[CUBLAS] Update wrapppers to use the ILP64 API

[amontoison]

I checked the symbols with nm -D .../libcusolver.so and it seems that they are in the library.

[github-actions]

Your PR requires formatting changes to meet the project's style guidelines.
Please consider running Runic (git runic master) to apply these changes.

Click here to view the suggested changes.

diff --git a/lib/cublas/wrappers.jl b/lib/cublas/wrappers.jl
index 07c584860..510d0fbde 100644
--- a/lib/cublas/wrappers.jl
+++ b/lib/cublas/wrappers.jl
@@ -400,8 +400,10 @@ for (fname, fname_64, elty) in ((:cublasSrotm_v2, :cublasSrotm_v2_64, :Float32),
 end
 
 ## rotmg
-for (fname, fname_64, elty) in ((:cublasSrotmg_v2, :cublasSrotmg_v2_64, :Float32),
-                                (:cublasDrotmg_v2, :cublasSrotmg_v2_64, :Float64))
+for (fname, fname_64, elty) in (
+        (:cublasSrotmg_v2, :cublasSrotmg_v2_64, :Float32),
+        (:cublasDrotmg_v2, :cublasSrotmg_v2_64, :Float64),
+    )
     @eval begin
         function rotmg!(d1::$elty,
                         d2::$elty,
@@ -1120,7 +1122,7 @@ end
 ## (GE) general matrix-matrix multiplication
 for (fname, fname_64, elty) in ((:cublasDgemm_v2, :cublasDgemm_v2_64, :Float64),
                                 (:cublasSgemm_v2, :cublasSgemm_v2_64, :Float32),
-                                (:cublasHgemm, :cublasHgemm_64, :Float16),
+        (:cublasHgemm, :cublasHgemm_64, :Float16),
                                 (:cublasZgemm_v2, :cublasZgemm_v2_64, :ComplexF64),
                                 (:cublasCgemm_v2, :cublasCgemm_v2_64, :ComplexF32))
     @eval begin
@@ -1531,7 +1533,7 @@ end
 ## (GE) general matrix-matrix multiplication batched
 for (fname, fname_64, elty) in ((:cublasDgemmBatched, :cublasDgemmBatched_64, :Float64),
                                 (:cublasSgemmBatched, :cublasSgemmBatched_64, :Float32),
-                                (:cublasHgemmBatched, :cublasHgemmBatched_64, :Float16),
+        (:cublasHgemmBatched, :cublasHgemmBatched_64, :Float16),
                                 (:cublasZgemmBatched, :cublasZgemmBatched_64, :ComplexF64),
                                 (:cublasCgemmBatched, :cublasCgemmBatched_64, :ComplexF32))
     @eval begin
@@ -1598,7 +1600,7 @@ end
 ## (GE) general matrix-matrix multiplication strided batched
 for (fname, fname_64, elty) in ((:cublasDgemmStridedBatched, :cublasDgemmStridedBatched_64, :Float64),
                                 (:cublasSgemmStridedBatched, :cublasSgemmStridedBatched_64, :Float32),
-                                (:cublasHgemmStridedBatched, :cublasHgemmStridedBatched_64, :Float16),
+        (:cublasHgemmStridedBatched, :cublasHgemmStridedBatched_64, :Float16),
                                 (:cublasZgemmStridedBatched, :cublasZgemmStridedBatched_64, :ComplexF64),
                                 (:cublasCgemmStridedBatched, :cublasCgemmStridedBatched_64, :ComplexF32))
     @eval begin
@@ -1950,10 +1952,12 @@ end
 
