tinygrad commit 5b60a09ab0a7f20c3426e505251c48cac020d1fa
device CPU
HCQ2 1
measured interval steady_ppo_start to steady_ppo_end
wall time 235.549 ms
captured kernels 150

kernel family                         aggregate     calls   kernel share
r_2500_32_4_4_32_4                   139.22 ms        12        59.39%
r_32_32_4_4_2500_4                    77.43 ms         6        33.03%
r_128_2500_4                           7.384 ms        9         3.15%
r_128_2_4_2500_4                       5.355 ms        2         2.28%
r_2500_32_4_4_8                        1.657 ms        2         0.71%
r_128_2_2500_4                         1.327 ms        1         0.57%

The CPU kernel timeline totals 234.46 ms.
The two largest dense families total 216.65 ms, or 92.42 percent.

Command:
VIZ=1 HCQ2=1 DEV=CPU NUM_CPU_THREADS=1 PYTHONPATH=/tmp/tinygrad-upstream-5b60a09 \
  /opt/homebrew/bin/python3.13 profile_cpu.py --worker tiny --test ppo \
  --threads 1 --warmup 5 --samples 1 --profile-steady

Summary command:
python -m tinygrad.viz.cli -s CPU \
  --interval steady_ppo_start steady_ppo_end -t 30 \
  --profile-path /path/to/profile.pkl