PantheonSim / GPUs / NVIDIA GeForce RTX 5090

Simulated NVIDIA GeForce RTX 5090: a GPU profile for CUDA and ROCm testing

PantheonSim, the GPU simulator for CUDA and ROCm, simulates the NVIDIA GeForce RTX 5090 (Blackwell, 12.0) as the profile nvidia/rtx5090. A 32 GB Blackwell GeForce flagship at compute capability 12.0. Its profile is built from public documents and not yet measured against a physical card.

The profile at a glance

These are the numbers a program can ask this simulated GPU, and the values nvidia-smi show for it. They come from the profile file in the simulator's repository, pinned to one commit of it.

Property	Value
Profile id	`nvidia/rtx5090`
Vendor and architecture	NVIDIA, Blackwell
Compute capability	`12.0`
Device memory	31.8 GiB, GDDR
Streaming multiprocessors	170
Warp size	32
Max threads per multiprocessor	1536
Shared memory per block (opt-in maximum)	99 KiB
L2 cache	96 MiB
Memory bus width	512-bit
Power limit (profile)	575 W
Maximum clocks (profile)	2964 MHz shader, 14001 MHz memory
ECC memory	no
PCIe link reported	Gen5 x8
Tensor or matrix cores	yes
Verified (measured on a physical card)	not yet measured

What this GPU is for

A 32 GB Blackwell GeForce flagship at compute capability 12.0.

Blackwell is split across several compute capabilities in these profiles: 10.0 for the B200 and GB200, 10.3 for the B300, 11.0 for Jetson Thor, 12.0 for the GeForce RTX 50 series and the RTX PRO 6000 cards, and 12.1 for the GB10. Only the RTX PRO 6000 Server Edition was read from a physical card.

Among the 37 NVIDIA profiles it ranks 24th by device memory (31.8 GiB) and 5th by SMs (170), counting the largest as first.

Is this profile verified?

No. This profile is marked not verified: it is built from public documents only, and nothing in it was measured on a physical card. Some figures are derived or copied from the nearest profile where the vendor publishes none, so treat them as approximate, and memory sizes as upper bounds. Programs still run on it, and it still answers every call as the architecture's rules require, but a value such as a clock or a limit may differ from a real card's.

Simulate this GPU

Build the simulator (see the docs) or open the playground, then:

nvcc -cudart shared -arch=compute_120 my_app.cu -o my_app
vgpu run --gpu nvidia/rtx5090 ./my_app
vgpu shell -y --gpu nvidia/rtx5090 --count 4
vgpu info --gpu nvidia/rtx5090

This profile is compute capability 12.0. A kernel built for compute_75 runs on it and on every other NVIDIA profile; a kernel built for a newer architecture is refused by a profile that is older than it. Blackwell targets need CUDA 12.8 or newer. Link the CUDA runtime shared (-cudart shared) so the simulator can stand in for it.

To use it every time, put gpu: nvidia/rtx5090 in a pantheonsim.yaml settings file, or in a GitHub Actions job give the PantheonSim action gpu: nvidia/rtx5090 (see CUDA and HIP tests in GitHub Actions). The simulator checks behaviour and has no timing model, so a run on a physical GPU before a release still counts. The CUDA guide explains the rest.

Where the numbers come from

Each profile file begins with a note on its sources. This is the note for NVIDIA GeForce RTX 5090, as written in the file:

Source note for nvidia/rtx5090

Limits from NVIDIA's public documentation for compute capability 12.0 (the CUDA C++ Programming Guide's per-capability table) and the RTX 5090 spec sheet; the framebuffer, power limit and links from the benchmark database. Not characterized on hardware yet. verified: false.

The GPU index lists every profile, and this one is not yet measured.

Other Blackwell profiles: