a2-ultragpu-1g by Google Cloud Platform

a2-ultragpu-1g is a Accelerator Optimized: 1 NVIDIA A100 80GB GPU, 12 vCPUs, 170GB RAM, 1 local SSD server offered by Google Cloud Platform with 12 vCPUs, 170 GiB of memory and 402 GB of storage. The pricing starts at 1.7932 USD per hour.
12 vCPU
170 GiB Memory
402 GB Storage
1 GPU
Spare SCore
9,485
(All-cores)
1,201
(Single-core)

Specifications

Server Metadata

Vendor ID
gcp
Server ID
1020012
Name
a2-ultragpu-1g
Description
Accelerator Optimized: 1 NVIDIA A100 80GB GPU, 12 vCPUs, 170GB RAM, 1 local SSD
Family
a2
Hw Virt
Average Time To Start
27
Status
active
Observed At
2026-10-26T14:46:46.830680

Availability

REGION / IDSPOTONDEMAND
Council Bluffs (US) / us-central1
3.0081 USD/h 5.0688 USD/h
Columbus (US) / us-east5
1.7932 USD/h 5.5649 USD/h
Eemshaven (NL) / europe-west4
3.0683 USD/h 5.5809 USD/h
Ashburn (US) / us-east4
2.529 USD/h 5.7077 USD/h
Jurong West (SG) / asia-southeast1
3.6163 USD/h 6.2477 USD/h

Processor

vCPUs
12
CPU Allocation
Dedicated
CPU Cores
6
CPU Speed
2.2 GHz
CPU Architecture
x86_64
CPU Manufacturer
Intel
CPU Family
Xeon
CPU L1D Cache
32 KiB
CPU L1D Cache Total
192 KiB
CPU L1I Cache
32 KiB
CPU L1I Cache Total
192 KiB
CPU L2 Cache
1 MiB
CPU L2 Cache Total
6 MiB
CPU L3 Cache
39 MiB
CPU L3 Cache Total
39 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, hle, avx2, smep, bmi2, erms, invpcid, rtm, mpx, avx512f, avx512dq, rdseed, adx, smap, clflushopt, clwb, avx512cd, avx512bw, avx512vl, xsaveopt, xsavec, xgetbv1, xsaves, arat, avx512_vnni, md_clear, arch_capabilities
Ecpus
7.9
Scalability
65.83

System Resources and Accelerators

MEMORY
Memory Amount
170 GiB
Memory Amount Actual
170 GiB
ACCELERATORS
Accelerator Count
1
Accelerator Memory Min
80 GiB
Accelerator Memory Total
80 GiB
Accelerator Type
GPU
Accelerator Manufacturer
NVIDIA
Accelerator Family
Ampere
Accelerator Model
A100
Accelerators
  • NVIDIA Ampere NVIDIA A100-SXM4-80GB (Memory amount: 81920, Firmware version: 590.48.01, BIOS version: 92.00.94.00.04, Clock rate: 1410)
STORAGE
Storage Size
402 GB
Storage Type
nvme ssd
Storages
  • 402 GB nvme ssd
NETWORK
Inbound Traffic
0 GB/month
Outbound Traffic
0 GB/month
IPv4
0

CPU and System Topology

Server Description

An accelerator-optimized platform pairing an NVIDIA A100 GPU with generous system memory for demanding machine learning and inference workloads.

GPU AcceleratedMemory Optimized

Google Cloud Platform a2-ultragpu-1g is an accelerator-optimized server designed for GPU-intensive workloads. It features an Intel Xeon x86_64 CPU with 12 vCPUs, 170.0 GB of system memory, a 402 GB local NVMe SSD, and a single NVIDIA Ampere A100 GPU with 80 GB of VRAM. Performance benchmarks demonstrate a significant contrast between CPU and GPU capabilities. The server achieves top-tier results in large language model inference, processing 499.41 tokens per second for prompt processing on a 70B model. However, its single-core CPU performance and single-threaded decompression tasks fall into the bottom 25% tier. Multi-core CPU operations and memory bandwidth perform in the middle 50% tier. This hardware profile is highly optimized for deep learning, LLM inference, and parallelized accelerator tasks, whereas general-purpose single-threaded processing represents a primary tradeoff.

Economics

Average Price per Region

Prices per Zone

Lowest Prices

Performance

Workload Profiles

0.55Score
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) 931,881 ops/sec 1,919,289 ops/sec 0.49 50.00% 50.00% -30.00%
Redis RPS (pipeline=16, SET) 5,898,989 ops/sec 11,728,058 ops/sec 0.49 20.00% 20.00% -13.00%
PassMark Memory Mark (composite) 2,073 2,416 0.88 10.00% 10.00% -1.30%
Memory bandwidth (read, 16 MB ~ L3) 69,855 MB/sec 108,466 MB/sec 0.64 10.00% 10.00% -4.40%
PassMark single-thread CPU 1,690 Mops/s 2,365 Mops/s 0.71 10.00% 10.00% -3.40%

GPU Bandwidth and Latency

Node (all GPUs)
Node baseline directional bandwidth
5.3 GB/s
Node GPU-to-host bandwidth
13.2 GB/s
Node host-to-GPU bandwidth
12.3 GB/s
Single GPU slot
Per-GPU GPU-to-host bandwidth
13.2 GB/s
Per-GPU host-to-GPU bandwidth
12.3 GB/s
Per-GPU host-device latency
1,258 ns
Benchmark Environment
Driver Version
 535.183.01
CUDA Runtime Version
 12.6
nvbandwidth Version
 v0.5
GPU Count
 1
P2P Supported
No
Efficiency
SM/CE copy efficiency
1.00

