View a markdown version of this page

Specifications for Amazon EC2 accelerated computing instances - Amazon EC2

Specifications for Amazon EC2 accelerated computing instances

Accelerated computing instances use hardware accelerators, or co-processors, to perform functions, such as floating point number calculations, graphics processing, or data pattern matching, more efficiently than is possible in software running on CPUs.

For information on previous generation instance types of this category, such as G3 instances, see Specifications for Amazon EC2 previous generation instances.

Pricing

For pricing information, see Amazon EC2 On-Demand Pricing.

Instance families and instance types

Instance family Available instance types
DL1 dl1.24xlarge
DL2q dl2q.24xlarge
F2 f2.6xlarge | f2.12xlarge | f2.48xlarge
G4ad g4ad.xlarge | g4ad.2xlarge | g4ad.4xlarge | g4ad.8xlarge | g4ad.16xlarge
G4dn g4dn.xlarge | g4dn.2xlarge | g4dn.4xlarge | g4dn.8xlarge | g4dn.12xlarge | g4dn.16xlarge | g4dn.metal
G5 g5.xlarge | g5.2xlarge | g5.4xlarge | g5.8xlarge | g5.12xlarge | g5.16xlarge | g5.24xlarge | g5.48xlarge
G5g g5g.xlarge | g5g.2xlarge | g5g.4xlarge | g5g.8xlarge | g5g.16xlarge | g5g.metal
G6 g6.xlarge | g6.2xlarge | g6.4xlarge | g6.8xlarge | g6.12xlarge | g6.16xlarge | g6.24xlarge | g6.48xlarge
G6e g6e.xlarge | g6e.2xlarge | g6e.4xlarge | g6e.8xlarge | g6e.12xlarge | g6e.16xlarge | g6e.24xlarge | g6e.48xlarge
G6f g6f.large | g6f.xlarge | g6f.2xlarge | g6f.4xlarge
Gr6 gr6.4xlarge | gr6.8xlarge
Gr6f gr6f.4xlarge
G7 g7.2xlarge | g7.4xlarge | g7.8xlarge | g7.12xlarge | g7.24xlarge | g7.48xlarge
G7e g7e.2xlarge | g7e.4xlarge | g7e.8xlarge | g7e.12xlarge | g7e.24xlarge | g7e.48xlarge
Inf1 inf1.xlarge | inf1.2xlarge | inf1.6xlarge | inf1.24xlarge
Inf2 inf2.xlarge | inf2.8xlarge | inf2.24xlarge | inf2.48xlarge
P4d p4d.24xlarge
P4de p4de.24xlarge
P5 p5.4xlarge | p5.48xlarge
P5e p5e.48xlarge
P5en p5en.48xlarge
P6-B200 p6-b200.48xlarge
P6-B300 p6-b300.48xlarge
P6e-GB200 p6e-gb200.36xlarge
Trn1 trn1.2xlarge | trn1.32xlarge
Trn1n trn1n.32xlarge
Trn2 trn2.3xlarge | trn2.48xlarge
Trn2u trn2u.48xlarge
VT1 vt1.3xlarge | vt1.6xlarge | vt1.24xlarge

