g2-standard-32 by Google Cloud Platform

g2-standard-32 is a Graphics Optimized: 1 NVIDIA L4 GPU, 32 vCPUs, 128GB RAM server offered by Google Cloud Platform with 32 vCPUs, 128 GiB of memory and 0 GB of storage. The pricing starts at 0.753 USD per hour.
32 vCPU
128 GiB Memory
1 GPU
Spare SCore
25,296
(All-cores)
1,195
(Single-core)

Specifications

Server Metadata

Vendor ID
gcp
Server ID
1600032
Name
g2-standard-32
Description
Graphics Optimized: 1 NVIDIA L4 GPU, 32 vCPUs, 128GB RAM
Family
g2
Hw Virt
Average Time To Start
20
Status
active
Observed At
2026-10-19T17:30:25.513890

Availability

REGION / IDSPOTONDEMAND
Ashburn (US) / us-east4
0.9224 USD/h 1.7287 USD/h
Council Bluffs (US) / us-central1
1.0377 USD/h 1.7344 USD/h
The Dalles (US) / us-west1
1.0377 USD/h 1.7344 USD/h
Moncks Corner (US) / us-east1
1.0377 USD/h 1.7344 USD/h
Mumbai (IN) / asia-south1
0.9196 USD/h 1.8054 USD/h
Eemshaven (NL) / europe-west4
1.0937 USD/h 1.8229 USD/h
Montréal (CA) / northamerica-northeast1
1.083 USD/h 1.9097 USD/h
Toronto (CA) / northamerica-northeast2
1.0856 USD/h 1.9097 USD/h
St. Ghislain (BE) / europe-west1
1.1459 USD/h 1.9097 USD/h
Las Vegas (US) / us-west4
1.169 USD/h 1.9534 USD/h
London (GB) / europe-west2
1.1848 USD/h 1.9747 USD/h
Changhua County (TW) / asia-east1
1.2032 USD/h 2.0082 USD/h
Frankfurt (DE) / europe-west3
1.2204 USD/h 2.044 USD/h
Jurong West (SG) / asia-southeast1
1.284 USD/h 2.1397 USD/h
Seoul (KR) / asia-northeast3
0.753 USD/h 2.2273 USD/h
Tokyo (JP) / asia-northeast1
1.1427 USD/h 2.2273 USD/h
Zurich (CH) / europe-west6
1.3895 USD/h 2.3219 USD/h
Dammam (SA) / me-central2
1.5659 USD/h 2.775 USD/h

Processor

vCPUs
32
CPU Allocation
Dedicated
CPU Cores
16
CPU Speed
2.2 GHz
CPU Architecture
x86_64
CPU Manufacturer
Intel
CPU Family
Xeon
CPU L1D Cache
32 KiB
CPU L1D Cache Total
512 KiB
CPU L1I Cache
32 KiB
CPU L1I Cache Total
512 KiB
CPU L2 Cache
1 MiB
CPU L2 Cache Total
16 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
21.2
Scalability
66.25

System Resources and Accelerators

MEMORY
Memory Amount
128 GiB
Memory Amount Actual
128 GiB
ACCELERATORS
Accelerator Count
1
Accelerator Memory Min
24 GiB
Accelerator Memory Total
24 GiB
Accelerator Type
GPU
Accelerator Manufacturer
NVIDIA
Accelerator Family
Ada Lovelace
Accelerator Model
L4
Accelerators
  • NVIDIA Ada Lovelace NVIDIA L4 (Memory amount: 23034, Firmware version: 590.48.01, BIOS version: 95.04.29.00.07, Clock rate: 2040)
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 server balancing dedicated GPU power for language model inference with standard multi-core x86 compute resources.

    GPU AcceleratedGeneral Purpose

    Google Cloud Platform g2-standard-32 is a graphics-optimized instance featuring 32 vCPUs, 128.0 GB of RAM, and a dedicated NVIDIA L4 GPU with 24 GB of VRAM. Powered by a 2.2 GHz Intel Xeon processor, this x86_64 server delivers a balanced 4.0 GB of memory per core but lacks local storage. Performance benchmarks highlight a significant tradeoff: the instance achieves top-tier results in LLM inference for small and medium models, yet exhibits weak single-core CPU performance in stress-ng and Geekbench tests. Multi-core CPU performance, memory bandwidth, and database operations fall within the average range. Qualitatively, the server provides cost-efficient GPU acceleration for workloads that can leverage the Ada Lovelace architecture. It is best suited for AI inference, image processing, and ray tracing, while being less optimal for single-threaded CPU tasks.

