p3dn.24xlarge by Amazon Web Services

p3dn.24xlarge is a GPU accelerated [Instance store volumes] [Network and EBS optimized] Gen3 24xlarge server offered by Amazon Web Services with 96 vCPUs, 768 GiB of memory and 1.8 TB of storage. The pricing starts at 3.5277 USD per hour.
96 vCPU
768 GiB Memory
1.8 TB Storage
8 GPU
Spare SCore
80,919
(All-cores)
1,379
(Single-core)

Specifications

Server Metadata

Vendor ID
aws
Name
p3dn.24xlarge
Description
GPU accelerated [Instance store volumes] [Network and EBS optimized] Gen3 24xlarge
Family
p3dn
Hw Virt
Status
active
Observed At
2026-10-23T21:38:48.905113

Availability

REGION / IDSPOTONDEMAND
Northern Virgina (US) / us-east-1
4.3527 USD/h 31.212 USD/h
Oregon (US) / us-west-2
8.4446 USD/h 31.212 USD/h
Dublin (IE) / eu-west-1
8.2842 USD/h 33.711 USD/h
Tokyo (JP) / ap-northeast-1
4.2783 USD/h 42.783 USD/h

Processor

vCPUs
96
Hypervisor
nitro
CPU Allocation
Dedicated
CPU Cores
48
CPU Speed
2.5 GHz
CPU Architecture
x86_64
CPU Manufacturer
Intel
CPU Family
Xeon
CPU Model
8175M
CPU L1D Cache
32 KiB
CPU L1D Cache Total
2 MiB
CPU L1I Cache
32 KiB
CPU L1I Cache Total
2 MiB
CPU L2 Cache
1 MiB
CPU L2 Cache Total
48 MiB
CPU L3 Cache
33 MiB
CPU L3 Cache Total
66 MiB
CPU Flags
fpu, vme, de, pse, tsc, msr, pae, mce, cx8, apic, sep, mtrr, pge, mca, cmov, pat, pse36, clflush, mmx, fxsr, sse, sse2, ss, ht, syscall, nx, pdpe1gb, rdtscp, lm, constant_tsc, arch_perfmon, rep_good, nopl, xtopology, nonstop_tsc, cpuid, aperfmperf, tsc_known_freq, pni, pclmulqdq, monitor, ssse3, fma, cx16, pcid, sse4_1, sse4_2, x2apic, movbe, popcnt, tsc_deadline_timer, aes, xsave, avx, f16c, rdrand, hypervisor, lahf_lm, abm, 3dnowprefetch, pti, fsgsbase, tsc_adjust, bmi1, avx2, smep, bmi2, erms, invpcid, mpx, avx512f, avx512dq, rdseed, adx, smap, clflushopt, clwb, avx512cd, avx512bw, avx512vl, xsaveopt, xsavec, xgetbv1, xsaves, ida, arat, pku, ospke
Ecpus
58.7
Scalability
61.15

System Resources and Accelerators

MEMORY
Memory Amount
768 GiB
Memory Amount Actual
768 GiB
Memory Generation
DDR4
Memory Speed
2666 Mhz
ACCELERATORS
Accelerator Count
8
Accelerator Memory Min
32 GiB
Accelerator Memory Total
256 GiB
Accelerator Type
GPU
Accelerator Manufacturer
NVIDIA
Accelerator Family
Volta
Accelerator Model
V100
Accelerators
  • NVIDIA Volta Tesla V100-SXM2-32GB (Memory amount: 32768, BIOS version: 88.00.8F.00.0F, Clock rate: 1530)
  • NVIDIA Volta Tesla V100-SXM2-32GB (Memory amount: 32768, BIOS version: 88.00.8F.00.0F, Clock rate: 1530)
  • NVIDIA Volta Tesla V100-SXM2-32GB (Memory amount: 32768, BIOS version: 88.00.8F.00.0F, Clock rate: 1530)
  • NVIDIA Volta Tesla V100-SXM2-32GB (Memory amount: 32768, BIOS version: 88.00.8F.00.0F, Clock rate: 1530)
  • NVIDIA Volta Tesla V100-SXM2-32GB (Memory amount: 32768, BIOS version: 88.00.8F.00.0F, Clock rate: 1530)
  • NVIDIA Volta Tesla V100-SXM2-32GB (Memory amount: 32768, BIOS version: 88.00.8F.00.0F, Clock rate: 1530)
  • NVIDIA Volta Tesla V100-SXM2-32GB (Memory amount: 32768, BIOS version: 88.00.8F.00.0F, Clock rate: 1530)
  • NVIDIA Volta Tesla V100-SXM2-32GB (Memory amount: 32768, BIOS version: 88.00.8F.00.0F, Clock rate: 1530)
STORAGE
Storage Size
1800 GB
Storage Type
nvme ssd
Storages
  • 900 GB nvme ssd
  • 900 GB nvme ssd
NETWORK
Network Speed Baseline
100 Gbps
Network Speed Max
100 Gbps
Network Storage Speed Baseline
19 Gbps
Network Storage Speed Max
19 Gbps
Inbound Traffic
0 GB/month
Outbound Traffic
0 GB/month
IPv4
0

CPU and System Topology

Server Description

A high-density GPU instance featuring eight accelerators, high-speed NVMe storage, and top-tier large language model inference performance.

