@[implemented_by _private.Sparkle.Core.JITLoop.0.Sparkle.Core.JITLoop.loopMemoJITImpl]

opaque Sparkle.Core.JITLoop.Signal.loopMemoJIT {dom : Domain.DomainConfig} {α : Type} [Inhabited α] (jitCppPath : String) (wireNames : Array String) (loadMem : JIT.JITHandle → IO Unit := fun (x : JIT.JITHandle) => pure ()) (reconstruct : JIT.JITHandle → Array UInt64 → IO α) : IO (Signal.Signal dom α)

def Sparkle.Core.JITLoop.JIT.run (handle : JIT.JITHandle) (cycles : Nat) (wireIndices : Array UInt32) (callback : Nat → Array UInt64 → IO Bool) : IO Unit

Run JIT simulation for up to cycles cycles with a per-cycle callback. No state caching — O(1) memory. The callback receives the cycle number and an array of wire values (indexed by wireNames order). Return false from the callback to stop early.

Parameters:

• handle: Pre-loaded JIT handle
• cycles: Maximum number of cycles to run
• wireIndices: Pre-resolved wire indices (from JIT.findWire)
• callback: Called each cycle with (cycle, wireValues); return false to stop

Equations

• One or more equations did not get rendered due to their size.

def Sparkle.Core.JITLoop.JIT.resolveWires (handle : JIT.JITHandle) (wireNames : Array String) : IO (Array UInt32)

Resolve an array of wire names to their JIT indices. Throws if any wire name is not found.

Equations

• One or more equations did not get rendered due to their size.

def Sparkle.Core.JITLoop.JIT.resolveRegs (handle : JIT.JITHandle) (regNames : Array String) : IO (Array UInt32)

Resolve an array of register names to their JIT indices. Throws if any register name is not found.

Equations

• One or more equations did not get rendered due to their size.

def Sparkle.Core.JITLoop.JIT.runOptimized (handle : JIT.JITHandle) (cycles : Nat) (wireIndices : Array UInt32) (oracle : JIT.JITHandle → Nat → Array UInt64 → IO (Option Nat)) (callback : Nat → Array UInt64 → IO Bool) : IO Nat

Run JIT simulation with an oracle callback for cycle-skipping. Same as JIT.run but with an additional oracle that receives the JITHandle directly and can manipulate JIT state (setReg, memsetWord) before returning a skip count.

Parameters:

• handle: Pre-loaded JIT handle
• cycles: Maximum number of cycles to run
• wireIndices: Pre-resolved wire indices (from JIT.resolveWires)
• oracle: Called each cycle with (handle, cycle, wireValues); return some skipCount to skip forward (oracle handles its own state mutations), or none for normal tick.
• callback: Called each cycle with (cycle, wireValues); return false to stop

Returns: the number of cycles actually executed

Equations

• One or more equations did not get rendered due to their size.