a3-edgegpu-8g by Google Cloud Platform

a3-edgegpu-8g is a Accelerator Optimized: 8 NVIDIA H100 GPU, 208 vCPUs, 1872GB RAM server offered by Google Cloud Platform with 208 vCPUs, 1872 GiB of memory and 6.43 TB of storage. The pricing starts at 10.7923 USD per hour.
208 vCPU
1872 GiB Memory
6.43 TB Storage
8 GPU
Spare SCore
167,331
(All-cores)
2,043
(Single-core)

Specifications #

Server Metadata #

Vendor ID
gcp
Server ID
1720367
Name
a3-edgegpu-8g
Description
Accelerator Optimized: 8 NVIDIA H100 GPU, 208 vCPUs, 1872GB RAM
Family
GPU Machines
Series
a3
Hw Virt
Average Time To Start
39
Status
active
Observed At
2026-11-17T04:14:20.739261

Availability #

Region / IDSpotOndemand
Council Bluffs (US) / us-central1
52.7005 USD/h
88.49 USD/h
The Dalles (US) / us-west1
52.7005 USD/h
88.49 USD/h
Ashburn (US) / us-east4
50.4984 USD/h
88.5558 USD/h
Mumbai (IN) / asia-south1
46.8537 USD/h
92.1348 USD/h
Toronto (CA) / northamerica-northeast2
24.6298 USD/h
97.4356 USD/h
Columbus (US) / us-east5
38.0027 USD/h
99.5832 USD/h
Las Vegas (US) / us-west4
46.7023 USD/h
99.649 USD/h
London (GB) / europe-west2
23.818 USD/h
100.9181 USD/h
Paris (FR) / europe-west9
35.1232 USD/h
102.6484 USD/h
Frankfurt (DE) / europe-west3
59.5458 USD/h
104.4314 USD/h
Eemshaven (NL) / europe-west4
67.4966 USD/h
112.5405 USD/h
Seoul (KR) / asia-northeast3
57.2789 USD/h
113.6492 USD/h
Jurong West (SG) / asia-southeast1
47.9831 USD/h
114.2783 USD/h
Melbourne (AU) / australia-southeast2
10.7923 USD/h
115.0699 USD/h
Tokyo (JP) / asia-northeast1
23.9411 USD/h
126.6434 USD/h

Processor #

vCPUs
208
CPU Allocation
Dedicated
CPU Cores
104
CPU Speed
2.7 GHz
CPU Architecture
x86_64
CPU Manufacturer
Intel
CPU Family
Xeon
CPU Model
8481C
CPU L1D Cache
48 KiB
CPU L1D Cache Total
5 MiB
CPU L1I Cache
32 KiB
CPU L1I Cache Total
3 MiB
CPU L2 Cache
2 MiB
CPU L2 Cache Total
208 MiB
CPU L3 Cache
105 MiB
CPU L3 Cache Total
210 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, rep_good, nopl, xtopology, nonstop_tsc, cpuid, tsc_known_freq, pni, pclmulqdq, ssse3, fma, cx16, pcid, sse4_1, sse4_2, x2apic, movbe, popcnt, aes, xsave, avx, f16c, rdrand, hypervisor, lahf_lm, abm, 3dnowprefetch, ssbd, ibrs, ibpb, stibp, ibrs_enhanced, fsgsbase, tsc_adjust, bmi1, avx2, smep, bmi2, erms, invpcid, rtm, avx512f, avx512dq, rdseed, adx, smap, avx512ifma, clflushopt, clwb, avx512cd, sha_ni, avx512bw, avx512vl, xsaveopt, xsavec, xgetbv1, xsaves, avx_vnni, avx512_bf16, arat, avx512vbmi, umip, avx512_vbmi2, gfni, vaes, vpclmulqdq, avx512_vnni, avx512_bitalg, avx512_vpopcntdq, la57, rdpid, cldemote, movdiri, movdir64b, fsrm, md_clear, serialize, tsxldtrk, amx_bf16, avx512_fp16, amx_tile, amx_int8, arch_capabilities
Ecpus
81.9
Scalability
39.38

