aboutsummaryrefslogtreecommitdiff
path: root/examples
diff options
context:
space:
mode:
authorOnur Gündüz <ogunduz@gmail.com>2026-06-28 07:19:48 +0200
committerOnur Gündüz <ogunduz@gmail.com>2026-06-28 07:19:48 +0200
commitbc940168825da277f5897c89a5591d6fab2315c5 (patch)
tree0a84ed6b845e29a4c6d383d058c902ee8ad36fbd /examples
parent9807b449c66b06b708766753a3d38ea9022537d8 (diff)
downloadcolabctl-bc940168825da277f5897c89a5591d6fab2315c5.tar.gz
colabctl-bc940168825da277f5897c89a5591d6fab2315c5.zip
Add CUDA vector addition example
Diffstat (limited to 'examples')
-rw-r--r--examples/cuda_vec_add.py49
1 files changed, 49 insertions, 0 deletions
diff --git a/examples/cuda_vec_add.py b/examples/cuda_vec_add.py
new file mode 100644
index 0000000..cc9c4f5
--- /dev/null
+++ b/examples/cuda_vec_add.py
@@ -0,0 +1,49 @@
+import subprocess
+
+src = r"""
+#include <stdio.h>
+
+__global__ void vec_add(float *a, float *b, float *c, int n) {
+ int i = blockIdx.x * blockDim.x + threadIdx.x;
+ if (i < n) c[i] = a[i] + b[i];
+}
+
+int main() {
+ const int N = 1 << 20;
+ float *ha = new float[N], *hb = new float[N], *hc = new float[N];
+ for (int i = 0; i < N; i++) { ha[i] = (float)i; hb[i] = (float)(N - i); }
+
+ float *da, *db, *dc;
+ cudaMalloc(&da, N * sizeof(float));
+ cudaMalloc(&db, N * sizeof(float));
+ cudaMalloc(&dc, N * sizeof(float));
+ cudaMemcpy(da, ha, N * sizeof(float), cudaMemcpyHostToDevice);
+ cudaMemcpy(db, hb, N * sizeof(float), cudaMemcpyHostToDevice);
+
+ vec_add<<<(N + 255) / 256, 256>>>(da, db, dc, N);
+ cudaDeviceSynchronize();
+
+ cudaMemcpy(hc, dc, N * sizeof(float), cudaMemcpyDeviceToHost);
+ printf("c[0] = %.0f\n", hc[0]);
+ printf("c[512] = %.0f\n", hc[512]);
+ printf("c[N-1] = %.0f\n", hc[N - 1]);
+ printf("All == N? %s\n", (hc[0] == N && hc[512] == N && hc[N-1] == N) ? "yes" : "no");
+
+ delete[] ha; delete[] hb; delete[] hc;
+ cudaFree(da); cudaFree(db); cudaFree(dc);
+}
+"""
+
+with open('/tmp/vadd.cu', 'w') as f:
+ f.write(src)
+
+r = subprocess.run(['nvcc', '-O2', '-o', '/tmp/vadd', '/tmp/vadd.cu'],
+ capture_output=True, text=True)
+if r.returncode != 0:
+ print("nvcc stderr:", r.stderr)
+else:
+ print("Compiled OK")
+ out = subprocess.run(['/tmp/vadd'], capture_output=True, text=True)
+ print(out.stdout)
+ if out.stderr:
+ print("run stderr:", out.stderr)