Skip to content

Commit 8ffa451

Browse files
PerkzZhengdominicshanshan
PerkzZheng
authored and
dominicshanshan
committed
[https://nvbugs/5394685][fix] proper fix for the accuracy issue in 2CTA MLA kernels (NVIDIA#6941)
Signed-off-by: Perkz Zheng <67892460+PerkzZheng@users.noreply.github.com>
1 parent cdcaeaa commit 8ffa451

File tree

312 files changed

+919
-924
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

312 files changed

+919
-924
lines changed

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:b9eb2afd61aebdbc44340dc64fd7cea0841e8830f2148e61a62cf24e0524a837
3-
size 675649
2+
oid sha256:7d52b82a2edd9ae0b2acb1f4bc1349978e22c4ed1107ee045a592b44c9c6a72b
3+
size 682309

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:0489ad85a996e7644c0928607156ed9a5628d3d82abd4bdb1f434afb7a9e62e7
3-
size 640621
2+
oid sha256:07b1297d91824fa2725a354d9df206300df49dfb3a729a026ff651a83e55867c
3+
size 647281

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:6f2533381882d99e651f99c676626fb82bc94a7655330ddeedb8ea64b063308d
3-
size 655917
2+
oid sha256:2d43fc63094483328c0dfc2942e974bf1f0a6b7afde62b6ebfc874677fe6c36e
3+
size 660997

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:344c9ac994de57fc15bb3d9cb00190eff4f5574d76e0d80c95a89c6b91982b62
3-
size 620099
2+
oid sha256:840bea09bf19531420d397055139a8b2612a528d1120f4e8ddc4fd5d184e6d31
3+
size 625969

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:258cc90789698c5372107627acdc637e15a51f6ca56da03ce97c721a7381a15b
3-
size 664485
2+
oid sha256:2899d9173ecc0af9b1b653bf3c90afb86e92f5fd6c5a13d2970f309a6ea6a4d7
3+
size 671935

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:3d1a4de8b24d104421be65700e55c7ba65079b3aa0deef8fab76dfbd75507be9
3-
size 625509
2+
oid sha256:d2438d61847c4d144b756ff2a0d280a3920b06fbf45dfdcbecad19bba97db58c
3+
size 632959

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:891b1c2a60bfbb42eabc9e12b8fe8ec9184dccb24612806577fe6ea3bee8aac2
3-
size 644751
2+
oid sha256:be24a7a704c1a00e37a4ed18afe59555eb6796f5f10fca9103e39dc3eb853e3a
3+
size 650623

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:1a457952848822892441c92db5fa4d8d922584e941a791b5c0adea5cb54d3ecf
3-
size 605776
2+
oid sha256:6cf683762ae90059c5bb48806e9e058292a45144fb1fb2c538c64b23a663cd19
3+
size 611646

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:1129357347e8fc8adda5a2e39819e0e4a730491bfcde0e0764d08e756b8f8d17
3-
size 611908
2+
oid sha256:384f342f4fdfaacfcecc74fa58c1a1e5a041021bb4d1dcf40087d170f46fbde7
3+
size 619359

‎cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/cubin/FmhaSm100Kernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen_cubin.cpp‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,3 @@
11
version https://git-lfs.github.com/spec/v1
2-
oid sha256:0c14156039a7b74626ae26b674c5df95784f63496202b57aa9c42a299e8af637
3-
size 551546
2+
oid sha256:274395bbefe58c2558d4534c57e5755e55c07f2810e25bab11d67d667cdda45b
3+
size 555838

0 commit comments

Comments
 (0)