wu-arch/vortex
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Commit Graph

  • 0ad87bde81 Implement WU architecture support wu-architecture abnerhexu 2026-05-25 19:25:05 +08:00
  • 323ed7d7e9 Update Vortex core for Blackwell tensor instructions abnerhexu 2026-05-06 14:50:54 +08:00
  • cb912d3b8b Add Blackwell tensor RTL scaffolding abnerhexu 2026-04-25 10:15:31 +08:00
  • f1d0fac518 Change to 8-core Volta/Ampere config Hansung Kim 2025-01-28 22:36:58 -08:00
  • c8529c4339 Disable EXT_T_HOPPER atm for flash runs Hansung Kim 2024-11-08 21:52:52 -08:00
  • cf000afc8f tensor: Remove unused a[2] and a[3] ports for FP32 DPU Hansung Kim 2024-11-08 14:34:47 -08:00
  • 8dc2a25e32 oopsie 3 Richard Yan 2024-11-02 14:52:43 -07:00
  • d794e055b6 oopsie 2 Richard Yan 2024-11-02 14:51:52 -07:00
  • ed61418ebf oopsie Richard Yan 2024-11-01 02:55:38 -07:00
  • 2e3ea060a5 gate operand read Richard Yan 2024-11-01 02:44:35 -07:00
  • ef902614ff tensor: Fix race in inflight_tensor counter Hansung Kim 2024-10-29 14:10:04 -07:00
  • 1013a74abd tensor: Switch back to hopper + 4 cores Hansung Kim 2024-10-28 23:39:39 -07:00
  • 19876ab9fd tensor: Fix wrong writeback bit Hansung Kim 2024-10-28 21:47:25 -07:00
  • 8a66b5ed89 tensor: Connect SMEM addr/rf IO Hansung Kim 2024-10-28 19:42:02 -07:00
  • 4376bd33a2 tensor: Decode rs1/rs2 of HGMMA for smem addresses Hansung Kim 2024-10-28 19:41:37 -07:00
  • 72db04cec0 tensor: Switch to 8cores, non-hopper config Hansung Kim 2024-10-27 19:47:22 -07:00
  • 3e67ddd6c6 tensor: Properly guard tc_rf_if for non-hopper Hansung Kim 2024-10-27 17:55:09 -07:00
  • 1bc4afe2bb tensor: Bore tensor regfile IO to execute units Hansung Kim 2024-10-24 20:32:18 -07:00
  • c88fd89f1f tensor: Don't make initiate_valid depend on ready Hansung Kim 2024-10-24 17:47:54 -07:00
  • b64e53ff02 Merge branch 'rtl' of github.com:hansungk/vortex-private into rtl Richard Yan 2024-10-24 16:51:22 -07:00
  • 155cbb0abc tc rf read port Richard Yan 2024-10-24 16:51:15 -07:00
  • 40565de8cd tensor: Fix initiate sync with meta queue when !commit.ready Hansung Kim 2024-10-24 16:41:54 -07:00
  • 3ebeb43568 tensor: Fix inflight_tensor decrement, add under/overflow checks Hansung Kim 2024-10-24 14:36:29 -07:00
  • 8337488ed3 tensor: Don't check invalid writeback reg for ghost writes Hansung Kim 2024-10-24 14:32:58 -07:00
  • e855a47295 Add missing commit_if.tensor bit inits Hansung Kim 2024-10-24 13:28:30 -07:00
  • c77a25c968 tensor: Add missing HOPPER guard Hansung Kim 2024-10-23 20:33:45 -07:00
  • 78df981366 tensor: Simply metadata queue Hansung Kim 2024-10-22 22:01:18 -07:00
  • 69cbbdd89b tensor: Consider inflight ops for HGMMA blocking Hansung Kim 2024-10-22 21:32:33 -07:00
  • 98eb7cb594 tensor: Block both HGMMA/HGMMA_WAIT at scoreboard Hansung Kim 2024-10-22 21:10:55 -07:00
  • 83979c3341 tensor: Fully connect writeback IO Hansung Kim 2024-10-22 20:17:00 -07:00
  • 47dff74d3a tensor: Fix commit/metadata logic for HGMMA Hansung Kim 2024-10-22 20:01:37 -07:00
  • 3abaaff16f tensor: Fix tag and data assignment for p0/p1 bus Hansung Kim 2024-10-22 17:47:04 -07:00
  • 8a8f682194 tensor: Bore smem IO from core to tensor core Hansung Kim 2024-10-22 17:42:30 -07:00
