Fix PR comments - int->int64 + add perf test for ABS unary ops

Author: YavorGIvanov

ggml/src/ggml-cuda/unary.cu

@@ -100,9 +100,9 @@ static __global__ void unary_op_kernel_noncont(
     const int64_t ne0, const int64_t ne1, const int64_t ne2, const int64_t ne3,
     const int64_t nb0_x, const int64_t nb1_x, const int64_t nb2_x, const int64_t nb3_x,
     const int64_t nb0_d, const int64_t nb1_d, const int64_t nb2_d, const int64_t nb3_d,
-    const int k) {
+    const int64_t k) {
 
-    const int i = blockDim.x*blockIdx.x + threadIdx.x;
+    const int64_t i = blockDim.x*blockIdx.x + threadIdx.x;
 
     if (i >= k) {
         return;

tests/test-backend-ops.cpp

@@ -5611,6 +5611,9 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_perf() {
 
     test_cases.emplace_back(new test_mean(GGML_TYPE_F32, {256, 256, 3, 1}));
 
+    test_cases.emplace_back(new test_unary((ggml_unary_op) GGML_UNARY_OP_ABS, GGML_TYPE_F32, {256, 256, 3, 1}, 0));
+    test_cases.emplace_back(new test_unary((ggml_unary_op) GGML_UNARY_OP_ABS, GGML_TYPE_F32, {256, 256, 3, 1}, 1));
+
     return test_cases;
 }