Instance family summary

Instance family Hypervisor Processor type (architecture) Metal instances available Dedicated Hosts support Spot support Hibernation support Supported operating systems
DL1 Nitro v3 Intel (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Linux
DL2q Nitro v3 Intel (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Linux
F2 Nitro v4 AMD (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Linux
G4ad Nitro v3 AMD (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Windows | Linux
G4dn Nitro v3 Intel (x86_64) ✓ Yes ✓ Yes ✓ Yes ✗ No Windows | Linux
G5 Nitro v3 AMD (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Windows | Linux
G5g Nitro v2 AWS Graviton (arm64) ✓ Yes ✓ Yes ✓ Yes ✗ No Linux
G6 Nitro v4 AMD (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Windows | Linux
G6e Nitro v4 AMD (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Windows | Linux
G6f Nitro v4 AMD (x86_64) ✗ No ✗ No ✓ Yes ✗ No Windows | Linux
Gr6 Nitro v4 AMD (x86_64) ✗ No ✗ No ✓ Yes ✗ No Windows | Linux
Gr6f Nitro v4 AMD (x86_64) ✗ No ✗ No ✓ Yes ✗ No Windows | Linux
G7 Nitro v6 Intel (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Windows | Linux
G7e Nitro v6 Intel (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Windows | Linux
Inf1 Nitro v3 Intel (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Linux
Inf2 Nitro v4 AMD (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Linux
P4d Nitro v3 Intel (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Linux
P4de Nitro v3 Intel (x86_64) ✗ No ✗ No ✓ Yes ✗ No Linux
P5 Nitro v4 AMD (x86_64) ✗ No ✗ No ✓ Yes ✗ No Windows (p5.4xlarge only) | Linux 1
P5e Nitro v4 AMD (x86_64) ✗ No ✗ No ✓ Yes ✗ No Linux
P5en Nitro v5 Intel (x86_64) ✗ No ✗ No ✓ Yes ✗ No Linux
P6-B200 Nitro v6 Intel (x86_64) ✗ No ✗ No ✓ Yes ✗ No Linux
P6-B300 Nitro v6 Intel (x86_64) ✗ No ✗ No ✓ Yes ✗ No Linux
P6e-GB200 Nitro v5 NVIDIA Grace (arm64) ✗ No ✗ No ✗ No ✗ No Linux
Trn1 Nitro v4 Intel (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Linux
Trn1n Nitro v4 Intel (x86_64) ✗ No ✗ No ✓ Yes ✗ No Linux
Trn2 Nitro v5 Intel (x86_64) ✗ No ✗ No ✓ Yes ✗ No Linux
Trn2u Nitro v5 Intel (x86_64) ✗ No ✗ No ✗ No ✗ No Linux
VT1 Nitro v3 Intel (x86_64) ✗ No ✓ Yes ✓ Yes ✗ No Linux
Note

1 p5.4xlarge supports both Windows and Linux operating systems. p5.48xlarge supports Linux operating systems only.

