Lint fix

.github/workflows/python-package-conda.yml

@@ -25,6 +25,10 @@ jobs:
         python -m virtualenv venv
         source venv/bin/activate
         pip install -r requirements.txt
+        git clone https://github.com/cdonovick/peak.git
+        pip install -e peak
+        git clone https://github.com/StanfordAHA/lassen.git
+        pip install -e lassen
         pip install -e .
         echo $VIRTUAL_ENV/bin >> $GITHUB_PATH
     - name: Test all (non SuiteSparse and Frostt) tests with pytest
@@ -44,12 +48,12 @@ jobs:
       run: |
         conda install flake8
         # stop the build if there are Python syntax errors or undefined names
-        flake8 . --count --select=E9,F63,F7,F82 --show-source --statistics --exclude venv
+        flake8 . --count --select=E9,F63,F7,F82 --show-source --statistics --exclude venv,peak,lassen
         # exit-zero treats all errors as warnings. The GitHub editor is 127 chars wide
-        flake8 . --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics --exclude venv
+        flake8 . --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics --exclude venv,peak,lassen
     - name: Python style checking flake8
       run: |
-        flake8 . --count --select=E,W --statistics --ignore=W503,W504 --max-line-length=127 --exclude venv
+        flake8 . --count --select=E,W --statistics --ignore=W503,W504 --max-line-length=127 --exclude venv,peak,lassen
     - name: Check SAM Simulator generating script
       run: |
         make tests

.gitignore

@@ -49,6 +49,13 @@ suitesparse-bench/
 sam-outputs/
 compiler/benchmark/
 
+# Generated SAM simulator tests
+*/sim/test/apps/test_*.py
+
+# Tensor files
+*.mtx
+*.tns
+
 # Temporary or generated tensor directories 
 tmp_mat*/
 tiles/

Makefile

@@ -31,20 +31,20 @@ endif
 
 ifeq ("$(NEVA)","ON")
 	CMD := OMP_PROC_BIND=true LD_LIBRARY_PATH=compiler/build/lib/:$(LD_LIBRARY_PATH) numactl -C 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 -m 0 compiler/build/taco-bench $(BENCHFLAGS)
-	export SUITESPARSE_PATH=/nobackup/owhsu/sparse-datasets/suitesparse/
-	export FROSTT_PATH=/nobackup/owhsu/sparse-datasets/frostt/
-	export SUITESPARSE_FORMATTED_PATH=/nobackup/owhsu/sparse-datasets/suitesparse-formatted
-	export FROSTT_FORMATTED_TACO_PATH=/nobackup/owhsu/sparse-datasets/frostt-formatted/taco-tensor
-	export FROSTT_FORMATTED_PATH=/nobackup/owhsu/sparse-datasets/frostt-formatted
-	export TACO_TENSOR_PATH=/nobackup/owhsu/sparse-datasets
+	# export SUITESPARSE_PATH=/nobackup/owhsu/sparse-datasets/suitesparse/
+	# export FROSTT_PATH=/nobackup/owhsu/sparse-datasets/frostt/
+	# export SUITESPARSE_FORMATTED_PATH=/nobackup/owhsu/sparse-datasets/suitesparse-formatted
+	# export FROSTT_FORMATTED_TACO_PATH=/nobackup/owhsu/sparse-datasets/frostt-formatted/taco-tensor
+	# export FROSTT_FORMATTED_PATH=/nobackup/owhsu/sparse-datasets/frostt-formatted
+	# export TACO_TENSOR_PATH=/nobackup/owhsu/sparse-datasets
 else ifeq ("$(LANKA)", "ON")
 	CMD := OMP_PROC_BIND=true LD_LIBRARY_PATH=compiler/build/lib/:$(LD_LIBRARY_PATH) numactl -C 0,2,4,6,8,10,24,26,28,30,32,34 -m 0 compiler/build/taco-bench $(BENCHFLAGS)
-	export SUITESPARSE_PATH=/data/scratch/changwan/florida_all
-	export FROSTT_PATH=/data/scratch/owhsu/datasets/frostt
-	export TACO_TENSOR_PATH=/data/scratch/owhsu/datasets
-	export SUITESPARSE_FORMATTED_PATH=/data/scratch/owhsu/datasets/suitesparse-formatted
-	export FROSTT_FORMATTED_TACO_PATH=/data/scratch/owhsu/datasets/frostt-formatted/taco-tensor
-	export FROSTT_FORMATTED_PATH=/data/scratch/owhsu/datasets/frostt-formatted
+	# export SUITESPARSE_PATH=/data/scratch/changwan/florida_all
+	# export FROSTT_PATH=/data/scratch/owhsu/datasets/frostt
+	# export TACO_TENSOR_PATH=/data/scratch/owhsu/datasets
+	# export SUITESPARSE_FORMATTED_PATH=/data/scratch/owhsu/datasets/suitesparse-formatted
+	# export FROSTT_FORMATTED_TACO_PATH=/data/scratch/owhsu/datasets/frostt-formatted/taco-tensor
+	# export FROSTT_FORMATTED_PATH=/data/scratch/owhsu/datasets/frostt-formatted
 else
 	CMD := LD_LIBRARY_PATH=compiler/build/lib/:$(LD_LIBRARY_PATH) compiler/build/taco-bench $(BENCHFLAGS)
 endif
@@ -114,14 +114,6 @@ suitesparse-formats: guard-SUITESPARSE_FORMATTED_PATH guard-SUITESPARSE_PATH
 frostt-formats: taco/build guard-FROSTT_FORMATTED_PATH guard-FROSTT_PATH
 	./scripts/formatting/generate_frostt_formats.sh
 