System Resources and Accelerators #

Memory
Memory Amount
2 TB
Memory Amount Actual
2 TB
Accelerators
Accelerator Count
8
Accelerator Memory Min
80 GiB
Accelerator Memory Total
640 GiB
Accelerator Type
GPU
Accelerator Manufacturer
NVIDIA
Accelerator Family
Hopper
Accelerator Model
H100
Accelerators
  • NVIDIA Hopper NVIDIA H100 80GB HBM3 (Memory amount: 81559, Firmware version: 590.48.01, BIOS version: 96.00.CF.00.01, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H100 80GB HBM3 (Memory amount: 81559, Firmware version: 590.48.01, BIOS version: 96.00.CF.00.01, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H100 80GB HBM3 (Memory amount: 81559, Firmware version: 590.48.01, BIOS version: 96.00.CF.00.01, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H100 80GB HBM3 (Memory amount: 81559, Firmware version: 590.48.01, BIOS version: 96.00.CF.00.01, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H100 80GB HBM3 (Memory amount: 81559, Firmware version: 590.48.01, BIOS version: 96.00.CF.00.01, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H100 80GB HBM3 (Memory amount: 81559, Firmware version: 590.48.01, BIOS version: 96.00.CF.00.01, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H100 80GB HBM3 (Memory amount: 81559, Firmware version: 590.48.01, BIOS version: 96.00.CF.00.01, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H100 80GB HBM3 (Memory amount: 81559, Firmware version: 590.48.01, BIOS version: 96.00.CF.00.01, Clock rate: 1980)
Storage
Storage Size
6432 GB
Storage Type
nvme ssd
Storages
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
  • 402 GB nvme ssd
Compatible Storage IDs
30014, 30017, 30012, 30013, 30016, 30007, 30002
Network
Inbound Traffic
0 GB/month
Outbound Traffic
0 GB/month
IPv4
0

CPU and System Topology #

Server Description #

An accelerator-optimized platform combining eight high-performance GPUs, massive system memory, and high-speed local NVMe storage for intensive parallel computing.

Google Cloud Platform a3-edgegpu-8g is an accelerator-optimized server featuring eight NVIDIA Hopper H100 GPUs with 640 GB of total VRAM, 208 vCPUs on an Intel Xeon 8481C processor, and 1872.0 GB of system memory. Built on the x86_64 architecture, it includes 6432 GB of local NVMe SSD storage. Benchmark results place this server in the top 10% for multi-core CPU performance, memory bandwidth, and database operations, with Redis SET performance reaching 177.5 million operations per second. The processor supports advanced instruction sets including AVX-512 and Advanced Matrix Extensions (AMX) for accelerated matrix mathematics. This server is engineered for high-density parallel computing tasks, making it highly suitable for large-scale machine learning, large language model inference, real-time database operations, and complex scientific simulations.

Economics #

Average Price per Region #

Prices per Zone #

Lowest Prices #

Performance #

Workload Profiles #

7Score
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) 26,065,542 ops/sec 1,933,252 ops/sec 14 50.00% 50.00% +270.00%
Redis RPS (pipeline=16, SET) 123,452,084 ops/sec 11,814,424 ops/sec 10 20.00% 20.00% +58.00%
PassMark Memory Mark (composite) 2,352 2,408 1 10.00% 10.00% 0.00%
Memory bandwidth (read, 16 MB ~ L3) 461,425 MB/sec 108,674 MB/sec 4.2 10.00% 10.00% +15.00%
PassMark single-thread CPU 3,134 Mops/s 2,365 Mops/s 1.3 10.00% 10.00% +2.70%

GPU Bandwidth and Latency #

Node (all GPUs)
Node baseline directional bandwidth
97.0 GB/s
Node GPU-to-host bandwidth
114.1 GB/s
Node host-to-GPU bandwidth
110.7 GB/s
Peer-to-peer
GPU-to-GPU P2P duplex bandwidth
383.5 GB/s
GPU-to-GPU P2P gather bandwidth
398.8 GB/s
GPU-to-GPU P2P latency
1,254 ns
GPU-to-GPU P2P bandwidth
388.8 GB/s
Single GPU slot
Per-GPU GPU-to-host bandwidth
28.6 GB/s
Per-GPU host-to-GPU bandwidth
27.7 GB/s
Per-GPU host-device latency
1,215 ns
Benchmark Environment
Driver Version
535.183.01
CUDA Version (driver)
12.6
nvbandwidth Version
v0.5
GPU Count
8
P2P Supported
Yes
Efficiency
SM/CE copy efficiency
0.91

Memory Bandwidth#

Compression #

OpenSSL #

Geekbench Single-Core #

Score: 2,121

Geekbench Multi-Core #

Score: 19,185

Passmark CPU Scores #

BENCHMARKSCORE
Mark
125,152
Compression
2,693,757
Encryption
168,381
Extended Instructions
191,152
Floating Point Maths
569,539
Integer Maths
798,990
Physics
14,782
Prime Numbers
1,023
Single Threaded
3,134
String Sorting
344,477

Passmark Memory Scores #

BENCHMARKSCORE
Memory Mark
2,352
Database Operations
34,439
Memory Latency
71
Memory Read Cached
27,110
Memory Read Uncached
9,944
Memory Write
8,391

