tpoisonooo/how-to-optimize-gemm

Issues

• 运行make run （MMult_cuda_2)时报错 CUDA error

  #23 opened 2 years ago by cloud9wj
  5

• cuda版本非m=n=k运算出错

  #19 opened 2 years ago by seth-lu
  1

• 关于测试浮点峰值的问题

  #15 opened 2 years ago by jeezrick
  2

• how to overlap the share2register and computing process?

  #14 opened 3 years ago by YijiaZhao
  6

• cuda代码中错误

  #17 opened 2 years ago by liuxubit
  2

• Why 2 kernels in armv8/MMult_4x4_13?

  #8 opened 3 years ago by Sh-Zh-7
  4

• fp32 gemm存在微小误差

  #16 opened 2 years ago by anaivebird
  2

• question about gflops benchmark

  #6 opened 3 years ago by Wanger-SJTU
  2

• about ldg32_nc_0

  #13 opened 3 years ago by YijiaZhao
  3

• In cuda v12, the a_sts_addr and a_base are redundant?

  #12 opened 3 years ago by Arsmart123
  0

• v11, line 20: why we define a p but not used it? line 29, why we use long type?

  #11 opened 3 years ago by Arsmart123
  4

• Row-majored or Column-majored

  #4 opened 5 years ago by LeoLai930603
  2

• packing analysis

  #1 opened 6 years ago by tpoisonooo
  0