Promote latest kernel artifacts to main
Browse files- README.md +0 -9
- benchmarks/benchmark.py +2 -0
- build/{torch212-cxx11-cu132-x86_64-linux/_flashrt_qkv_cache_rope_cuda_fed80c9.abi3.so → torch211-cxx11-cu128-x86_64-linux/_flashrt_qkv_cache_rope_cuda_7db5ebe.abi3.so} +2 -2
- build/torch211-cxx11-cu128-x86_64-linux/_ops.py +3 -3
- build/torch211-cxx11-cu128-x86_64-linux/metadata.json +1 -1
- build/{torch211-cxx11-cu128-x86_64-linux/_flashrt_qkv_cache_rope_cuda_fed80c9.abi3.so → torch211-cxx11-cu130-x86_64-linux/_flashrt_qkv_cache_rope_cuda_7db5ebe.abi3.so} +2 -2
- build/torch211-cxx11-cu130-x86_64-linux/_ops.py +3 -3
- build/torch211-cxx11-cu130-x86_64-linux/metadata.json +1 -1
- build/{torch211-cxx11-cu130-x86_64-linux/_flashrt_qkv_cache_rope_cuda_fed80c9.abi3.so → torch212-cxx11-cu130-x86_64-linux/_flashrt_qkv_cache_rope_cuda_7db5ebe.abi3.so} +2 -2
- build/torch212-cxx11-cu130-x86_64-linux/_ops.py +3 -3
- build/torch212-cxx11-cu130-x86_64-linux/metadata.json +1 -1
- build/{torch212-cxx11-cu130-x86_64-linux/_flashrt_qkv_cache_rope_cuda_fed80c9.abi3.so → torch212-cxx11-cu132-x86_64-linux/_flashrt_qkv_cache_rope_cuda_7db5ebe.abi3.so} +2 -2
- build/torch212-cxx11-cu132-x86_64-linux/_ops.py +3 -3
- build/torch212-cxx11-cu132-x86_64-linux/metadata.json +1 -1
README.md
DELETED
|
@@ -1,9 +0,0 @@
|
|
| 1 |
-
# flashrt/flashrt-qkv-cache-rope
|
| 2 |
-
|
| 3 |
-
This repository is a compatibility mirror for older `kernels` clients
|
| 4 |
-
that resolve repositories through the default Hugging Face model repo API.
|
| 5 |
-
|
| 6 |
-
Canonical Kernel Hub repo: https://huggingface.co/kernels/flashrt/flashrt-qkv-cache-rope
|
| 7 |
-
|
| 8 |
-
Do not edit this mirror by hand. It is generated from the Kernel Hub
|
| 9 |
-
`vN` branches and contains the same `build/**` artifacts.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
benchmarks/benchmark.py
CHANGED
|
@@ -894,6 +894,7 @@ def main() -> None:
|
|
| 894 |
if args.shapes in ("smoke", "all"):
|
| 895 |
results.append(run_joint3(ops, "joint3_small", 64, 8, 4, 8, 128, args))
|
| 896 |
results.append(run_kvcache_gqa(ops, "pi05_decoder_gqa_kvcache", 1, 10, 8, 1, 256, args))
|
|
|
|
| 897 |
if args.shapes in ("headline", "all"):
|
| 898 |
results.append(run_joint3(ops, "joint3_vla", 2520, 16, 16, 24, 128, args))
|
| 899 |
results.append(run_decode_q(ops, "decode_q_stage_h24", 24, args))
|
|
@@ -901,6 +902,7 @@ def main() -> None:
|
|
| 901 |
results.append(run_decode_kv(ops, "decode_kvwrite_devpos_h8", 8, True, args))
|
| 902 |
if args.shapes == "headline":
|
| 903 |
results.append(run_kvcache_gqa(ops, "pi05_decoder_gqa_kvcache", 1, 10, 8, 1, 256, args))
|
|
|
|
| 904 |
|
| 905 |
for r in results:
|
| 906 |
print(
|
|
|
|
| 894 |
if args.shapes in ("smoke", "all"):
|
| 895 |
results.append(run_joint3(ops, "joint3_small", 64, 8, 4, 8, 128, args))
|
| 896 |
results.append(run_kvcache_gqa(ops, "pi05_decoder_gqa_kvcache", 1, 10, 8, 1, 256, args))
|
| 897 |
+
results.append(run_kvcache_gqa(ops, "pi05_thor_decoder_gqa_kvcache", 1, 50, 8, 1, 256, args))
|
| 898 |
if args.shapes in ("headline", "all"):
|
| 899 |
results.append(run_joint3(ops, "joint3_vla", 2520, 16, 16, 24, 128, args))
|
| 900 |
results.append(run_decode_q(ops, "decode_q_stage_h24", 24, args))
|
|
|
|
| 902 |
results.append(run_decode_kv(ops, "decode_kvwrite_devpos_h8", 8, True, args))
|
| 903 |
if args.shapes == "headline":
|
| 904 |
results.append(run_kvcache_gqa(ops, "pi05_decoder_gqa_kvcache", 1, 10, 8, 1, 256, args))
|
| 905 |
+
results.append(run_kvcache_gqa(ops, "pi05_thor_decoder_gqa_kvcache", 1, 50, 8, 1, 256, args))
|
| 906 |
|
| 907 |
for r in results:
|
| 908 |
print(
|
build/{torch212-cxx11-cu132-x86_64-linux/_flashrt_qkv_cache_rope_cuda_fed80c9.abi3.so → torch211-cxx11-cu128-x86_64-linux/_flashrt_qkv_cache_rope_cuda_7db5ebe.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d84eebb1e3a1291e0f7aeeda723efd357119c828a842acd5b1cde13c690ede36
