PantheonSim / GPUs / NVIDIA RTX PRO 4500 Blackwell Server Edition

Simulated NVIDIA RTX PRO 4500 Blackwell Server Edition: a GPU profile for CUDA and ROCm testing

PantheonSim, the GPU simulator for CUDA and ROCm, simulates the NVIDIA RTX PRO 4500 Blackwell Server Edition (Blackwell, 12.0) as the profile nvidia/rtx-pro-4500. The Server Edition of the RTX PRO 4500 Blackwell, a 32 GB GB203 card with 82 SMs. The profile was read from a physical card, an AWS g7 instance. Its profile is verified, which means it was read from 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/rtx-pro-4500`
Vendor and architecture	NVIDIA, Blackwell
Compute capability	`12.0`
Device memory	31.4 GiB
Streaming multiprocessors	82
Warp size	32
Max threads per multiprocessor	1536
Shared memory per block (opt-in maximum)	99 KiB
L2 cache	64 MiB
Memory bus width	256-bit
Power limit (profile)	165 W
Maximum clocks (profile)	2415 MHz shader, 12501 MHz memory
Tensor or matrix cores	yes
Verified (measured on a physical card)	yes

What this GPU is for

The Server Edition of the RTX PRO 4500 Blackwell, a 32 GB GB203 card with 82 SMs. The profile was read from a physical card, an AWS g7 instance.

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 4500 and 6000 cards, and 12.1 for the GB10. Only the RTX PRO 4500 Server Edition and the RTX PRO 6000 Server Edition were read from physical cards.

Among the 38 NVIDIA profiles it ranks 25th by device memory (31.4 GiB) and 24th by SMs (82), counting the largest as first.

Is this profile verified?

Yes. This profile is marked verified, which means its values were read from a physical NVIDIA RTX PRO 4500 Blackwell Server Edition by the simulator's characterization tools, not copied from a data sheet. Verified describes the profile's numbers: it says nothing about speed, because the simulator has no timing model.

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/rtx-pro-4500 ./my_app
vgpu shell -y --gpu nvidia/rtx-pro-4500 --count 4
vgpu info --gpu nvidia/rtx-pro-4500

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/rtx-pro-4500 in a pantheonsim.yaml settings file, or in a GitHub Actions job give the PantheonSim action gpu: nvidia/rtx-pro-4500 (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 RTX PRO 4500 Blackwell Server Edition, as written in the file:

Source note for nvidia/rtx-pro-4500

Generated by nvidia/tools/characterize.cu from a physical device. Every value below was read from the hardware, not a datasheet. Card: an AWS g7.2xlarge (driver 595.91.07, CUDA 13.2, 2026-10-18). Not recorded here because the run did not read them: ECC, memory type, PCIe link and how the SMs group into GPCs (no `layout:`, so a cluster launch spreads SMs evenly). The L2 size is the datasheet's.

The GPU index lists every profile, and this one is measured on a physical card.

Other Blackwell profiles: