diff options
| author | Tim Hall <tim.hall@arm.com> | 2022-05-19 12:36:58 +0100 |
|---|---|---|
| committer | tim.hall <tim.hall@arm.com> | 2022-05-19 15:56:19 +0000 |
| commit | cda4fcb0fd3e9766a161cf3e5aa7c3283e7f7c9e (patch) | |
| tree | 2ca560bcf290bf88ab7a0058098df794486ab528 /ethosu/vela/scheduler.py | |
| parent | 8bc7a652607a771e234fda6b05275542ff0fc072 (diff) | |
| download | ethos-u-vela-cda4fcb0fd3e9766a161cf3e5aa7c3283e7f7c9e.tar.gz | |
MLBEDSW-6563: networks failing with memory area exceeded in vela3.4.0.rc2
- For allocations that have a hard memory limit the Hill Climb allocator
should be given more attempts to find a solution that would fit
- The fix is to use a memory limit when there is a hard constraint, and
a minimum iteration count, reset on every improvement, when there is a soft
constraint
- Added maximum number iterations CLI option
Signed-off-by: Tim Hall <tim.hall@arm.com>
Change-Id: I19ff53a0b68412de280263626778a3102cbe52fa
Diffstat (limited to 'ethosu/vela/scheduler.py')
| -rw-r--r-- | ethosu/vela/scheduler.py | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/ethosu/vela/scheduler.py b/ethosu/vela/scheduler.py index 0b79402a..1e311758 100644 --- a/ethosu/vela/scheduler.py +++ b/ethosu/vela/scheduler.py @@ -1312,6 +1312,7 @@ def _update_tensor_allocation(nng: Graph, arch: ArchitectureFeatures, options): tensor_allocator=options.tensor_allocator, verbose_allocation=options.verbose_allocation, cpu_tensor_alignment=options.cpu_tensor_alignment, + hillclimb_max_iterations=options.hillclimb_max_iterations, ) |