Two days ago AMD revealed a beast in the market that has everyone in shock. Not because of its specs but Because of its anomalous design, 50% bigger than NAVI.
Designated to be placed in big messes of cables found in most data centers, the M100 is a significant number-crunching GPO.
This card doesn’t connect with your monitor, to give you a boost in gaming which is a bummer as it provides no graphic output or any sort of fixed-function graphic blocks, providing 120 compute units.
But it doesn’t fail to modify and fasten your AI workloads and HPC. Due to the latest “Matric Core”, this device converts 23.1 TFLOPS of FP32 performance to 46.1 with the help of AMD’s new edition of wavefront-level instruction called MFMA instructions.
The valor of this new device doesn’t stop here as it further amazes us with its whopping 32GB HBMe memory providing a peculiar bandwidth of 1.23Tb/s and a PCle 4.0 support consuming only 300 W TDP, similar to the yet to be launched RX 6900 XT.
Additionally, the reason for this placid TDP, surprisingly for a chip of such size is because most of the silicon has been removed to establish space for extra number-crunching kits.