NM Profiler : Update visualize_trace.py #370
Conversation
varun-sundar-rabindranath
force-pushed
the
varun/visualize-trace-updates
branch
from
ced0868
to
2b9a9f3
Compare
| def shorten_plot_legend_strings(legend, max_char_len: int): | ||
| for t in legend.get_texts(): | ||
| t.set_text( | ||
| trim_string_back(abbreviate_known_names(t.get_text()), | ||
| max_char_len)) |
There was a problem hiding this comment.
Lucas had this already :)
There was a problem hiding this comment.
I see now that it was there before this PR. The "thank you" still stands given how wide some of the previous plot legends were :)
| ''' | ||
|
|
||
|
|
||
| def group_trace_by_operations(trace_df: pd.DataFrame) -> pd.DataFrame: |
There was a problem hiding this comment.
nice this is super useful, thanks!
|
They Y-label in the |
Hey Lucas. Yes, I noticed that and fixed it. Sorry, should have mentioned it somewhere. |
|
Migrated all changes including all of the layer-by-layer profiling code to neuralmagic#3 |
Update visualize trace utility.
ignore_samplerarg, and instead add afold_json_nodearg - This argument collapses the specified JSON tree so the plot has less clutter.Usage:
python3 neuralmagic/tools/profiler/visualize_trace.py --json-trace profiler_fp8_trace.json --output-directory ./kernel --level kernelThis command produce 2 output files :
kernel/prefill.pngandkernel/decode_steps.pngwhich are stacked-bar graph plots. In these plots the operations are grouped together by high-level concepts such asgemms,attention,rms-normetc.python3 neuralmagic/tools/profiler/visualize_trace.py --json-trace profiler_fp8_trace.json --output-directory ./module --level module --plot-metric pct_cuda_timeThis command also produces 2 output files :
module/prefill.pngandmodule/decode_steps.pngwhich are stacked-bar graph plots. In these plots the bars sum up to a 100 as the requested plot metric ispct_cuda_time