Skip to content

Clean up python tooling and update README #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
sandlbn merged 3 commits into from
Jul 31, 2026
Merged

Clean up python tooling and update README #2

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
b43cfb9
Add logo and clean up python files
sandlbn Jul 30, 2026
8da09cd
Update logo
sandlbn Jul 30, 2026
6f71bfd
Update for 1024 tile
sandlbn Jul 31, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,11 @@
# xe-fuse

<img src="assets/xe-fuse-logo.png" alt="xe-fuse banner" width="100%"/>

[![OpenSSF Scorecard](https://api.scorecard.dev/projects/github.com/IntelLabs/Xe-Fuse/badge)](https://scorecard.dev/viewer/?uri=github.com/IntelLabs/Xe-Fuse)

> ⚠️ **Disclaimer**: This project is currently in active development. The code is **not stable** and **not intended for use in production environments**. Interfaces, features, and behaviors are subject to change without notice.

GEMM epilogue fusion framework for Intel Xe GPUs, built on sycl-tla.

Fuses memory-bound Transformer operations (RMSNorm, SwiGLU, RoPE, GeLU, residual-add, etc.) into GEMM epilogues — the ops execute on data still in registers from the accumulator, avoiding separate kernel launches and global memory round-trips.
Expand Down
Binary file added assets/xe-fuse-logo.png
View file
Open in desktop
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
170 changes: 111 additions & 59 deletions autotune/generate_kernel.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,8 +44,6 @@

import argparse
import json
import os
import sys
from datetime import datetime
from pathlib import Path

