For the millions still running GTX 1080 Ti or Tesla P100 accelerators, this is a sunset notice. New CUDA toolkit versions will still compile for these architectures, but driver-level optimizations — and critical security patches — will cease after 2027.
The NVCC compiler now defaults to Zstd for "fatbins," leading to smaller binary sizes and faster load times for complex AI applications. cuda driver release news exclusive
For HPC applications utilizing oversubscription (allocating more memory than physically available on the GPU): For the millions still running GTX 1080 Ti
The driver appears to reserve more SM resources for potential compute kernels, hurting pure raster scenarios. NVIDIA’s solution? A new control flag in nvidia-smi . By default, it’s set to “balanced” – but gamers may want “low_latency” to claw back performance. By default, it’s set to “balanced” – but
Here is everything you need to know.