DeepSeek/DeepEP
1
0
Fork 0
mirror of https://github.com/deepseek-ai/DeepEP synced 2025-06-26 18:28:11 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

fix: not output result in some linux system

Browse Source
This commit is contained in:
fujianhao.fjh
2025-04-10 18:18:30 +08:00
parent 42494864ba
commit 0f80da8458
3 changed files with 20 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
tests/test_internode.py
View File

@@ -76,7 +76,7 @@ def test_main(num_sms: int, local_rank: int, num_local_ranks: int, num_ranks: in
t = bench(lambda: buffer.get_dispatch_layout(topk_idx, num_experts))[0]
if local_rank == 0:
print(f'[layout] Kernel performance: {t * 1000:.3f} ms', flush=True)
print()
print('', flush=True)
group.barrier()
time.sleep(1)
@@ -163,7 +163,7 @@ def test_main(num_sms: int, local_rank: int, num_local_ranks: int, num_ranks: in
if local_rank == 0:
print(' passed', flush=True)
if local_rank == 0:
print()
print('', flush=True)
# Tune dispatch performance
best_dispatch_results = None
@@ -180,10 +180,10 @@ def test_main(num_sms: int, local_rank: int, num_local_ranks: int, num_ranks: in
if t < best_time:
best_time, best_results = t, (num_sms, nvl_chunk_size, rdma_chunk_size)
if local_rank == 0:
print(f'[tuning] SMs {num_sms}, NVL chunk {nvl_chunk_size}, RDMA chunk {rdma_chunk_size}: {rdma_send_bytes / 1e9 / t:.2f} GB/s (RDMA), {nvl_recv_bytes / 1e9 / t:.2f} GB/s (NVL) ')
print(f'[tuning] SMs {num_sms}, NVL chunk {nvl_chunk_size}, RDMA chunk {rdma_chunk_size}: {rdma_send_bytes / 1e9 / t:.2f} GB/s (RDMA), {nvl_recv_bytes / 1e9 / t:.2f} GB/s (NVL) ', flush=True)
if local_rank == 0:
print(f'[tuning] Best dispatch ({"FP8" if isinstance(current_x, tuple) else "BF16"}): SMs {best_results[0]}, NVL chunk {best_results[1]}, RDMA chunk {best_results[2]}: {rdma_send_bytes / 1e9 / best_time:.2f} GB/s (RDMA), {nvl_recv_bytes / 1e9 / best_time:.2f} GB/s (NVL)')
print()
print(f'[tuning] Best dispatch ({"FP8" if isinstance(current_x, tuple) else "BF16"}): SMs {best_results[0]}, NVL chunk {best_results[1]}, RDMA chunk {best_results[2]}: {rdma_send_bytes / 1e9 / best_time:.2f} GB/s (RDMA), {nvl_recv_bytes / 1e9 / best_time:.2f} GB/s (NVL)', flush=True)
print('', flush=True)
if isinstance(current_x, tuple):
# Gather FP8 the best config from rank 0
@@ -206,13 +206,13 @@ def test_main(num_sms: int, local_rank: int, num_local_ranks: int, num_ranks: in
tune_args = {'x': recv_x, 'handle': handle, 'config': config}
t = bench(lambda: buffer.combine(**tune_args))[0]
if local_rank == 0:
print(f'[tuning] SMs {num_sms}, NVL chunk {nvl_chunk_size}, RDMA chunk {rdma_chunk_size}: {combine_bf16_rdma_recv_bytes / 1e9 / t:.2f} GB/s (RDMA), {combine_bf16_nvl_send_bytes / 1e9 / t:.2f} GB/s (NVL) ')
print(f'[tuning] SMs {num_sms}, NVL chunk {nvl_chunk_size}, RDMA chunk {rdma_chunk_size}: {combine_bf16_rdma_recv_bytes / 1e9 / t:.2f} GB/s (RDMA), {combine_bf16_nvl_send_bytes / 1e9 / t:.2f} GB/s (NVL) ', flush=True)
if t < best_time:
best_time, best_results = t, (num_sms, nvl_chunk_size, rdma_chunk_size)
if local_rank == 0:
print(f'[tuning] Best combine: SMs {best_results[0]}, NVL chunk {best_results[1]}, RDMA chunk {best_results[2]}: {combine_bf16_rdma_recv_bytes / 1e9 / best_time:.2f} GB/s (RDMA), {combine_bf16_nvl_send_bytes / 1e9 / best_time:.2f} GB/s (NVL)')
print()
print(f'[tuning] Best combine: SMs {best_results[0]}, NVL chunk {best_results[1]}, RDMA chunk {best_results[2]}: {combine_bf16_rdma_recv_bytes / 1e9 / best_time:.2f} GB/s (RDMA), {combine_bf16_nvl_send_bytes / 1e9 / best_time:.2f} GB/s (NVL)', flush=True)
print('', flush=True)
# noinspection PyUnboundLocalVariable
@@ -231,7 +231,7 @@ def test_loop(local_rank: int, num_local_ranks: int):
for i in (24, ):
test_main(i, local_rank, num_local_ranks, num_ranks, num_nodes, rank, buffer, group)
if local_rank == 0:
print()
print('', flush=True)
# Test compatibility with low latency functions
if test_ll_compatibility: