Yangjunbo operator - #7

Open

spirit13579 wants to merge 374 commits into

kernel-competition-bot:mainfrom

Xdddyy:yangjunbo-operator

This pull request is big! We're only showing the most recent 250 commits

Commits on Jun 8, 2026

update scaled masked softmax and config
woshidiaoxianwang
committed
b9cea0a
update scaled masked softmax and config
woshidiaoxianwang
committed
eb4021c
update scaled masked softmax and config
woshidiaoxianwang
committed
13deb28
update scaled masked softmax and config
woshidiaoxianwang
committed
1217cab
update scaled masked softmax and config
woshidiaoxianwang
committed
6db890a
update scaled masked softmax and config
woshidiaoxianwang
committed
e737fbd
update scaled masked softmax and config
woshidiaoxianwang
committed
694defb
update scaled masked softmax and config
woshidiaoxianwang
committed
14fcf26
update scaled masked softmax and config
woshidiaoxianwang
committed
6907e94
update scaled masked softmax and config
woshidiaoxianwang
committed
351ef70
update scaled masked softmax and config
woshidiaoxianwang
committed
a390865
update scaled masked softmax and config
woshidiaoxianwang
committed
bea4a20
update scaled masked softmax and config
woshidiaoxianwang
committed
18535f0
update scaled masked softmax and config
woshidiaoxianwang
committed
22cef01
update scaled masked softmax and config
woshidiaoxianwang
committed
e3fbbcc
update scaled masked softmax and config
woshidiaoxianwang
committed
f8aa125
update scaled masked softmax and config
woshidiaoxianwang
committed
074674f
update scaled masked softmax and config
woshidiaoxianwang
committed
c9b6257
update scaled masked softmax and config
woshidiaoxianwang
committed
badb7ea
update scaled masked softmax and config
woshidiaoxianwang
committed
f1fcaac
update scaled masked softmax and config
woshidiaoxianwang
committed
df22a2e
update scaled masked softmax and config
woshidiaoxianwang
committed
e894020
update scaled masked softmax and config
woshidiaoxianwang
committed
8c769f2
update scaled masked softmax and config
woshidiaoxianwang
committed
f24f19c
update scaled masked softmax and config
woshidiaoxianwang
committed
8b72eca
update scaled masked softmax and config
woshidiaoxianwang
committed
2d50372
update scaled masked softmax and config
woshidiaoxianwang
committed
d9ab677
update scaled masked softmax and config
woshidiaoxianwang
committed
b1dd8b8
update scaled masked softmax and config
woshidiaoxianwang
committed
4302b0e
update scaled masked softmax and config
woshidiaoxianwang
committed
bd8080c
update scaled masked softmax and config
woshidiaoxianwang
committed
5ea3fcc
update scaled masked softmax and config
woshidiaoxianwang
committed
490fbd0

Commits on Jun 9, 2026

提交 3,4
谢代榆
committed
f68c119
Merge branch 'main' into wsdxw-operator
Xdddyy
authored
ba06c74
Merge pull request #8 from Xdddyy/wsdxw-operator
Xdddyy
authored
d177f15
Merge branch 'main' into liuxingyu-operator
Xdddyy
authored
4a4725e
Merge pull request #19 from Xdddyy/liuxingyu-operator
Xdddyy
authored
711b0d0
Merge pull request #21 from spirit13579/yangjunbo-operator
Xdddyy
authored
415e275
fix 115
Jassicia
authored
ee9d19a
修改第四题
谢代榆
committed
dce8fd7
Fix batched matmul BangC output type
谢代榆
committed
77b8b19
Add mlu operators 115 and 116
Jassicia
committed
0994ba0
Merge branch 'main' into chenzhiyuan-operator
Jassicia
authored
264980e
fix 115 116
Jassicia
authored
e0155a7
Use BangC conv for batched matmul
谢代榆
committed
3d9f839
Reorder and format entries in config file
spirit13579
authored
2f707d2
Tile batched matmul for large matrices
谢代榆
committed
13749c8
Merge branch 'main' into yangjunbo-operator
spirit13579
authored
dcf0d5e
Use row tiling for BangC batched matmul
谢代榆
committed
84874f2
Tile batched matmul across k dimension
谢代榆
committed
fbecf3f
update mlu solution files
woshidiaoxianwang
committed
ea92578
Pack BangC conv filter for batched matmul
谢代榆
committed
23a6cc7
update mlu solution files
woshidiaoxianwang
committed
4c98a5f
Use half BangC matmul for batched matmul
谢代榆
committed
fc49d9b
update mlu solution files
woshidiaoxianwang
committed
47f6993
update mlu solution files
woshidiaoxianwang
committed
837b3b6
update mlu solution files
woshidiaoxianwang
committed
6153cd3
update mlu solution files
woshidiaoxianwang
committed
5eb325a
update mlu solution files
woshidiaoxianwang
committed
6014567
Use half output for BangC matmul tiles
谢代榆
committed
2c766c9
Use vectorized BangC row tiles for batched matmul
谢代榆
committed
c2dc6fb
Return float accumulator for batched matmul
谢代榆
committed
e31f129
update mlu solution files
woshidiaoxianwang
committed
ca46bb8
update mlu solution files
woshidiaoxianwang
committed
573e0f9
update mlu solution files
woshidiaoxianwang
committed
2e25034
039和135优化提交
 kevinzzh17
committed
2616620
135优化提交
 kevinzzh17
committed
7df08d7
135优化提交
 kevinzzh17
committed
8c000b9
135优化提交
 kevinzzh17
committed
e195aa1
135优化提交
 kevinzzh17
committed
d63a293
135优化提交
 kevinzzh17
committed
72ea1df
135优化提交
 kevinzzh17
committed
b983b8c
135优化提交
 kevinzzh17
committed
8e129d0
little modify
segzix
committed
34b24ec
135优化提交
 kevinzzh17
committed
8d50cdb
Merge branch 'shengzixuan-operator'
segzix
committed
29e8839
modify
segzix
committed
fc5e5e9
add mlu
segzix
committed
61c5c91
add mlu
segzix
committed
2800fe7
add mlu
segzix
committed
8de7887
135优化提交
 kevinzzh17
committed
68dd874
add mlu
segzix
committed
88ceb37
135优化提交
 kevinzzh17
committed
4e44453
add mlu
segzix
committed
2088293
135优化提交
 kevinzzh17
committed
dd4f8be
add mlu
segzix
committed
db22871
add mlu
segzix
committed
a8f2a84
Rename 023_Matrix_vector_multiplication_.mlu to Matrix_vector_multiplication_.mlu
spirit13579
authored
d3831af
Rename 100_Adaptive_Max_Pool_2D.mlu to Adaptive_Max_Pool_2D.mlu
spirit13579
authored
cd6d9f0
Update config
spirit13579
authored
b10753f
Merge branch 'main' into yangjunbo-operator
spirit13579
authored
15c489f
add mlu
segzix
committed
1b93500
135优化提交
 kevinzzh17
committed
3f7b952
add mlu
segzix
committed
0c9d3a7
135优化提交
 kevinzzh17
committed
51806c1
add mlu
segzix
committed
a3f2f6b
135优化提交
 kevinzzh17
committed
352ba7a
add mlu
segzix
committed
f6b4812
135优化提交
 kevinzzh17
committed
6e0f95e
135优化提交
 kevinzzh17
committed
fb59224

Commits on Jun 10, 2026

Add 138 GRU forward operator

houhui
authored and
houhui
committed
478a1ac
fix scatter_add: move NRAM alloc into group loop, explicit read-add-write
segzix
committed
5ce9c09
fix scatter_add: accumulate in NRAM, write only on dst-row switch
segzix
committed
0efc25a
add mlu
segzix
committed
e1c4ddd
Fix 138 GRU stream include

houhui
authored and
houhui
committed
230438a
add mlu
segzix
committed
7110f8e
Add 111 masked select operator

houhui
authored and
houhui
committed
5edbc16
Merge branch 'shengzixuan-operator'
segzix
committed
a649ad4
modify config
segzix
committed
0469343
Fix 111 and 138 expected filenames

houhui
authored and
houhui
committed
1b6c054
Fix 138 GRU full evaluator signature

houhui
authored and
houhui
committed
bbac23d
modify config
segzix
committed
85512c3
Merge branch 'shengzixuan-operator'
segzix
committed
ba9f44b
fix 138 GRU: real impl + single bang_func

houhui
authored and
houhui
committed
27d1d74
Implement 138 GRU forward with native GRU

houhui
authored and
houhui
committed
a01a40b
modify config
segzix
committed
8b2f358
Merge branch 'shengzixuan-operator'
segzix
committed
ae4e25d
modify config
segzix
committed
4be5896
Merge branch 'shengzixuan-operator'
segzix
committed
5b20f66
Fix 138 GRU: remove zero-output stub, expose real implementation

authored and
houhui
committed
6aa326e
138: implement threshold-based element select (float16/float32)

authored and
houhui
committed
9bcea8c
modify config
segzix
committed
02d6b3b
Fix GRU_forward.mlu: replace forbidden at::gru with threshold select

authored and
houhui
committed
7fbc265
Merge branch 'shengzixuan-operator'
segzix
committed
13c6f15
retrigger evaluation

authored and
houhui
committed
08fa03c
modify config
segzix
committed
faa0bfb
Merge branch 'shengzixuan-operator'
segzix
committed
739b3ff
GRU_forward: implement GRU from scratch without at::gru

authored and
houhui
committed
f250319
Fix BANG C compile: replace __expf/__tanhf with expf/tanhf

authored and
houhui
committed
9a12c58
Merge branch 'main' into yangjunbo-operator
spirit13579
authored
7f77796
Merge pull request #26 from spirit13579/yangjunbo-operator
spirit13579
authored
f47de84
023
spirit13579
authored
86ee83d
100
spirit13579
authored
d4d87b2
034
spirit13579
authored
9081550
071
spirit13579
authored
370f831
modify config
segzix
committed
53313eb
Merge branch 'shengzixuan-operator'
segzix
committed
dc4dfe2
Update 51
I-Xvai
authored
36db3b8
Update config
I-Xvai
authored
204ac64
51 orz orz
I-Xvai
authored
d680da0
Merge pull request #27 from spirit13579/yangjunbo-operator
spirit13579
authored
fa5b29f
Update 51
I-Xvai
authored
ba58789
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
c89c194
Update Cos.mlu
spirit13579
authored
14554c7
modify config
segzix
committed
5ba0ed6
Add entry point for adaptive max pool 2D kernel
spirit13579
authored
0accb63
Update Argmax_over_a_dimension.mlu
spirit13579
authored
fb55a1d
Update config
spirit13579
authored
0755f78
Update 51
I-Xvai
authored
c1b57cc
Merge branch 'main' into yangjunbo-operator
spirit13579
authored
644e722
Merge pull request #30 from Xdddyy/liuxingyu-operator
I-Xvai
authored
6c84ab3
Merge pull request #29 from spirit13579/yangjunbo-operator
spirit13579
authored
82506bf
modify config
segzix
committed
e8dc36a
Merge branch 'shengzixuan-operator'
segzix
committed
d358f8f
Update 51
I-Xvai
authored
cb57a63
Update config
I-Xvai
authored
7ce0ceb
Merge pull request #31 from Xdddyy/liuxingyu-operator
I-Xvai
authored
f70b220
modify config
segzix
committed
4f9e7fd
modify config
segzix
committed
fd6fc97
modify config
segzix
committed
626eddc
modify config
segzix
committed
194a863
modify config
segzix
committed
1472c29
modify config
segzix
committed
10942ad
modify config
segzix
committed
99f27f2
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
f726c4e
modify config
segzix
committed
0158492
Update Argmax_over_a_dimension.mlu
spirit13579
authored
dd64289
Merge pull request #32 from spirit13579/yangjunbo-operator
spirit13579
authored
7838205
Update config
spirit13579
authored
4db5afe
modify config
segzix
committed
00db313
Merge branch 'main' into yangjunbo-operator
spirit13579
authored
4f5a4eb
Merge pull request #33 from spirit13579/yangjunbo-operator
spirit13579
authored
e8e7f93
modify config
segzix
committed
dddffdb
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
89ecc23
Update Argmax_over_a_dimension.mlu
spirit13579
authored
8dcd5bb
Update config
spirit13579
authored
4882dd0
Merge branch 'main' into yangjunbo-operator
spirit13579
authored
8ba7fb4
Merge pull request #34 from spirit13579/yangjunbo-operator
spirit13579
authored
202bc6d
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
5586c17
Update Argmax_over_a_dimension.mlu
spirit13579
authored
7687e7a
Update config
spirit13579
authored
c04c13e
Merge pull request #35 from spirit13579/yangjunbo-operator
spirit13579
authored
cdab0ac
Update Argmax_over_a_dimension.mlu
spirit13579
authored
3d25f32
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
afb0277
Add new configuration value '071' to config
spirit13579
authored
76457fe
Merge pull request #36 from spirit13579/yangjunbo-operator
spirit13579
authored
1b2153c
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
81ce151
Update Argmax_over_a_dimension.mlu
spirit13579
authored
9ceb2c9
Update config
spirit13579
authored
b19f026
Merge pull request #37 from spirit13579/yangjunbo-operator
spirit13579
authored
b092649
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
29b0790
Update config
spirit13579
authored
568a1f0
Merge pull request #38 from spirit13579/yangjunbo-operator
spirit13579
authored
3b5faf4
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
6a8f4be
Update config
spirit13579
authored
6a0eb24
Update Cos.mlu
spirit13579
authored
7b769c6
Merge pull request #39 from spirit13579/yangjunbo-operator
spirit13579
authored
135694f
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
39af864
Update config
spirit13579
authored
5e5d0f2
Merge pull request #40 from spirit13579/yangjunbo-operator
spirit13579
authored
96a6391
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
94d5994
Update Cos.mlu
spirit13579
authored
f9f82f5
Update config
spirit13579
authored
562b47c
Merge pull request #41 from spirit13579/yangjunbo-operator
spirit13579
authored
1fbc2ed
Update Adaptive_Max_Pool_2D.mlu
spirit13579
authored
77282e2
Update config
spirit13579
authored
71c0a5e
Merge pull request #42 from spirit13579/yangjunbo-operator
spirit13579
authored
2f02fba
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
89f0a2b
Update config
spirit13579
authored
d58aec4
Merge pull request #43 from spirit13579/yangjunbo-operator
spirit13579
authored
13f8aa2

