Search results

Bug: Incorrect cross-rank GPU attribution during multinode merged trace analysis, indefinite hangs

Status: Open.

#581 In AMD-AGI/TraceLens;

· brieflynn opened on Apr 3, 2026

Extend perf report to include idle time histogram

effort: M2-3 days2-3 days

Status: Open.

Task

#578 In AMD-AGI/TraceLens;

· devalshahamd opened on Apr 2, 2026

[Feature] Provide model architecture (including attention sparsity pattern) as an input file for improved roofline analysis

Status: Open.

#567 In AMD-AGI/TraceLens;

· ajassani opened on Mar 25, 2026

[Documentation] Doc/ section in readme for latest features

Status: Open.

#565 In AMD-AGI/TraceLens;

· ajassani opened on Mar 25, 2026

Feature: Create an automated way to update reference traces

Status: Open.

#561 In AMD-AGI/TraceLens;

· gabeweisz opened on Mar 24, 2026

Add RoPE and CrossEntropy categories

Status: Open.

#543 In AMD-AGI/TraceLens;

· gphuang opened on Mar 18, 2026

Add MoE_comm category for dispatch/combine ops

Status: Open.

#542 In AMD-AGI/TraceLens;

· gphuang opened on Mar 18, 2026

Add SSM category for Mamba scan/conv ops

Status: Open.

#541 In AMD-AGI/TraceLens;

· gphuang opened on Mar 18, 2026

CPU Idle and Device Synch Analysis

effort: M2-3 days2-3 days

Major FeatureCompletely new featureCompletely new feature

Status: Open.

#532 In AMD-AGI/TraceLens;

· tsrikris opened on Mar 16, 2026

Comprehensive tests for reports

effort: M2-3 days2-3 days

Status: Open.

Feature

#530 In AMD-AGI/TraceLens;

· unagiboi opened on Mar 13, 2026

Add tests related to PR #512

Status: Open.

#523 In AMD-AGI/TraceLens;

· ajassani opened on Mar 11, 2026

Add coverage for DeepEP* expert-parallel ops (Communication / EP)

perf_modelAdd performance model for calculating TFLOPS/s and TB/sAdd performance model for calculating TFLOPS/s and TB/s

Status: Open.

Feature

#518 In AMD-AGI/TraceLens;

· gphuang opened on Mar 9, 2026