gemm_hls
gemm_hls copied to clipboard
spcl
Scalable systolic array-based matrix-matrix multiplication implemented in Vivado HLS for Xilinx FPGAs.
Vitis 2021.2 has broken the repository by wrongly inferring a loop-carried dependency on writing `aSplit` in `Memory.cpp`. The dependence pragma does not fix this. Verification is broken by this (https://github.com/spcl/gemm_hls/issues/27)....
Hi I reproduce the project following under command lines ``` mkdir build cd build cmake ../ -DMM_DATA_TYPE=float -DMM_PARALLELISM_N=32 -DMM_PARALLELISM_M=8 -DMM_MEMORY_TILE_SIZE_N=512 -DMM_MEMORY_TILE_SIZE_M=512 -DMM_ADD_RESOURCE=FAddSub_nodsp -DMM_MULT_RESOURCE=FMul_nodsp make make hw ``` Then run like...
I would like to use this project as a drop in GEMM function. This would necessitate the implementation of support for TransA, TransB and to a lesser importance (to me...
Hi, Running a matrix with the size of 16384 on a DGEMM build returns the following errors: ``` [XRT] ERROR: unable to sync BO: Input/output error [XRT] ERROR: Profiling info...
Hi, I was wondering what is the meaning of DMM_MEMORY_TILE_SIZE_N and DMM_MEMORY_TILE_SIZE_M parameters. Additionally, how will these parameters affect the performance? For example, in the given default parameters, -DMM_PARALLELISM_N=32 -DMM_PARALLELISM_M=8...
