derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
30 lines
2.2 KiB
Plaintext
30 lines
2.2 KiB
Plaintext
# RUN: %stinkytofu-opt --arch gfx1250 %s --StinkyBuildImplicitDependencyPass --ds-read-order=AscendingCache --StinkyDAGSchedulerPass --print-output
|
|
#
|
|
# AscendingCache zigzag: AABB → ABBA across WMMA groups.
|
|
# Original: A0A0 A1A1 B0B0 B1B1. A0+B0 feed WMMA0, A1+B1 feed WMMA1.
|
|
# Zigzag alternates direction: group 0 ascending (A0 B0), group 1 descending (B1 A1).
|
|
# Result: A0A0 B0B0 B1B1 A1A1 — B0 and B1 adjacent for cache reuse.
|
|
#
|
|
# CHECK: v[8:11]{{.*}}ds_load_b128
|
|
# CHECK: v[12:15]{{.*}}ds_load_b128
|
|
# CHECK: v[60:63]{{.*}}ds_load_b128
|
|
# CHECK: v[64:67]{{.*}}ds_load_b128
|
|
# CHECK: v[68:71]{{.*}}ds_load_b128
|
|
# CHECK: v[72:75]{{.*}}ds_load_b128
|
|
# CHECK: v[16:19]{{.*}}ds_load_b128
|
|
# CHECK: v[20:23]{{.*}}ds_load_b128
|
|
|
|
st.func @test() {
|
|
^entry:
|
|
v[8:11] = "st.ds_load_b128"(v100) { issueCycles = 1, latencyCycles = 56, mod.ds = { na = 1, offset = 0, gds = false }, mod.memtoken = { tokens = [0] } }
|
|
v[12:15] = "st.ds_load_b128"(v100) { issueCycles = 1, latencyCycles = 56, mod.ds = { na = 1, offset = 32, gds = false }, mod.memtoken = { tokens = [0] } }
|
|
v[16:19] = "st.ds_load_b128"(v100) { issueCycles = 1, latencyCycles = 56, mod.ds = { na = 1, offset = 8192, gds = false }, mod.memtoken = { tokens = [0] } }
|
|
v[20:23] = "st.ds_load_b128"(v100) { issueCycles = 1, latencyCycles = 56, mod.ds = { na = 1, offset = 8224, gds = false }, mod.memtoken = { tokens = [0] } }
|
|
v[60:63] = "st.ds_load_b128"(v101) { issueCycles = 1, latencyCycles = 56, mod.ds = { na = 1, offset = 0, gds = false }, mod.memtoken = { tokens = [0] } }
|
|
v[64:67] = "st.ds_load_b128"(v101) { issueCycles = 1, latencyCycles = 56, mod.ds = { na = 1, offset = 32, gds = false }, mod.memtoken = { tokens = [0] } }
|
|
v[68:71] = "st.ds_load_b128"(v101) { issueCycles = 1, latencyCycles = 56, mod.ds = { na = 1, offset = 256, gds = false }, mod.memtoken = { tokens = [0] } }
|
|
v[72:75] = "st.ds_load_b128"(v101) { issueCycles = 1, latencyCycles = 56, mod.ds = { na = 1, offset = 288, gds = false }, mod.memtoken = { tokens = [0] } }
|
|
v[0:7] = "st.v_wmma_f32_16x16x16_bf16"(v[8:15], v[60:67], v[0:7]) { mod.mfma = { reuseA = false, reuseB = false } }
|
|
v[0:7] = "st.v_wmma_f32_16x16x16_bf16"(v[16:23], v[68:75], v[0:7]) { mod.mfma = { reuseA = false, reuseB = false } }
|
|
}
|