m6g / m7g / m8g / m9g 16xlarge (64 vCPU & 256 GiB RAM)

Four Graviton generations compared at the 16xlarge size (64 vCPU, 256 GiB RAM).

	M6G.16XLARGE 💰 🥇	M7G.16XLARGE	M8G.16XLARGE	M9G.16XLARGE 💪
Vendor
$Core	599,264/USD Best (ondemand):59,919/USD Best (spot):599,264/USD	310,134/USD Best (ondemand):52,448/USD Best (spot):310,134/USD	287,264/USD Best (ondemand):53,169/USD Best (spot):287,264/USD	173,845/USD Best (ondemand):40,867/USD Best (spot):173,845/USD
SCore All	97021	97878	109160	127950
SCore Single	1519	1533	1707	2006
Best spot price	0.1619 USD ap-south-2b	0.3156 USD ap-east-2a	0.38 USD af-south-1a	0.736 USD us-east-2a
Best ondemand price	1.6192 USD ap-south-1a	1.8662 USD ap-south-1a	2.0531 USD ap-south-1a	3.1309 USD us-east-1d
Metadata
Family	m6g	m7g	m8g	m9g
Hw Virt
Average Time To Start				16
Status	active	active	active	active
CPU
vCPUs	64	64	64	64
Hypervisor	nitro	nitro	nitro	nitro
CPU Allocation	Dedicated	Dedicated	Dedicated	Dedicated
CPU Cores	64	64	64	64
CPU Speed	2.5 GHz	2.6 GHz	2.8 GHz	2.8 GHz
CPU Architecture	arm64	arm64	arm64	arm64
CPU Manufacturer	AWS	AWS	AWS	AWS
CPU Family	ARMv8	ARMv8		ARMv9
CPU Model	Neoverse-N1	Neoverse-V1	Neoverse-V2	Neoverse-V3
CPU L1D Cache	64 KiB	64 KiB	64 KiB	64 KiB
CPU L1D Cache Total	4096 KiB	4096 KiB	4096 KiB	4096 KiB
CPU L1I Cache	64 KiB	64 KiB	64 KiB	64 KiB
CPU L1I Cache Total	4096 KiB	4096 KiB	4096 KiB	4096 KiB
CPU L2 Cache	1024 KiB	1024 KiB	2048 KiB	2048 KiB
CPU L2 Cache Total	65536 KiB	65536 KiB	131072 KiB	131072 KiB
CPU L3 Cache	32768 KiB	32768 KiB	36864 KiB	49152 KiB
CPU L3 Cache Total	32768 KiB	32768 KiB	36864 KiB	98304 KiB
Ecpus	63.9	63.9	64	63.8
Scalability	99.84	99.84	100	99.69
Memory
Memory Amount	256 GiB	256 GiB	256 GiB	256 GiB
Memory Amount Actual	257024 MiB	258048 MiB	258048 MiB	258048 MiB
GPU
GPU Count	0	0	0	0
GPU Memory Min	0 GiB	0 GiB	0 GiB	0 GiB
GPU Memory Total	0 GiB	0 GiB	0 GiB	0 GiB
Storage
Storage Size	0 GB	0 GB	0 GB	0 GB
Network
Network Speed Baseline	25 Gbps	30 Gbps	30 Gbps	34 Gbps
Network Speed Max	25 Gbps	30 Gbps	30 Gbps	34 Gbps
Network Storage Speed Baseline	19 Gbps	20 Gbps	20 Gbps	24 Gbps
Network Storage Speed Max	19 Gbps	20 Gbps	20 Gbps	24 Gbps
Inbound Traffic	0 GB/month	0 GB/month	0 GB/month	0 GB/month
Outbound Traffic	0 GB/month	0 GB/month	0 GB/month	0 GB/month
IPv4	0	0	0	0
Memory Bandwidth

Compression

OpenSSL

Geekbench

PassMark (CPU)

String Sorting	175.83K	246.17K	293.31K	339.65K
Single Threaded	1.11K	1.56K	1.94K	2.46K
Prime Numbers	251.48	340.52	691.82	1.13K
Physics	4.11K	8.64K	16.13K	19.07K
Mark	28.11K	34.60K	44.11K	60.84K
Integer Maths	252.61K	305.24K	333.77K	391.06K
Floating Point Maths	191.66K	303.58K	387.70K	488.78K
Extended Instructions	39.87K	52.97K	56.41K	86.07K
Encryption	8.74K	8.89K	12.05K	19.05K
Compression	304.72K	338.40K	425.57K	595.39K
PassMark (Memory)

Memory Write	12.55K	19.70K	21.29K	25.03K
Memory Read Uncached	9.19K	19.02K	20.23K	24.66K
Memory Read Cached	9.48K	19.68K	19.52K	22.14K
Memory Mark	1.85K	3.15K	3.32K	4.59K
Memory Latency	70.03	49.88	46.57	25.66
Database Operations	20.46K	31.33K	47.38K	58.75K
Stress-ng div16 Raw Scores per vCPU

Stress-ng Relative Multicore Performance per vCPU

LLM Inference Speed

Static Web Server

Redis

Further Benchmarks
Workload profile: Web Server
Workload profile: Multimodal LLM Inference
Workload profile: Relational Database
Workload profile: Data Analysis
Workload profile: Compute Heavy Applications
Workload profile: CI/CD Build
Workload profile: Cache Intensive
stress-ng CPU all
stress-ng div16 multi-core
stress-ng div16 single-core
BogoMips
View Server

m6g / m7g / m8g / m9g 16xlarge (64 vCPU & 256 GiB RAM)

Metadata

CPU

Memory

GPU

Storage

Network

Memory Bandwidth

Compression

OpenSSL

Geekbench

PassMark (CPU)

PassMark (Memory)

Stress-ng div16 Raw Scores per vCPU

Stress-ng Relative Multicore Performance per vCPU

LLM Inference Speed

Static Web Server

Redis

Further Benchmarks

Embed Chart