TensorRT-RTX 1.4 Support Matrix#

GPU Architecture and Precision Support#

The following table lists NVIDIA hardware and the precision modes that each hardware supports.

Table 1 Supported Hardware#
CUDA Compute Capability	GPU Architecture	Example Devices	FP32	FP16	BF16	INT8 WoQ GEMM	INT4 WoQ GEMM	FP8	FP4 GEMM
12.0	NVIDIA Blackwell	NVIDIA RTX 5090	Yes	Yes	Yes	Yes	Yes	Yes	Yes
8.9	NVIDIA Ada Lovelace	NVIDIA RTX 4090	Yes	Yes	Yes	Yes	Yes	Yes	No
8.6	NVIDIA Ampere	NVIDIA RTX 3090	Yes	Yes	Yes	Yes	Yes	No	No
7.5	NVIDIA Turing	NVIDIA RTX 2080Ti	Yes [1]	Yes	No	No	No	No	No

Note

Engines built with TensorRT-RTX are by default portable across devices with compute capabilities 8.6, 8.9, and 12.0, if the model precisions are supported on the target architecture.
By default, TensorRT-RTX does not create portable engines that are compatible with Turing when targeting other compute capabilities, as this can impact performance. Instead of relying on multi-device portable engines that include Turing support, we encourage developers to build optimized, device-specific engines for Turing using appropriate API flags.
Operator support varies across numeric types.

The following table shows which GPU compute capabilities are supported on each platform.

Table 2 Supported Compute Capabilities per Platform#
Platform	Compute Capability
Linux x86-64	7.5 8.6 8.9 12.0
Windows x64	7.5 8.6 8.9 12.0

The following table shows the required compiler and Python versions for building and using TensorRT-RTX on each supported platform.

Table 3 List of Supported Platforms per Software Version#
Platform	Compiler Version	Tested CUDA Version	TensorRT-RTX Python API [2]	TensorRT-RTX C++ API
Rocky Linux 8.9 x86-64 Ubuntu 22.04 x86-64 Ubuntu 24.04 x86-64	gcc 11.2.1	12.9, 13.2 [3]	Yes	Yes
Windows 10 x64 Windows 11 x64	MSVC 2019 v16.9.2	12.9, 13.2 [3]	Yes	Yes

Footnotes