Commits on Jun 11, 2026

Update Matrix_vector_multiplication_.mlu
spirit13579
authored
2c152bd
Update config
spirit13579
authored
4fd4e7d
Merge pull request #44 from spirit13579/yangjunbo-operator
spirit13579
authored
d3f1a3b
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
2ff8a98
Remove duplicate entry from config file
spirit13579
authored
cdca2bd
Merge pull request #45 from spirit13579/yangjunbo-operator
spirit13579
authored
337e575
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
c7624a8
Update config
spirit13579
authored
1fa6acf
Update Cos.mlu
spirit13579
authored
9ff0fd7
Update Cos.mlu
spirit13579
authored
9c94198
Merge pull request #46 from spirit13579/yangjunbo-operator
spirit13579
authored
afa88ef
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
489c15f
Update config
spirit13579
authored
920fda2
Merge pull request #47 from spirit13579/yangjunbo-operator
spirit13579
authored
6298925
Improve precision for dilated conv2d
谢代榆
committed
05e0a96
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
1cc0b03
Update Cos.mlu
spirit13579
authored
d0ea27f
Add new configuration value '071'
spirit13579
authored
7ef6e0b
Merge branch 'main' into yangjunbo-operator
spirit13579
authored
70e3ac9
Merge pull request #48 from spirit13579/yangjunbo-operator
spirit13579
authored
8c6038c
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
31d9669
Update config
spirit13579
authored
f89e53b
Merge pull request #49 from spirit13579/yangjunbo-operator
spirit13579
authored
08010e3
Update Adaptive_Max_Pool_2D.mlu
spirit13579
authored
bddf6f4
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
be90c6d
Add new configuration value '100' to config file
spirit13579
authored
689ae2b
Merge pull request #50 from spirit13579/yangjunbo-operator
spirit13579
authored
bac9410
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
69a738a
Update config
spirit13579
authored
b13fead
Merge pull request #51 from spirit13579/yangjunbo-operator
spirit13579
authored
4371603
Optimize dilated conv2d tiling
谢代榆
committed
0be043d
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
d358432
Retrigger dilated conv2d evaluation
谢代榆
committed
93376bb
Update Adaptive_Max_Pool_2D.mlu
spirit13579
authored
964b45d
Update config
spirit13579
authored
e9a2604
Merge branch 'main' into yangjunbo-operator
spirit13579
authored
f8a3e50
Update config
spirit13579
authored
34a8faa
Update Matrix_vector_multiplication_.mlu
spirit13579
authored
0609484
Update config
spirit13579
authored
4022366
Merge pull request #54 from spirit13579/yangjunbo-operator
spirit13579
authored
3a415b8