a2-highgpu-2g by Google Cloud Platform

a2-highgpu-2g is a Accelerator Optimized: 2 NVIDIA Tesla A100 GPUs, 24 vCPUs, 170GB RAM server offered by Google Cloud Platform with 24 vCPUs, 170 GiB of memory and 0 GB of storage. The pricing starts at 3.4795 USD per hour.
24 vCPU
170 GiB Memory
2 GPU
Spare SCore
18,976
(All-cores)
1,202
(Single-core)

Specifications

Server Metadata

Vendor ID
gcp
Server ID
1000024
Name
a2-highgpu-2g
Description
Accelerator Optimized: 2 NVIDIA Tesla A100 GPUs, 24 vCPUs, 170GB RAM
Family
a2
Hw Virt
Average Time To Start
57
Status
active
Observed At
2026-11-05T01:32:27.791318

Availability

REGION / IDSPOTONDEMAND
Council Bluffs (US) / us-central1
3.9783 USD/h 7.3468 USD/h
The Dalles (US) / us-west1
3.9783 USD/h 7.3468 USD/h
Moncks Corner (US) / us-east1
3.9783 USD/h 7.3468 USD/h
Eemshaven (NL) / europe-west4
3.4795 USD/h 7.4959 USD/h
Las Vegas (US) / us-west4
3.6531 USD/h 7.8687 USD/h
Jurong West (SG) / asia-southeast1
4.8045 USD/h 8.0275 USD/h
Tel Aviv (IL) / me-west1
4.8488 USD/h 8.0814 USD/h
Seoul (KR) / asia-northeast3
4.8239 USD/h 8.0992 USD/h
Tokyo (JP) / asia-northeast1
4.8349 USD/h 8.0992 USD/h
Salt Lake City (US) / us-west3
4.6607 USD/h 8.6501 USD/h

Processor

vCPUs
24
CPU Allocation
Dedicated
CPU Cores
12
CPU Speed
2.2 GHz
CPU Architecture
x86_64
CPU Manufacturer
Intel
CPU Family
Xeon
CPU L1D Cache
32 KiB
CPU L1D Cache Total
384 KiB
CPU L1I Cache
32 KiB
CPU L1I Cache Total
384 KiB
CPU L2 Cache
1 MiB
CPU L2 Cache Total
12 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
15.8
Scalability
65.83

System Resources and Accelerators

MEMORY
Memory Amount
170 GiB
Memory Amount Actual
170 GiB
ACCELERATORS
Accelerator Count
2
Accelerator Memory Min
40 GiB
Accelerator Memory Total
80 GiB
Accelerator Type
GPU
Accelerator Manufacturer
NVIDIA
Accelerator Family
Ampere
Accelerator Model
A100
Accelerators
  • NVIDIA Ampere NVIDIA A100-SXM4-40GB (Memory amount: 40960, Firmware version: 590.48.01, BIOS version: 92.00.45.00.03, Clock rate: 1410)
  • NVIDIA Ampere NVIDIA A100-SXM4-40GB (Memory amount: 40960, Firmware version: 590.48.01, BIOS version: 92.00.45.00.03, Clock rate: 1410)
STORAGE
Storage Size
0 GB
Storages
    NETWORK
    Inbound Traffic
    0 GB/month
    Outbound Traffic
    0 GB/month
    IPv4
    0

    CPU and System Topology

    Server Description

    An accelerator-optimized platform pairing dual NVIDIA A100 GPUs with 170 GB of memory for high-throughput parallel computing and LLM inference.

    GPU AcceleratedMemory Optimized

    Google Cloud Platform a2-highgpu-2g is an accelerator-optimized server configured with two NVIDIA Ampere A100 GPUs providing 80 GB of total VRAM, 24 vCPUs on an Intel Xeon x86_64 processor, and 170.0 GB of system memory. It does not include local storage. Performance benchmarks indicate a distinct tradeoff: single-core CPU execution is relatively weak, scoring in the bottom 25%, and multi-core CPU performance is average. However, the instance delivers top-tier performance in LLM inference benchmarks, particularly for prompt processing and text generation on medium to large models. Qualitatively, the high resource density of GPUs and memory makes this instance cost-efficient for specialized parallel workloads rather than general-purpose compute. It is highly suited for deep learning, large language model inference, and GPU-accelerated data science tasks.