 ## (TR) Triangular matrix and vector multiplication and solution
 for (mmname, mmname_64, elty) in
-        ((:cublasDtrmm_v2, :cublasDtrmm_v2_64, :Float64),
-         (:cublasStrmm_v2, :cublasStrmm_v2_64, :Float32),
-         (:cublasZtrmm_v2, :cublasZtrmm_v2_64, :ComplexF64),
-         (:cublasCtrmm_v2, :cublasCtrmm_v2_64, :ComplexF32))
+    (
+        (:cublasDtrmm_v2, :cublasDtrmm_v2_64, :Float64),
+        (:cublasStrmm_v2, :cublasStrmm_v2_64, :Float32),
+        (:cublasZtrmm_v2, :cublasZtrmm_v2_64, :ComplexF64),
+        (:cublasCtrmm_v2, :cublasCtrmm_v2_64, :ComplexF32),
+    )
     @eval begin
         # Note: CUBLAS differs from BLAS API for trmm
         #   BLAS: inplace modification of B
@@ -1987,10 +1991,12 @@ for (mmname, mmname_64, elty) in
 end
 
 for (smname, smname_64, elty) in
-        ((:cublasDtrsm_v2, :cublasDtrsm_v2_64, :Float64),
-         (:cublasStrsm_v2, :cublasStrsm_v2_64, :Float32),
-         (:cublasZtrsm_v2, :cublasZtrsm_v2_64, :ComplexF64),
-         (:cublasCtrsm_v2, :cublasCtrsm_v2_64, :ComplexF32))
+    (
+        (:cublasDtrsm_v2, :cublasDtrsm_v2_64, :Float64),
+        (:cublasStrsm_v2, :cublasStrsm_v2_64, :Float32),
+        (:cublasZtrsm_v2, :cublasZtrsm_v2_64, :ComplexF64),
+        (:cublasCtrsm_v2, :cublasCtrsm_v2_64, :ComplexF32),
+    )
     @eval begin
         function trsm!(side::Char,
                        uplo::Char,

[github-actions]