Expand All @@ -68,16 +66,21 @@
"",
"typename b::MulOp<>::Arguments mul_args{};",
"",
"typename EVT::Arguments evt_args{accum_args, scale_args, mul_args};"
]
"typename EVT::Arguments evt_args{accum_args, scale_args, mul_args};",
],
},
"k3": {
"name": "K3_RoPE",
"evt_description": "D = RoPE(acc, cos_sin)",
"tile_shape": "_256, _256, _32",
"evt_typedefs": "using EVT = b::RoPE<float>;",
"aux_data": [
{"name": "cos_sin", "type": "float", "shape": "M * N * L", "init_seed": 2024}
{
"name": "cos_sin",
"type": "float",
"shape": "M * N * L",
"init_seed": 2024,
}
],
"evt_args": [
"typename b::AuxLoad<float>::Arguments cs_args;",
Expand All @@ -88,8 +91,8 @@
"",
"typename xe_fuse::XeRoPECompute::Arguments rope_args{};",
"",
"typename EVT::Arguments evt_args{cs_args, rope_args};"
]
"typename EVT::Arguments evt_args{cs_args, rope_args};",
],
},
"k4": {
"name": "K4_RmsNormRoPE",
Expand All @@ -98,7 +101,12 @@
"evt_typedefs": "using EVT = b::RoPEComposed<b::ScaleRows<b::Acc, TileShape, float>, float>;",
"aux_data": [
{"name": "scale", "type": "float", "shape": "M * L", "init_seed": 42},
{"name": "cos_sin", "type": "float", "shape": "M * N * L", "init_seed": 2024}
{
"name": "cos_sin",
"type": "float",
"shape": "M * N * L",
"init_seed": 2024,
},
],
"evt_args": [
"// child 0: ScaleRows (inner tree)",
Expand All @@ -118,8 +126,8 @@
" cute::Stride<int64_t, cute::Int<1>, int64_t>{}, make_shape(M, N, L));",
"",
"typename xe_fuse::XeRoPEComputeTwoChild::Arguments rope_args{};",
"typename EVT::Arguments evt_args{rms_args, cs_args, rope_args};"
]
"typename EVT::Arguments evt_args{rms_args, cs_args, rope_args};",
],
},
"k4v2": {
"name": "K4v2_RoPEScaled",
Expand All @@ -128,7 +136,12 @@
"evt_typedefs": "using EVT = b::RoPEScaled<TileShape, float, float>;",
"aux_data": [
{"name": "scale", "type": "float", "shape": "M * L", "init_seed": 42},
{"name": "cos_sin", "type": "float", "shape": "M * N * L", "init_seed": 2024}
{
"name": "cos_sin",
"type": "float",
"shape": "M * N * L",
"init_seed": 2024,
},
],
"evt_args": [
"// child 0: ColBroadcast<R>",
Expand All @@ -147,17 +160,22 @@
"// root: XeRoPEScaledCompute (merged)",
"typename xe_fuse::XeRoPEScaledCompute::Arguments rope_args{};",
"",
"typename EVT::Arguments evt_args{scale_args, cs_args, rope_args};"
]
"typename EVT::Arguments evt_args{scale_args, cs_args, rope_args};",
],
},
"k0a": {
"name": "K0a_ResidualGamma",
"evt_description": "D = gamma[n] * (acc + residual)",
"tile_shape": "_256, _256, _32",
"evt_typedefs": "using EVT = b::ScaleCols<b::AddResidual<bf16>, TileShape, float>;",
"aux_data": [
{"name": "residual", "type": "bf16", "shape": "M * N * L", "init_seed": 2021},
{"name": "gamma", "type": "float", "shape": "N * L", "init_seed": 99}
{
"name": "residual",
"type": "bf16",
"shape": "M * N * L",
"init_seed": 2021,
},
{"name": "gamma", "type": "float", "shape": "N * L", "init_seed": 99},
],
"evt_args": [
"// RowBroadcast<gamma> (child 0 of outer Mul)",
Expand All @@ -178,8 +196,8 @@
"",
"// Outer Mul",
"typename b::MulOp<>::Arguments mul_args{};",
"typename EVT::Arguments evt_args{gamma_args, inner_args, mul_args};"
]
"typename EVT::Arguments evt_args{gamma_args, inner_args, mul_args};",
],
},
"k2": {
"name": "K2_RmsNormSwiGLU",
Expand All @@ -201,8 +219,8 @@
"",
"// Outer: SwiGLU",
"typename xe_fuse::XePairwiseCompute<xe_fuse::SwiGLUFn>::Arguments swiglu_args{};",
"typename EVT::Arguments evt_args{rms_args, swiglu_args};"
]
"typename EVT::Arguments evt_args{rms_args, swiglu_args};",
],
},
"k2_geglu": {
"name": "K2_RmsNormGeGLU",
Expand All @@ -224,8 +242,8 @@
"",
"// Outer: GeGLU",
"typename xe_fuse::XePairwiseCompute<xe_fuse::GeGLUFn>::Arguments geglu_args{};",
"typename EVT::Arguments evt_args{rms_args, geglu_args};"
]
"typename EVT::Arguments evt_args{rms_args, geglu_args};",
],
},
"k1v2": {
"name": "K1v2_ScaleRowsMerged",
Expand All @@ -245,8 +263,8 @@
"// root: XeScaleRowsCompute (merged)",
"typename xe_fuse::XeScaleRowsCompute::Arguments visitor_args{};",
"",
"typename EVT::Arguments evt_args{scale_args, visitor_args};"
]
"typename EVT::Arguments evt_args{scale_args, visitor_args};",
],
},
"k2v2": {
"name": "K2v2_SwiGLUScaled",
Expand All @@ -266,8 +284,8 @@
"// root: XeScaleRowsSwiGLUCompute (merged)",
"typename xe_fuse::XeScaleRowsSwiGLUCompute::Arguments visitor_args{};",
"",
"typename EVT::Arguments evt_args{scale_args, visitor_args};"
]
"typename EVT::Arguments evt_args{scale_args, visitor_args};",
],
},
"k2v2_geglu": {
"name": "K2v2_GeGLUScaled",
Expand All @@ -287,8 +305,8 @@
"// root: XeScaleRowsGeGLUCompute (merged)",
"typename xe_fuse::XeScaleRowsGeGLUCompute::Arguments visitor_args{};",
"",
"typename EVT::Arguments evt_args{scale_args, visitor_args};"
]
"typename EVT::Arguments evt_args{scale_args, visitor_args};",
],
},
"w8a8_dequant": {
"name": "W8A8_Dequant",
Expand All @@ -304,7 +322,12 @@
"evt_typedefs": "using EVT = b::DequantW8A8<TileShape, float, float>;",
"aux_data": [
{"name": "scale_token", "type": "float", "shape": "M * L", "init_seed": 42},
{"name": "scale_channel", "type": "float", "shape": "N * L", "init_seed": 99}
{
"name": "scale_channel",
"type": "float",
"shape": "N * L",
"init_seed": 99,
},
],
"evt_args": [
"// Inner Mul: Acc * scale_token[m] (ColBroadcast, Idx=0)",
Expand All @@ -325,8 +348,8 @@
"// Compose: EVT = Mul(Mul(Acc, ColBcast), RowBcast)",
"using InnerMul = b::Mul<b::Acc, b::ColBroadcast<0, TileShape, float>>;",
"typename InnerMul::Arguments inner_args{accum_args, token_args, inner_mul_args};",
"typename EVT::Arguments evt_args{inner_args, channel_args, outer_mul_args};"
]
"typename EVT::Arguments evt_args{inner_args, channel_args, outer_mul_args};",
],
},
"w8a8_dequant_biased": {
"name": "W8A8_DequantBiased",
Expand All @@ -342,8 +365,13 @@
"evt_typedefs": "using EVT = b::DequantW8A8Biased<TileShape, float, float, float>;",
"aux_data": [
{"name": "scale_token", "type": "float", "shape": "M * L", "init_seed": 42},
{"name": "scale_channel", "type": "float", "shape": "N * L", "init_seed": 99},
{"name": "bias", "type": "float", "shape": "N * L", "init_seed": 77}
{
"name": "scale_channel",
"type": "float",
"shape": "N * L",
"init_seed": 99,
},
{"name": "bias", "type": "float", "shape": "N * L", "init_seed": 77},
],
"evt_args": [
"// Inner Mul: Acc * scale_token[m] (ColBroadcast, Idx=0)",
Expand Down Expand Up @@ -373,9 +401,9 @@
"using DequantMul = b::DequantW8A8<TileShape, float, float>;",
"typename InnerMul::Arguments inner_args{accum_args, token_args, inner_mul_args};",
"typename DequantMul::Arguments dequant_args{inner_args, channel_args, mid_mul_args};",
"typename EVT::Arguments evt_args{dequant_args, bias_args, add_args};"
]
}
"typename EVT::Arguments evt_args{dequant_args, bias_args, add_args};",
],
},
}


