issues
search
nod-ai
/
iree-amd-aie
IREE plugin repository for the AMD AIE accelerator
Apache License 2.0
69
stars
30
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Enable 4x4 AIE cores for matmul
#920
yzhang93
opened
5 hours ago
0
[Experiment, do not review] Function outlining tracker
#919
newling
opened
8 hours ago
0
[DistributeL1Allocations] Fix logic for conv and correctly propagate errors
#918
newling
opened
8 hours ago
0
[CI] Fix for bug in `aie-compilation-flags` option
#917
newling
closed
4 hours ago
0
Add control code scf.forall to scf.for pass
#916
jtuyls
closed
11 hours ago
2
Add pass to assign tiles to logical objectFifos
#915
jtuyls
closed
9 hours ago
2
Split distribute-cores-and-objectfifos pass in 2
#914
newling
closed
14 hours ago
3
[Utils] Finding ancestor in block
#913
newling
closed
1 day ago
0
Fix bug in PushToQueue address generation
#912
jtuyls
closed
1 day ago
1
[CI] run.py improvements
#911
newling
closed
1 day ago
0
Refactor LogicalObjFifoSplittingUtils
#910
yzhang93
closed
2 days ago
0
[CI] Add new shape and with/without ukernel for benchmark tracking
#909
newling
closed
17 hours ago
1
Add pass that inserts infinite loops around core blocks
#908
jtuyls
closed
3 days ago
2
Coalescing loops
#907
newling
opened
5 days ago
0
Remove conv_fill_spec_pad.mlir test
#906
yzhang93
closed
6 days ago
0
Bump IREE to 5318fce9dcc809b13420ab9ff22c4c4320e26746
#905
yzhang93
closed
6 days ago
2
[AMDAIEAccessToAcquireRelease] Insert acquire/release pairs in same scope
#904
newling
closed
12 hours ago
4
Split L2 input and output objectFifos for memTile/shimTile distribution
#903
yzhang93
closed
1 day ago
3
CI system for windows shall use the NPU driver released by Microsoft windows
#902
dezhiAmd
closed
1 week ago
3
Debug build fails on windows OS
#901
dezhiAmd
opened
1 week ago
1
[LinalgFunctionOutlining] Enable/disable function outlining using pass flag
#900
Abhishek-Varma
closed
1 week ago
2
[POC] Unpeel
#899
newling
closed
3 days ago
6
[AMDAIENoneAccessToTemporaryBuffer] Insert new alloc at block start to avoid SSA error
#898
newling
closed
1 week ago
0
[xrt-lite] bump cts matmul to 1024x1024x1024xi32
#897
makslevental
closed
1 week ago
0
[Vectorization] Remove coalescing from insert-loops-for-vectorization
#896
Abhishek-Varma
closed
1 week ago
0
Two WIP regression fixes (outlining + colaesing)
#895
newling
closed
3 days ago
3
[AMDAIEVectorization] Do not coalesce the 3 loops
#894
newling
closed
1 week ago
1
Revert "Increase the K tile size in L1 for matmul ops"
#893
yzhang93
closed
1 week ago
0
Invasively print kernel execution time, percolate to CI output
#892
newling
closed
1 week ago
1
[Tests] Improve help and failure mode
#891
newling
closed
1 week ago
0
[xrt-lite] Update cts test to use objectFifo with bf16.
#890
newling
closed
3 days ago
6
Bump IREE to 8e5f218a1d73c39ccdebb4a1d19225d5329cc6f0
#889
yzhang93
closed
1 week ago
5
[FuncOutline] Fix function outlining to skip linalg.fill/linalg.skip
#888
Abhishek-Varma
closed
1 week ago
2
[Test] Update run.py to run only `target_device` specific tests
#887
Abhishek-Varma
closed
1 week ago
1
Uncompiled file: remove or fix cmake?
#886
newling
closed
2 weeks ago
2
[Testing] Move some tests from run_matmul_test.sh to run.py
#885
newling
opened
2 weeks ago
5
[aievec] Don't add alignment ops if already aligned
#884
newling
closed
2 weeks ago
0
Vectorized matmul performance regression - function inlining
#883
jtuyls
opened
2 weeks ago
5
Vectorized matmul performance regression - transfer reads
#882
jtuyls
closed
2 weeks ago
2
[CI][Test] Add Linux Strix job
#881
jtuyls
closed
2 weeks ago
1
[CI][Test] Add `target_device` flag
#880
jtuyls
closed
2 weeks ago
0
[Strix][Ukernel][ObjectFifo] Add Strix ukernel support
#879
Abhishek-Varma
closed
1 week ago
6
Test Ukernel update
#878
Abhishek-Varma
closed
2 weeks ago
1
Update xdna-driver link in README.md
#877
jtuyls
closed
2 weeks ago
0
[Testing] Make it possible to run individual tests
#876
newling
closed
2 weeks ago
3
[Strix][KernelDispatch][Ukernel] Enable Matmul intrinsics for Strix
#875
Abhishek-Varma
closed
2 weeks ago
0
Add hardware-aware canonicalization
#874
jtuyls
closed
2 weeks ago
3
Control of loop unrolling
#873
newling
closed
2 weeks ago
1
[WIP] Remove libc workaround
#872
newling
closed
2 weeks ago
1
[aievec] ShiftOp folder
#871
newling
closed
2 weeks ago
0
Next