The bend release is making me feel like I’m watching a collective AI psychosis. Apparently you can call theorems laws, claim you’re fast without any substantial benchmarks, and profit!? I really hope I’m the one missing something.
10
2
136
28,528
@Keleesssss send the file where it underperforms and your setup please
2
1
257
14,840
@VictorTaelin My initial reaction was because I believed your benchmarks were either cherry picked or reward hacked, because I simply find the parallelism claims excessive.I played a bit, here's BFS with a different set of parameters, Bend GPU slower than sequential C.
gist.github.com/alpaylan/77abd…
gist.github.com/alpaylan/77abd…
2
0
24
8,870
@Keleesssss wait, 16% slower than C is what you meant by "under-performing"?
I... concede. you're correct.
under that metric, Bend definitely under-performs
I... concede. you're correct.
under that metric, Bend definitely under-performs
Sep 19, 2026 · 12:56 AM UTC
3
0
221
28,440
@Keleesssss just to be clear, your code is sequential. you didn't even *enable* parallelism. of course the GPU will under-perform. GPUs are slow at sequential computations. that's just how GPUs work. that's on Huang, not me. yet, your Bend program is still almost as fast as C! come on
3
2
284
7,589
