Tag: ThunderX2

We discuss the impact of the general-purpose Marvell ThunderX3 getting canceled and how the Arm server industry has pivoted to grow

At Hot Chips 32 (2020) we get a deeper look into the Marvell ThunderX3. Something interesting is that we are now seeing Arm v. Arm marketing efforts

We discuss the opportunity Arm has with cloud service providers and why recent developments are making Arm server chips more compelling than ever

We discuss how the per-socket Licenseageddon we predicted is happening as VMware overhauls their licensing model to be socket and core based

At SC19, Gigabyte showed off new AMD EPYC 7002 series GPU compute servers as well as a Marvell ThunderX2 CUDA on Arm development platform

At Supercomputing 2019, NVIDIA CUDA on Arm is moving into beta with broad Arm server ecosystem support. This dream for many is becoming a reality

We look at Arm-on-Arm (AoA) performance to see how many Raspberry Pi 4 4GB nodes one Marvell ThunderX2 64 core and 256 thread server replaces and the cost

NVIDIA announces that it will fully support Arm for GPU accelerated computing for HPC and AI workloads porting CUDA-X and tools to the architecture

New 1U and 2U Marvell ThunderX2 servers are available with built-in SAS controllers and 10GbE networking paired with dual 32 core 128 thread Arm CPUs

Gigabyte launches two new Cavium ThunderX2 servers each supporting dual sockets and up to 32 cores and 128 threads per socket. The Gigabyte R181-T90 is the 1U offering and the Gigabyte R281-T91 is the 2U offering. Both have large amounts of PCIe I/O, IPMI, and redundant power supplies