|
| 3 |
+
size 3337896
|
build/torch211-cxx11-cu128-x86_64-linux/_ops.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _flashrt_qkv_cache_rope_cuda_7db5ebe
|
| 3 |
+
ops = torch.ops._flashrt_qkv_cache_rope_cuda_7db5ebe
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
+
return f"_flashrt_qkv_cache_rope_cuda_7db5ebe::{op_name}"
|
build/torch211-cxx11-cu128-x86_64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "flashrt-qkv-cache-rope",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 1 |
{
|
| 2 |
"name": "flashrt-qkv-cache-rope",
|
| 3 |
+
"id": "_flashrt_qkv_cache_rope_cuda_7db5ebe",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
build/{torch211-cxx11-cu128-x86_64-linux/_flashrt_qkv_cache_rope_cuda_fed80c9.abi3.so → torch211-cxx11-cu130-x86_64-linux/_flashrt_qkv_cache_rope_cuda_7db5ebe.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:daae4e6e4e813f1a043028b435c33ba18b965942adb3177e58ec5aafaeab1a53
|
| 3 |
+
size 3234736
|
build/torch211-cxx11-cu130-x86_64-linux/_ops.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _flashrt_qkv_cache_rope_cuda_7db5ebe
|
| 3 |
+
ops = torch.ops._flashrt_qkv_cache_rope_cuda_7db5ebe
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
+
return f"_flashrt_qkv_cache_rope_cuda_7db5ebe::{op_name}"
|
build/torch211-cxx11-cu130-x86_64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "flashrt-qkv-cache-rope",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 1 |
{
|
| 2 |
"name": "flashrt-qkv-cache-rope",
|
| 3 |
+
"id": "_flashrt_qkv_cache_rope_cuda_7db5ebe",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
build/{torch211-cxx11-cu130-x86_64-linux/_flashrt_qkv_cache_rope_cuda_fed80c9.abi3.so → torch212-cxx11-cu130-x86_64-linux/_flashrt_qkv_cache_rope_cuda_7db5ebe.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8face78625291d76ed099d8cd865545f77776ab282df85c85c91f8a3e3fc3835
|
| 3 |
+
size 3240672
|
build/torch212-cxx11-cu130-x86_64-linux/_ops.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _flashrt_qkv_cache_rope_cuda_7db5ebe
|
| 3 |
+
ops = torch.ops._flashrt_qkv_cache_rope_cuda_7db5ebe
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
+
return f"_flashrt_qkv_cache_rope_cuda_7db5ebe::{op_name}"
|
build/torch212-cxx11-cu130-x86_64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "flashrt-qkv-cache-rope",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 1 |
{
|
| 2 |
"name": "flashrt-qkv-cache-rope",
|
| 3 |
+
"id": "_flashrt_qkv_cache_rope_cuda_7db5ebe",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
build/{torch212-cxx11-cu130-x86_64-linux/_flashrt_qkv_cache_rope_cuda_fed80c9.abi3.so → torch212-cxx11-cu132-x86_64-linux/_flashrt_qkv_cache_rope_cuda_7db5ebe.abi3.so}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd43519d16be254d6fef61be47a1ed52c0049c1c2f02b655224645749fdbde97
|
| 3 |
+
size 3232480
|
build/torch212-cxx11-cu132-x86_64-linux/_ops.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
import torch
|
| 2 |
-
from . import
|
| 3 |
-
ops = torch.ops.
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
-
return f"
|
|
|
|
| 1 |
import torch
|
| 2 |
+
from . import _flashrt_qkv_cache_rope_cuda_7db5ebe
|
| 3 |
+
ops = torch.ops._flashrt_qkv_cache_rope_cuda_7db5ebe
|
| 4 |
|
| 5 |
def add_op_namespace_prefix(op_name: str):
|
| 6 |
"""
|
| 7 |
Prefix op by namespace.
|
| 8 |
"""
|
| 9 |
+
return f"_flashrt_qkv_cache_rope_cuda_7db5ebe::{op_name}"
|
build/torch212-cxx11-cu132-x86_64-linux/metadata.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"name": "flashrt-qkv-cache-rope",
|
| 3 |
-
"id": "
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|
|
|
|
| 1 |
{
|
| 2 |
"name": "flashrt-qkv-cache-rope",
|
| 3 |
+
"id": "_flashrt_qkv_cache_rope_cuda_7db5ebe",
|
| 4 |
"version": 1,
|
| 5 |
"license": "Apache-2.0",
|
| 6 |
"python-depends": [],
|