Try the release 2025.2.0 of oneAPI

[amontoison]

cc @michel2323

Update: All tests passed but we are not using the release 2025.2.0. The artifacts seem to be used.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 81.93%. Comparing base (cacd3c2) to head (88eb3e3).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #511   +/-   ##
=======================================
  Coverage   81.93%   81.93%           
=======================================
  Files          44       44           
  Lines        2530     2530           
=======================================
  Hits         2073     2073           
  Misses        457      457           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[maleadt]

Maybe because the Conda build fails for some reason:

2025-07-21 16:59:46 UTC
	Failed to precompile Conda [8f4d0f93-b110-5947-807f-2305c1781a2d] to "/root/.cache/julia-buildkite-plugin/depots/05c310bb-cd1d-4bbf-ad62-61f2372c55f0/compiled/v1.11/Conda/jl_KoEqGO".
2025-07-21 16:59:46 UTC
	ERROR: LoadError: ArgumentError: Path to conda environment is not valid: /root/.cache/julia-buildkite-plugin/depots/05c310bb-cd1d-4bbf-ad62-61f2372c55f0/conda/3/x86_64

[amontoison]

@michel2323, I will need your help to update libsycl.so.
It seems that Intel changed something for the release 2025.2.0.

[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/deps/generate_helpers.jl b/deps/generate_helpers.jl
index 01f162d..5e715db 100644
--- a/deps/generate_helpers.jl
+++ b/deps/generate_helpers.jl
@@ -1,6 +1,6 @@
 non_parametric_routines = ["init_matrix_handle", "release_matrix_handle", "set_matrix_property",
 "init_matmat_descr", "release_matmat_descr", "set_matmat_data", "get_matmat_data", "matmat",
-"omatcopy", "sort_matrix", "optimize_gemv", "optimize_gemm", "optimize_trmv", "optimize_trsv", "optimize_trsm",
+    "omatcopy", "sort_matrix", "optimize_gemv", "optimize_gemm", "optimize_trmv", "optimize_trsv", "optimize_trsm",
 "init_omatconvert_descr", "release_omatconvert_descr", "init_omatadd_descr", "release_omatadd_descr",
 "omatconvert_buffer_size", "omatconvert_analyze", "omatconvert_get_nnz", "omatconvert",
 "omatadd_buffer_size", "omatadd_analyze", "omatadd_get_nnz"]
diff --git a/deps/generate_interfaces.jl b/deps/generate_interfaces.jl
index a36fd0a..470c451 100644
--- a/deps/generate_interfaces.jl
+++ b/deps/generate_interfaces.jl
@@ -241,7 +241,7 @@ function generate_headers(library::String, filename::Vector{String}, output::Str
         # The routine "optimize_trsm" has two versions.
         suffix = ""
         (name_routine == "optimize_trsm") && occursin("columns", header) && (suffix = "_advanced")
-        (name_routine == "optimize_gemm") && occursin("columns", header) && (suffix = "_advanced")
+                (name_routine == "optimize_gemm") && occursin("columns", header) && (suffix = "_advanced")
         name_routine ∈ ("set_csr_data", "set_coo_data") && occursin("int64_t", header) && (suffix = "_64")
         occursin("batch", name_routine) && !occursin("**", header) && (suffix = "_strided")
 
diff --git a/lib/support/liboneapi_support.jl b/lib/support/liboneapi_support.jl
index 9b5858f..29bc6f1 100644
--- a/lib/support/liboneapi_support.jl
+++ b/lib/support/liboneapi_support.jl
@@ -6613,23 +6613,31 @@ function onemklXsparse_release_matmat_descr(p_desc)
 end
 
 function onemklXsparse_init_omatconvert_descr(device_queue, p_descr)
-    @ccall liboneapi_support.onemklXsparse_init_omatconvert_descr(device_queue::syclQueue_t,
-                                                                  p_descr::Ptr{omatconvert_descr_t})::Cint
+    return @ccall liboneapi_support.onemklXsparse_init_omatconvert_descr(
+        device_queue::syclQueue_t,
+        p_descr::Ptr{omatconvert_descr_t}
+    )::Cint
 end
 
