Commit Graph

68 Commits

Author SHA1 Message Date
Craig Raw
a9c38ceae2 fix progress map: concurrent scans with same scan_key no longer clobber each other
Some checks failed
Main Extension Distribution Pipeline / Build extension binaries (push) Has been cancelled
Main Extension Distribution Pipeline / Code Quality Check (push) Has been cancelled
2026-05-07 13:32:05 +02:00
Craig Raw
488482d948 add total_rows parameter to ufsecp_scan for smooth ufsecp_progress reporting
Some checks are pending
Main Extension Distribution Pipeline / Build extension binaries (push) Waiting to run
Main Extension Distribution Pipeline / Code Quality Check (push) Waiting to run
2026-05-06 17:05:18 +02:00
Craig Raw
380c34da4f skip cuda backend when any visible device is below sm_80 so opencl fallback engages
Some checks are pending
Main Extension Distribution Pipeline / Build extension binaries (push) Waiting to run
Main Extension Distribution Pipeline / Code Quality Check (push) Waiting to run
2026-05-05 12:23:47 +02:00
Craig Raw
e37cc5a999 upgrade ufsecp to v3.68.0 2026-05-05 12:13:11 +02:00
Craig Raw
e7194bc280 remove legacy gpu dispatch paths now that all backends pass spend bytes per-batch 2026-05-05 11:41:43 +02:00
Craig Raw
7a1570615b fix cuda multi-thread races: per-batch scan-wnaf and spend-keys via kernel params 2026-05-05 11:30:33 +02:00
Craig Raw
a3b5bc5bee fix opencl multi-thread races: per-batch spend buffer, eager lut init 2026-05-05 09:49:41 +02:00
Craig Raw
d3829b2a4f fix multi-thread races: emit matches per-thread, move metal spend buffer to per-batch state
Some checks are pending
Main Extension Distribution Pipeline / Build extension binaries (push) Waiting to run
Main Extension Distribution Pipeline / Code Quality Check (push) Waiting to run
2026-05-04 17:36:47 +02:00
Craig Raw
c2f903a659 upgrade ufsecp to v3.64 and switch batch inversion from device to threadgroup memory
Some checks failed
Main Extension Distribution Pipeline / Build extension binaries (push) Has been cancelled
Main Extension Distribution Pipeline / Code Quality Check (push) Has been cancelled
2026-04-09 12:46:01 +02:00
Craig Raw
9a2648aaee upgrade ufsecp to v3.61
Some checks failed
Main Extension Distribution Pipeline / Build extension binaries (push) Has been cancelled
Main Extension Distribution Pipeline / Code Quality Check (push) Has been cancelled
2026-04-06 14:48:25 +02:00
Craig Raw
036cab0179 update readme with progress function
Some checks are pending
Main Extension Distribution Pipeline / Build extension binaries (push) Waiting to run
Main Extension Distribution Pipeline / Code Quality Check (push) Waiting to run
2026-04-06 08:57:49 +02:00
Craig Raw
49f4f03c88 load opencl dynamically at runtime
Some checks failed
Main Extension Distribution Pipeline / Code Quality Check (push) Has been cancelled
Main Extension Distribution Pipeline / Build extension binaries (push) Has been cancelled
2026-03-31 14:14:12 +02:00
Craig Raw
52380c374d downgrade cuda runtime to v12.8.1 for blackwell support with maximum compatibility 2026-03-31 12:00:02 +02:00
Craig Raw
d2f77a1391 add ufsecp_progress function
Some checks are pending
Main Extension Distribution Pipeline / Build extension binaries (push) Waiting to run
Main Extension Distribution Pipeline / Code Quality Check (push) Waiting to run
2026-03-30 13:49:11 +02:00
Craig Raw
