Ask HN: Why don't we just use hashcat to evaluate GPU performance
I ran the following on a M4 Pro (20 core GPU) in MacOS vs a 4070 Super in Windows 11 using WSL and was kinda shocked to see the dramatic difference in GPU performance.
## generate a random sha256 8 alphanumeric
python3 - <<'PY' > hashes-sha256.txt
import secrets
import string
import hashlib
alphabet = string.ascii_letters + string.digits
password = ''.join(secrets.choice(alphabet) for _ in range(8))
print(hashlib.sha256(password.encode()).hexdigest())
hashcat -m 1400 -a 3 -O \
--potfile-path eight-alphanumeric-sha256.pot \
-1 '?l?u?d' \
hashes-sha256.txt '?1?1?1?1?1?1?1?1'
1 thread
That's fine if you're looking at running hashcat or some other computation like that, but tells you nothing about gaming performance.
Fair, I get the performance of DX12 or OpenGL may vary widely, but I was thinking for AI applications though further research on my part showed a dramatic math difference.
Sha256 is mostly integer additions, bit rotations, XOR, AND operations. While AI workload are matrix multiplication and accumulation. I sort of answered my own question then. :)