diff --git a/laser/lux/ast/ast_codegen.nim b/laser/lux/ast/ast_codegen.nim
index 0170166..8638e46 100644
--- a/laser/lux/ast/ast_codegen.nim
+++ b/laser/lux/ast/ast_codegen.nim
@@ -23,15 +23,9 @@ proc codegen*(
     of Input:
       return params[ast.symId]
     of IntImm:
-      if arch == ArchGeneric:
-        return newLit(ast.intVal)
-      else:
-        return newCall(SimdTable[arch][simdBroadcast], newLit(ast.intVal))
+      return newCall(SimdTable[arch][simdBroadcast], newLit(ast.intVal))
     of FloatImm:
-      if arch == ArchGeneric:
-        return newLit(ast.floatVal)
-      else:
-        return newCall(SimdTable[arch][simdBroadcast], newLit(ast.intVal))
+      return newCall(SimdTable[arch][simdBroadcast], newLit(ast.intVal))
     of Output, LVal:
       let sym = newIdentNode(ast.symLVal)
       if ast.id in visited:
@@ -92,16 +86,10 @@ proc codegen*(
       stmts.add lhsStmt
       stmts.add rhsStmt
 
-      if arch == ArchGeneric:
-        case ast.kind
-        of Add: callStmt.add newidentNode"+"
-        of Mul: callStmt.add newidentNode"*"
-        else: raise newException(ValueError, "Unreachable code")
-      else:
-        case ast.kind
-        of Add: callStmt.add SimdTable[arch][simdAdd]
-        of Mul: callStmt.add SimdTable[arch][simdMul]
-        else: raise newException(ValueError, "Unreachable code")
+      case ast.kind
+      of Add: callStmt.add SimdTable[arch][simdAdd]
+      of Mul: callStmt.add SimdTable[arch][simdMul]
+      else: raise newException(ValueError, "Unreachable code")
 
       callStmt.add lhs
       callStmt.add rhs
diff --git a/laser/lux/platforms.nim b/laser/lux/platforms.nim
index 45b7c99..2e4f67f 100644
--- a/laser/lux/platforms.nim
+++ b/laser/lux/platforms.nim
@@ -3,6 +3,13 @@
 # Distributed under the Apache v2 License (license terms are at http://www.apache.org/licenses/LICENSE-2.0).
 # This file may not be copied, modified, or distributed except according to those terms.
 
+import ./platforms/platform_common
+
 when defined(i386) or defined(x86_64):
   import ./platforms/platform_x86
-  export platform_x86
+
+export SimdPrimitives
+export SimdArch, SimdAlignment, SimdTable
+
+func elemsPerVector*(arch: SimdArch, T: typedesc): int =
+  SimdWidth[arch] div sizeof(T)
diff --git a/laser/lux/platforms/platform_common.nim b/laser/lux/platforms/platform_common.nim
new file mode 100644
index 0000000..64b203e
--- /dev/null
+++ b/laser/lux/platforms/platform_common.nim
@@ -0,0 +1,43 @@
+# Laser
+# Copyright (c) 2018 Mamy André-Ratsimbazafy
+# Distributed under the Apache v2 License (license terms are at http://www.apache.org/licenses/LICENSE-2.0).
+# This file may not be copied, modified, or distributed except according to those terms.
+
+import
+  # Standard library
+  macros
+
+type
+  SimdPrimitives* = enum
+    simdSetZero
+    simdBroadcast
+    simdLoadA
+    simdLoadU
+    simdStoreA
+    simdStoreU
+    simdAdd
+    simdMul
+    simdFma
+    simdType
+
+template noop*(scalar: untyped): untyped =
+  scalar
+
+template unreachable*(): untyped =
+  {.error: "Unreachable".}
+
+proc genericPrimitives*: array[SimdPrimitives, NimNode] =
+  # Use a proc instead of a const
+  # to workaround https://github.com/nim-lang/Nim/issues/11668
+  [
+    simdSetZero:   bindSym"unreachable",
+    simdBroadcast: bindSym"noop",
+    simdLoadA:     bindSym"unreachable",
+    simdLoadU:     bindSym"unreachable",
+    simdStoreA:    bindSym"unreachable",
+    simdStoreU:    bindSym"unreachable",
+    simdAdd:       bindSym"+",
+    simdMul:       bindSym"*",
+    simdFma:       bindSym"unreachable",
+    simdType:      bindSym"unreachable"
+  ]
diff --git a/laser/lux/platforms/platform_x86.nim b/laser/lux/platforms/platform_x86.nim
index d1caa3c..9cb361d 100644
--- a/laser/lux/platforms/platform_x86.nim
+++ b/laser/lux/platforms/platform_x86.nim
@@ -9,21 +9,10 @@ import
   # Standard library
   macros,
   # Internal
+  ./platform_common,
   ../../simd
 
 type
-  SimdPrimitives* = enum
-    simdSetZero
-    simdBroadcast
-    simdLoadA
-    simdLoadU
-    simdStoreA
-    simdStoreU
-    simdAdd
-    simdMul
-    simdFma
-    simdType
-
   SimdArch* = enum
     ArchGeneric,
     x86_SSE,
@@ -58,7 +47,7 @@ template avx_fma_fallback(a, b, c: m128): m128 =
 
 proc genSimdTableX86(): array[SimdArch, array[SimdPrimitives, NimNode]] =
 
-  let sse: array[SimdPrimitives, NimNode] = [
+  let sse = [
     simdSetZero:   bindSym"mm_setzero_ps",
     simdBroadcast: bindSym"mm_set1_ps",
     simdLoadA:     bindSym"mm_load_ps",
@@ -88,7 +77,7 @@ proc genSimdTableX86(): array[SimdArch, array[SimdPrimitives, NimNode]] =
   avx_fma[simdFma] = bindSym"mm256_fmadd_ps"
 
   result = [
-    ArchGeneric: default(array[SimdPrimitives, NimNode]),
+    ArchGeneric: genericPrimitives(),
     x86_SSE: sse,
     x86_AVX: avx,
     x86_AVX_FMA: avx_fma