Instances›AWS EC2›g3›g3.16xlarge
Compute
vCPUs64
Memory488 GiB
Mem / vCPU7.63 GiB
ProcessorXeon E5-2686 v4
Clock speed2.3 GHz
Architecturex86_64
Storage
TypeEBS only
BandwidthN/A
ThroughputN/A
IOPS0
Networking
Performance20 Gigabit
Max ENIsN/A
IPv6Yes
GPUNVIDIA Tesla M60
Family sizes
g3.4xlarge16 vCPU · 122 GiB
g3.8xlarge32 vCPU · 244 GiB
g3.16xlarge64 vCPU · 488 GiB
g3.16xlarge

g3.16xlarge is AWS's GPU EC2 size at 64 vCPU and 488 GiB. In the name, "g" is graphics, "3" is the generation and "16xlarge" is the size within the g3 family. It runs on Intel Xeon E5-2686 v4 (Broadwell) at 2.3 GHz, and AWS introduced it in July 2017. AWS has since shipped g4ad, g5, g6 and g7; g4ad.16xlarge is the direct replacement at this size. Rackspace Spot runs one GPU class, gpu.vs2.xlargeplusplus — a single NVIDIA A30 with 24 vCPU and 64 GiB, in San Jose only. Against this instance's 4 × NVIDIA Tesla M60 that is a downgrade per card, honest for inference and not for the training this instance was bought for — it prices $6.14/hr on AWS in US West (N. California), from $0.191/hr on Rackspace Spot.

Match qualityWeaker GPU

g3.16xlarge puts 4 GPUs (32 GiB VRAM) in one machine. Spot has a single GPU class and no multi-GPU node, so there is no shape here that places 4 cards behind one host. Per card the NVIDIA A30 (Ampere, 24 GB VRAM) is a step down from what g3.16xlarge carries, and the aggregate is not reachable at all.

That makes this honest for data-parallel inference and embarrassingly parallel batch work, where each card is independent, and dishonest for anything that shards a model across cards or synchronises every step. The catch is location rather than hardware: US West, San Jose is the only Spot region that offers it, so latency to your users and egress to your data are the things to check.

Pricing by region
1 hr
RegionAWS on-demandAWS Spot1-yr Reserved3-yr ReservedRackspace SpotYou save
US West, San Jose$6.136$0.191
$5.945 savings
When to use it
Good for
  • ●Remote graphics workstations and rendering
  • ●CUDA-accelerated batch jobs
  • ●Model inference and embedding generation
  • ●Workloads that fit inside 488 GiB of RAM
  • ●Parallel work that can occupy 64 vCPU
  • ●Network-heavy services — AWS rates this size at 20 Gigabit
Avoid if
  • ●Starting a new deployment. g4ad.16xlarge is the newer generation at this size.
  • ●The working set needs more than 488 GiB of RAM
  • ●More than 7.63 GiB of RAM per vCPU is needed — that ratio is fixed for the whole family
  • ●The work is single-threaded. Most of these 64 vCPU would sit idle.
  • ●Local NVMe or instance-store disk is a hard requirement
  • ●Anything the CPU can do fast enough. GPU hours are the most expensive on the price list.
$4,479/mo on AWS on-demand against $139/mo on Rackspace Spot for the equivalent — market-priced, not list-priced.
  • Spot sells spare capacity at market prices. It is the cheapest compute on the internet.
  • Prices are set by supply and demand. You pay what the market decides.
  • Compute is delivered as fully managed Kubernetes. All infrastructure is taken care of.
  • Egress costs $0.01/GB after the free allowance. AWS charges $0.09/GB.
Build your Cloudspace
Frequently asked questions