Issues

• Support ECL on Linux in the CI matrix

  infra
  communal-software/cl-tensor#35 · Filter by author takeiteasy opened on Oct 21, 2026

• Support f64 device matrix products, gather and fusion

  dtypeenhancementupstream
  communal-software/cl-tensor#32 · Filter by author takeiteasy opened on Oct 19, 2026

• Measure and reduce GPU fusion plan lookup contention

  perf
  communal-software/cl-tensor#31 · Filter by author takeiteasy opened on Oct 18, 2026

• Dispatch concatenate and stack on GPU tensors

  enhancement
  communal-software/cl-tensor#29 · Filter by author takeiteasy opened on Oct 18, 2026

• Support complex device tensor storage and operations

  dtypeenhancementupstream
  communal-software/cl-tensor#28 · Filter by author takeiteasy opened on Oct 18, 2026
  1

• Reduce GPU synchronization and destination staging

  perf
  communal-software/cl-tensor#22 · Filter by author takeiteasy opened on Oct 17, 2026
  1

• Allow per-call executor overrides

  apienhancement
  communal-software/cl-tensor#20 · Filter by author takeiteasy opened on Oct 13, 2026

• Support per-reducer axis sets in fused expressions

  apienhancement
  communal-software/cl-tensor#19 · Filter by author takeiteasy opened on Oct 13, 2026

• Parallelize single dot products and whole-tensor reductions

  enhancementperf
  communal-software/cl-tensor#18 · Filter by author takeiteasy opened on Oct 13, 2026

• Support fused mask outputs and mask reductions

  dtypeenhancement
  communal-software/cl-tensor#17 · Filter by author takeiteasy opened on Oct 13, 2026

• Support integer and complex tensor expression fusion

  dtypeenhancement
  communal-software/cl-tensor#16 · Filter by author takeiteasy opened on Oct 13, 2026

• Support fused mean and widened reduction accumulation

  dtypeenhancement
  communal-software/cl-tensor#15 · Filter by author takeiteasy opened on Oct 13, 2026

• Execute fused expressions through upstream strided N-D kernels

  blockedenhancementperfupstream
  communal-software/cl-tensor#14 · Filter by author takeiteasy opened on Oct 13, 2026

• Bound fusion result staging and irregular reduction-row packing

  perf
  communal-software/cl-tensor#13 · Filter by author takeiteasy opened on Oct 13, 2026

• Bound take! scratch for disjoint destinations

  perf
  communal-software/cl-tensor#11 · Filter by author takeiteasy opened on Oct 10, 2026

• Bound remaining accessible-storage fallback packing

  perf
  communal-software/cl-tensor#10 · Filter by author takeiteasy opened on Oct 10, 2026
  2

• Investigate incomplete or crashing normalization reference checks on experimental ARM64 CCL

  buglowpri
  communal-software/cl-tensor#9 · Filter by author takeiteasy opened on Oct 10, 2026
  1

• Reduce normalization workspace and portable f64 stage allocation

  perf
  communal-software/cl-tensor#8 · Filter by author takeiteasy opened on Oct 10, 2026
  1

• Reduce workspace for irregular axis reductions

  lowpriperf
  communal-software/cl-tensor#7 · Filter by author takeiteasy opened on Oct 10, 2026

• Avoid unnecessary snapshots for mixed Lisp and foreign storage

  lowpriperf
  communal-software/cl-tensor#6 · Filter by author takeiteasy opened on Oct 10, 2026
  2

• Reduce workspace for general output-layout validation

  lowpriperf
  communal-software/cl-tensor#5 · Filter by author takeiteasy opened on Oct 10, 2026
  2

• Autograd

  autogradenhancementlowpri
  communal-software/cl-tensor#3 · Filter by author takeiteasy opened on Oct 10, 2026
  1