Optimization for miranda

• Optimize cava and edge detection schedules (CPU schedules are multi-core now).
• Add fork-extend pass to make fork dimensions multiple of some number (closes #28 (closed)).
• Make GCM add extra dimensions to per-thread arrays in fork-joins.
• Make GCM allocate extra memory for calls in parallel, make RT backend use different memories per parallel call.
• Make not finding xdot binary crash the compiler.
• Some misc. fixes.
• Add conv-like fork-join test.

juno_samples/rodinia/cfd/src/main.rs

@@ -151,7 +151,7 @@ fn cfd_harness(args: CFDInputs) {
         pre_euler,
     } = args;
 
-    assert!(block_size % 8 == 0, "Hercules expects all arrays to be 32-byte aligned, cfd uses structs of arrays that are annoying to deal with if the block_size is not a multiple of 8");
+    assert!(block_size % 16 == 0, "Hercules expects all arrays to be 64-byte aligned, cfd uses structs of arrays that are annoying to deal with if the block_size is not a multiple of 16");
 
     let FarFieldConditions {
         ff_variable,
@@ -245,7 +245,7 @@ fn test_euler() {
     cfd_harness(CFDInputs {
         data_file: "data/fvcorr.domn.097K".to_string(),
         iterations: 1,
-        block_size: 8,
+        block_size: 16,
         pre_euler: false,
     });
 }
@@ -255,7 +255,7 @@ fn test_pre_euler() {
     cfd_harness(CFDInputs {
         data_file: "data/fvcorr.domn.097K".to_string(),
         iterations: 1,
-        block_size: 8,
+        block_size: 16,
         pre_euler: true,
     });
 }