task_id
stringlengths 5
11
| query_token_num
int64 151
4.39k
| Engineering Domain
stringclasses 9
values | prompt
stringlengths 640
112k
| contributor
stringlengths 8
50
|
|---|---|---|---|---|
YH_03
| 207
|
Digital Hardware Design
|
Write a synthesizable C function for matrix multiplication (GEMM) using Vitis HLS, with the following specifications:
• Function signature: void gemm(float A[M_SIZE][K_SIZE], float B[K_SIZE][N_SIZE], float C[M_SIZE][N_SIZE])
• Use #define for M_SIZE, N_SIZE, and K_SIZE as 1024.
• Perform the matrix multiplication: C = A \times B
• Apply the following interface pragmas for HLS:
• Use AXI4 master interfaces (m_axi) for matrices A, B, and C (with separate bundles for each).
• Use an AXI4-Lite slave interface (s_axilite) for function control and return.
• Add #pragma HLS PIPELINE II=1 to pipeline the innermost loop for high throughput.
• Ensure the code is synthesizable with Vitis HLS and suitable for implementation on the VCK5000 board
|
Yingbing Huang, Greg Jun
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.