-.PHONY: env
-env:
-	export SUITESPARSE_PATH=/nobackup/owhsu/sparse-datasets/suitesparse/
-	export FROSTT_PATH=/nobackup/owhsu/sparse-datasets/frostt/
-	export SUITESPARSE_FORMATTED_PATH=/nobackup/owhsu/sparse-datasets/suitesparse-formatted
-	export FROSTT_FORMATTED_TACO_PATH=/nobackup/owhsu/sparse-datasets/frostt-formatted/taco-tensor
-	export FROSTT_FORMATTED_PATH=/nobackup/owhsu/sparse-datasets/frostt-formatted
-
 .PHONY: pydepends
 pydepends:
 	conda env export > environment.yml

compiler/baseline.cpp

@@ -315,7 +315,8 @@ enum SuiteSparseOp {
     SDDMM = 4,
     MATTRANSMUL = 5,
     RESIDUAL = 6,
-    MMADD = 7
+    MMADD = 7,
+    MMMUL = 8
 };
 
 std::string opName(SuiteSparseOp op) {
@@ -341,6 +342,9 @@ std::string opName(SuiteSparseOp op) {
         case MMADD: {
             return "mmadd";
         }
+	case MMMUL: {
+	    return "mmmul"
+	}
         default:
             return "";
     }
@@ -467,6 +471,13 @@ static void bench_suitesparse(benchmark::State &state, SuiteSparseOp op, int fil
                 result(i, j) = ssTensor(i, j) + otherShifted(i, j);
                 break;
             }
+            case MMMUL: {
+                result = Tensor<int64_t>("result", ssTensor.getDimensions(), ssTensor.getFormat(), fill_value);
+
+                IndexVar i, j, k;
+                result(i, j) = ssTensor(i, j) * otherShifted(i, j);
+                break;
+            }
             case MATTRANSMUL: {
                 result = Tensor<int64_t>("result", {DIM1}, Format(Sparse), fill_value);
 
@@ -516,4 +527,5 @@ static void bench_suitesparse(benchmark::State &state, SuiteSparseOp op, int fil
     // TODO: need to fix for DCSC for this
     TACO_BENCH_ARGS(bench_suitesparse, mat_mattransmul, MATTRANSMUL);
     TACO_BENCH_ARGS(bench_suitesparse, matmul_spmm, SPMM);
+    TACO_BENCH_ARGS(bench_suitesparse, mat_elemmul, MMMUL);
 

compiler/sam-kernels.sh

@@ -41,6 +41,12 @@ GEN_KERNEL_NAMES=(
   mat_spacc_simple
   vec_sd_compression_WRONG
   vec_ds_compression_WRONG
+  vec_sub
+  mat_mask_tri
+  mat_mask_tri_partial
+  mat_vecmul_iter
+  mat_vecmul_iter_short
+  tensor3_website_expr
 )
 
