Implement Inline Caching for Dynamic Dispatch

Monomorphic Call Sites

Dynamic languages call methods with unknown targets. Inline caching stores the last seen callee at the call site; monomorphic sites jump direct after one comparison. JSC and V8 use ICs for obj.method() dispatch.

Inline Cache Structure

Call site stub checks receiver's class/map against cached class. Match: jump to cached function. Miss: call runtime to resolve and patch cache.

c
Loading…

Monomorphic: one cached target, fastest
Polymorphic: small table of pairs
Megamorphic: give up, always runtime
Prototype changes invalidate caches

Connecting to real compilers

Production compilers embed this step inside a longer pipeline. GCC flows through cpp, cc1, assembly, and ld; Clang uses the driver, Sema, LLVM IR passes, and a target backend. LLVM bitcode, JVM bytecode, and WASM are other familiar IRs at the same layer. The exercise isolates one pass so you can test it alone before chaining it to the next stage.

Why for this exercise

You will implement inline caching for dynamic dispatch. This exercise asks you to patch call sites with class checks and direct jumps to cached callee code on the monomorphic fast path.

Implement a simple inline cache simulator in C.

Requirements:

Define a structure representing a call site with a cached target
Implement monomorphic inline cache: check receiver type, call cached function on hit
Implement polymorphic inline cache: check 2-4 cached targets in sequence
Simulate type transitions and measure cache hit/miss rates
Show the code path for cache hit (fast) vs cache miss (slow)
Print statistics: hits, misses, hit rate
Demonstrate transition from monomorphic to polymorphic to megamorphic
Explain why megamorphic sites are slow

Test:

Show cache hit path
Show cache miss path
Show hit rate statistics
Show transition to megamorphic

Monomorphic Call Sites

Inline Cache Structure

Call site stub checks receiver's class/map against cached class. Match: jump to cached function. Miss: call runtime to resolve and patch cache.

c
Loading…

Monomorphic: one cached target, fastest
Polymorphic: small table of pairs
Megamorphic: give up, always runtime
Prototype changes invalidate caches

Connecting to real compilers

Why for this exercise

You will implement inline caching for dynamic dispatch. This exercise asks you to patch call sites with class checks and direct jumps to cached callee code on the monomorphic fast path.

Implement a simple inline cache simulator in C.

Requirements:

Define a structure representing a call site with a cached target
Implement monomorphic inline cache: check receiver type, call cached function on hit
Implement polymorphic inline cache: check 2-4 cached targets in sequence
Simulate type transitions and measure cache hit/miss rates
Show the code path for cache hit (fast) vs cache miss (slow)
Print statistics: hits, misses, hit rate
Demonstrate transition from monomorphic to polymorphic to megamorphic
Explain why megamorphic sites are slow

Test:

Show cache hit path
Show cache miss path
Show hit rate statistics
Show transition to megamorphic

Topic Overview

Monomorphic Call Sites

Inline Cache Structure

Connecting to real compilers

Why for this exercise

Implementation Task

Progressive Hints

Resources

Open Source

Implement Inline Caching for Dynamic Dispatch

Topic Overview

Monomorphic Call Sites

Inline Cache Structure

Connecting to real compilers

Why for this exercise

Implementation Task

Progressive Hints

Resources

Open Source