  • 9131558950 tensor: Connect Chisel-generated TensorCoreDecoupled module Hansung Kim 2024-10-22 15:15:38 -07:00
  • 32ccdeef01 Merge branch 'tensor-decoupled' into rtl Hansung Kim 2024-10-21 22:57:07 -07:00
  • 0f06afc3ef Update doc Hansung Kim 2024-10-21 22:37:20 -07:00
  • cde8da1f3b add tag to tc smem interface Richard Yan 2024-10-17 14:48:39 -07:00
  • 4dcbc31a88 tensor: Separate async commit from tensor commit Hansung Kim 2024-10-11 21:32:20 -07:00
  • 717fe7ff29 tensor: Fix FSM when commit not ready Hansung Kim 2024-10-11 20:24:31 -07:00
  • 2934b1bd94 tensor: Split execution module from pipeline logic Hansung Kim 2024-10-11 20:09:09 -07:00
  • f7f23e0c05 tensor: Doc update Hansung Kim 2024-10-11 18:00:36 -07:00
  • 42b9d23f83 tensor: Write release logic for hgmma Hansung Kim 2024-10-11 17:27:51 -07:00
  • 408a9b5d2a tensor: Write stall logic for hgmma_wait Hansung Kim 2024-10-11 17:18:01 -07:00
  • 72f9dedce3 tensor: Disable micro-ops for hopper Hansung Kim 2024-10-11 15:59:31 -07:00
  • 100d69ef21 Doc update on accumulator regs Hansung Kim 2024-10-11 15:47:58 -07:00
  • d9ad4809ec Add 'tensor' bit to commit_if and writeback_if Hansung Kim 2024-10-11 15:41:16 -07:00
  • 58c9761829 Revert decode change for hopper Hansung Kim 2024-10-09 21:53:04 -07:00
  • 7ab14445f0 tensor: Test many-commit per execute with an FSM Hansung Kim 2024-10-07 21:29:44 -07:00
  • e8ca4677df Remove old code for pending_instr underflow fix Hansung Kim 2024-10-07 20:21:35 -07:00
  • 4cac1adf7d Add dummy code for decoupled Hopper tensor core Hansung Kim 2024-10-07 17:10:59 -07:00
  • 8bf7f39f04 add tensor core memory interface Richard Yan 2024-10-07 02:56:38 -07:00
  • da54162241 tensor: Add FP16 parameter and expose to VX_core Hansung Kim 2024-09-10 15:25:48 -07:00
  • a968bdd69b tensor: Fix HALF_PRECISION to 1 Hansung Kim 2024-09-08 01:43:21 -07:00
  • 3f8c28c7d6 sync rf, x0 fix Richard Yan 2024-09-05 16:49:01 -07:00
  • 2b1a9b7c16 tensor: Rename & docs Hansung Kim 2024-08-23 16:21:45 -07:00
  • 45f6ae5aad tensor: Doc comments Hansung Kim 2024-08-20 14:46:40 -07:00
  • 20faf87b80 tensor: Rename halves_buf to reduce confusion Hansung Kim 2024-08-19 16:42:02 -07:00
  • 789d873e19 Disable reduce_unit for timing optimization Hansung Kim 2024-08-16 15:28:56 -07:00
  • 715539b2c3 Guard trace printf in mem_scheduler for synthesis Hansung Kim 2024-08-15 06:09:39 -07:00
  • 119c52004e Enable LSU dedup in VX_platform.vh Hansung Kim 2024-08-15 13:39:43 -07:00
  • 1410b39143 Disable trace during the very start of simulation Hansung Kim 2024-08-13 16:01:12 -07:00
  • d39e24643d tensor: Parameterize fedp for fp16/fp32 Hansung Kim 2024-08-12 20:01:56 -07:00
  • 15e93e01d8 tensor: Split packed fp16 and wire correctly to DPU Hansung Kim 2024-08-07 11:16:38 -07:00
  • d4d18c2823 tensor: spurious assert, doc, remove unused param Hansung Kim 2024-07-27 20:53:56 -07:00
  • 4e0dcdadac tensor: Share B operand buffer between threadgroups Hansung Kim 2024-07-27 20:32:44 -07:00
  • 7ad3f64528 tensor: Remove old ready_reg DPI code Hansung Kim 2024-07-27 17:36:02 -07:00
  • 01f6024a76 tensor: Split flops into structural module Hansung Kim 2024-07-26 16:26:48 -07:00
  • 7f43bab0aa tensor: Parameterize result buffer depth Hansung Kim 2024-07-25 16:31:45 -07:00
