issue with appearing comma instead of the decimal point in generated Gcode.
2022-06-01 21:18:52 +02:00
Lukas Matena
c12eff19d8 Fixed a possible deadlock:
The thread counter should be modified under a mutex, atomic is not enough here
2021-10-18 16:24:15 +02:00
Vojtech Bubnik
f9a5ee725d Follow-up to ae7d6db1d961f85b8e2788d707cef03de7e28fdb
Exporting G-code on a worker thread did not work correctly as the worker
threads were using user's locale, not "C" locale.
The "C" locale is newly enforced to TBB worker threads by
name_tbb_thread_pool_threads_set_locale()
2021-09-13 13:04:12 +02:00
Vojtech Bubnik
3a5a0575ef Fixed after removing TBB deprecated APIs 2021-06-23 15:24:05 +02:00
Roman Beránek
e13535f822
drop deprecated TBB components (#6590)
Quite some time ago, many of the TBB components were deprecated in favor
of their near-equivalents in the STL or, in the case of task_scheduler_init,
were broken up and reconstituted under a less ad-hoc logic. Every time a header
file marked deprecated gets included, a rather loud warning is emitted, which
leads to a complete TBB's domination over the stderr stream during build time,
making it harder to notice _legitimate_ warnings.

Instead of merely muting the output with TBB_SUPPRESS_DEPRECATED_MESSAGES,
perform a genuine migration away from the deprecated components with the added
benefit of achieving a source compatibility with oneTBB, the successor to TBB
which has dropped the deprecated API for good.

What got replaced for what?

| Deprecated				| Replacement					|
| ------------------------------------- | --------------------------------------------- |
| `tbb::atomic`				| `std::atomic`					|
| `tbb::mutex`				| `std::mutex`					|
| `tbb::mutex::scoped_lock`		| `std::scoped_lock<std::mutex>`		|
| `tbb::mutex::scoped_lock` (empty)	| `std::unique_lock<std::mutex>` (deferred)	|
| `tbb::task_scheduler_init`		| `tbb::global_control`				|
| `tbb::this_thread`			| `std::this_thread`				|

Signed-off-by: Roman Beranek <roman.beranek@prusa3d.com>
2021-06-23 11:48:48 +02:00
Vojtech Bubnik
b002c1485d Fix of PrusaSlicer hangs on Linux when run with non standard task affinity #5661 2021-01-04 09:03:49 +01:00
Yuri D'Elia
44f71f0ed1 Include <atomic> for std::atomic where needed
Fixes build on linux using clang 11
2020-12-13 19:41:47 +01:00
Vojtech Bubnik
ce020781d3 Refactored loading of Get/SetThreadDescription() on Windows:
These new API functions are not available on Windows 7 and on older Windows 10,
thus they are newly loaded dynamically and the functions using it retur
a bool indicating whether the functionality is supported or not.
Also the OSX variants that are not supported newly return false instead
of throwing an exception.

Fixes #4972 #4974
2020-10-26 08:09:17 +01:00
Lukas Matena
5670f5ffb0 Fix couple of build issues on gcc 2020-10-23 07:51:58 +02:00
Vojtech Bubnik
d8f45ff1d8 Workarounds and documentation of OSX posix incompatibilities 2020-10-22 14:57:50 +02:00
Vojtech Bubnik
723406dfea Workaround for OSX non-compliant implementation of
pthread_getname_np / pthread_setname_np
2020-10-22 14:45:15 +02:00
Vojtech Bubnik
0d2c31d0e4 and again a fix on Windows 2020-10-22 14:35:20 +02:00
Vojtech Bubnik
6e2a5419cc Fixes of previous commit on Posix platforms 2020-10-22 14:29:40 +02:00
Vojtech Bubnik
81b6883710 Thread names shortened to 15 characters to fit Posix norm.
Added get_current_thread_name()
2020-10-22 14:11:08 +02:00
Vojtech Bubnik
6531061d09 Support for naming slicer own threads for debugging.
The following threads are named with this commit:
slic3r_main, slic3r_BackgroundSlicingProcess,
slic3r_tbbpool_xx_yy where xx is a one based index of the TTB thread
and yy is the platform thread ID.
2020-10-22 13:54:15 +02:00