change the type of alpha and beta from int8 to int32 for quant_dot

src/include/migraphx/op/quant_dot.hpp

@@ -18,8 +18,8 @@ namespace op {
 
 struct quant_dot
 {
-    int8_t alpha = 1;
-    int8_t beta  = 1;
+    int32_t alpha = 1;
+    int32_t beta  = 1;
 
     template <class Self, class F>
     static auto reflect(Self& self, F f)

src/targets/cpu/gemm.cpp

@@ -119,7 +119,7 @@ void migemm(
 }
 
 void migemm(
-    const argument& c_arg, const argument& a_arg, const argument& b_arg, int8_t alpha, int8_t beta)
+    const argument& c_arg, const argument& a_arg, const argument& b_arg, int32_t alpha, int32_t beta)
 {
     migemm_tpl(c_arg, a_arg, b_arg, alpha, beta);
 }

src/targets/cpu/include/migraphx/cpu/gemm.hpp

@@ -11,7 +11,7 @@ namespace cpu {
 void migemm(
     const argument& c_arg, const argument& a_arg, const argument& b_arg, float alpha, float beta);
 void migemm(
-    const argument& c_arg, const argument& a_arg, const argument& b_arg, int8_t alpha, int8_t beta);
+    const argument& c_arg, const argument& a_arg, const argument& b_arg, int32_t alpha, int32_t beta);
 
 } // namespace cpu
 } // namespace MIGRAPHX_INLINE_NS

src/targets/cpu/lowering.cpp

@@ -563,7 +563,7 @@ struct cpu_quant_gemm
         }
 
         // 2 input arguments
-        int8_t beta = 0;
+        int32_t beta = 0;
         migemm(result, arg_0, arg_1, op.alpha, beta);
 
         return result;

src/targets/gpu/quant_gemm.cpp

@@ -94,7 +94,7 @@ argument miopen_quant_gemm::compute(context& ctx,
     }
 
     bool is_3inputs = (args.size() == 4);
-    int8_t beta     = 0;
+    int32_t beta     = 0;
     if(is_3inputs)
     {
         beta = op.beta;