  • f3afd4a6f9 Hardcode NUM_THREADS/.. only when SYNTHESIS Hansung Kim 2024-07-23 13:58:34 -07:00
  • ed247e21bb Merge branch 'rtl' of https://github.com/hansungk/vortex-private into rtl Richard Yan 2024-07-20 23:37:58 -07:00
  • 7d422cc9b0 pre-submission changes Richard Yan 2024-07-20 23:33:56 -07:00
  • 14b811f334 Update doc Hansung Kim 2024-07-19 16:39:05 -07:00
  • 4b093e3ff7 tensor: Mark PARTIAL_BW on power impact Hansung Kim 2024-06-26 14:25:00 -07:00
  • 9a6fe79bd3 VX_operands_dup: Add counter for RF read/write accesses Hansung Kim 2024-06-22 16:35:23 -07:00
  • fb973a51b6 core_wrapper: Only terminate when core 0 is finished; more slack time Hansung Kim 2024-06-22 16:34:42 -07:00
  • 46fe1897bf VX_platform.vh: Undefine FIRESIM by default Hansung Kim 2024-06-22 16:34:08 -07:00
  • d4f6f8a257 Set NUM_ALU_BLOCKS=2, NUM_FPU_BLOCKS=1 Hansung Kim 2024-06-22 16:33:42 -07:00
  • a9b75dd492 Set default to 4cores/8barriers in VX_config.{h,vh} Hansung Kim 2024-06-12 20:51:15 -07:00
  • 86deaa8e07 Give some slack time for other cores to finish Hansung Kim 2024-06-12 09:46:51 -07:00
  • 1833e8a176 Merge branch 'rtl' of https://github.com/hansungk/vortex-private into rtl Richard Yan 2024-06-12 02:17:01 -07:00
  • 7947df8a6c config change, move ucode Richard Yan 2024-06-12 02:15:08 -07:00
  • 5218292b6f core_wrapper: Use finished and !reset to determine termination Hansung Kim 2024-06-11 16:27:23 -07:00
  • de10d5a957 Don't print from mem_scheduler in reset Hansung Kim 2024-06-09 21:07:22 -07:00
  • 5d5e4a468c Merge remote-tracking branch 'refs/remotes/origin/rtl' into rtl Hansung Kim 2024-06-09 15:58:32 -07:00
  • a47389fc0e Merge branch 'rtl' of https://github.com/hansungk/vortex-private into rtl Richard Yan 2024-06-09 15:15:31 -07:00
  • 67a13410fd gate level sim changes Richard Yan 2024-06-09 15:15:01 -07:00
  • 1bacbb839f Add GPR_DUPLICATED to synthesis in VX_platform.vh Hansung Kim 2024-06-09 14:00:34 -07:00
  • 874a3bf194 Doc changes Hansung Kim 2024-06-09 13:41:00 -07:00
  • 12f8722dd5 Shush display Hansung Kim 2024-06-03 13:04:09 -07:00
  • 9caafb2d8a tensor: Decode rd of macro-op to designate additional accumulator Hansung Kim 2024-05-31 19:17:56 -07:00
  • 0ebbb8e223 tensor: Fix perf counter; comment out dpi Hansung Kim 2024-05-31 00:32:32 -07:00
  • 73293061ea tensor: Enlarge metadata queue Hansung Kim 2024-05-30 23:21:23 -07:00
  • 52bb827a46 Handle BLOCK_SIZE != 1 in dispatch_unit Hansung Kim 2024-05-30 23:20:21 -07:00
  • a02773eb92 Add more efficient dispatch_unit Hansung Kim 2024-05-30 21:55:42 -07:00
  • 574cc0e5f0 tensor: Document configuring queue depths Hansung Kim 2024-05-30 18:32:27 -07:00
  • 83f9f6d84f tensor: Fix sync for dpu warp queue as well Hansung Kim 2024-05-30 18:22:36 -07:00
  • 0a032ab400 tensor: Fix out-of-sync enqueue to dpu and metadata queue Hansung Kim 2024-05-30 18:03:04 -07:00
  • 97f37b1c75 tensor: Add commit stall injection for debugging Hansung Kim 2024-05-30 18:00:26 -07:00
  • 06e0f901ff tensor: Handle backpressure from metadata queue Hansung Kim 2024-05-30 17:34:39 -07:00
  • dfb2276657 tensor: Remove redundant issue queue outside pdu Hansung Kim 2024-05-30 17:29:59 -07:00