You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
- Adds nvjitlink as a CUDA requirement to CI
- Borrow Jitify1 demangle for gcc
- Prevent jitify2::KernelData from being destroyed after context.
- Preload FLAMEGPU headers
- This only reduces time from 6.8s to 4.1s (Windows/CUDA 12.0) and can't easily extend it to system headers.
- Split up program load so we can serialise linked program.
- Quick windows test shows it to be much faster to deserialize.
- Specify -D_FILE_OFFSET_BITS=64 for when required for Jitify2 via try_compile
- Remove getKnownHeaders()
- Disables Werror from some CI workflows
- Add compile error string to exception
- Jitify2: pin the commit to the HEAD of the jitify2 branch from August 2025, including nodiscard fix
- CMake: Specify the jitify2 include directory as SYSTEM to attempt warning suppression on windows
- Either JITIFY_VERBOSE_ERRORS must be defined to a truthy value, or .info() must be called on the ErrorMsg for the header log to be included.
- see jitify2::make_compilation_error_msg for the implementation
- Readme: Remove the known issue about RTC performance for CUDA >= 12.2
- CMake: apply jitify2#146 patches prior to it being merged
Co-authored-by Robert Chisholm <r.chisholm@sheffield.ac.uk>
Co-authored-by Peter Heywood <p.heywood@sheffield.ac.uk>
Copy file name to clipboardExpand all lines: README.md
-1Lines changed: 0 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -373,4 +373,3 @@ For a full list of known issues pleases see the [Issue Tracker](https://github.c
373
373
374
374
+ Warnings and a loss of performance due to hash collisions in device code ([#356](https://github.com/FLAMEGPU/FLAMEGPU2/issues/356))
375
375
+ Multiple known areas where performance can be improved (e.g. [#449](https://github.com/FLAMEGPU/FLAMEGPU2/issues/449), [#402](https://github.com/FLAMEGPU/FLAMEGPU2/issues/402))
376
-
+ CUDA 12.2+ suffers from poor RTC compilation times, to be fixed in a future release ([#1118](https://github.com/FLAMEGPU/FLAMEGPU2/issues/1118)).
# Jitify2 on linux requires uses F_OFD_SETLKW or -D_FILE_OFFSET_BITS=64.
192
+
# Cannot detect the presence of this from CMAKE_SYSTEM_VERSION when in a container, so try to compile with it and if an error occurs then we must define this option.
193
+
# If required, sets a cmake cache internal variable FLAMEGPU__FILE_OFFSET_BITS_64_REQUIRED to true
# Set the cahce var too, to ensure it appears in the GUI.
45
+
# Set the cache var too, to ensure it appears in the GUI.
46
46
set(FLAMEGPU_VISUALISATION_ROOT "${FLAMEGPU_VISUALISATION_ROOT}"CACHESTRING"Path to local copy of the FLAMEGPU2-visualiser repository, rather than CMake-based fetching")
0 commit comments