Skip to content

Actions: PrismML-Eng/llama.cpp

Actions

Server

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
server.yml
125 workflow runs
125 workflow runs

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
llama/ggml: Hadamard weight-fold runtime (prism.hadamard GGUF contract, FWHT to 2048) Server #179: Pull request #119 opened by bri-prism
20m 55s hadamard-gguf-contract
hadamard-gguf-contract
20m 55s
cuda: AMD RDNA4 Q1_0/Q2_0 — HIP-path vec_dots (+37%/+16% decode), opt-in quant dedup, opt-in hipBLASLt prefill routes Server #178: Pull request #116 synchronize by The-Monk
Action required The-Monk:rdna4-q1q2
The-Monk:rdna4-q1q2
Action required
cuda: AMD RDNA4 Q1_0/Q2_0 — HIP-path vec_dots (+37%/+16% decode), opt-in quant dedup, opt-in hipBLASLt prefill routes Server #177: Pull request #116 synchronize by The-Monk
Action required The-Monk:rdna4-q1q2
The-Monk:rdna4-q1q2
Action required
cuda: AMD RDNA4 Q1_0/Q2_0 — HIP-path vec_dots (+37%/+16% decode), opt-in quant dedup, opt-in hipBLASLt prefill routes Server #176: Pull request #116 synchronize by The-Monk
Action required The-Monk:rdna4-q1q2
The-Monk:rdna4-q1q2
Action required
cuda: AMD RDNA4 Q1_0/Q2_0 — HIP-path vec_dots (+37%/+16% decode), opt-in quant dedup, opt-in hipBLASLt prefill routes Server #175: Pull request #116 synchronize by The-Monk
Action required The-Monk:rdna4-q1q2
The-Monk:rdna4-q1q2
Action required
metal: fix the recurrent-state snapshot ring's CPU fallback and wide-row set_rows Server #174: Pull request #118 synchronize by bri-prism
19m 20s fix/metal-im2col-dispatch-wide-set-rows
fix/metal-im2col-dispatch-wide-set-rows
19m 20s
metal: fix the recurrent-state snapshot ring's CPU fallback and wide-row set_rows Server #173: Pull request #118 synchronize by bri-prism
3m 34s fix/metal-im2col-dispatch-wide-set-rows
fix/metal-im2col-dispatch-wide-set-rows
3m 34s
metal: fix the recurrent-state snapshot ring's CPU fallback and wide-row set_rows Server #172: Pull request #118 opened by bri-prism
19m 8s fix/metal-im2col-dispatch-wide-set-rows
fix/metal-im2col-dispatch-wide-set-rows
19m 8s
cuda: AMD RDNA4 Q1_0/Q2_0 — HIP-path vec_dots (+37%/+16% decode), opt-in quant dedup, opt-in hipBLASLt prefill routes Server #171: Pull request #116 synchronize by The-Monk
Action required The-Monk:rdna4-q1q2
The-Monk:rdna4-q1q2
Action required
cuda: AMD RDNA4 Q1_0/Q2_0 — HIP-path vec_dots (+37%/+16% decode), opt-in quant dedup, opt-in hipBLASLt prefill routes Server #169: Pull request #116 synchronize by The-Monk
Action required The-Monk:rdna4-q1q2
The-Monk:rdna4-q1q2
Action required
cuda: AMD RDNA4 Q1_0/Q2_0 — HIP-path vec_dots (+37%/+16% decode), opt-in quant dedup, opt-in hipBLASLt prefill routes Server #168: Pull request #116 synchronize by The-Monk
Action required The-Monk:rdna4-q1q2
The-Monk:rdna4-q1q2
Action required
cuda: AMD RDNA4 Q1_0/Q2_0 — HIP-path vec_dots (+37%/+16% decode), opt-in quant dedup, opt-in hipBLASLt prefill routes Server #167: Pull request #116 opened by The-Monk
Action required The-Monk:rdna4-q1q2
The-Monk:rdna4-q1q2
Action required
prism-v6: dspark Path A, converge on upstream dflash with log-SNR conditioning and drafter-own embeddings Server #166: Pull request #114 synchronize by bri-prism
21m 31s prism-v6-dspark-dflash
prism-v6-dspark-dflash
21m 31s
prism-v6: complete Q2_0_g128 switch coverage in ops.cpp (fixes -Werror=switch CI) Server #165: Pull request #115 opened by bri-prism
18m 11s prism-v6-g128-switch-coverage
prism-v6-g128-switch-coverage
18m 11s
prism-v6: fix Q2_0_g128 ARM link failure and Metal shader compile Server #164: Pull request #113 synchronize by khosravipasha
21m 57s prism-v6-g128-arm-metal-fixes
prism-v6-g128-arm-metal-fixes
21m 57s
prism-v6: dspark Path A, converge on upstream dflash with log-SNR conditioning and drafter-own embeddings Server #163: Pull request #114 synchronize by khosravipasha
21m 39s prism-v6-dspark-dflash
prism-v6-dspark-dflash
21m 39s
prism-v6: fix Q2_0_g128 ARM link failure and Metal shader compile Server #162: Pull request #113 synchronize by khosravipasha
7m 28s prism-v6-g128-arm-metal-fixes
prism-v6-g128-arm-metal-fixes
7m 28s
cuda: MoE prefill — fused SwiGLU epilogue and fused weighted reduction Server #161: Pull request #107 synchronize by bri-prism
24m 10s perf/moe-prefill-cda
perf/moe-prefill-cda
24m 10s
cuda: enable the Hopper wgmma Q1_0/Q2_0 prefill path by default Server #160: Pull request #106 synchronize by bri-prism
23m 39s perf/hopper-q1-default-on
perf/hopper-q1-default-on
23m 39s
cuda: MoE prefill — fused SwiGLU epilogue and fused weighted reduction Server #159: Pull request #107 synchronize by bri-prism
19m 18s perf/moe-prefill-cda
perf/moe-prefill-cda
19m 18s
cuda: MoE prefill — fused SwiGLU epilogue and fused weighted reduction Server #158: Pull request #107 opened by bri-prism
16m 49s perf/moe-prefill-cda
perf/moe-prefill-cda
16m 49s
cuda: enable the Hopper wgmma Q1_0/Q2_0 prefill path by default Server #157: Pull request #106 opened by bri-prism
20m 8s perf/hopper-q1-default-on
perf/hopper-q1-default-on
20m 8s
speculative: window dspark drafter staging to its trained position range Server #156: Pull request #98 synchronize by thadreber-web
Action required thadreber-web:fix/dspark-windowed-staging
thadreber-web:fix/dspark-windowed-staging
Action required
server: clear slot state on task cancellation Server #155: Pull request #103 synchronize by Copilot AI
Action required copilot/fix-cuda-illegal-memory-access
copilot/fix-cuda-illegal-memory-access
Action required
mtmd: add lanczos resize method [no release] (#26341) Server #154: Commit 5f55650 pushed by khosravipasha
17m 26s master
master
17m 26s