diff options
| author | William Isaksson <william.isaksson@arm.com> | 2023-07-25 09:43:02 +0000 |
|---|---|---|
| committer | Fredrik Svedberg <fredrik.svedberg@arm.com> | 2023-07-31 20:43:19 +0000 |
| commit | e4d576772c0e0a1c7ab1e6bdcf5be8964994fbbc (patch) | |
| tree | fd822034c381944e8c4fca81f618c2330d0bba62 | |
| parent | 2ff8c45528cb38c28c7b7ac40bbeb3c8a374934d (diff) | |
| download | ethos-u-vela-e4d576772c0e0a1c7ab1e6bdcf5be8964994fbbc.tar.gz | |
MLBEDSW-7846: Number of CPU Ops reported is wrong
- Added support for multiple npu subgraphs to have the same cpu output tensor
Change-Id: I2e787306dd64af9b03cdf2bacb4c9ff7119f6c49
Signed-off-by: William Isaksson <william.isaksson@arm.com>
| -rw-r--r-- | ethosu/vela/extract_npu_subgraphs.py | 32 |
1 files changed, 24 insertions, 8 deletions
diff --git a/ethosu/vela/extract_npu_subgraphs.py b/ethosu/vela/extract_npu_subgraphs.py index 5e9a5b5..dcc8687 100644 --- a/ethosu/vela/extract_npu_subgraphs.py +++ b/ethosu/vela/extract_npu_subgraphs.py @@ -110,17 +110,21 @@ def rewrite_tensor_cpu_producer_npu_consumers( def rewrite_tensor_npu_producer_cpu_consumers( - orig_tens, call_ps, startup_init_ps, npu_subgraph, cpu_subgraph, subgraph_for_pass + orig_tens, call_ps, npu_subgraph, cpu_subgraph, subgraph_for_pass, multiple_npu_sg_have_same_cpu_out_tens ): + if multiple_npu_sg_have_same_cpu_out_tens: + new_tens = orig_tens + orig_tens = orig_tens.src_tensor + else: + new_tens = orig_tens.clone("") + orig_tens.name = orig_tens.name + "_cpu" + new_tens.ops = [] - new_tens = orig_tens.clone("") - orig_tens.name = orig_tens.name + "_cpu" npu_subgraph.output_tensors.append(orig_tens) call_ps.outputs.append(new_tens) call_ps.primary_op.outputs.append(new_tens) - new_tens.ops = [call_ps.primary_op] - + new_tens.ops.append(call_ps.primary_op) # Elementwise op can not overwrite ifm if input is used by many consumers if orig_tens in npu_subgraph.input_tensors and len(orig_tens.consumers()) > 1: new_tens.ifm_write_protected = True @@ -235,16 +239,28 @@ def extract_subgraph(nng, orig_sg, arch): dest_sgs = [subgraph_for_pass[op.scheduled_pass] for op in tens.consumers() if op is not None] need_rewrite = False + multiple_npu_sg_have_same_cpu_out_tens = False + output_tensor = tens for sg in dest_sgs: if sg != curr_sg: need_rewrite = True break - if tens in orig_sg.output_tensors: - need_rewrite = True + for orig_out_tens in orig_sg.output_tensors: + if tens == orig_out_tens: + need_rewrite = True + elif tens.equivalence_id == orig_out_tens.equivalence_id: + need_rewrite = True + multiple_npu_sg_have_same_cpu_out_tens = True + output_tensor = orig_out_tens if need_rewrite: rewrite_tensor_npu_producer_cpu_consumers( - tens, call_pass[curr_sg], startup_init_passes[curr_sg], curr_sg, orig_sg, subgraph_for_pass + output_tensor, + call_pass[curr_sg], + curr_sg, + orig_sg, + subgraph_for_pass, + multiple_npu_sg_have_same_cpu_out_tens, ) for tens in curr_sg.output_tensors: |