What's New
- Added hardware support for AMD RX 9000 series GPUs
- Added NVIDIA Hopper and Blackwell data center GPU support
- Added support for NVIDIA 50xx series GPUs
- Added CUDA 12 support
- Added Apple Metal performance optimizations
- Added macOS Tahoe and Sequoia support with pre-compiled shaders
- Removed macOS AMD GPU support
- Use of newest Redshift code, with the following areas modified:
- Bump mapping improvements
- Improved convergence for volume GI rendering
- Instanced mesh lights now consume less memory
- Reduced texture processing time
- Upgrade to OCIO 2.3.2
- Upgrade to OIDN 2.3.3
- Upgrade to OpenColorIO 2.4.2
- Semantics of Core and Thread notation fixed in Cinebench UI
- Single Core SMT measurement option for CPUs with this feature on Windows added
- Update to newer Clang compiler versions on macOS (Xcode 26) and Windows (clang v19), leading to small performance improvements for RS-CPU on x86 and Arm architectures
- New Maxon branding
- New range for Cinebench scores. With several improvements in place and newer compilers used in Redshift, Cinebench got faster (with the same scene) on every hardware. Therefore, we moved the score calculation into a different (higher) range, avoiding customer confusion or mix-up with Cinebench 2024 results.
Comments
0 comments
Article is closed for comments.