Search results

Transposing 128x128 blocks is suboptimal

performance

Status: Open.

#7815 In triton-lang/triton;

· danthe3rd opened on Aug 11, 2025

use triton kernel for training need warm up？

performance

Status: Open.

#7717 In triton-lang/triton;

· mdy666 opened on Jul 31, 2025

If we perform a load without executing a matmul operation, the memory access won’t be coalesced.

needs reproducer

performance

Status: Open.

#7202 In triton-lang/triton;

· woongjoonchoi opened on Jun 17, 2025

AMD/MI300X performance is lacking compared to torch.matmul

performance

Status: Open.

#7199 In triton-lang/triton;

· mobicham opened on Jun 17, 2025

Triton 3.3 Performance Regression on Small Gemms

performance

Status: Open.

#7096 In triton-lang/triton;

· axelfeldmann opened on Jun 6, 2025

release3.3 03-matrix-multiplication performance drop on A100

performance

Status: Open.

#7085 In triton-lang/triton;

· GaryYuyjl opened on Jun 6, 2025

Matmul tutorial (03) perf could be reduced by half on H100 with a small "trivial" change

performance

Status: Open.

#6825 In triton-lang/triton;

· pchen7e2 opened on May 15, 2025

tl.dot on transposed matrix tries to rearrange matrix in shared memory

performance

Status: Open.

#6569 In triton-lang/triton;

· saagarjha opened on Apr 23, 2025

Matmul performance on A100

performance

Status: Open.

#6523 In triton-lang/triton;

· QuqqU opened on Apr 17, 2025

The position of <code>if-else</code> significantly affect performance, which is unexpected.

performance

Status: Open.

#6491 In triton-lang/triton;

· wenhaoli-xmu opened on Apr 15, 2025

Triton BF16 mode is slower compared to FP32 mode.

performance

Status: Open.

#6427 In triton-lang/triton;

· wenhaoli-xmu opened on Apr 8, 2025

[AMD] BufferOps require additional annotation for grid dependent information

performance

Status: Open.

#6398 In triton-lang/triton;

· njriasan opened on Apr 4, 2025