Performance specifications

Instance type Memory (GiB) Processor vCPUs CPU cores Threads per core Accelerators Accelerator memory
DL1
dl1.24xlarge 768.00 Intel Xeon Platinum 8275CL 96 48 2 8 x Habana Gaudi HL-205 GPU 256 GiB (8 x 32 GiB)
DL2q
dl2q.24xlarge 768.00 Intel Xeon Cascade Lake 96 48 2 8 x Qualcomm Qualcomm AI100 inference accelerator 125 GiB (8 x 15 GiB)
F2
f2.6xlarge 256.00 AMD EPYC 7R13 24 12 2 1 x Xilinx Virtex UltraScale+ (VU47P) FPGA 80 GiB (1 x 80 GiB)
f2.12xlarge 512.00 AMD EPYC 7R13 48 24 2 2 x Xilinx Virtex UltraScale+ (VU47P) FPGA 160 GiB (2 x 80 GiB)
f2.48xlarge 2048.00 AMD EPYC 7R13 192 96 2 8 x Xilinx Virtex UltraScale+ (VU47P) FPGA 640 GiB (8 x 80 GiB)
G4ad
g4ad.xlarge 16.00 2nd Gen AMD EPYC 7R32 4 2 2 1 x AMD Radeon Pro V520 GPU 8 GiB (1 x 8 GiB)
g4ad.2xlarge 32.00 2nd Gen AMD EPYC 7R32 8 4 2 1 x AMD Radeon Pro V520 GPU 8 GiB (1 x 8 GiB)
g4ad.4xlarge 64.00 2nd Gen AMD EPYC 7R32 16 8 2 1 x AMD Radeon Pro V520 GPU 8 GiB (1 x 8 GiB)
g4ad.8xlarge 128.00 2nd Gen AMD EPYC 7R32 32 16 2 2 x AMD Radeon Pro V520 GPU 16 GiB (2 x 8 GiB)
g4ad.16xlarge 256.00 2nd Gen AMD EPYC 7R32 64 32 2 4 x AMD Radeon Pro V520 GPU 32 GiB (4 x 8 GiB)
G4dn
g4dn.xlarge 16.00 Intel Xeon P-8259L 4 2 2 1 x NVIDIA T4 GPU 16 GiB (1 x 16 GiB)
g4dn.2xlarge 32.00 Intel Xeon P-8259L 8 4 2 1 x NVIDIA T4 GPU 16 GiB (1 x 16 GiB)
g4dn.4xlarge 64.00 Intel Xeon P-8259L 16 8 2 1 x NVIDIA T4 GPU 16 GiB (1 x 16 GiB)
g4dn.8xlarge 128.00 Intel Xeon P-8259L 32 16 2 1 x NVIDIA T4 GPU 16 GiB (1 x 16 GiB)
g4dn.12xlarge 192.00 Intel Xeon P-8259L 48 24 2 4 x NVIDIA T4 GPU 64 GiB (4 x 16 GiB)
g4dn.16xlarge 256.00 Intel Xeon P-8259L 64 32 2 1 x NVIDIA T4 GPU 16 GiB (1 x 16 GiB)
g4dn.metal 384.00 Intel Xeon P-8259L 96 48 2 8 x NVIDIA T4 GPU 128 GiB (8 x 16 GiB)
G5
g5.xlarge 16.00 2nd Gen AMD EPYC 7R32 4 2 2 1 x NVIDIA A10G GPU 22 GiB (1 x 22 GiB)
g5.2xlarge 32.00 2nd Gen AMD EPYC 7R32 8 4 2 1 x NVIDIA A10G GPU 22 GiB (1 x 22 GiB)
g5.4xlarge 64.00 2nd Gen AMD EPYC 7R32 16 8 2 1 x NVIDIA A10G GPU 22 GiB (1 x 22 GiB)
g5.8xlarge 128.00 2nd Gen AMD EPYC 7R32 32 16 2 1 x NVIDIA A10G GPU 22 GiB (1 x 22 GiB)
g5.12xlarge 192.00 2nd Gen AMD EPYC 7R32 48 24 2 4 x NVIDIA A10G GPU 89 GiB (4 x 22 GiB)
g5.16xlarge 256.00 2nd Gen AMD EPYC 7R32 64 32 2 1 x NVIDIA A10G GPU 22 GiB (1 x 22 GiB)
g5.24xlarge 384.00 2nd Gen AMD EPYC 7R32 96 48 2 4 x NVIDIA A10G GPU 89 GiB (4 x 22 GiB)
g5.48xlarge 768.00 2nd Gen AMD EPYC 7R32 192 96 2 8 x NVIDIA A10G GPU 178 GiB (8 x 22 GiB)
G5g
g5g.xlarge 8.00 AWS Graviton2 Processor 4 4 1 1 x NVIDIA T4g GPU 16 GiB (1 x 16 GiB)
g5g.2xlarge 16.00 AWS Graviton2 Processor 8 8 1 1 x NVIDIA T4g GPU 16 GiB (1 x 16 GiB)
g5g.4xlarge 32.00 AWS Graviton2 Processor 16 16 1 1 x NVIDIA T4g GPU 16 GiB (1 x 16 GiB)
g5g.8xlarge 64.00 AWS Graviton2 Processor 32 32 1 1 x NVIDIA T4g GPU 16 GiB (1 x 16 GiB)
