aboutsummaryrefslogtreecommitdiff
Commit message (Expand)AuthorAgeFilesLines
...
* cuda: pipeline bert flash kv stagingOnur Gündüz2026-06-241-179/+649
* cuda: widen bert flash key tile to 64, hoist q fragmentsOnur Gündüz2026-06-241-100/+146
* cuda: trim bert reduced scratchOnur Gündüz2026-06-241-73/+213
* server: bound merged batches by token budgetOnur Gündüz2026-06-242-6/+46
* server: fix tokenizer data race under concurrencyOnur Gündüz2026-06-243-22/+8
* server: make tokenize/render overlap adaptive, drop env switchesOnur Gündüz2026-06-244-146/+50
* chore: document server diagnostic env switchesOnur Gündüz2026-06-241-0/+12
* server: tokenize requests off the worker threadOnur Gündüz2026-06-243-23/+209
* server: render inline when idle, defer only under backlogOnur Gündüz2026-06-243-6/+35
* server: pack all compatible requests into token-budget batchesOnur Gündüz2026-06-241-20/+46
* server: render embedding responses off workerOnur Gündüz2026-06-236-16/+238
* cuda: stage bert flash kv directly from qkv16Onur Gündüz2026-06-231-73/+143
* cuda: store bert qkv in 16-bit for flash batchesOnur Gündüz2026-06-231-40/+120
* cuda: fuse bert qkv bias via cublaslt epilogue (f32 output)Onur Gündüz2026-06-231-40/+59
* cuda: cache the cublaslt gate-up matmul plan per shapeOnur Gündüz2026-06-231-77/+114
* cuda: fuse bert gate-up bias+gelu via cublaslt epilogueOnur Gündüz2026-06-232-66/+226
* cuda: write bert residual delta 16-bit on the residual16 pathOnur Gündüz2026-06-231-10/+155
* chore: clang-format libffwd/cuda.cuOnur Gündüz2026-06-231-93/+92
* cuda: add bert ffn16 intermediate pathOnur Gündüz2026-06-235-170/+907
* cuda: auto-enable bert residual16 for long batchesOnur Gündüz2026-06-231-8/+12
* cuda: gate bert bf16 residual streamOnur Gündüz2026-06-231-33/+172
* cuda: store weight dtype per contextOnur Gündüz2026-06-222-48/+58
* cuda: prune attention experimentsOnur Gündüz2026-06-224-1357/+102
* cuda: add opt-in BERT flash attention for head_dim 32Onur Gündüz2026-06-221-49/+66
* cuda: make bf16 BERT flash attention the default on the head_dim-64 pathOnur Gündüz2026-06-221-6/+9
* cuda: add gated bf16 flash attention for the BERT shape (head_dim 64)Onur Gündüz2026-06-221-6/+237
* cuda: auto-enable bf16-activation BERT path above a token-count gateOnur Gündüz2026-06-221-9/+18
* cuda: add gated bf16-activation BERT pathOnur Gündüz2026-06-221-48/+142
* cuda: prepack streaming K as bf16Onur Gündüz2026-06-221-44/+65
* cuda: promote tc3 to the default Qwen bf16 attention pathOnur Gündüz2026-06-221-16/+32
* cuda: add qwen tc3 ldmatrix attentionOnur Gündüz2026-06-221-8/+460
* cuda: add K/V-sharing bf16 tensor-core attention (tc2)Onur Gündüz2026-06-221-30/+227
* cuda: add qwen bf16 tensor-core flash attentionOnur Gündüz2026-06-221-4/+459
* cuda: add register-resident streaming attention behind the flagOnur Gündüz2026-06-221-4/+172
* cuda: BERT bf16-compute GEMMs skip the f32 weight widenOnur Gündüz2026-06-211-6/+14
* cuda: add gated streaming attention prototypeOnur Gündüz2026-06-211-0/+383
* cuda: fuse BERT elementwise kernelsOnur Gündüz2026-06-211-73/+96
* cuda: fix bf16 BERT batched-attention buffer-layout mismatchOnur Gündüz2026-06-212-4/+22
* build: source all.bash checks from tests, drop precision-drift toolOnur Gündüz2026-06-213-9/+3
* devtools: fix container-build-test.sh (python3 dep, clean-tree abort)Onur Gündüz2026-06-211-2/+12
* build: add check-sanitize (ASan/UBSan hermetic run) to all.bashOnur Gündüz2026-06-211-8/+44
* build: -Werror=implicit-function-declaration to catch missing prototypesOnur Gündüz2026-06-211-1/+5
* tests: make qwen3 parity token-count check best-effortOnur Gündüz2026-06-211-6/+12
* build: define _DEFAULT_SOURCE so glibc declares strdup under -std=c11Onur Gündüz2026-06-211-0/+5
* devtools: per-backend binaries in compare_model_embeddings for cpu-vs-cudaOnur Gündüz2026-06-211-11/+33
* cuda: restore embed_tokens member name (build fix)Onur Gündüz2026-06-211-6/+6
* chore: clang-format cuda.cuOnur Gündüz2026-06-211-39/+34
* bench: select per-backend binary in bench_matrixOnur Gündüz2026-06-211-4/+45
* build: fix checks aggregate under bash 3.2 (empty array + set -u)Onur Gündüz2026-06-211-5/+9
* devtools: analyze bench sources with the tools/server include pathOnur Gündüz2026-06-211-1/+2