Memory Bandwidth

Compression

OpenSSL

Geekbench Single-Core

Score: 1,058

Geekbench Multi-Core

Score: 5,325

Passmark CPU Scores

BENCHMARKSCORE
Mark
 10,455
Compression
 141,277
Encryption
 4,350
Extended Instructions
 8,881
Floating Point Maths
 17,047
Integer Maths
 31,681
Physics
 1,131
Prime Numbers
 61
Single Threaded
 1,690
String Sorting
 16,624

Passmark Memory Scores

BENCHMARKSCORE
Memory Mark
 2,073
Database Operations
 3,679
Memory Latency
 62
Memory Read Cached
 20,813
Memory Read Uncached
 7,733
Memory Write
 7,754

Stress-ng Stressors

CPU integerSingleMultiScaling
atomic
 197.42 192.87 0.98×
bitops
 92,226 634,383 6.88×
branch
 131,264 1,190,796 9.07×
funccall
 463 2,851 6.16×
funcret
 9,626 82,298 8.55×
hash
 107,412 925,521 8.62×
intmath
 18,979 170,116 8.96×
nop
 162,180 843,881 5.20×
prime
 268,820 1,817,468 6.76×
rotate
 81,726 681,298 8.34×
str
 642,890 4,168,233 6.48×
wcs
 391,724 2,574,296 6.57×
CryptographySingleMultiScaling
crypt
 2,077 12,132 5.84×
CompressionSingleMultiScaling
jpeg
 513.18 3,680 7.17×
zlib
 48.66 399.04 8.20×
HypervisorSingleMultiScaling
clock
 26,747 159,469 5.96×
clone
 291.95 1,457 4.99×
context
 2,240 15,717 7.02×
cpu-sched
 63.5 745.7 11.74×
eventfd
 45,654 538,732 11.80×
fault
 9,380 70,009 7.46×
fork
 3,559 24,442 6.87×
getrandom
 38,153 278,772 7.31×
hrtimers
 219,676 1,341,225 6.11×
mprotect
 887,093 1,669,859 1.88×
mtx
 9,730,677 239,241,111 24.59×
munmap
 2,342 17,943 7.66×
null
 249,391 1,558,487 6.25×
pipe
 817,772 3,729,119 4.56×
priv-instr
 483,694 3,259,856 6.74×
resched
 2,015,666 584,939 0.29×
switch
 68,421 1,772,348 25.90×
timermix
 162,133 1,427,232 8.80×
tlb-shootdown
 10,110 22,347 2.21×
tsc
 849,477 7,089,381 8.35×
usersyscall
 330,481 2,220,696 6.72×
vdso
 70,522,699 592,315,700 8.40×
vfork
 25,242 132,956 5.27×
x86syscall
 2,905,337 18,040,058 6.21×
yield
 5,980,897 6,147,069 1.03×
zero
 1,721,094 11,127,996 6.47×
CPU float/SIMDSingleMultiScaling
eigen
 13,768 100,376 7.29×
expmath
 2,278 20,862 9.16×
fma
 510,549 4,970,158 9.73×
fp
 110.4 791.46 7.17×
fractal
 4.2 37 8.81×
hyperbolic
 2,386 22,250 9.33×
logmath
 3,857 34,057 8.83×
matrix
 2,608 25,001 9.59×
matrix-3d
 589.84 2,242 3.80×
monte-carlo
 118.46 835.77 7.06×
mpfr
 1,031 6,638 6.44×
powmath
 1,337 12,320 9.21×
trig
 1,456 11,569 7.94×
veccmp
 13,269 82,806 6.24×
vecfp
 1,150 8,638 7.51×
vecmath
 1,862 17,590 9.45×
vecshuf
 480.23 3,093 6.44×
vecwide
 20,981 143,230 6.83×
AlgorithmsSingleMultiScaling
bitonicsort
 6.2 43.2 6.97×
bsearch
 205.75 1,398 6.80×
bubblesort
 0.6 2.4 4.00×
fibsearch
 208.27 1,430 6.87×
heapsort
 7.4 52.8 7.14×
hsearch
 4,171 30,768 7.38×
insertionsort
 7 40.8 5.83×
malloc
 1,906,387 18,581,369 9.75×
mcontend
 121.5 2,308 18.99×
mergesort
 19.8 160.39 8.10×
qsort
 12.6 86.6 6.87×
skiplist
 4,151 35,339 8.51×
tree
 1.2 7.2 6.00×
MemorySingleMultiScaling
cache
 258,914 446,814 1.73×
cachehammer
 332,499 3,524,372 10.60×
cacheline
 32.49 347.63 10.70×
l1cache
 193.56 1,259 6.50×
memcpy
 295.94 1,904 6.43×
memrate
 1.16 6.82 5.88×
memthrash
 126.55 392.74 3.10×
ptr-chase
 20,443,881 107,251,028 5.25×
spinmem
 191.19 463.25 2.42×
stream
 24.36 2,147 88.13×
vm
 50,586 292,323 5.78×

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
a2-highgpu-1g 1285 GiB1
a2-highgpu-2g 24170 GiB2
a2-ultragpu-2g 24340 GiB2
a2-highgpu-4g 48340 GiB4
a2-ultragpu-4g 48680 GiB4
a2-highgpu-8g 96680 GiB8
a2-ultragpu-8g 961360 GiB8

Similar Servers

INSTANCEVENDORvCPUsMEMORYAccelerators

a2-ultragpu-1g FAQs