-
Notifications
You must be signed in to change notification settings - Fork 76
Pull requests: hw-native-sys/simpler
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fix: support head_dim=256 in paged attention kernels
#1925
opened Aug 20, 2026 by
doraemonmj
Contributor
Loading…
Update: reversing the task wake-up wakee
#1924
opened Aug 20, 2026 by
raphael-s-steiner
Contributor
Loading…
Optimize: pack HBG Graph replay data into one arena block
#1923
opened Aug 20, 2026 by
sunkaixuan2018
Contributor
Loading…
Fix: initialize DSv4 decode buffers on device and bound hc_head_linear to valid rows
#1922
opened Aug 20, 2026 by
ChaoZheng109
Collaborator
Loading…
Refactor: move DSv4 scale reads from the orchestrator into the kernels
code health
Technical debt, robustness, code quality
enhancement
New feature or request
#1921
opened Aug 20, 2026 by
lwDavid
Contributor
Loading…
Fix: prevent paged-attention lane poison cascades
#1918
opened Aug 20, 2026 by
doraemonmj
Contributor
Loading…
Refactor: collapse HBG single-ring machinery
#1917
opened Aug 20, 2026 by
zmnobug
Contributor
Loading…
4 of 5 tasks
host_build_graph: record Definitions concurrently, one per Graph identity
#1916
opened Aug 20, 2026 by
ChaoWao
Collaborator
Loading…
3 of 4 tasks
Fix: remove redundant scene run configuration
#1915
opened Aug 20, 2026 by
yanghaoran29
Contributor
Loading…
3 of 4 tasks
test(a5): classify the deadlock head without racing the publication window
#1914
opened Aug 20, 2026 by
ChaoWao
Collaborator
Loading…
3 tasks done
host_build_graph: carry dispatch predicates through Graph Execution
#1912
opened Aug 20, 2026 by
ChaoZheng109
Collaborator
Loading…
5 tasks done
host_build_graph: route TensorCreateInfo initial values through the host view
#1905
opened Aug 20, 2026 by
ChaoZheng109
Collaborator
Loading…
Support: add l2-orchestrator-standalone bench for the L2 submit path
#1903
opened Aug 19, 2026 by
network4agent
Loading…
Perf: copy all Graph submission PODs to the device in one bulk H2D
#1890
opened Aug 19, 2026 by
yanghaoran29
Contributor
Loading…
5 of 6 tasks
Add MPI direct L3 transport, runtime, and vector-add example.
#1888
opened Aug 19, 2026 by
xl1123
Contributor
Loading…
Fix: gate disabled host spans before formatting
#1875
opened Aug 18, 2026 by
indigo1973
Contributor
Loading…
Perf: pin Graph PODs and H2D each layer synchronously
#1872
opened Aug 18, 2026 by
yanghaoran29
Contributor
Loading…
2 tasks
Add: test host input refresh across L3 dispatches
#1856
opened Aug 17, 2026 by
vegetabledoww
Contributor
Loading…
Perf: reuse retained temp for HBG bind.args staging
#1854
opened Aug 17, 2026 by
yanghaoran29
Contributor
Loading…
7 tasks done
Fix: support scope stats in host-built graphs
#1840
opened Aug 14, 2026 by
vegetabledoww
Contributor
•
Draft
tensormap_and_ringbuffer: 1-hop transitive reduction of redundant WAIT edges
#1830
opened Aug 13, 2026 by
ChaoZheng109
Collaborator
Loading…
Fix: track reader anti-dependencies in TensorMap
#1810
opened Aug 12, 2026 by
yanghaoran29
Contributor
•
Draft
[Optimization] Replace wiring with polling-based task readiness test (~17% median device speedup)
#1137
opened Jun 24, 2026 by
SergioMartin86
Contributor
Loading…
6 of 7 tasks
ProTip!
Follow long discussions with comments:>50.