Stress-ng Stressors #

CPU integerSingleMultiScaling
atomic
217.13 27.77 0.13×
bitops
157,018 14,580,462 92.86×
branch
146,504 17,747,554 121.14×
funccall
781.99 72,420 92.61×
funcret
13,170 1,567,266 119.00×
hash
188,792 20,845,112 110.41×
intmath
68,630 6,383,652 93.02×
nop
286,424 18,234,020 63.66×
prime
497,063 40,848,373 82.18×
rotate
105,928 8,956,163 84.55×
str
958,216 93,013,725 97.07×
wcs
730,672 56,412,300 77.21×
CryptographySingleMultiScaling
crypt
3,587 251,126 70.01×
CompressionSingleMultiScaling
jpeg
882.94 75,048 85.00×
zlib
56.02 7,836 139.89×
HypervisorSingleMultiScaling
clock
22,245 196,693 8.84×
clone
423.46 12,924 30.52×
context
5,483 714,431 130.31×
cpu-sched
59.09 13,024 220.42×
eventfd
66,652 2,522,736 37.85×
fault
32,241 163,417 5.07×
fork
3,906 51,827 13.27×
getrandom
62,676 6,492,150 103.58×
hrtimers
436,708 17,730,568 40.60×
mprotect
592,957 13,992,465 23.60×
mtx
29,736,196 5,191,988,065 174.60×
munmap
3,121 62,968 20.17×
null
945,235 109,926,296 116.30×
pipe
693,968 83,471,114 120.28×
priv-instr
867,080 6,983,020 8.05×
resched
9,458,481 24,250,786 2.56×
switch
90,653 6,610,897 72.93×
timermix
235,111 27,500,073 116.97×
tlb-shootdown
7,223 152,084 21.05×
tsc
615,809 81,222,231 131.90×
usersyscall
748,687 6,986,022 9.33×
vdso
95,553,558 11,442,427,775 119.75×
vfork
23,730 109,262 4.60×
x86syscall
10,560,350 1,363,042,917 129.07×
yield
154,096,958 183,468,862 1.19×
zero
6,798,873 818,235,619 120.35×
CPU float/SIMDSingleMultiScaling
eigen
17,022 1,601,392 94.08×
expmath
3,737 397,812 106.46×
fma
1,460,457 139,883,039 95.78×
fp
326.39 36,510 111.86×
fractal
7.4 785.44 106.14×
hyperbolic
4,300 455,251 105.88×
logmath
5,524 656,175 118.78×
matrix
6,500 479,326 73.74×
matrix-3d
634.77 15,710 24.75×
monte-carlo
311.46 36,414 116.91×
mpfr
1,837 138,929 75.61×
powmath
2,049 230,841 112.66×
trig
2,807 252,116 89.81×
veccmp
27,374 2,025,355 73.99×
vecfp
2,402 211,153 87.92×
vecmath
4,163 425,099 102.11×
vecshuf
4,063 605,889 149.13×
vecwide
86,127 7,222,693 83.86×
AlgorithmsSingleMultiScaling
bitonicsort
9.4 769.98 81.91×
bsearch
302.75 27,958 92.35×
bubblesort
1 83.2 83.20×
fibsearch
325.47 30,520 93.77×
heapsort
10.6 978.17 92.28×
hsearch
8,713 737,192 84.60×
insertionsort
12.8 849.99 66.41×
malloc
2,728,947 125,314,122 45.92×
mcontend
53.46 25,702 480.76×
mergesort
32.6 3,045 93.42×
qsort
23.2 2,118 91.27×
skiplist
5,779 674,095 116.65×
tree
1.6 106.73 66.71×
MemorySingleMultiScaling
cache
212,338 8,621,235 40.60×
cachehammer
186,697 7,869,858 42.15×
cacheline
24.18 2,214 91.55×
l1cache
1,178 23,919 20.31×
memcpy
470.93 32,168 68.31×
memrate
0.12 7.89 65.75×
memthrash
452.88 1,013 2.24×
ptr-chase
26,392,441 1,432,722,895 54.29×
spinmem
253.19 8,603 33.98×
stream
6.49 40,161 6188.08×
vm
120,884 311,920 2.58×

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
a3-highgpu-1g 26234 GiB1
a3-highgpu-2g 52468 GiB2
a3-highgpu-4g 104936 GiB4
a3-highgpu-8g 2081872 GiB8
a3-megagpu-8g 2081872 GiB8
a3-edgegpu-8g-nolssd 2081872 GiB8
a3-ultragpu-8g 2242952 GiB8

Similar Servers #

INSTANCEVENDORvCPUsMEMORYAccelerators

a3-edgegpu-8g FAQs #