Amazon EC2 offers a massive variety of instance types optimized to fit in the accommodation of varying combinations of CPU, memory, storage and networking capacity. They give the user flexibility to select the suitable mixture of resources for application. 

    Among multiple EC2 Instance Types, Compute Optimized and Accelerated Computing are worth mentioning due to their prime importance.

    Compute Optimized

    Compute-optimized instances are best for compute-bound applications that are benefitted from high performances processors. Instances which belong to this family are optimized for batch processing workloads. 

    C5 Instances: 

    These instances are well optimized for intensive-compute workloads. They deliver high performances at low price.

    Features Of C5:

    • Provide a choice of processors based on the instances’ size
    • 24xlarge instance size offers 96 vCPUs, and 192 GiB of memory
    • C5, C5d 12xlarge, 24xlarge, and metal instance sizes attribute customized 2nd generation Intel Xeon
    • Scalable Processors (Cascade Lake) with a sustained all core Turbo frequency of 3.6GHz and a single-core turbo frequency of up to 3.9GHz.
    • Needs HVM AMIs including drivers for ENA and NVMe
    • Elastic Network Adapter (ENA) offers C5 instances up to 25 Gbps of network bandwidth and up to 14 Gbps of bandwidth 
    • AWS Nitro System powered, a combination of dedicated hardware and lightweight hypervisor
    • With C5d instances, local NVMe-based SSDs are physically linked to the host server and provide block-level storage that is coupled to the lifetime of the C5 instance

     C5n Instances: 

    C5n instances are for high compute applications along with High-Performance Computing (HPC) workloads, data lakes, and network appliances like firewalls and routers which can take advantage of the improved network and packet rate performance. They offer up to 100 Gbps network bandwidth and increased memory. C5n.18xlarge instances give support to Elastic Fabric Adapter (EFA), a network interface for Amazon EC2 instances that allows the customers to run applications demanding high levels of inter-node communications, such as High-Performance Computing (HPC) applications with the use of the Message Passing Interface (MPI), at scale on AWS.

    Features:

    • Run each core up to 3.5 GHz using Intel Turbo Boost Technology
    • 3.0 GHz Intel Xeon Platinum processors with instruction set of Intel Advanced Vector Extension 512 (AVX-512) 
    • Larger size of instance, c5n.18xlarge, offering 72 vCPUs and memory of 192 GiB 
    • Network bandwidth that increases up to 100 Gbps, providing enhanced performance for network-intensive applications.
    • Requires HVM AMIs including drivers for ENA and NVMe
    • Support of EFA on c5n.18xlarge instances
    • 33% higher memory footprint in comparison to C5 instances
    • AWS Nitro System powered

    C4

    C4 instances are generated for intensive-compute workloads and give high performance at a low cost per compute ratio.

    Features: 

    • High-frequency Intel Xeon E5-2666 v3 (Has well) processors optimized for EC2
    • Higher networking performance with Enhanced Networking providing support to Intel 82599 VF
    • EBS-optimized (default) for increased storage performance at zero extra cost
    • Needs Amazon VPC, Amazon EBS and 64-bit HVM AMIs

    Accelerated Computing:

    The accelerated computing instances utilize hardware accelerators and coprocessors for efficiently executing various functions like calculations of a floating-point number, processing of graphics, matching of data pattern, etc.

     

    P3 instances

    The newest generation of ‘general-purpose GPU instances” are P3 instances.

    The Features Are As Follows:

    • Till 8 NVIDIA Tesla V100 GPUs each of which pairs 5,120 CUDA Cores and 640 Tensor Cores
    • High frequency processors of Intel Xeon E5-2686 v4 (Broadwell) for p3.2xlarge, p3.8xlarge, p3.16xlarge, etc.
    • High-frequency processors of 2.5 GHz (base) Intel Xeon P-8175M for p3dn.24xlarge
    • Provides support to NVLink concerning peer-to-peer GPU communication
    • Delivers an aggregate network bandwidth up to 100 Gbps

    P2 Instances

    These are meant for the ‘General Purpose GPU compute applications

    Features Are As Follows:

    • Processors are high-frequency Intel Xeon E5-2686 v4 (Broad well) 
    • NVIDIA K80 GPUs (high performance) each having 2,496 parallel processing cores and 12GiB GPU memory
    • Give supports to GPUDirect™ concerning peer-to-peer GPU communications
    • Delivers enhanced networking through the use of Elastic Network Adapter (ENA) with aggregate network bandwidth (up to 25 Gbps) within specific Placement Group
    • EBS optimized

    Inf1 Instances

    These are generated from the ground until the support machine learning interface applications.

    Features:

    • 16 AWS Inferentia Chips maximum
    • AWS Neuron SDK
    • High frequency “2nd Gen Intel® Xeon® Scalable” processors
    • Networking till 100 Gbps

    G4 Instances

    Graphics-intended workloads and machine learning interfaces are accelerated.

    Features:

    • Processors:- 2nd Generation Intel Xeon Scalable (Cascade Lake) 
    • NVIDIA T4 Tensor Core GPUs
    • Networking Up to 100 Gbps 
    • NVMe storage Up to 1.8 TB

    G3 Instances

    Graphics-intensive applications optimized

    Features Are As Follows:

    • Processors:- High-frequency Intel Xeon E5-2686 v4 (Broadwell) 
    • NVIDIA Tesla M60 GPUs, each having 2048 parallel processing cores and video memory (8 GiB)
    • Enables the features of NVIDIA GRID Virtual Workstation 
    • Enables capabilities of NVIDIA GRID Virtual Application 
    • The GPU features support 18 H.264 1080p30 streams and up to 10 H.265 (HEVC) 1080p30 streams
    • Provides Enhanced Networking via the Elastic Network Adapter (ENA) containing 25 Gbps of aggregate network bandwidth

    F1 instances

    F1 instances provide customizable hardware acceleration with gate arrays (FPGAs) that are field programmable.

    Features:

    • Processors:- High-frequency Intel Xeon E5-2686 v4 (Broadwell) 
    • NVMe SSD Storage
    • Enhanced Networking

    AWS EC2 has brought an extensive amount of CPU capacity with zero requirements for hardware. It eases your applications’ development and lets you execute at a faster rate. 

    Richard is an experienced tech journalist and blogger who is passionate about new and emerging technologies. He provides insightful and engaging content for Connection Cafe and is committed to staying up-to-date on the latest trends and developments.