Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Revise Code - Fix MSCCL Example #17

Merged
merged 1 commit into from
Sep 19, 2022
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
fix msccl sample
  • Loading branch information
yzygitzh committed Sep 19, 2022
commit ff7f8e36a1f101d5506a136001ecd5af40f35006
35 changes: 0 additions & 35 deletions msccl_samples/npkit_launcher.sh
Original file line number Diff line number Diff line change
Expand Up @@ -34,39 +34,4 @@ export MSCCL_NUM_WARMUPS="0"
# Number of nccl-tests iterations.
export MSCCL_NUM_ITERS="10"

NPKIT_FLAGS_CPU_PREFIX="-DENABLE_NPKIT"
NPKIT_FLAGS_GPU_PREFIX="-DENABLE_NPKIT -DENABLE_NPKIT_EVENT_TIME_SYNC_CPU -DENABLE_NPKIT_EVENT_TIME_SYNC_GPU"

# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_ALL_REDUCE_RING_ENTRY -DENABLE_NPKIT_EVENT_ALL_REDUCE_RING_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_ALL_REDUCE_TREE_UPDOWN_ENTRY -DENABLE_NPKIT_EVENT_ALL_REDUCE_TREE_UPDOWN_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_ALL_REDUCE_TREE_SPLIT_ENTRY -DENABLE_NPKIT_EVENT_ALL_REDUCE_TREE_SPLIT_EXIT"

# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_COPY_SEND_ENTRY -DENABLE_NPKIT_EVENT_COPY_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_DIRECT_COPY_SEND_ENTRY -DENABLE_NPKIT_EVENT_DIRECT_COPY_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_DIRECT_RECV_ENTRY -DENABLE_NPKIT_EVENT_DIRECT_RECV_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_DIRECT_RECV_COPY_SEND_ENTRY -DENABLE_NPKIT_EVENT_DIRECT_RECV_COPY_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_DIRECT_RECV_REDUCE_COPY_SEND_ENTRY -DENABLE_NPKIT_EVENT_DIRECT_RECV_REDUCE_COPY_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_DIRECT_SEND_ENTRY -DENABLE_NPKIT_EVENT_DIRECT_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_DIRECT_SEND_FROM_OUTPUT_ENTRY -DENABLE_NPKIT_EVENT_DIRECT_SEND_FROM_OUTPUT_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_RECV_ENTRY -DENABLE_NPKIT_EVENT_RECV_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_RECV_COPY_SEND_ENTRY -DENABLE_NPKIT_EVENT_RECV_COPY_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_RECV_REDUCE_COPY_ENTRY -DENABLE_NPKIT_EVENT_RECV_REDUCE_COPY_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_RECV_REDUCE_COPY_SEND_ENTRY -DENABLE_NPKIT_EVENT_RECV_REDUCE_COPY_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_RECV_REDUCE_SEND_ENTRY -DENABLE_NPKIT_EVENT_RECV_REDUCE_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_SEND_ENTRY -DENABLE_NPKIT_EVENT_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_SEND_FROM_OUTPUT_ENTRY -DENABLE_NPKIT_EVENT_SEND_FROM_OUTPUT_EXIT"

# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_PRIM_SIMPLE_WAIT_PEER_ENTRY -DENABLE_NPKIT_EVENT_PRIM_SIMPLE_WAIT_PEER_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_PRIM_SIMPLE_REDUCE_OR_COPY_MULTI_ENTRY -DENABLE_NPKIT_EVENT_PRIM_SIMPLE_REDUCE_OR_COPY_MULTI_EXIT"

# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_PRIM_LL_WAIT_SEND_ENTRY -DENABLE_NPKIT_EVENT_PRIM_LL_WAIT_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_PRIM_LL_DATA_PROCESS_ENTRY -DENABLE_NPKIT_EVENT_PRIM_LL_DATA_PROCESS_EXIT"

# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_PRIM_LL128_WAIT_SEND_ENTRY -DENABLE_NPKIT_EVENT_PRIM_LL128_WAIT_SEND_EXIT"
# export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_PRIM_LL128_DATA_PROCESS_ENTRY -DENABLE_NPKIT_EVENT_PRIM_LL128_DATA_PROCESS_EXIT"

# export NPKIT_FLAGS=${NPKIT_FLAGS_CPU_PREFIX}" -DENABLE_NPKIT_EVENT_NET_SEND_ENTRY -DENABLE_NPKIT_EVENT_NET_SEND_EXIT -DENABLE_NPKIT_EVENT_NET_RECV_ENTRY -DENABLE_NPKIT_EVENT_NET_RECV_EXIT"

export NPKIT_FLAGS=${NPKIT_FLAGS_GPU_PREFIX}" -DENABLE_NPKIT_EVENT_PRIM_LL_DATA_PROCESS_ENTRY -DENABLE_NPKIT_EVENT_PRIM_LL_DATA_PROCESS_EXIT -DENABLE_NPKIT_EVENT_MSCCL_REDUCE_ENTRY -DENABLE_NPKIT_EVENT_MSCCL_REDUCE_EXIT"

bash npkit_runner.sh
2 changes: 1 addition & 1 deletion msccl_samples/npkit_runner.sh
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,7 @@ msccl_algo_xml="${NPKIT_SRC_DIR}/msccl_samples/msccl_algo_sample.xml"
# Build MSCCL with NPKit
cd ${MSCCL_SRC_DIR}
make clean
make -j src.build NPKIT_FLAGS="${NPKIT_FLAGS}"
make -j src.build NPKIT=1

# Clean existing results
rm -rf ${NPKIT_RUN_DIR}
Expand Down
12 changes: 11 additions & 1 deletion msccl_samples/npkit_trace_generator.py
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,16 @@ def parse_npkit_event_header(npkit_event_header_path):
line_idx += 1
return npkit_event_def

def trim_event_name(event_type):
list_event_type_name = event_type.split("_")
if "NPKIT" in list_event_type_name:
list_event_type_name.remove("NPKIT")
if "EVENT" in list_event_type_name:
list_event_type_name.remove("EVENT")
if "ENTRY" in list_event_type_name:
list_event_type_name.remove("ENTRY")
return "_".join(list_event_type_name)

def parse_gpu_clock_scale(gpu_clock_file_path):
with open(gpu_clock_file_path, 'r') as f:
freq_in_khz = f.read()
Expand Down Expand Up @@ -85,7 +95,7 @@ def parse_gpu_event_file(npkit_dump_dir, npkit_event_def, rank, buf_idx, gpu_clo
if event_type not in event_type_to_seq:
event_type_to_seq[event_type] = 0
gpu_events[-1].update({
'name': event_type,
'name': trim_event_name(event_type),
'cat': 'GPU',
'args': {
'rank': rank,
Expand Down