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.

requested review from @aaronjc4

assigned to @rarbore2

added 1 commit

• 789d72b1 - whoops

Compare with previous version

added 1 commit

• 3fe286cc - whoops

Compare with previous version

changed the description

added 1 commit

• 90a737cf - oops, parallel launch of functions containing allocations causes data

Compare with previous version

added 1 commit

• ae8f17eb - Look for arrays that are used in parallel

Compare with previous version

added 1 commit

• 65aae3b1 - fix cfd to use proper alignment

Compare with previous version

added 1 commit

• 37567e75 - Add extra dimensions to arrays when they'd cause a race otherwise

Compare with previous version

added 1 commit

• 3d1e5b15 - Outline constants as well

Compare with previous version

added 1 commit

• 2b60a19f - Allocate more memory for calls inside fork-joins

Compare with previous version

added 1 commit

• af317577 - Use different allocations in parallel calls

Compare with previous version

changed the description

merged

mentioned in commit 52efbe6e

mentioned in issue #29 (closed)