I see that CUDA Toolkit 12.8 added Blackwell architecture options listed as sm_100, sm_101, and sm_120 I know how these naming scheme’s usually work sm_70 = CC 7.0 = Volta (V100, Titan V, etc) sm_80 = CC 8.0 = Ampere (A100, etc) sm_89 = CC 8.9 = Ada (RTX 4090, etc) now sm_100 and sm_101 seem ob…

thanks for the reply. i looked through the comments and links there, including your wikipedia link. but the answer still seems ambiguous. wikipedia lists it as 12.0, but TechPowerUp reports it as 10.1 in their specs. both are third party sources, but equally reputable IMO. do you have a more concre…

I don’t have any NVIDIA documentation to refer to yet. Blackwell information is still flowing out.

Thank you. I was able to get some other users to confirm 12.0 also with GPUz So 10.1 is actually the unknown device. Maybe that project digits thing?

Hello. Why compilation for sm_100, 101 and 120 so slow and start to use 255 registers, more stack and no cmem? Compiling entry function ‘Z14comp_keys_compPtPjPmS0’ for ‘sm_86’ ptxas info : Function properties for Z14comp_keys_compPtPjPmS0 4240 bytes stack frame, 0 bytes spill stores, 0 bytes s…

Yes, I faced that problem too. [image] Registers management 8.9 vs 9.0+ CUDA Programming and Performance Hello community, I think I understand the conceptual changes in register file organization between 8.9 and 9.0 and I understand that different allocation g…

On sm_90 all ok as on sm_89 and earlier. I discovered workaround to use sm_100, sm_101, sm_120 without huge and slow ptx and cubin size if using compute_100 and later. Just compile it with compute_89. compute_89, sm_100 and so on.

CUDA Toolkit 12.8 what GPU is 'sm_120'?

Accelerated Computing CUDA CUDA NVCC Compiler

Robert_Crovella January 31, 2025, 3:54pm 2

This may be of interest.

Topic		Replies	Views
Kernel 2x slower when compiling with sm12 and above CUDA Programming and Performance	3	2468	May 24, 2012
Choosing the right "-arch" flag for NVCC? CUDA NVCC Compiler	2	278	March 9, 2026
Nvcc fatal : Unsupported gpu architecture 'sm_12' CUDA NVCC Compiler cuda	2	474	May 13, 2025
Nvcc 13.2 spilling many registers for cc 120 CUDA NVCC Compiler nvcc	1	92	March 23, 2026
question Regarding ptxas info and some weird problem CUDA Programming and Performance	7	1041	November 23, 2010
AM Arch for RTX 4070 Ti CUDA-GDB	1	2596	April 4, 2023
GTX480 faster if compiled with sm12 istead of sm20 CUDA Programming and Performance	0	10818	August 3, 2010
Thread block clustering in Blackwell GPUs CUDA Programming and Performance clustering	21	4838	January 29, 2025
Kernel error: forward compatibility was attempted on non supported HW CUDA Setup and Installation cuda	5	479	August 7, 2025
CUDA Toolkit 12.0 Released for General Availability Technical Blog	0	649	December 12, 2022

CUDA Toolkit 12.8 what GPU is 'sm_120'?

Related topics