 HAND_KERNEL_NAMES=(
@@ -83,6 +89,12 @@ TACO_ARGS=(
   "X(j,k)=B(i,j,k) -f=X:ss -f=B:sss"
   "x(i)=b(i) -f=b:s -f=x:d"
   "x(i)=b(i) -f=b:d -f=x:s"
+  "x(i)=b(i)-c(i) -f=b:s -f=c:s"
+  "x=B(i,j)*C(i,k)*D(k,j) -f=B:ss -f=C:ss -f=D:ss:1,0 -s=reorder(i,j,k)"
+  "X(i,j)=B(i,j)*C(i,k)*D(k,j) -f=X:ss -f=B:ss -f=C:ss -f=D:ss:1,0 -s=reorder(i,j,k)"
+  "x(i)=B(i,j)*C(j,k)*D(k,l)*E(l,m)*f(m) -f=x:s -f=B:ss -f=C:ss -f=D:ss -f=E:ss -f=f:s -s=reorder(i,j,k,l,m)"
+  "x(i)=B(i,j)*C(j,k)*d(k) -f=x:s -f=B:ss -f=C:ss -f=d:s -s=reorder(i,j,k)"
+  "x=B(i)*C(j)*D(i,j,k)*E(j,l)*F(l,m,n) -f=B:s -f=C:s -f=D:sss -f=E:ss -f=F:sss -s=reorder(i,j,k,l,m,n)"
 )
 
