Skip to content

Actions: warpfront/redline

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
116 workflow runs
116 workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
docs(claims): add the multi-queue tier as Figure 1b, shape-gated CI #65: Commit ab94cd8 pushed by Kaden-Schutt
48s master
master
48s
docs: claims discipline -- overhead reduction is not a compute claim CI #64: Commit 65b4bdf pushed by Kaden-Schutt
46s master
master
46s
docs: the field data lands on the model; ~4% was the wrong anchor CI #63: Commit 343a275 pushed by Kaden-Schutt
43s master
master
43s
docs: record the reject verdict and reconcile with prior work CI #62: Commit 59bc26b pushed by Kaden-Schutt
56s master
master
56s
docs: llama-shaped decode -- interposer worth ~4%, multi-queue worth … CI #61: Commit 2ef12b8 pushed by Kaden-Schutt
46s master
master
46s
docs: correction -- redline recovers 91-99% of HIP's gap to Vulkan CI #60: Commit 8ff8169 pushed by Kaden-Schutt
52s master
master
52s
docs: gfx1151 was hiding the largest interposer win; nine cells, one … CI #59: Commit 9213448 pushed by Kaden-Schutt
54s master
master
54s
fix(hipgraph): bind the app's GPU by PCI id, and latch failed replays CI #58: Commit 67c94e0 pushed by Kaden-Schutt
45s master
master
45s
docs: interposer ignores the app's device; capability limit, not a ha… CI #57: Commit 2501b98 pushed by Kaden-Schutt
1m 3s master
master
1m 3s
fix(hipgraph): bounded bundle-arch matching, and print why a match fa… CI #56: Commit a6c9019 pushed by Kaden-Schutt
53s master
master
53s
docs: interposer across three parts; the shipped default is wrong on … CI #55: Commit ab6844e pushed by Kaden-Schutt
44s master
master
44s
fix(pm4): refuse RDNA1/RDNA2, which silently executed nothing CI #54: Commit 83889d6 pushed by Kaden-Schutt
51s master
master
51s
docs: interposer measured end to end; compounding confirmed with a co… CI #53: Commit fa1d27c pushed by Kaden-Schutt
51s master
master
51s
docs(lanes): one integer per device is the wrong shape, by measurement CI #52: Commit dec9b83 pushed by Kaden-Schutt
56s master
master
56s
feat(hipgraph): segmented multi-queue lowering for the interposer CI #51: Commit 6879b7c pushed by Kaden-Schutt
51s master
master
51s
docs: gfx1100 control confirms the queue policy is inert where 4 was … CI #50: Commit 455ff07 pushed by Kaden-Schutt
47s master
master
47s
fix(aql): restore selector re-export dropped by the segment export CI #49: Commit caff539 pushed by Kaden-Schutt
49s master
master
49s
docs: suite was flattering redline 2.6x; queue policy and provenance … CI #48: Commit 528d992 pushed by Kaden-Schutt
39s master
master
39s
docs: withdraw gfx1030 PM4 numbers; suite gains queue policy + proven… CI #47: Commit 49c8582 pushed by Kaden-Schutt
47s master
master
47s
fix(dispatch_floor): flush to system scope before the gate reads CI #46: Commit 4b16f3d pushed by Kaden-Schutt
46s master
master
46s
feat(segment): DAG-to-lane mapping that refuses to fake a split CI #45: Commit 0a6d2d0 pushed by Kaden-Schutt
44s master
master
44s
feat(lanes): per-device queue-lane policy, and gate the multi-queue arm CI #44: Commit ad2321b pushed by Kaden-Schutt
47s master
master
47s
feat(dispatch_floor): multi-queue PM4 arm, and it widens the lead CI #43: Commit 72ff005 pushed by Kaden-Schutt
47s master
master
47s
docs: the queue:chain ratio is 1:1, at a device-specific width CI #42: Commit 588bf22 pushed by Kaden-Schutt
51s master
master
51s
bench(dispatch): queue-width cliff, and PM4 measured against tuned hi… CI #41: Commit d280d45 pushed by Kaden-Schutt
53s master
master
53s