    Economics

    Average Price per Region

    Prices per Zone

    Lowest Prices

    Performance

    Workload Profiles

    0.93Score
    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) 1,844,266 ops/sec 1,933,252 ops/sec 0.95 50.00% 50.00% -2.50%
    Redis RPS (pipeline=16, SET) 11,433,281 ops/sec 11,794,699 ops/sec 0.92 20.00% 20.00% -1.70%
    PassMark Memory Mark (composite) 2,483 2,410 1 10.00% 10.00% 0.00%
    Memory bandwidth (read, 16 MB ~ L3) 114,321 MB/sec 108,490 MB/sec 1 10.00% 10.00% 0.00%
    PassMark single-thread CPU 1,693 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
    		22.3 GB/s
    Node GPU-to-host bandwidth
    		26.4 GB/s
    Node host-to-GPU bandwidth
    		24.7 GB/s
    Peer-to-peer
    GPU-to-GPU P2P duplex bandwidth
    		262.0 GB/s
    GPU-to-GPU P2P gather bandwidth
    		276.8 GB/s
    GPU-to-GPU P2P latency
    		1,216 ns
    GPU-to-GPU P2P bandwidth
    		277.0 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
    		1,228 ns
    Benchmark Environment
    Driver Version
    		 535.183.01
    CUDA Version (driver)
    		 12.6
    nvbandwidth Version
    		 v0.5
    GPU Count
    		 2
    P2P Supported
    Yes
    Efficiency
    SM/CE copy efficiency
    		1.00

    Memory Bandwidth

    Compression

    OpenSSL

    Geekbench Single-Core

    Score: 1,066

    Geekbench Multi-Core

    Score: 8,247

    Passmark CPU Scores

    BENCHMARKSCORE
    Mark
    		 19,022
    Compression
    		 278,622
    Encryption
    		 8,685
    Extended Instructions
    		 18,027
    Floating Point Maths
    		 34,100
    Integer Maths
    		 63,469
    Physics
    		 1,952
    Prime Numbers
    		 102
    Single Threaded
    		 1,693
    String Sorting
    		 32,865

    Passmark Memory Scores

    BENCHMARKSCORE
    Memory Mark
    		 2,483
    Database Operations
    		 7,693
    Memory Latency
    		 56
    Memory Read Cached
    		 20,838
    Memory Read Uncached
    		 10,561
    Memory Write
    		 9,923