 mkdir -p $dir

compiler/sam-outputs/dot/mat_mask_tri.gv

@@ -0,0 +1,62 @@
+digraph SAM {
+    comment="x=none,B=ss01,C=ss01,D=ss10"
+    22 [comment="type=fiberlookup,index=i,tensor=B,mode=0,format=compressed,src=true,root=true" label="FiberLookup i: B0\ncompressed" color=green4 shape=box style=filled type="fiberlookup" index="i" tensor="B" mode="0" format="compressed" src="true" root="true"]
+    21 [comment="type=intersect,index=i" label="intersect i" color=purple shape=box style=filled type="intersect" index="i"]
+    20 [comment="type=repsiggen,index=i" label="RepeatSignalGenerator i" color=cyan3 shape=box style=filled type="repsiggen" index="i"]
+    19 [comment="type=repeat,index=i,tensor=D,root=true" label="Repeat i: D" color=cyan2 shape=box style=filled type="repeat" index="i" tensor="D" root="true"]
+    18 [comment="type=fiberlookup,index=j,tensor=D,mode=1,format=compressed,src=true,root=false" label="FiberLookup j: D1\ncompressed" color=green4 shape=box style=filled type="fiberlookup" index="j" tensor="D" mode="1" format="compressed" src="true" root="false"]
+    16 [comment="type=intersect,index=j" label="intersect j" color=purple shape=box style=filled type="intersect" index="j"]
+    15 [comment="type=repsiggen,index=j" label="RepeatSignalGenerator j" color=cyan3 shape=box style=filled type="repsiggen" index="j"]
+    14 [comment="type=repeat,index=j,tensor=C,root=false" label="Repeat j: C" color=cyan2 shape=box style=filled type="repeat" index="j" tensor="C" root="false"]
+    12 [comment="type=fiberlookup,index=k,tensor=C,mode=1,format=compressed,src=true,root=false" label="FiberLookup k: C1\ncompressed" color=green4 shape=box style=filled type="fiberlookup" index="k" tensor="C" mode="1" format="compressed" src="true" root="false"]
+    11 [comment="type=intersect,index=k" label="intersect k" color=purple shape=box style=filled type="intersect" index="k"]
+    10 [comment="type=repsiggen,index=k" label="RepeatSignalGenerator k" color=cyan3 shape=box style=filled type="repsiggen" index="k"]
+    9 [comment="type=repeat,index=k,tensor=B,root=false" label="Repeat k: B" color=cyan2 shape=box style=filled type="repeat" index="k" tensor="B" root="false"]
+    6 [comment="type=arrayvals,tensor=B" label="Array Vals: B" color=green2 shape=box style=filled type="arrayvals" tensor="B"]
+    5 [comment="type=mul" label="Mul" color=brown shape=box style=filled type="mul"]
+    4 [comment="type=mul" label="Mul" color=brown shape=box style=filled type="mul"]
+    3 [comment="type=reduce" label="Reduce" color=brown shape=box style=filled type="reduce"]
+    2 [comment="type=reduce" label="Reduce" color=brown shape=box style=filled type="reduce"]
+    1 [comment="type=reduce" label="Reduce" color=brown shape=box style=filled type="reduce"]
+    0 [comment="type=fiberwrite,mode=vals,tensor=x,size=1,sink=true" label="FiberWrite Vals: x" color=green3 shape=box style=filled type="fiberwrite" tensor="x" mode="vals" size="1" sink="true"]
+    7 [comment="type=arrayvals,tensor=C" label="Array Vals: C" color=green2 shape=box style=filled type="arrayvals" tensor="C"]
+    8 [comment="type=arrayvals,tensor=D" label="Array Vals: D" color=green2 shape=box style=filled type="arrayvals" tensor="D"]
+    13 [comment="type=fiberlookup,index=k,tensor=D,mode=0,format=compressed,src=true,root=false" label="FiberLookup k: D0\ncompressed" color=green4 shape=box style=filled type="fiberlookup" index="k" tensor="D" mode="0" format="compressed" src="true" root="false"]
+    17 [comment="type=fiberlookup,index=j,tensor=B,mode=1,format=compressed,src=true,root=false" label="FiberLookup j: B1\ncompressed" color=green4 shape=box style=filled type="fiberlookup" index="j" tensor="B" mode="1" format="compressed" src="true" root="false"]
+    23 [comment="type=fiberlookup,index=i,tensor=C,mode=0,format=compressed,src=true,root=true" label="FiberLookup i: C0\ncompressed" color=green4 shape=box style=filled type="fiberlookup" index="i" tensor="C" mode="0" format="compressed" src="true" root="true"]
+    22 -> 21 [label="crd_in-B" style=dashed type="crd" comment="in-B"]
+    21 -> 20 [label="crd" style=dashed type="crd"]
+    20 -> 19 [label="repsig" style=dotted type="repsig"]
+    19 -> 18 [label="ref" style=bold type="ref"]
+    18 -> 16 [label="crd_in-D" style=dashed type="crd" comment="in-D"]
+    16 -> 15 [label="crd" style=dashed type="crd"]
+    15 -> 14 [label="repsig" style=dotted type="repsig"]
+    14 -> 12 [label="ref" style=bold type="ref"]
+    12 -> 11 [label="crd_in-C" style=dashed type="crd" comment="in-C"]
+    11 -> 10 [label="crd" style=dashed type="crd"]
+    10 -> 9 [label="repsig" style=dotted type="repsig"]
+    9 -> 6 [label="ref" style=bold type="ref"]
+    6 -> 5 [label="val" type="val"]
+    5 -> 4 [label="val" type="val"]
+    4 -> 3 [label="val" type="val"]
+    3 -> 2 [label="val" type="val"]
+    2 -> 1 [label="val" type="val"]
+    1 -> 0 [label="val" type="val"]
+    11 -> 7 [label="ref_out-C" style=bold type="ref" comment="out-C"]
+    7 -> 5 [label="val" type="val"]
+    11 -> 8 [label="ref_out-D" style=bold type="ref" comment="out-D"]
+    8 -> 4 [label="val" type="val"]
+    12 -> 11 [label="ref_in-C" style=bold type="ref" comment="in-C"]
+    16 -> 9 [label="ref_out-B" style=bold type="ref" comment="out-B"]
+    16 -> 13 [label="ref_out-D" style=bold type="ref" comment="out-D"]
+    13 -> 11 [label="crd_in-D" style=dashed type="crd" comment="in-D"]
+    13 -> 11 [label="ref_in-D" style=bold type="ref" comment="in-D"]
+    18 -> 16 [label="ref_in-D" style=bold type="ref" comment="in-D"]
+    21 -> 17 [label="ref_out-B" style=bold type="ref" comment="out-B"]
+    17 -> 16 [label="crd_in-B" style=dashed type="crd" comment="in-B"]
+    17 -> 16 [label="ref_in-B" style=bold type="ref" comment="in-B"]
+    21 -> 14 [label="ref_out-C" style=bold type="ref" comment="out-C"]
+    22 -> 21 [label="ref_in-B" style=bold type="ref" comment="in-B"]
+    23 -> 21 [label="crd_in-C" style=dashed type="crd" comment="in-C"]
+    23 -> 21 [label="ref_in-C" style=bold type="ref" comment="in-C"]
+}