PCI Express graphics card RTX4090m

HDMI

Product overview
RTX4090m 16GD6 3PH is a NVIDIA Ada Lovelace‑architecture flagship PCIe graphics card with 16GB GDDR6 memory and 9,728 CUDA cores. It uses an ATX 2‑slot form factor and provides 3×DisplayPort 1.4a plus 1×HDMI 2.1 outputs supporting up to 8K@60Hz, suitable for high‑end workstations and AI inference platforms.

Key features

• NVIDIA Ada Lovelace architecture — NVIDIA GeForce RTX4090 Max‑Q GPU
• 16GB GDDR6, 256‑bit memory interface, 576GB/s memory bandwidth
• 9,728 CUDA cores, 304 Tensor Cores, 76 RT Cores
• PCIe 4.0 x16 full‑speed interface
• ATX / 2‑slot form factor (267 × 111 × 42 mm)
• 3×DisplayPort 1.4a + 1×HDMI 2.1, supports 7680×4320@60Hz
• TDP 150W (max 240W), requires 1×8‑pin external power
• Supports Windows 10/11 and Linux 64‑bit

Detailed description
The RTX4090m 16GD6 3PH integrates the NVIDIA GeForce RTX4090 Max‑Q GPU based on the Ada Lovelace architecture. With 9,728 CUDA cores, 304 Tensor Cores and 76 RT Cores, the card delivers up to 44.5 TFLOPS FP32 compute performance. The 16GB GDDR6 memory on a 256‑bit bus and 576GB/s bandwidth enables handling of large game engines, 8K video post‑production and complex AI inference tasks.

The card conforms to ATX 2‑slot dimensions (267×111×42 mm) and connects via PCIe 4.0 x16. It offers 3 DisplayPort 1.4a and 1 HDMI 2.1 output, each supporting 7680×4320@60Hz. Power consumption is specified at 150W TDP (up to 240W peak) and requires a single 8‑pin external power connector. Supported operating systems include Windows 10/11 and Linux 64‑bit. Operating temperature range is 0°C to 45°C. Typical applications include high‑end gaming systems, professional visual workstations and edge AI compute nodes.

Specifications
Model: RTX4090m 16GD6 3PH (FA11)
Product type: PCIe flagship graphics card
GPU architecture: NVIDIA Ada Lovelace
GPU model: NVIDIA GeForce RTX4090 Max‑Q
Memory capacity: 16GB
Memory type: GDDR6
Memory bus: 256‑bit
Memory bandwidth: 576GB/s
Core clock: 1620 MHz (Boost 2040 MHz)
Memory clock: 2250 MHz
CUDA cores: 9728
Tensor Cores: 304
RT Cores: 76
TMUs: 304
ROPs: 112
SM count: 76
L1 cache: 128 KB (per SM)
L2 cache: 64 MB
PCIe interface: PCI Express 4.0 x16 (by 16x)
Board spec: ATX / 2‑slot
Dimensions: 267 × 111 × 42 mm
Display outputs: 3×DisplayPort 1.4a + 1×HDMI 2.1
Maximum resolution: 7680×4320@60Hz (DP/HDMI)
Maximum displays: 4
TDP: 150W (max 240W)
External power: 1×8‑pin
Pixel fillrate: 163.0 GPixel/s
Texture fillrate: 442.3 GTexel/s
FP32 (single precision): 44.5 TFLOPS
FP64 (double precision): 706.1 GFLOPS
DirectX: 12 Ultimate (12_2)
OpenGL: 4.6
OpenCL: 3.0
Vulkan: 1.3
Shader Model: 6.7
CUDA version: 8.9
Operating systems: Windows 10 64‑bit, Windows 11, Linux 64‑bit
Operating temperature: 0°C ~ 45°C
Storage temperature: -20°C ~ 75°C
Operating humidity: 0% ~ 95% (non‑condensing)
Storage humidity: 10% ~ 90%