56fd68e7dd
CCCL (CUDA C++ Core Libraries) provides: - CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk) - Thrust: high-level parallel algorithms (transform_reduce, sort, scan) - libcudacxx: CUDA C++ standard library (atomics, barriers, memory) - cudax: experimental features (memory resources, allocators) - Tuning policies: per-SM hardware-specific algorithm parameters Competition optimization vectors mapped to CCCL: - Output TPS (83% weight): warp_reduce, block_reduce, device_topk - Input TPS (14% weight): device_scan, block_load, prefetch - Cache TPS (3% weight): prefix caching strategy patterns - Memory (0.9 util): pooled/cached/buddy allocators Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only) License: Apache-2.0
23 lines
949 B
Plaintext
23 lines
949 B
Plaintext
CHECK: [step 0] initial array
|
|
CHECK-NEXT: 1 1 1 1
|
|
CHECK-NEXT: 1 1 1 1
|
|
CHECK-NEXT: 1 1 1 1
|
|
CHECK-NEXT: [step 1] scan horizontally
|
|
CHECK-NEXT: 1 2 3 4
|
|
CHECK-NEXT: 1 2 3 4
|
|
CHECK-NEXT: 1 2 3 4
|
|
CHECK-NEXT: [step 2] transpose array
|
|
CHECK-NEXT: 1 1 1
|
|
CHECK-NEXT: 2 2 2
|
|
CHECK-NEXT: 3 3 3
|
|
CHECK-NEXT: 4 4 4
|
|
CHECK-NEXT: [step 3] scan transpose horizontally
|
|
CHECK-NEXT: 1 2 3
|
|
CHECK-NEXT: 2 4 6
|
|
CHECK-NEXT: 3 6 9
|
|
CHECK-NEXT: 4 8 12
|
|
CHECK-NEXT: [step 4] transpose the transpose
|
|
CHECK-NEXT: 1 2 3 4
|
|
CHECK-NEXT: 2 4 6 8
|
|
CHECK-NEXT: 3 6 9 12
|