Expand All @@ -387,7 +415,9 @@ def generate_aux_allocations(aux_data: list[dict]) -> str:
name = aux["name"]
shape = aux["shape"]
seed = aux.get("init_seed", 2020)
lines.append(f" cutlass::DeviceAllocation<{ctype}> block_{name}(static_cast<size_t>({shape}));")
lines.append(
f" cutlass::DeviceAllocation<{ctype}> block_{name}(static_cast<size_t>({shape}));"
)
lines.append(f" initialize_block(block_{name}, {seed});")
return "\n".join(lines)

Expand All @@ -399,6 +429,7 @@ def generate_cpp(spec: dict, defaults: dict | None = None) -> str:
if template_path.exists():
try:
from jinja2 import Template

with open(template_path) as f:
tmpl = Template(f.read())

Expand All @@ -411,8 +442,10 @@ def generate_cpp(spec: dict, defaults: dict | None = None) -> str:
align_cd = spec.get("alignment_cd", 8)
make_gemm_extra = ""
if align_ab != 8 or align_cd != 8:
make_gemm_extra = (f",\n cutlass::layout::RowMajor, cutlass::layout::RowMajor, "
f"{align_ab}, {align_cd}")
make_gemm_extra = (
f",\n cutlass::layout::RowMajor, cutlass::layout::RowMajor, "
f"{align_ab}, {align_cd}"
)

return tmpl.render(
kernel_name=spec["name"],
Expand Down Expand Up @@ -465,8 +498,10 @@ def generate_cpp_inline(spec: dict, defaults: dict | None = None) -> str:
# MakeGemm template args beyond the 7 positional defaults
make_gemm_extra = ""
if align_ab != 8 or align_cd != 8:
make_gemm_extra = (f",\n cutlass::layout::RowMajor, cutlass::layout::RowMajor, "
f"{align_ab}, {align_cd}")
make_gemm_extra = (
f",\n cutlass::layout::RowMajor, cutlass::layout::RowMajor, "
f"{align_ab}, {align_cd}"
)

