For display and output, M3A1000-PP, M3A2000-VY, and M3A4500-WP all support up to 4x DisplayPort (DP) and 8K (7680 × 4320) resolution.
M3A1000-PP provides 6.66 TFLOPS peak FP32 performance and has 2048 CUDA cores, 16 RT cores, 64 Tensor cores, and 4GB GDDR6 memory, while M3A2000-VY offers 8.25 TFLOPS peak FP32 performance and has 2560 CUDA cores, 20 RT cores and 80 Tensor cores, and 8GB GDDR6 memory.
M3A4500-WP, with the highest computer power among all new models, provides 18.55 TFLOPS peak FP32 performance, and has 5888 CUDA cores, 80 RT cores, 184 Tensor cores, and 16GB GDDR6 memory. The MXM GPU module, as well as M3A2000-VY, supports error correction code (ECC) memory for controlling errors in data over unreliable or unstable communication channels.
The new MXM GPU modules, with their compact designs, can be easily integrated into a variety of AI systems to boost the computing performance and run complex AI inference tasks. For these GPU modules, Aetina provides optional conformal coating service to make the modules capable of functioning normally in harsh environments, and ensures 5-year availability of the modules as well as exclusive technical supports to help its clients save maintenance costs and time for their long-term AI projects.