Google Cloud offers a range of Arm powered servers in Compute Engine throughthe N4A,C4A bare metal(Preview), C4A, and A4Xmachine series. Arm architecture is optimized for powerefficiency, and as a result can yield better price for performance.

Arm processors are common in standard servers due to their power efficiency ascompared to x86 servers. Mobile phones and laptops are examples of devices thatrun on an Arm processor. With an Arm CPU's reduced instruction set, fewerinstructions equals greater performance speed with lower battery and powerconsumption.

N4A uses Google's Axion Arm processor with the Neoverse N3 processor. C4A usesthe Axion Arm processor based on the Arm Neoverse V2 processor. The Neoverse V2is the first V-series CPU to have Armv9 performance, power, and securityenhancements. It is designed for high performance computing, machine learning,and general-purpose cloud computing. Consider using N4A or C4A general-purposeArm virtual machines (VMs) for any of the following purposes:

• Run compute-intensive workloads that require the ability to scaleusage quickly when needed.
• Optimize for price-performance on Arm-compatible workloads.
• Build on modern, open source software stacks.
• Develop and test mobile or embedded systems which use an Arm CPU.
• Evaluate whether your workload is suitable for an Arm CPU.

To use GPUs with an Arm-based CPU, choose either the A4X Max or A4X machineseries, which runs on the NVIDIA rack scale architecture. Instances in thesemachine series have attached NVIDIA Grace Blackwell Superchips. These machineseries are optimized for massively parallelized Compute Unified DeviceArchitecture (CUDA) compute workloads, such as machine learning (ML) and highperformance computing (HPC).

A4X Max and A4X machine series

Caution: TheCompute Engine Service Level Agreement (SLA)doesn't apply to the A4X Max and A4X machine series.

TheA4X Max and A4X machine serieshave both Arm-based CPUs and attached GPUs.Both machine types in this series have two sockets with NVIDIAGrace Arm CPUs connected to four GPUs with fast chip-to-chip (NVLink C2C)communication.

• A4X Max: offers machine types with up to 144 vCPUs and 960 GB ofmemory. It uses NVIDIA GB300 Superchips, which have B300 GPUs that offer279 GB memory per GPU. A4X Max is available as a bare metal instancein thea4x-maxgpu-4g-metal machine type.
• A4X: offers machine types that have up to 140 vCPUs and 884 GB ofmemory. It uses NVIDIA GB200 Superchips, which have B200 GPUs that offer186 GB memory per GPU. A4X is available in thea4x-highgpu-4g machinetype.

Storage options for A4X Max and A4X instances

A4X Max and A4X can be used with Google Cloud Hyperdisk attached storage and comes with12,000 GiB ofLocal SSD. Compute Engineautomatically attaches the Local SSD disks to your instances during instancecreation.

OS images

A4X Max and A4X instances support public Arm-basedOS images.You can alsocreate custom imagesusing a public Arm-based OS image.

N4A machine series

N4A is thelatest Google Axion-based VM, built on Neoverse N3 CPU. N4A offers machine typeswith up to 64 vCPUs and 512 GB of DDR5 memory and Titaniuminfrastructure processing unit (IPU) support. N4A uses next-generationdynamic resource management and isavailable instandard,highmem, andhighcpu machine types with the optionto customize your machine type and add extended memory.

N4A supports standard networking up to 50 Gbps with thegVNIC networking interface. N4A alsosupports the NVMe disk interface with Hyperdisk Balanced, Hyperdisk Balanced High Availability, and Hyperdisk Throughput storage.

C4A machine series

C4A is the firstArm-based VM built on Google's AxionArm64-based CPU. C4A offers machine types with up to 72 vCPUs and 576 GB ofDDR5 memory. C4A is available instandard,highmem, andhighcpumachine types.

C4A offers a(Preview)bare metalhighmem 96 vCPU machine type with 768 GB of DDR5 memory.

C4A is built onTitanium which usesnetwork offloads and enables per VM Tier_1 networking performance of up to 100 Gbps with thegVNIC networking interface.C4A also supports the NVMe and IDPF disk interfaces with Hyperdiskvolumes.

Simultaneous multithreading

For the C4A machine series, each vCPU is backed by a single core with nosimultaneous multithreading (SMT).Thus, C4A VMs deliver greater performance per vCPU compared to a VM with SMTenabled. While SMT provides benefits to certain workloads, single-threaded coresare ideal for compute-intensive workloads because the processes can access theentire core instead of sharing it with other processes.

OS images

C4A and N4A VMs support public Arm-basedOS images.You can alsocreate custom imagesusing a publicly-available Arm-based image.

Tau T2A machine series

TheTau T2A Armmachine series runs on the 64 core Ampere Altra Arm processor at 3.0 GHzall-core frequency. Tau T2A makes it possible to run workloads that run best,or exclusively, on Arm.

TheTau T2A machine serieshas predefined machine types of up to 48 physical cores with 4 GB of memory pervCPU. Tau T2A machine types run within a singleNUMAnode.

Tau T2A machine types support only the NVMe interface for storage, andGoogle virtual NIC (gVNIC) fornetworking. Virtio-Net and SCSI interfaces are not supported. Allpublicly-available ArmOS imagesare configured to use the NVMe and gVNIC interfaces.gVNIC is a network interface that is designed specifically forCompute Engine. It provides better performance and supports highernetwork bandwidths and throughput.

For this machine series, each vCPU is backed by a single core with nosimultaneous multithreading (SMT).

Workload recommendations

The C4A machine series is an excellent choice for a wide range of scale-out andcompute-intensive workloads, especially when price performance is a key concern.Consider C4A when you are deploying workloads such as the following:

• ML data processing
• ML inferencing and model serving
• App serving, web serving, and game serving
• Embedded systems development
• Development on CI/CD on Arm
• Video and image encoding, transcoding, and processing
• Digital advertising exchanges and serving
• Cache servers
• Computational drug discovery
• Android development
• Autonomous or conventional automotive software development

What's next

• Review the specifications and features of theA4X Max and A4X machine series.
• Review the specifications for theN4A machine series.
• Review the specifications for theC4A machine series.
• Learn about availableCPU platforms for Google Cloud.
• Create and start a Compute Engine instanceusing an Arm OS image.