KytyPlus v2.3 – FSR Upscaler + Compatibility Push
Tag: v2.3 · License: GPL-2.0
Downloads
- Windows: KytyPlus-Windows-x64.zip
- macOS: KytyPlus-macOS-x86_64.zip
- Linux: KytyPlus-Linux-x86_64.zip
Key Changes
FSR 1.0 Upscaler (All GPUs)
A new spatial upscaler that makes games look sharper when scaled up to your monitor, with no GPU brand restriction — works on AMD, NVIDIA, and Intel equally.
- Upscales the internal game resolution to your window/monitor size using edge-aware directional filtering, then applies adaptive sharpening for a crisp result.
- Four quality presets to balance image quality vs. performance:
- Ultra Quality — subtle upscale, minimal performance cost
- Quality — good balance (default)
- Balanced — bigger performance gain, slight softness
- Performance — maximum FPS uplift, visible quality trade-off
- Adjustable sharpening slider (0.0–1.0) to dial in how crisp the image looks.
- "Enable via the config file (set upscaler_method to Fsr1). Falls back gracefully if your GPU can't handle it."
Broader Game Engine Compatibility
Significant new HLE modules and graphics improvements that expand the range of titles that boot further:
- New game engine modules for scripting, physics, collision, archive reading, and texture streaming.
- Audio output, controller input, save data, and system service HLE added.
- Texture decode acceleration for BC7 and ASTC formats used by modern titles.
- Improved PM4 command processing benefits Unreal Engine 4/5 and other engine families.
iGPU / Handheld-Friendly Optimizations
Several under-the-hood improvements aimed at getting better performance on integrated GPUs (Radeon 780M, Intel Arc iGPU) and lower-end hardware:
- Resolution scaling — render at half or quarter resolution internally, then upscale to your screen. Major bandwidth and VRAM savings on shared-memory systems.
- Texture LOD bias — skip the highest-resolution mip levels to reduce memory pressure on iGPUs with limited VRAM.
- UMA staging bypass — eliminates redundant memory copies on unified-memory architectures.
- CPU timing throttle — matches guest console clock expectations to prevent timing-related glitches.
Smoother Frame Delivery
- Pipeline state tracking has been optimized to reduce redundant GPU commands, lowering driver overhead per frame.
- First-encounter stutter from pipeline compilation is further reduced with an improved cache that skips rebinding identical pipelines.