 function onemklXsparse_release_omatconvert_descr(device_queue, descr)
-    @ccall liboneapi_support.onemklXsparse_release_omatconvert_descr(device_queue::syclQueue_t,
-                                                                     descr::omatconvert_descr_t)::Cint
+    return @ccall liboneapi_support.onemklXsparse_release_omatconvert_descr(
+        device_queue::syclQueue_t,
+        descr::omatconvert_descr_t
+    )::Cint
 end
 
 function onemklXsparse_init_omatadd_descr(device_queue, p_omatadd_desc)
-    @ccall liboneapi_support.onemklXsparse_init_omatadd_descr(device_queue::syclQueue_t,
-                                                              p_omatadd_desc::Ptr{omatadd_descr_t})::Cint
+    return @ccall liboneapi_support.onemklXsparse_init_omatadd_descr(
+        device_queue::syclQueue_t,
+        p_omatadd_desc::Ptr{omatadd_descr_t}
+    )::Cint
 end
 
 function onemklXsparse_release_omatadd_descr(device_queue, omatadd_desc)
-    @ccall liboneapi_support.onemklXsparse_release_omatadd_descr(device_queue::syclQueue_t,
-                                                                 omatadd_desc::omatadd_descr_t)::Cint
+    return @ccall liboneapi_support.onemklXsparse_release_omatadd_descr(
+        device_queue::syclQueue_t,
+        omatadd_desc::omatadd_descr_t
+    )::Cint
 end
 
 function onemklXsparse_omatcopy(device_queue, transpose_val, spMat_in, spMat_out)
@@ -6695,19 +6703,25 @@ function onemklXsparse_optimize_trsv(device_queue, uplo_val, opA, diag_val, A)
 end
 
 function onemklXsparse_optimize_gemm(device_queue, opA, A)
-    @ccall liboneapi_support.onemklXsparse_optimize_gemm(device_queue::syclQueue_t,
-                                                         opA::onemklTranspose,
-                                                         A::matrix_handle_t)::Cint
-end
-
-function onemklXsparse_optimize_gemm_advanced(device_queue, layout_val, opA, opB, A,
-                                              columns)
-    @ccall liboneapi_support.onemklXsparse_optimize_gemm_advanced(device_queue::syclQueue_t,
-                                                                  layout_val::onemklLayout,
-                                                                  opA::onemklTranspose,
-                                                                  opB::onemklTranspose,
-                                                                  A::matrix_handle_t,
-                                                                  columns::Int64)::Cint
+    return @ccall liboneapi_support.onemklXsparse_optimize_gemm(
+        device_queue::syclQueue_t,
+        opA::onemklTranspose,
+        A::matrix_handle_t
+    )::Cint
+end
+
+function onemklXsparse_optimize_gemm_advanced(
+        device_queue, layout_val, opA, opB, A,
+        columns
+    )
+    return @ccall liboneapi_support.onemklXsparse_optimize_gemm_advanced(
+        device_queue::syclQueue_t,
+        layout_val::onemklLayout,
+        opA::onemklTranspose,
+        opB::onemklTranspose,
+        A::matrix_handle_t,
+        columns::Int64
+    )::Cint
 end
 
 function onemklXsparse_optimize_trsm(device_queue, uplo_val, opA, diag_val, A)
diff --git a/test/onemkl.jl b/test/onemkl.jl
index e5b6541..a4fd56c 100644
--- a/test/onemkl.jl
+++ b/test/onemkl.jl
@@ -1132,7 +1132,7 @@ end
 
                     alpha = rand(T)
                     beta = rand(T)
-                    oneMKL.sparse_optimize_gemm!(transa, dA)
+                        oneMKL.sparse_optimize_gemm!(transa, dA)
                     oneMKL.sparse_gemm!(transa, transb, alpha, dA, dB, beta, dC)
                     @test alpha * opa(A) * opb(B) + beta * C ≈ collect(dC)
                 end

[amontoison]

Done in #513