return f"""\
// Auto-generated xe-fuse kernel benchmark
Expand Down Expand Up @@ -675,46 +710,54 @@ def generate_standalone_cpp(spec: dict, defaults: dict | None = None) -> str:
],
"op_code": [
"xe_fuse::standalone::scale_rows(q, block_D.get(), block_scale.get(), M, N, L);"
]
],
},
"sa_residual_gamma": {
"name": "SA_ResidualGamma",
"evt_description": "D[m,n] = gamma[n] * (D + residual) (standalone)",
"standalone": True,
"aux_data": [
{"name": "residual", "type": "bf16", "shape": "M * N * L", "init_seed": 2021},
{"name": "gamma", "type": "float", "shape": "N * L", "init_seed": 99}
{
"name": "residual",
"type": "bf16",
"shape": "M * N * L",
"init_seed": 2021,
},
{"name": "gamma", "type": "float", "shape": "N * L", "init_seed": 99},
],
"op_code": [
"xe_fuse::standalone::residual_gamma(q, block_D.get(), block_residual.get(),",
" block_gamma.get(), M, N, L);"
]
" block_gamma.get(), M, N, L);",
],
},
"sa_swiglu": {
"name": "SA_SwiGLU",
"evt_description": "D = SwiGLU(D) pairwise (standalone)",
"standalone": True,
"aux_data": [],
"op_code": [
"xe_fuse::standalone::swiglu(q, block_D.get(), M, N, L);"
]
"op_code": ["xe_fuse::standalone::swiglu(q, block_D.get(), M, N, L);"],
},
"sa_rope_scaled": {
"name": "SA_RoPEScaled",
"evt_description": "D = RoPE(D * scale[m], cos_sin) (standalone)",
"standalone": True,
"aux_data": [
{"name": "scale", "type": "float", "shape": "M * L", "init_seed": 42},
{"name": "cos_sin", "type": "float", "shape": "M * N * L", "init_seed": 2024},
{"name": "tmp", "type": "bf16", "shape": "M * N * L", "init_seed": 0}
{
"name": "cos_sin",
"type": "float",
"shape": "M * N * L",
"init_seed": 2024,
},
{"name": "tmp", "type": "bf16", "shape": "M * N * L", "init_seed": 0},
],
"op_code": [
"q.memcpy(block_tmp.get(), block_D.get(), static_cast<size_t>(M) * N * L * sizeof(bf16));",
"q.wait();",
"xe_fuse::standalone::rope_scaled(q, block_D.get(), block_tmp.get(),",
" block_scale.get(), block_cos_sin.get(), M, N, L);"
]
}
" block_scale.get(), block_cos_sin.get(), M, N, L);",
],
},
}


Expand All @@ -723,17 +766,25 @@ def main():
all_presets = {**PRESETS, **STANDALONE_PRESETS}

parser.add_argument("--spec", help="Path to kernel spec JSON file")
parser.add_argument("--preset", choices=list(all_presets.keys()),
help="Use a built-in kernel preset")
parser.add_argument(
"--preset",
choices=list(all_presets.keys()),
help="Use a built-in kernel preset",
)
parser.add_argument("--output", "-o", help="Output .cpp path")
parser.add_argument("--m", type=int, default=4096)
parser.add_argument("--n", type=int, default=4096)
parser.add_argument("--k", type=int, default=4096)
parser.add_argument("--iterations", type=int, default=200)
parser.add_argument("--tile", type=str, default=None,
help="Override tile shape, e.g. '128x256x32' or 'auto'")
parser.add_argument("--list-presets", action="store_true",
help="List available presets and exit")
parser.add_argument(
"--tile",
type=str,
default=None,
help="Override tile shape, e.g. '128x256x32' or 'auto'",
)
parser.add_argument(
"--list-presets", action="store_true", help="List available presets and exit"
)
args = parser.parse_args()

if args.list_presets:
Expand Down Expand Up @@ -762,6 +813,7 @@ def main():
if args.tile:
if args.tile == "auto":
from tile_selector import select_tile

kernel_tag = args.preset.split("_")[0] if args.preset else "bare"
spec["tile_shape"] = select_tile(args.m, args.n, args.k, kernel_tag)
else:
Expand Down
Loading
Loading