g5g.16xlarge 128.00 AWS Graviton2 Processor 64 64 1 2 x NVIDIA T4g GPU 32 GiB (2 x 16 GiB)
g5g.metal 128.00 AWS Graviton2 Processor 64 64 1 2 x NVIDIA T4g GPU 32 GiB (2 x 16 GiB)
G6
g6.xlarge 16.00 AMD EPYC 7R13 4 2 2 1 x NVIDIA L4 GPU 22 GiB (1 x 22 GiB)
g6.2xlarge 32.00 AMD EPYC 7R13 8 4 2 1 x NVIDIA L4 GPU 22 GiB (1 x 22 GiB)
g6.4xlarge 64.00 AMD EPYC 7R13 16 8 2 1 x NVIDIA L4 GPU 22 GiB (1 x 22 GiB)
g6.8xlarge 128.00 AMD EPYC 7R13 32 16 2 1 x NVIDIA L4 GPU 22 GiB (1 x 22 GiB)
g6.12xlarge 192.00 AMD EPYC 7R13 48 24 2 4 x NVIDIA L4 GPU 89 GiB (4 x 22 GiB)
g6.16xlarge 256.00 AMD EPYC 7R13 64 32 2 1 x NVIDIA L4 GPU 22 GiB (1 x 22 GiB)
g6.24xlarge 384.00 AMD EPYC 7R13 96 48 2 4 x NVIDIA L4 GPU 89 GiB (4 x 22 GiB)
g6.48xlarge 768.00 AMD EPYC 7R13 192 96 2 8 x NVIDIA L4 GPU 178 GiB (8 x 22 GiB)
G6e
g6e.xlarge 32.00 AMD EPYC 7R13 4 2 2 1 x NVIDIA L40S GPU 44 GiB (1 x 44 GiB)
g6e.2xlarge 64.00 AMD EPYC 7R13 8 4 2 1 x NVIDIA L40S GPU 44 GiB (1 x 44 GiB)
g6e.4xlarge 128.00 AMD EPYC 7R13 16 8 2 1 x NVIDIA L40S GPU 44 GiB (1 x 44 GiB)
g6e.8xlarge 256.00 AMD EPYC 7R13 32 16 2 1 x NVIDIA L40S GPU 44 GiB (1 x 44 GiB)
g6e.12xlarge 384.00 AMD EPYC 7R13 48 24 2 4 x NVIDIA L40S GPU 178 GiB (4 x 44 GiB)
g6e.16xlarge 512.00 AMD EPYC 7R13 64 32 2 1 x NVIDIA L40S GPU 44 GiB (1 x 44 GiB)
g6e.24xlarge 768.00 AMD EPYC 7R13 96 48 2 4 x NVIDIA L40S GPU 178 GiB (4 x 44 GiB)
g6e.48xlarge 1536.00 AMD EPYC 7R13 192 96 2 8 x NVIDIA L40S GPU 357 GiB (8 x 44 GiB)
G6f
g6f.large 8.00 AMD EPYC 7R13 2 1 2 0.125 x NVIDIA L4 GPU 2.79 GiB
g6f.xlarge 16.00 AMD EPYC 7R13 4 2 2 0.125 x NVIDIA L4 GPU 2.79 GiB
g6f.2xlarge 32.00 AMD EPYC 7R13 8 4 2 0.25 x NVIDIA L4 GPU 5.59 GiB
g6f.4xlarge 64.00 AMD EPYC 7R13 16 8 2 0.5 x NVIDIA L4 GPU 11.18 GiB
Gr6
gr6.4xlarge 128.00 AMD EPYC 7R13 16 8 2 1 x NVIDIA L4 GPU 22 GiB (1 x 22 GiB)
gr6.8xlarge 256.00 AMD EPYC 7R13 32 16 2 1 x NVIDIA L4 GPU 22 GiB (1 x 22 GiB)
Gr6f
gr6f.4xlarge 128.00 AMD EPYC 7R13 16 8 2 0.5 x NVIDIA L4 GPU 11.18 GiB
G7
g7.2xlarge 32.00 Intel Xeon Granite Rapids 8 8 1 1 x NVIDIA RTX PRO 4500 GPU 32 GiB (1 x 32 GiB)
g7.4xlarge 64.00 Intel Xeon Granite Rapids 16 16 1 1 x NVIDIA RTX PRO 4500 GPU 32 GiB (1 x 32 GiB)
g7.8xlarge 128.00 Intel Xeon Granite Rapids 32 32 1 1 x NVIDIA RTX PRO 4500 GPU 32 GiB (1 x 32 GiB)
g7.12xlarge 192.00 Intel Xeon Granite Rapids 48 48 1 2 x NVIDIA RTX PRO 4500 GPU 64 GiB (2 x 32 GiB)
g7.24xlarge 384.00 Intel Xeon Granite Rapids 96 96 1 4 x NVIDIA RTX PRO 4500 GPU 128 GiB (4 x 32 GiB)
g7.48xlarge 768.00 Intel Xeon Granite Rapids 192 192 1 8 x NVIDIA RTX PRO 4500 GPU 256 GiB (8 x 32 GiB)
G7e
g7e.2xlarge 64.00 Intel Xeon Emerald Rapids 8 4 2 1 x NVIDIA RTX PRO Server 6000 GPU 96 GiB (1 x 96 GiB)
g7e.4xlarge 128.00 Intel Xeon Emerald Rapids 16 8 2 1 x NVIDIA RTX PRO Server 6000 GPU 96 GiB (1 x 96 GiB)