CUDA.jl Benchmarks

Benchmark suite	Current: ace28c3	Previous: c05359d	Ratio
latency/precompile	42966135947 ns	42922336650.5 ns	1.00
latency/ttfp	7012136452 ns	7015168424 ns	1.00
latency/import	3573568156 ns	3571269514 ns	1.00
integration/volumerhs	9625284.5 ns	9608723 ns	1.00
integration/byval/slices=1	147213 ns	146920.5 ns	1.00
integration/byval/slices=3	426506 ns	425845 ns	1.00
integration/byval/reference	145291 ns	145020 ns	1.00
integration/byval/slices=2	286867 ns	286380 ns	1.00
integration/cudadevrt	103634 ns	103554 ns	1.00
kernel/indexing	14158.5 ns	14235 ns	0.99
kernel/indexing_checked	15011 ns	14711 ns	1.02
kernel/occupancy	670.632911392405 ns	672.5506329113924 ns	1.00
kernel/launch	2238.1111111111113 ns	2270.3333333333335 ns	0.99
kernel/rand	15214 ns	14669 ns	1.04
array/reverse/1d	19943 ns	19682 ns	1.01
array/reverse/2d	24868.5 ns	23613.5 ns	1.05
array/reverse/1d_inplace	10397 ns	10461 ns	0.99
array/reverse/2d_inplace	12028 ns	13212 ns	0.91
array/copy	20985 ns	20972 ns	1.00
array/iteration/findall/int	157586.5 ns	157808 ns	1.00
array/iteration/findall/bool	140117 ns	139837 ns	1.00
array/iteration/findfirst/int	165035.5 ns	164937 ns	1.00
array/iteration/findfirst/bool	158235 ns	165868 ns	0.95
array/iteration/scalar	72865 ns	73041 ns	1.00
array/iteration/logical	215963.5 ns	214850 ns	1.01
array/iteration/findmin/1d	46181 ns	46704 ns	0.99
array/iteration/findmin/2d	96327.5 ns	96962.5 ns	0.99
array/reductions/reduce/Int64/1d	43700.5 ns	46033 ns	0.95
array/reductions/reduce/Int64/dims=1	49010 ns	55193 ns	0.89
array/reductions/reduce/Int64/dims=2	62902.5 ns	62917 ns	1.00
array/reductions/reduce/Int64/dims=1L	89016 ns	88869 ns	1.00
array/reductions/reduce/Int64/dims=2L	88505 ns	87079 ns	1.02
array/reductions/reduce/Float32/1d	34730 ns	34606 ns	1.00
array/reductions/reduce/Float32/dims=1	41845 ns	43875 ns	0.95
array/reductions/reduce/Float32/dims=2	59974 ns	59705 ns	1.00
array/reductions/reduce/Float32/dims=1L	52527 ns	52260 ns	1.01
array/reductions/reduce/Float32/dims=2L	70389 ns	70051.5 ns	1.00
array/reductions/mapreduce/Int64/1d	43580 ns	42671.5 ns	1.02
array/reductions/mapreduce/Int64/dims=1	48210.5 ns	45980 ns	1.05
array/reductions/mapreduce/Int64/dims=2	62641 ns	62143.5 ns	1.01
array/reductions/mapreduce/Int64/dims=1L	89035 ns	88812 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	87202 ns	86818 ns	1.00
array/reductions/mapreduce/Float32/1d	34464 ns	34742 ns	0.99
array/reductions/mapreduce/Float32/dims=1	42028 ns	43090.5 ns	0.98
array/reductions/mapreduce/Float32/dims=2	60389 ns	60061 ns	1.01
array/reductions/mapreduce/Float32/dims=1L	52805 ns	52528 ns	1.01
array/reductions/mapreduce/Float32/dims=2L	70688 ns	70191 ns	1.01
array/broadcast	20331 ns	20155 ns	1.01
array/copyto!/gpu_to_gpu	12820 ns	11294 ns	1.14
array/copyto!/cpu_to_gpu	215251 ns	216503 ns	0.99
array/copyto!/gpu_to_cpu	283637 ns	284237 ns	1.00
array/accumulate/Int64/1d	124894.5 ns	125529 ns	0.99
array/accumulate/Int64/dims=1	83631 ns	84037 ns	1.00
array/accumulate/Int64/dims=2	158233 ns	159166 ns	0.99
array/accumulate/Int64/dims=1L	1719529 ns	1720376 ns	1.00
array/accumulate/Int64/dims=2L	967882 ns	968348 ns	1.00
array/accumulate/Float32/1d	109433 ns	109984 ns	0.99
array/accumulate/Float32/dims=1	80815 ns	81082 ns	1.00
array/accumulate/Float32/dims=2	147970 ns	148760 ns	0.99
array/accumulate/Float32/dims=1L	1618276 ns	1629307.5 ns	0.99
array/accumulate/Float32/dims=2L	698936 ns	701479 ns	1.00
array/construct	1305.6 ns	1287.2 ns	1.01
array/random/randn/Float32	44977 ns	44176 ns	1.02
array/random/randn!/Float32	25089 ns	24930 ns	1.01
array/random/rand!/Int64	27518 ns	27547 ns	1.00
array/random/rand!/Float32	8792.666666666666 ns	8724.666666666666 ns	1.01
array/random/rand/Int64	30018 ns	30114 ns	1.00
array/random/rand/Float32	12998 ns	13059 ns	1.00
array/permutedims/4d	60364.5 ns	60761 ns	0.99
array/permutedims/2d	54202 ns	54037 ns	1.00
array/permutedims/3d	55093 ns	54954 ns	1.00
array/sorting/1d	2755872.5 ns	2756544 ns	1.00
array/sorting/by	3354726 ns	3343249 ns	1.00
array/sorting/2d	1084530 ns	1080799 ns	1.00
cuda/synchronization/stream/auto	1026.3 ns	1040.3 ns	0.99
cuda/synchronization/stream/nonblocking	8291.400000000001 ns	7220 ns	1.15
cuda/synchronization/stream/blocking	805.3258426966293 ns	802.3333333333334 ns	1.00
cuda/synchronization/context/auto	1197.8 ns	1203.5 ns	1.00
cuda/synchronization/context/nonblocking	7976.1 ns	7276.700000000001 ns	1.10
cuda/synchronization/context/blocking	932.25 ns	900.4347826086956 ns	1.04

This comment was automatically generated by workflow using github-action-benchmark.