Hi,
I am debugging a kernel using cuda-gdb
I launch the kernel with 326 blocks each having 32 threads.
While debugging I can switch to any block from 0 to 111, but any request for blocks 112 to 326 fails ! I want to know why. Please help me figure out why. Is this a bug?
I am using following syntax
“cuda block 200” to change focus
Here is the log:
[Launch of CUDA Kernel 0 (kernel<<<(326,1,1),(32,1,1)>>>) on Device 0]
[Switching focus to CUDA kernel 0, grid 1, block (0,0,0), thread (0,0,0), device 0, sm 2, warp 0, lane 0]
Breakpoint 1, kernel<<<(326,1,1),(32,1,1)>>> (block_offset=0, positions=0x200100000, forces=0x200127200, target_list=0x200200000,
source_list=0x200144800, num_interaction_pairs=0x200200c00, source_start_array=0x200201200, debug_array=0x200300000) at gpu.cu:128
128 int blockId = block_offset + blockIdx.x;
(cuda-gdb) cuda block 10
[Switching focus to CUDA kernel 0, grid 1, block (10,0,0), thread (0,0,0), device 0, sm 10, warp 1, lane 0]
128 int blockId = block_offset + blockIdx.x;
(cuda-gdb) cuda block 100
[Switching focus to CUDA kernel 0, grid 1, block (100,0,0), thread (0,0,0), device 0, sm 10, warp 7, lane 0]
128 int blockId = block_offset + blockIdx.x;
b cuda block 200
Request cannot be satisfied. CUDA focus unchanged.
(cuda-gdb) cuda block 300
Request cannot be satisfied. CUDA focus unchanged.[/b]