Age | Commit message (Collapse) | Author | Lines | |
---|---|---|---|---|
2024-09-27 | README: -v; add references | Birte Kristina Friesel | -2/+49 | |
2024-08-23 | Make STREAM Microbenchmark work with SDK 2024.1.0 | Birte Kristina Friesel | -1/+1 | |
For whatever reason, -flto causes a "Heap Full" error even in an otherwise empty program: * thread #1, name = 'DPUthread0', stop reason = fault 1 (Heap Full) * frame #0: 0x80000350 dpu_code`mem_alloc_nolock(size=64) at alloc.c:32:5 frame #1: 0x80000170 dpu_code`main_kernel1 [inlined] mem_alloc(size=64) at alloc.c:52:21 frame #2: 0x80000168 dpu_code`main_kernel1 at add.c:73 frame #3: 0x80000308 dpu_code`main at add.c:42:12 frame #4: 0x80000050 dpu_code`__bootstrap at crt0.c:36:5 I do not know why this is the case. | ||||
2024-08-19 | HST-S: Add memcpy variant and update HBM eval script | Birte Kristina Friesel | -69/+139 | |
2024-08-15 | GEMV: measure each write (dpu_push_xfer call) separately | Birte Kristina Friesel | -17/+37 | |
2024-07-28 | TRNS DPU: Correctly initalize task so that it's actually repeatable | Birte Kristina Friesel | -0/+1 | |
2024-07-28 | TRNS host: set active_dpus_before at proper location | Birte Kristina Friesel | -1/+1 | |
2024-07-26 | TRNS: decrease number of repetitions | Birte Kristina Friesel | -5/+5 | |
2024-07-26 | GEMV: specify SDK version | Birte Kristina Friesel | -0/+2 | |
2024-07-26 | CPU-DPU: specify SDK version | Birte Kristina Friesel | -1/+5 | |
2024-07-26 | TS: re-add --resume | Birte Kristina Friesel | -12/+11 | |
2024-07-26 | TRNS host: do not needlessly re-allocate DPUs | Birte Kristina Friesel | -0/+1 | |
2024-07-26 | VA hbm: reduce number of repetitions | Birte Kristina Friesel | -1/+1 | |
2024-07-25 | VA nmc: explicitly source SDK | Birte Kristina Friesel | -0/+2 | |
2024-07-25 | TRNS: fix write1 documentation | Birte Kristina Friesel | -4/+4 | |
2024-07-25 | TRNS: specify memcpy node | Birte Kristina Friesel | -10/+32 | |
2024-07-25 | TRNS nmc: explicitly source SDK | Birte Kristina Friesel | -0/+2 | |
2024-07-24 | CPU-DPU: correctly log numa_node_rank | Birte Kristina Friesel | -9/+12 | |
2024-07-24 | CPU-DPU dimes24-transfer: decrease input size; do not pin memory for >20 ranks | Birte Kristina Friesel | -6/+8 | |
2024-07-23 | GEMV: specify memcpy cpu node | Birte Kristina Friesel | -45/+87 | |
2024-07-23 | VA: remove single-node with in/out on same node | Birte Kristina Friesel | -13/+3 | |
2024-07-23 | BS baseline: Add memcpy variant | Birte Kristina Friesel | -42/+153 | |
2024-07-22 | VA baseline: configurable memcpy NUMA binding | Birte Kristina Friesel | -18/+50 | |
2024-07-22 | VA dimes-hetsim-nmc: variable input node | Birte Kristina Friesel | -2/+2 | |
2024-07-18 | GEMV baseline: check for numa_alloc errors | Birte Kristina Friesel | -0/+7 | |
2024-07-18 | GEMV baseline: decrease repetitions | Birte Kristina Friesel | -1/+1 | |
2024-07-18 | GEMV baseline: ooopsie | Birte Kristina Friesel | -1/+1 | |
2024-07-18 | GEMV: add MEMCPY variant | Birte Kristina Friesel | -7/+89 | |
2024-07-18 | GEMV: move ifndef T out of NUMA block; set membind only once | Birte Kristina Friesel | -8/+10 | |
2024-07-18 | TRNS: Update HBM script | Birte Kristina Friesel | -11/+23 | |
2024-07-18 | TRNS dimes nmc: update for new baseline version | Birte Kristina Friesel | -18/+28 | |
2024-07-17 | TRNS baseline: Add NUMA_MEMCPY support | Birte Kristina Friesel | -56/+121 | |
2024-07-17 | VA: do not count move_pages towards timer | Birte Kristina Friesel | -1/+1 | |
2024-07-17 | VA hetsim nmc: nits | Birte Kristina Friesel | -5/+5 | |
2024-07-17 | VA memcpy: HBM support | Birte Kristina Friesel | -25/+63 | |
2024-07-17 | VA dimes benchmarks: re-introduce --resume; add memcpy baseline | Birte Kristina Friesel | -25/+31 | |
2024-07-17 | VA: Remove legacy run scripts | Birte Kristina Friesel | -136/+0 | |
2024-07-17 | VA: Add baseline variant with memcpy overhead (always use local input data) | Birte Kristina Friesel | -5/+81 | |
2024-07-17 | CPU-DPU dimes24-hetsim-transfer: decrease number of repetitions | Birte Kristina Friesel | -3/+3 | |
2024-07-17 | TRNS DPU version: more fine-grained latency output | Birte Kristina Friesel | -37/+65 | |
2024-07-17 | TRNS: document and update dimes-hetsim scripts | Birte Kristina Friesel | -20/+72 | |
2024-07-16 | SEL: consistent dfatool keys | Birte Kristina Friesel | -2/+2 | |
2024-07-16 | SEL hbm: fewer repetitions | Birte Kristina Friesel | -1/+1 | |
2024-07-16 | SEL NMC: log NUMA node of rank | Birte Kristina Friesel | -2/+24 | |
2024-07-15 | update gitignore | Birte Kristina Friesel | -0/+14 | |
2024-07-15 | RED: update hetsim scripts for tinos | Birte Kristina Friesel | -4/+4 | |
2024-07-15 | RED: add run-paper-strong-full | Birte Kristina Friesel | -0/+25 | |
2024-07-15 | SEL: Add DIMES HBM benchmark | Birte Kristina Friesel | -0/+30 | |
2024-07-15 | SEL: Address compiler warnings | Birte Kristina Friesel | -3/+5 | |
2024-07-15 | Revert "VA CPU version: log latency" | Birte Kristina Friesel | -2/+2 | |
This reverts commit 92b065344bfec4a529c821380ddbce65f92eee78. | ||||
2024-07-15 | Revert "GEMV CPU version: log latency" | Birte Kristina Friesel | -2/+2 | |
This reverts commit 4f9ab55e5278ee3b248b20f3541d289ebfb4d585. |