PPSSPP Git (2026/10/15) is compiled. PPSSPP is a fast and portable PSP emulator for Android, Windows, Mac, and Linux, written in C++.

PPSSPP Git Changelog:
* Merge pull request #22579 from hrydgard/ir-jit-hot-blocks
* IR JITs: Name the fixed code routines that jumps go to in the disassembly
* IR: Build a dot operand of one quad's lanes and constants as a vector
* ARM64 IR JIT: Join two mapped pairs into a quad with a ZIP1
* IR: Vec4 vmin, vmax, vslt, vsge and vi2f
* ARM64 IR JIT: Look up the block for a jump to a register at the jump
* IR: Compile on past conditional branches
* CPU fuzzer: Generate forward branches, and keep them intact when minimizing
* IR: Compute runs of VFPU dot products four at a time
* Unit tests: Add JitDotBench, timing the JIT's VFPU dot products
* ARM64 emitter: Add vector CLZ, LD4, UMULL by element, ABS and SLI
* JIT: Add --jit-profile, a sampling profiler for the CPU cores
* Merge pull request #22581 from hrydgard/resistance-crash-19755
* Resistance: Retribution: Add smoke trail nodes at most every 1/30s
* Resistance: Retribution: Stop smoke trails overflowing the display list
* Scan modules that only exist in a savestate for functions
* Merge pull request #22580 from eraycep/make-scrollbar-fully-opaque-while-held-down
* UI: Make scrollbar opaque while dragging
* Merge pull request #22577 from hrydgard/gpu-cull-speed
* CrossSIMD: The C++ fallback's AsVec3ByMatrix44 computes W
* Depth raster: Stop a row once it has left the triangle
* GPU: Cull-test draws four vertices at a time
* CrossSIMD: Load 8-bit vertex components without assuming alignment
* Merge pull request #22576 from hrydgard/cpu-generic-fixes
* Mac ARM64: Switch JIT memory between writable and executable per thread
* bltzal and friends test rs from before they link
* IR: vmtvc to a masked control register reads vs
* IR: Keep d in vcmov's lanes that don't move under a D prefix
* IR: Compare store offsets as signed in OptimizeLoadsAfterStores
* IR JITs: Flushing a host register that holds zero unmaps it
* IR: vf2i and vi2f go through temps when the result overlaps the source
* IR: Fix the matrix inits of transposed matrices with an offset
* IR: Don't propagate an FP copy into a result that covers its source
* IR: vc2i and vuc2i read their source at the size encoded
* VFPU: vfim converts its half float like vh2f
* IR: Fix vmscl of two transposed matrices with an offset
* Interpreter: Vector loads ignore the D prefix's write mask
* Headless: Decode vertices with the JIT where the host can, like the app
* Merge pull request #22573 from hrydgard/remove-fakejit
* AGENTS.md: Update submodules when starting work
* Remove FakeJit and the pre-IR JitBlockCache
* Merge pull request #22572 from hrydgard/ci-alpine-cache
* CI: Make the Alpine job's caches save
* Merge pull request #22568 from hrydgard/accurate-vdot-ir
* x86-64 IR JIT: Address the dot routine's operands from CTXREG's offset
* IR: Include prefix constants past the vector size in the accurate dot products
* x86-64 IR JIT: Call the dot routine with the operands in registers
* ARM32 IR JIT: Call a NEON dot routine for the accurate dot products
* Compat: Replace MoreAccurateVMMUL with ForceAccurateVdot
* VFPU: Make AccurateVdot the default
* ARM64 IR JIT: Call the dot routine with the operands in registers
* x86-64 JITs: Make the VFPU dot routine SSE2, after vfpu_dot_sse2
* ARM64 emitter: Add vector BIC, and use it for the dot routine's normal mask
* ARM64 JITs: Make the VFPU dot routine NEON, after clang's vfpu_dot_neon
* ARM64 emitter: Add vector ADD/SUB, UMAXP, USHL, UMULL/UMULL2, ADDV and USRA
* x86-64 IR JIT: Run Vec4VfpuDot through a shared routine
* ARM64 IR JIT: Run Vec4VfpuDot through a shared routine
* CPUEquivalence: Add PPSSPP_CPUFUZZ_ACCURATE_VDOT