GPU AcceleratedCompute OptimizedStorage & Database

Amazon Web Services p3dn.24xlarge is a GPU-accelerated instance designed for high-performance parallel computing. Powered by an Intel Xeon 8175M processor, it offers 96 vCPUs, 768.0 GB of DDR4 RAM, and eight NVIDIA Volta V100 GPUs with 256 GB of total VRAM. Storage is supported by 1800 GB of local NVMe SSD, while networking delivers a baseline bandwidth of 100 Gbps. Benchmarks reveal weak single-core performance but exceptional multi-core and memory bandwidth capabilities. Notably, the instance achieves top-tier results in LLM inference prompt processing and text generation for medium and large models. Given its massive GPU and memory density, it is highly suited for deep learning, large-scale LLM deployment, and high-throughput database operations.

Economics

Average Price per Region

Prices per Zone

Lowest Prices

Performance

Workload Profiles

2.9Score
Precomputed compound score for Cache Intensive workloads. A weighted average (geometric mean) of benchmark scores compared to their medians: score = ∏ (x_i / m_i)^(w_i / Σw). The score of 1.0 represents a synthetic baseline server with the median performance of each component benchmark; 0.5 means roughly half the performance; and 2.0 means twice the performance of that reference profile. Component weights: 50% Redis RPS (pipeline=1, SET), 20% Redis RPS (pipeline=16, SET), 10% PassMark Memory Mark (composite), 10% Memory bandwidth (read, 16 MB ~ L3), 10% PassMark single-thread CPU. Rationale for component selection: In-memory key-value store workload, mixing direct Redis performance metrics with memory speed and latency benchmarks, and single-core CPU performance profiles.
Component Score Weight Impact
Raw Reference Normalized Target Actual
Redis RPS (pipeline=1, SET) 8,097,875 ops/sec 1,919,289 ops/sec 4.3 50.00% 50.00% +110.00%
Redis RPS (pipeline=16, SET) 45,826,060 ops/sec 11,728,058 ops/sec 3.8 20.00% 20.00% +31.00%
PassMark Memory Mark (composite) 2,694 2,416 1.1 10.00% 10.00% +0.96%
Memory bandwidth (read, 16 MB ~ L3) 225,301 MB/sec 108,466 MB/sec 2.1 10.00% 10.00% +7.70%
PassMark single-thread CPU 1,959 Mops/s 2,365 Mops/s 0.83 10.00% 10.00% -1.80%

GPU Bandwidth and Latency

Node (all GPUs)
Node baseline directional bandwidth
39.6 GB/s
Node GPU-to-host bandwidth
49.9 GB/s
Node host-to-GPU bandwidth
49.6 GB/s
Peer-to-peer
GPU-to-GPU P2P duplex bandwidth
36.3 GB/s
GPU-to-GPU P2P gather bandwidth
92.9 GB/s
GPU-to-GPU P2P latency
1,185 ns
GPU-to-GPU P2P bandwidth
36.3 GB/s
Single GPU slot
Per-GPU GPU-to-host bandwidth
12.5 GB/s
Per-GPU host-to-GPU bandwidth
11.2 GB/s
Per-GPU host-device latency
1,187 ns
Benchmark Environment
Driver Version
 535.183.01
CUDA Runtime Version
 12.6
nvbandwidth Version
 v0.5
GPU Count
 8
P2P Supported
Yes
Efficiency
SM/CE copy efficiency
1.00

Memory Bandwidth

Compression

OpenSSL

Geekbench Single-Core

Score: 1,156

Geekbench Multi-Core

Score: 11,798

Passmark CPU Scores

BENCHMARKSCORE
Mark
 55,736
Compression
 1,182,319
Encryption
 37,026
Extended Instructions
 74,975
Floating Point Maths
 145,201
Integer Maths
 271,254
Physics
 5,013
Prime Numbers
 340
Single Threaded
 1,959
String Sorting
 136,905

Passmark Memory Scores

BENCHMARKSCORE
Memory Mark
 2,694
Database Operations
 21,503
Memory Latency
 53
Memory Read Cached
 23,522
Memory Read Uncached
 10,587
Memory Write
 10,168

Stress-ng Raw Scores

Stress-ng Relative Multicore Performance

LLM Inference Speed for Prompt Processing

LLM Inference Speed for Text Generation

Static Web Server

Redis

Alternatives

Servers of the Same Family

INSTANCEvCPUsMEMORYAccelerators

Similar Servers

INSTANCEVENDORvCPUsMEMORYAccelerators

p3dn.24xlarge FAQs