g3.4xlarge

The g3.4xlarge instance is in the GPU instance family with 16 vCPUs, 122 GiB of memory and up to 10 Gibps of bandwidth starting at $1.14 per hour.

Pricing

$1.140

On Demand

N/A

Spot

N/A

1-Year Reserved

N/A

3-Year Reserved

Family Sizes

SizevCPUsMemory (GiB)
g3.4xlarge16122
g3.8xlarge32244
g3.16xlarge64488

Instance Variants

Variant
g3
g3s

Instance Details

ComputeValue
vCPUs16
Memory (GiB)122
Memory per vCPU (GiB)7.63
Physical ProcessorIntel Xeon E5-2686 v4 (Broadwell)
Clock Speed (GHz)2.3 GHz
CPU Architecturex86_64
GPU
1
GPU Average Wattage0 W
GPU Architecture
NVIDIA Tesla M60
Video Memory (GiB)8
GPU Compute Capability ?5.2
FPGA
0
ffmpeg FPSN/A
CoreMark iterations/SecondN/A
NUMA ArchitectureValue
Uses NUMA Architecture ?N/A
NUMA Node CountN/A
Max NUMA DistanceN/A
Cores per NUMA Node (Avg)N/A
Threads per NUMA Node (Avg)N/A
Memory per NUMA Node (Avg MB)N/A
L3 Cache per NUMA Node (Avg MB)N/A
L3 Cache SharedN/A
NetworkingValue
Network Performance (Gibps)up to 10
Enhanced Networking
true
IPv6
true
Placement Group ?
StorageValue
EBS Optimized
false
Max Bandwidth (Mbps) on EBS0
Max Throughput (MB/s) on EBS0
Max I/O operations/second IOPS0
Baseline Bandwidth (Mbps) on EBS0
Baseline Throughput (MB/s) on EBS0
Baseline I/O operations/second IOPS0
Devices0
Swap Partition
false
NVME Drive ?
false
Disk Space (GiB)0
SSD
false
Initialize Storage
false
Instance Store Read IOPSN/A
Instance Store Write IOPSN/A
AmazonValue
Generation
current
Date Introduced2017-07-13
Instance Typeg3.4xlarge
FamilyGPU instance
NameG3 Graphics GPU Quadruple Extra Large
Elastic Map Reduce EMR
true
By Net service