g7e.8xlarge 256.00 Intel Xeon Emerald Rapids 32 16 2 1 x NVIDIA RTX PRO Server 6000 GPU 96 GiB (1 x 96 GiB)
g7e.12xlarge 512.00 Intel Xeon Emerald Rapids 48 24 2 2 x NVIDIA RTX PRO Server 6000 GPU 192 GiB (2 x 96 GiB)
g7e.24xlarge 1024.00 Intel Xeon Emerald Rapids 96 48 2 4 x NVIDIA RTX PRO Server 6000 GPU 384 GiB (4 x 96 GiB)
g7e.48xlarge 2048.00 Intel Xeon Emerald Rapids 192 96 2 8 x NVIDIA RTX PRO Server 6000 GPU 768 GiB (8 x 96 GiB)
Inf1
inf1.xlarge 8.00 Intel Xeon P-8259L 4 2 2 1 x AWS Inferentia inference accelerator 8 GiB (1 x 8 GiB)
inf1.2xlarge 16.00 Intel Xeon P-8259L 8 4 2 1 x AWS Inferentia inference accelerator 8 GiB (1 x 8 GiB)
inf1.6xlarge 48.00 Intel Xeon P-8259L 24 12 2 4 x AWS Inferentia inference accelerator 32 GiB (4 x 8 GiB)
inf1.24xlarge 192.00 Intel Xeon P-8259L 96 48 2 16 x AWS Inferentia inference accelerator 128 GiB (16 x 8 GiB)
Inf2
inf2.xlarge 16.00 AMD EPYC 7R13 4 2 2 1 x AWS Inferentia2 inference accelerator 32 GiB (1 x 32 GiB)
inf2.8xlarge 128.00 AMD EPYC 7R13 32 16 2 1 x AWS Inferentia2 inference accelerator 32 GiB (1 x 32 GiB)
inf2.24xlarge 384.00 AMD EPYC 7R13 96 48 2 6 x AWS Inferentia2 inference accelerator 192 GiB (6 x 32 GiB)
inf2.48xlarge 768.00 AMD EPYC 7R13 192 96 2 12 x AWS Inferentia2 inference accelerator 384 GiB (12 x 32 GiB)
P4d
p4d.24xlarge 1152.00 Intel Xeon Platinum 8275CL 96 48 2 8 x NVIDIA A100 GPU 320 GiB (8 x 40 GiB)
P4de
p4de.24xlarge 1152.00 Intel Xeon Platinum 8275CL 96 48 2 8 x NVIDIA A100 GPU 640 GiB (8 x 80 GiB)
P5
p5.4xlarge 256.00 AMD EPYC 7R13 16 8 2 1 x NVIDIA H100 GPU 80 GiB (1 x 80 GiB)
p5.48xlarge 2048.00 AMD EPYC 7R13 192 96 2 8 x NVIDIA H100 GPU 640 GiB (8 x 80 GiB)
P5e
p5e.48xlarge 2048.00 AMD EPYC 7R13 192 96 2 8 x NVIDIA H200 GPU 1128 GiB (8 x 141 GiB)
P5en
p5en.48xlarge 2048.00 Intel Xeon Sapphire Rapids 192 96 2 8 x NVIDIA H200 GPU 1128 GiB (8 x 141 GiB)
P6-B200
p6-b200.48xlarge 2048.00 Intel Xeon Emerald Rapids 192 96 2 8 x NVIDIA B200 GPU 1432 GiB (8 x 179 GiB)
P6-B300
p6-b300.48xlarge 4096.00 Intel Xeon Emerald Rapids 192 96 2 8 x NVIDIA B300 GPU 2148 GiB (8 x 268 GiB)
P6e-GB200
p6e-gb200.36xlarge 960.00 Nvidia Grace CPU 144 144 1 4 x NVIDIA B200 GPU 740 GiB (4 x 185 GiB)
Trn1
trn1.2xlarge 32.00 Intel Xeon Ice Lake 8375C 8 4 2 1 x AWS Trainium accelerators 32 GiB (1 x 32 GiB)
trn1.32xlarge 512.00 Intel Xeon Ice Lake 8375C 128 64 2 16 x AWS Trainium accelerators 512 GiB (16 x 32 GiB)
Trn1n
trn1n.32xlarge 512.00 Intel Xeon Ice Lake 128 64 2 16 x AWS Trainium accelerators 512 GiB (16 x 32 GiB)
Trn2
trn2.3xlarge 128.00 Intel Xeon Sapphire Rapids 12 6 2 1 x AWS Trainium2 accelerators 512 GiB (1 x 512 GiB)
trn2.48xlarge 2048.00 Intel Xeon Sapphire Rapids 192 96 2 16 x AWS Trainium2 accelerators 8192 GiB (16 x 512 GiB)
Trn2u
trn2u.48xlarge 2048.00 Intel Xeon Sapphire Rapids 192 96 2 ✗ No ✗ No
VT1
vt1.3xlarge 24.00 Intel Cascade Lake P-8259CL 12 6 2