* IR: Add Vec4VfpuDot, the VFPU's own dot product, and use it with AccurateVdot
* Interpreter: Use the accurate VFPU dot product when AccurateVdot is set
* Config: Add AccurateVdot, and --accurate-vdot to set it
* Merge pull request #22571 from hrydgard/vrnds-seed-from-vs
* VFPU: vrnds takes its seed from the vs field
* Merge pull request #22569 from hrydgard/readback-shadow
* D3D11: Include CommonWindows.h before the Windows SDK headers
* Add a ReadbackShadow compat flag, for games that read many pieces of a buffer
* Merge pull request #22570 from hrydgard/depth-raster-region-clip
* Depth raster: Clip to the drawing region as well as the scissor
* Merge pull request #22567 from hrydgard/cpu-generic-fixes
* x86-64 IR JIT: Keep NaNs through vsat0 and the [0:1] prefix
* VFPU: Decode prefixes with shared helpers
* IR: Leave the dot products with prefix constants past the vector size to the interpreter
* ArmDisasm test: Expect the ARMv7 sequences on ARMv7
* ARM emitter: Let VMOV_imm load ones-filled I_32 immediates
* CPU fuzzer: Add an instruction filter, and a script to run it in parallel
* CMake: List the headers that were missing
* Merge pull request #22561 from eraycep/fix-sdl-input-pixel-density
* SDL: Use window pixel density for input coordinates
* Merge pull request #22564 from hrydgard/arm32-ir-jit
* ARM32 IR JIT: Pass float helper arguments the softfp way where needed
* Remove the legacy ARM32 JIT
* ARM32 IR JIT: Vec4 operations
* ARM32 IR JIT: FPU arithmetic, compares, conversions and math helpers
* ARM32 IR JIT: Loads and stores
* ARM32 IR JIT: Integer arithmetic, logic, shifts, compares, mult and div
* ARM32: Start an IR-based JIT
* Add a headless armhf cross build, for testing ARM32 under qemu
* Merge pull request #22566 from hrydgard/fix-armv7-anycomparebits
* CrossSIMD: Fix AnyCompareBitsSet on ARMv7
* Merge pull request #22565 from hrydgard/dnas-decrypt
* Tools: Script to connect PSPLink to WSL
* Load game patches installed on a PSP, given its Fuse ID
* Implement sceKernelLoadModuleDNAS
* Merge pull request #22562 from hrydgard/ir-jit-fixes
* ARM64 IR JIT: Keep the address out of SCRATCH1 across flushes
* IR regcache: Load a dest that's a later lane of a source
* VFPU: Don't assert on mfv/mtv past the control registers
* Merge pull request #22563 from hrydgard/remove-legacy-arm64-x64-jits
* Remove the legacy (non-IR) x86-64 and ARM64 JITs
* Merge pull request #22560 from eraycep/add-fading-chat-button
* Merge branch 'master' into add-fading-chat-button
* UI: Fade chat button with touch controls
* Merge pull request #22558 from hrydgard/cpu-equiv-6
* Docs: CPU equivalence state, including ARM32
* ARM JIT: Leave matrix inits with a prefix to the interpreter, like the others
* ARM JIT: Fix mtvc masking and vsgn of denormals, and interpret vh2f
* ARM JIT: Gate VFPU ops on prefixes like the other JITs, and drop its vasin
* ARM JIT: Fix ins with msb below pos, ext past bit 31, and constant max/min
* Merge pull request #22559 from hrydgard/game-browser-xmb-list
* Search: Drop the focus from a view the filter hides
* Game browser: Switch existing installs to the large list on Games and Homebrew
* Game browser: Default to the large list for Games and Homebrew, bigger folders
* Game browser: Keep the search and focus when returning from a game
* Translations: Game browser view modes, "Icon size"
* Game browser: Keyboard focus for the info button, more left padding
* Game browser: Info button in the large list, long press with buttons
* Game browser: Add an XMB-style large list mode
* GitLab CI: Only run libretro pipelines for master, tags and manual runs

Download: PPSSPP Git (2026/10/15) x64
Download: PPSSPP Git (2026/10/15) Android
Source: Here