70c00e5feb update uf library to 3.50.0, and fix multi-GPU support 2026-03-30 11:48:54 +02:00
Craig Raw
cfdc1e98ab undefine int128 builtin on windows 2026-03-27 09:32:14 +02:00
Craig Raw
5d06a2a6d9 revert windows 4x64 compile change 2026-03-27 08:43:43 +02:00
Craig Raw
ba78f710b6 exclude test/audit targets 2026-03-27 08:05:18 +02:00
Craig Raw
dd92c0b64c switch back to upstream uf lib 2026-03-27 07:37:44 +02:00
Craig Raw
dbc18f39d5 use smaller stack frames on windows (4x64) 2026-03-26 16:23:48 +02:00
Craig Raw
df01b4271e use w=12 for cpu precompute table 2026-03-26 15:19:42 +02:00
Craig Raw
5fd08f3665 add opencl headers for windows build 2026-03-26 13:06:12 +02:00
Craig Raw
5d23aa1a98 fix: add opencl external headers to gpu backend include path 2026-03-26 12:50:51 +02:00
Craig Raw
242d374959 switch to fork of uf library 2026-03-26 12:12:01 +02:00
Craig Raw
c355230549 add workaround for stray brace issue 2026-03-26 10:39:06 +02:00
Craig Raw
9ddfc41f2e update uf lib to fix windows build issue 2026-03-25 16:44:42 +02:00
Craig Raw
d004b0d2cf further github workflow improvements 2026-03-25 16:03:36 +02:00
Craig Raw
2321e226c7 use clang-cl instead of msvc for windows builds 2026-03-25 15:28:48 +02:00
Craig Raw
6d3e7bb04a fix metal path after scalar_mul_generator_lut was removed 2026-03-25 11:47:05 +02:00
Craig Raw
cfa9cd6021 update uf library and implement unchecked variants on cuda path 2026-03-25 11:23:07 +02:00
Craig Raw
16f1099129 avoid building source graph 2026-03-20 15:34:54 +02:00
Craig Raw
c06a48cb65 fix formatting 2026-03-20 15:19:00 +02:00
Craig Raw
f40202af66 add full pipeline (phases 1-6) to metal backend with batch inversion and label matching on gpu 2026-03-20 15:07:26 +02:00
Craig Raw
3fed24022f add full pipeline (phases 1-6) to gpu with batch inversion for spend key matching 2026-03-20 14:47:33 +02:00
Craig Raw
48bcc4eea4 add predecomposed scan key and update metal fused kernels to match cuda/opencl pipeline 2026-03-20 13:24:47 +02:00
Craig Raw
89e800b925 change w=12 for smaller lut lookup table 2026-03-20 12:01:37 +02:00
Craig Raw
71fd968e19 update uf library and implement predecomposed scan key plan 2026-03-20 10:22:11 +02:00
Craig Raw
11a2ceea8d install opencl on linux builds 2026-03-19 14:28:59 +02:00
Craig Raw
67717e5d14 default to arch x86-64-v2 on x86_64 builds for intel nuc support 2026-03-19 13:30:20 +02:00
Craig Raw
f64a99a86d set min cmake target of 13.0 on macos 2026-03-19 12:11:23 +02:00
Craig Raw
c726e5fb89 fix long string limit compile issue on windows 2026-03-19 11:54:07 +02:00
Craig Raw
e0efcf8ff8 default to arch x86-64-v3 on x86_64 builds 2026-03-19 11:53:08 +02:00
Craig Raw
a12b852d90 auto enable cuda and opencl builds based on nvcc availability 2026-03-19 10:59:18 +02:00
Craig Raw
21ab7aeb5d fix gpu build workflow on windows/openssl 2 2026-03-19 10:39:26 +02:00
Craig Raw
41f7e4af3d fix gpu build workflow on windows/openssl 2026-03-19 10:29:01 +02:00
Craig Raw
61fdf3d4c9 fix gpu build workflow 2026-03-19 09:52:40 +02:00
Craig Raw
8c99b112f0 github actions: create gpu build and build with metal support by default 2026-03-19 09:47:59 +02:00
Craig Raw
d745654b82 fix formatting 2026-03-19 08:13:17 +02:00
Craig Raw
088e885e6b add lut support for metal path 2026-03-18 15:48:56 +02:00
Craig Raw
ee92d2f6aa add lut support for opencl path 2026-03-18 15:43:12 +02:00