Ares Git (2026/05/11) is compiled. ares is a multi-system emulator that began development on October 14th, 2004. It is a descendent of higan and bsnes, and focuses on accuracy and preservation.

ares currently emulates the following 27 hardware devices:
* Famicom + Famicom Disk System
* Super Famicom + Super Game Boy
* Nintendo 64 (under development)
* Game Boy + Game Boy Color
* Game Boy Advance + Game Boy Player
* SG-1000 + SC-3000
* Master System + Game Gear
* Mega Drive + Mega CD
* PC Engine + PC Engine CD + SuperGrafx
* MSX + MSX2
* ColecoVision
* Neo Geo Pocket + Neo Geo Pocket Color
* WonderSwan + WonderSwan Color + SwanCrystal + Pocket Challenge V2

Ares Git changelog:
* sfc: Improve autojoypad polling accuracy
* Update libchdr to version 0.3.0
* Update libchdr to version 0.3.0
* Update libchdr to version 0.3.0
* Update libchdr to version 0.3.0
* n64: improve CPU JIT documentation
* n64: decrease a bit jit interleaving (unfreeze conker)
* n64: fix interrupt generation to never happen mid-instruction
* n64: fix a few cases where interrupt latency was still higher than expected
* n64: change CPU JIT flush message
* n64: increase JIT interleaving and add more forced syncs on MMIOs
* n64: reduce interrupt latency to zero in JIT
* n64: in CPU JIT, correctly declare 6 scratch registers (and don't use 7th)
* n64: in CPU JIT, verify internal assumptions on block links
* n64: in CPU JIT, implement LL/SC/LLD/SCD
* n64: in CPU JIT, rewrite branches to use branchless code
* n64: in CPU JIT, extend blocks also beyond jump and links
* n64: bump serializer
* n64: in CPU jit, synchronize CPU/RSP on status and semaphore I/O
* n64: in CPU JIT, fix LWR.
* n64: in CPU JIT, compute homebrew dcache dirty mask after stores
* n64: CPU now has configurable interleaving with the rest of the hardware
* n64: split recompiler in three files
* n64: in CPU JIT, cleanup to remove redundant structure fields
* n64: revisit integration between JIT and gdb to improve performance
* n64: make sure events inserted in the queue affect JIT timings
* n64: in CPU JIT, fix a bug in TLB handling (eg: Linux)
* n64: in CPU JIT, simplify decoder removing dead code
* n64: in CPU JIT, blocks now extend through branches
* n64: avoid excessive JIT trashing by reducing the 4K section granularity
* n64: in CPU JIT, do not emit pipeline prologue for linear instructions
* n64: in CPU JIT, make epilogue code conditional in fast paths
* n64: in CPU JIT, return pipeline information from emit functions
* n64: in CPU JIT, add SDL/SDR/SWL/SWL/SWC1/SDC1
* n64: implement LWL, LWR, LDL, LDR, LWC1, LDC1 in JIT
* n64: in CPU JIT, use JIT-time PC for branches as much as possible
* n64: in FPU JIT, optimize comparisons
* n64: in FPU JIT, implement compare opcodes (first draft)
* n64: in FPU JIT, add MOV.S/MOV.D implementation
* n64: in FPU JIT, add comments and slight tweaks to underflow fixup
* n64: in FPU JIT, reduce input checks required for conversion opcodes
* n64: in FPU JIT, reduce fallbacks to C++ for conversions in most occurrences
* n64: in FPU JIT, simplify a bit conversion opcodes
* n64: fix a benign typo in FCVT_D_L
* n64: implement float/int conversions in FPU JIT
* n64: in FPU JIT, implement ALU double opcodes
* n64: refactoring and cleanup of FPU JIT opcode emission
* n64: in FPU JIT, implement FADD.S and FSUB.S
* n64: in FPU JIT, unity 1/2 arity codepaths
* n64: in FPU JIT, add support for MUL.S and DIV.S
* n64: in FPU JIT, implement passthrough of flag bits from host processor
* n64: simplify FPU JIT by moving some code into nall/recompiler
* n64: start implementation of FPU JIT for amd64 too
* n64: implement initial FPU JIT, for ARM64
* n64: in CPU JIT, remove the deferred pc/nextpc/pipeline materialization
* n64: refactor to remove obsolete decoder JIT flags
* n64: in JIT, add optimizations for accesses via $sp/$gp
* n64: implement also memory store opcodes in JIT
* n64: implement all aligned load opcodes in JIT
* n64: in JIT, remove confusing callf macro
* n64: optimize dcache to use a single word with both tag and valid bit
* n64: add JIT for LW as a first experiment
* n64 cpu: inline MFC1/MTC1/DMFC1/DMTC1 in the JIT when COP1 is enabled.
* n64: in CPU JIT, inline icache management
* nall: add mov128 and mem0 for absolute addressing
* n64: in CPU JIT, inline fastpath of icache hit check
* n64: in CPU JIT, simplify branch link management
* n64: JIT multiply/divide SPECIAL opcodes via SLJIT
* n64: lazily emit callf setup in CPU JIT
* n64: centralize JIT callf setup
* n64: JIT 64-bit shifts and trap opcodes
* n64: JIT more 64-bit integer ALU operations
* n64: optimize CPU JIT ALU slow paths
* n64: gate JIT pipeline state machinery on demand
* n64: defer nextpc materialization in JIT hot path
* n64: lazily commit ipu.pc in JIT loop
* n64: materialize CPU pipeline pc only when needed
* n64: flush deferred CPU cycles only on branch/helper opcodes
* n64: restore unconditional CPU JIT endblock checks
* n64: add virtual PC materialization in CPU JIT loop
* n64: avoid unconditional JIT step flush in hot loop
* n64: gate JIT EndBlock checks on branch and helper ops
* n64: remove CPU JIT metrics instrumentation
* n64: extend JIT branch linking to dual targets
* n64: add safe intra-section JIT block linking
* n64: refactor CPU JIT to introduce 4K sections and state key
* n64: add JIT documentation
* n64: convert remaining CPU branch opcodes to JIT
* n64: convert four CPU branches to JIT and silence offsetof warnings
* n64: convert four more CPU branches to JIT
* n64: convert BEQ to JIT
* n64: initial cleanup of CPU JIT, introduce opcode decoder

Download: Ares Git (2026/05/11) x64
Source:Here