AWS EC2 Instances›GPU Instance›g4dn›g4dn.16xlarge

Compared with Rackspace Spot

AWSRackspace Spot
Machineg4dn.16xlargegpu.vs2.xlargeplusplus
vCPU6424
Memory256 GiB64 GiB
Local disk1 × 900 GB NVMe SSDNone
On-demand / hr$5.222$7.250
Spot / hr$0.668$0.151
Prices in US West, San Jose

Compute

vCPUs64
Memory256 GiB
Mem / vCPU4.00 GiB
ProcessorXeon Family
Clock speed2.5 GHz
Architecturex86_64
GPU1
GPU Wattage0 W
GPU ArchitectureTuring
Video Memory16 GiB
FPGA0
ffmpeg FPS105
CoreMark / sec914,808

Storage

Type1 × 900 GB NVMe SSD
Bandwidth9,500 Mbps
Throughput1187.5 MB/s
IOPS40,000
Base Bandwidth9,500 Mbps
Base Throughput1187.5 MB/s
Base IOPS40,000
Devices1
NVMe DriveYes
SSDYes
Swap PartitionNo
Local Read IOPS250,000
Local Write IOPS200,000

Networking

Performance50 Gigabit
Max ENIs4
IPv6Yes
GPUNVIDIA T4 Tensor Core
Placement GroupSupported
TrunkingYes
Branch Interface118
IPs / ENI15
Baseline Bandwidth50 Gbps

NUMA Architecture

Uses NUMAYes
Node Count2
Max Distance21
Cores / Node16
Threads / Node32
Memory / Node127,577 MB
L3 Cache / Node35 MB
L3 Cache SharedYes

AWS Metadata

Instance typeg4dn.16xlarge
Full nameG4DN 16xlarge Extra Large
FamilyGPU instance
Variantg4dn
GenerationCurrent
Bare MetalNo
EMR supportYes
DedicatedSupported
Introduced2019-09-20
Family sizes

g4dn.16xlarge

g4dn.16xlarge is AWS's GPU EC2 size at 64 vCPU and 256 GiB. 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 1 × NVIDIA T4 Tensor Core that is a fair swap for inference and media work, because the A30 carries more memory and more throughput per card.

In US West (N. California), Rackspace Spot's Spot market price is about 77% below AWS Spot.

Server comparison caveats
Comparable GPU

The AWS g4dn.16xlarge pairs 1 Turing GPU (16 GiB VRAM) with 64 vCPU. Rackspace Spot's GPU server uses the NVIDIA A30 (Ampere, 24 GB VRAM) with 24 vCPU and 64 GiB of memory. That is newer hardware with more GPU memory, but 24 vCPU against 64. For inference, batch transcoding and single-card fine-tuning the card is the upgrade; anything CPU-bound running alongside it gets less to work with. The catch is location rather than hardware: US West, San Jose is the only Rackspace Spot region that offers it, so latency to your users and egress to your data are the things to check.

No local disk

The AWS g4dn.16xlarge comes with 1 × 900 GB NVMe SSD attached to the machine. Rackspace Spot servers use network storage only, so that disk doesn't come with it. If you were using it as fast scratch space, you'd move to a persistent volume attached to your Rackspace Spot server, or object storage, and both are slower to read.

Pricing by region

1 hr

Regions are Rackspace Spot locations.

RegionAWS on-demandAWS Spot1-yr Reserved3-yr ReservedRackspace on-demandRackspace SpotOn-demand price savingsSpot price savings
US West, San Jose$5.222$0.668$3.674$2.538$7.250$0.151AWS cheaper
$0.517 (77%)

Prices updated 17 Nov 2026.

When to use it

Good for
  • ●Video transcoding and image processing
  • ●Machine learning experimentation and fine-tuning
  • ●Remote graphics workstations and rendering
  • ●Workloads that fit inside 256 GiB of RAM
  • ●Jobs that want 1 × 900 GB of local NVMe scratch space
  • ●Network-heavy services (AWS rates this size at 50 Gigabit)
Avoid if
  • ●Starting a new deployment. g5.16xlarge is the newer generation at this size.
  • ●The working set needs more than 256 GiB of RAM
  • ●Anything the CPU can do fast enough. GPU hours are the most expensive on the price list.
  • ●More than 4.00 GiB of RAM per vCPU is needed, since that ratio is fixed for the whole family
  • ●The work is single-threaded. Most of these 64 vCPU would sit idle.
Rackspace Spot sells spare capacity at market prices. You pay what the market decides, not a list price.
  • Rackspace 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

How much does g4dn.16xlarge cost?

The AWS g4dn.16xlarge costs $5.22/hr on-demand in US West (N. California), about $3,812/mo running 24/7, or $0.668/hr at the AWS Spot price. On Rackspace Spot, gpu.vs2.xlargeplusplus is $7.25/hr on-demand, or $0.151/hr at the Spot price.

How much is g4dn.16xlarge per month?

At 730 hours a month, g4dn.16xlarge in US West (N. California) is $3,812/mo on AWS on-demand, or $487/mo at the AWS Spot price. The same capacity on Rackspace Spot (gpu.vs2.xlargeplusplus) is $5,293/mo on-demand, or $110/mo at the Spot price. Rackspace Spot prices move with demand, so the Spot figure is a typical rate rather than a fixed one.

Can Rackspace Spot run g4dn.16xlarge GPU workloads?

Rackspace Spot's entire GPU catalog is gpu.vs2.xlargeplusplus: one NVIDIA A30, 24 vCPU and 64 GiB, available in San Jose only. Against g4dn.16xlarge's 1 × NVIDIA T4 Tensor Core the A30 is comparable or better for inference and transcoding, which is where most of these hours actually go. The pricing table has one row for that reason.

Does the Rackspace Spot equivalent include local NVMe storage?

No. g4dn.16xlarge ships 1 × 900 GB of local NVMe SSD, rated 250,000 read IOPS. Rackspace Spot servers are network-backed, so a workload that leans on local disk throughput needs block storage or a re-architecture. Everything else about the match still holds.

Is g4dn.16xlarge still worth using?

For workloads already running on it, yes: it works and it is priced accordingly. For anything new, g5.16xlarge is the current answer and measures roughly 36% more CoreMark throughput at the same vCPU count. 3 newer generations of the g family have shipped since (g5, g6 and g7), which is the signal that matters for infrastructure you expect to keep running.

What is the difference between g4dn and g5?

g5 is the newer generation. It runs on AMD EPYC 7R32 against g4dn's Intel Xeon Family. On the CoreMark figures in this dataset, g5.16xlarge scores 1,246,915 per second against g4dn.16xlarge's 914,808. On-demand in US East (N. Virginia) the newer size is 6% cheaper. Neither generation changes the Rackspace Spot side: both map to the same Rackspace Spot servers.