    Stress-ng Stressors

    CPU integerSingleMultiScaling
    atomic
    		 196.82 176.81 0.90×
    bitops
    		 92,107 1,268,948 13.78×
    branch
    		 131,159 2,385,202 18.19×
    funccall
    		 463.25 5,707 12.32×
    funcret
    		 9,662 164,306 17.01×
    hash
    		 107,499 1,848,802 17.20×
    intmath
    		 19,300 340,768 17.66×
    nop
    		 162,668 1,688,568 10.38×
    prime
    		 270,590 3,637,855 13.44×
    rotate
    		 82,327 1,359,992 16.52×
    str
    		 642,061 7,613,147 11.86×
    wcs
    		 401,184 5,255,517 13.10×
    CryptographySingleMultiScaling
    crypt
    		 2,080 24,294 11.68×
    CompressionSingleMultiScaling
    jpeg
    		 513.55 6,838 13.32×
    zlib
    		 45.7 797.57 17.45×
    HypervisorSingleMultiScaling
    clock
    		 25,680 256,514 9.99×
    clone
    		 302.01 2,711 8.98×
    context
    		 2,244 31,266 13.93×
    cpu-sched
    		 74.55 1,543 20.69×
    eventfd
    		 46,778 557,472 11.92×
    fault
    		 9,375 133,891 14.28×
    fork
    		 3,508 43,757 12.47×
    getrandom
    		 38,164 558,030 14.62×
    hrtimers
    		 225,071 2,038,878 9.06×
    mprotect
    		 1,029,751 3,304,564 3.21×
    mtx
    		 8,851,959 494,470,262 55.86×
    munmap
    		 2,279 30,794 13.51×
    null
    		 251,155 3,097,039 12.33×
    pipe
    		 794,930 7,570,880 9.52×
    priv-instr
    		 483,782 6,468,454 13.37×
    resched
    		 3,821,317 1,774,549 0.46×
    switch
    		 76,437 3,519,818 46.05×
    timermix
    		 167,692 2,901,791 17.30×
    tlb-shootdown
    		 10,330 38,723 3.75×
    tsc
    		 849,644 14,190,085 16.70×
    usersyscall
    		 329,111 4,419,184 13.43×
    vdso
    		 70,460,799 1,189,919,617 16.89×
    vfork
    		 24,942 203,921 8.18×
    x86syscall
    		 2,881,811 35,909,521 12.46×
    yield
    		 11,221,927 11,899,278 1.06×
    zero
    		 1,728,446 22,118,678 12.80×
    CPU float/SIMDSingleMultiScaling
    eigen
    		 13,793 198,845 14.42×
    expmath
    		 2,278 41,829 18.37×
    fma
    		 510,662 9,954,223 19.49×
    fp
    		 99.95 1,584 15.85×
    fractal
    		 4.2 73.81 17.57×
    hyperbolic
    		 2,378 44,564 18.74×
    logmath
    		 3,857 68,186 17.68×
    matrix
    		 2,635 50,211 19.05×
    matrix-3d
    		 572.62 3,364 5.87×
    monte-carlo
    		 118.28 1,571 13.28×
    mpfr
    		 1,030 13,254 12.87×
    powmath
    		 1,349 24,701 18.31×
    trig
    		 1,459 23,146 15.86×
    veccmp
    		 13,279 165,343 12.45×
    vecfp
    		 1,150 17,035 14.81×
    vecmath
    		 1,867 35,239 18.87×
    vecshuf
    		 480.47 6,184 12.87×
    vecwide
    		 21,862 286,654 13.11×
    AlgorithmsSingleMultiScaling
    bitonicsort
    		 6.2 86.2 13.90×
    bsearch
    		 205.75 2,803 13.63×
    bubblesort
    		 0.6 4.8 8.00×
    fibsearch
    		 208.24 2,869 13.78×
    heapsort
    		 7.4 105.6 14.27×
    hsearch
    		 4,173 64,505 15.46×
    insertionsort
    		 7 81.6 11.66×
    malloc
    		 1,876,383 25,197,782 13.43×
    mcontend
    		 108.48 4,658 42.94×
    mergesort
    		 19.8 316.59 15.99×
    qsort
    		 12.6 172.99 13.73×
    skiplist
    		 4,158 70,439 16.94×
    tree
    		 1.2 14.39 11.99×
    MemorySingleMultiScaling
    cache
    		 320,837 654,801 2.04×
    cachehammer
    		 364,651 7,310,591 20.05×
    cacheline
    		 30.14 770.05 25.55×
    l1cache
    		 193.44 2,497 12.91×
    memcpy
    		 295.76 3,807 12.87×
    memrate
    		 1.17 8.99 7.68×
    memthrash
    		 127.25 722.32 5.68×
    ptr-chase
    		 16,482,173 233,426,796 14.16×
    spinmem
    		 189.59 822.72 4.34×
    stream
    		 26.5 6,185 233.40×
    vm
    		 51,959 295,353 5.68×

    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-ultragpu-1g 12170 GiB1
    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-highgpu-2g FAQs