    Economics

    Average Price per Region

    Prices per Zone

    Lowest Prices

    Performance

    Workload Profiles

    1.2Score
    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) 2,448,798 ops/sec 1,917,050 ops/sec 1.3 50.00% 50.00% +14.00%
    Redis RPS (pipeline=16, SET) 14,627,375 ops/sec 11,661,818 ops/sec 1.2 20.00% 20.00% +3.70%
    PassMark Memory Mark (composite) 2,588 2,417 1.1 10.00% 10.00% +0.96%
    Memory bandwidth (read, 16 MB ~ L3) 108,668 MB/sec 108,361 MB/sec 1 10.00% 10.00% 0.00%
    PassMark single-thread CPU 1,715 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
    		6.1 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
    		909 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: 1,062

    Geekbench Multi-Core

    Score: 9,347

    Passmark CPU Scores

    BENCHMARKSCORE
    Mark
    		 24,012
    Compression
    		 370,758
    Encryption
    		 11,564
    Extended Instructions
    		 24,434
    Floating Point Maths
    		 45,472
    Integer Maths
    		 84,600
    Physics
    		 2,291
    Prime Numbers
    		 127
    Single Threaded
    		 1,715
    String Sorting
    		 43,673

    Passmark Memory Scores

    BENCHMARKSCORE
    Memory Mark
    		 2,588
    Database Operations
    		 10,267
    Memory Latency
    		 52
    Memory Read Cached
    		 20,717
    Memory Read Uncached
    		 10,930
    Memory Write
    		 10,253

    Stress-ng Stressors

    CPU integerSingleMultiScaling
    atomic
    		 201.46 180.72 0.90×
    bitops
    		 92,267 1,691,789 18.34×
    branch
    		 131,413 3,181,778 24.21×
    funccall
    		 462.87 7,563 16.34×
    funcret
    		 9,614 217,847 22.66×
    hash
    		 107,576 2,452,997 22.80×
    intmath
    		 18,992 453,794 23.89×
    nop
    		 163,214 2,253,565 13.81×
    prime
    		 271,728 4,795,556 17.65×
    rotate
    		 82,431 1,807,779 21.93×
    str
    		 643,579 10,670,745 16.58×
    wcs
    		 410,437 6,561,144 15.99×
    CryptographySingleMultiScaling
    crypt
    		 2,080 32,227 15.49×
    CompressionSingleMultiScaling
    jpeg
    		 514.43 9,125 17.74×
    zlib
    		 32.4 1,113 34.35×
    HypervisorSingleMultiScaling
    clock
    		 25,028 294,574 11.77×
    clone
    		 291.96 3,518 12.05×
    context
    		 2,212 40,593 18.35×
    cpu-sched
    		 57.65 2,030 35.20×
    eventfd
    		 46,309 701,712 15.15×
    fault
    		 9,473 170,651 18.01×
    fork
    		 3,492 53,506 15.32×
    getrandom
    		 38,230 741,569 19.40×
    hrtimers
    		 152,162 3,320,785 21.82×
    mprotect
    		 1,010,669 4,317,379 4.27×
    mtx
    		 8,788,544 709,377,485 80.72×
    munmap
    		 2,274 38,037 16.73×
    null
    		 254,029 4,063,988 16.00×
    pipe
    		 827,340 8,958,296 10.83×
    priv-instr
    		 481,585 8,273,887 17.18×
    resched
    		 4,504,539 1,712,704 0.38×
    switch
    		 74,798 4,616,317 61.72×
    timermix
    		 168,669 3,779,631 22.41×
    tlb-shootdown
    		 10,354 44,484 4.30×
    tsc
    		 849,983 18,912,201 22.25×
    usersyscall
    		 330,035 5,744,516 17.41×
    vdso
    		 70,617,399 1,541,429,883 21.83×
    vfork
    		 23,811 210,791 8.85×
    x86syscall
    		 2,845,933 46,422,231 16.31×
    yield
    		 14,022,714 16,334,969 1.16×
    zero
    		 1,722,433 29,043,946 16.86×
    CPU float/SIMDSingleMultiScaling
    eigen
    		 13,795 265,067 19.21×
    expmath
    		 2,286 55,406 24.23×
    fma
    		 510,773 13,263,259 25.97×
    fp
    		 105.6 2,098 19.87×
    fractal
    		 4.2 98.02 23.34×
    hyperbolic
    		 2,380 59,155 24.85×
    logmath
    		 3,865 90,659 23.46×
    matrix
    		 2,595 65,631 25.30×
    matrix-3d
    		 604.18 3,236 5.36×
    monte-carlo
    		 118.04 2,079 17.61×
    mpfr
    		 1,034 17,285 16.72×
    powmath
    		 1,351 32,794 24.28×
    trig
    		 1,457 29,516 20.26×
    veccmp
    		 13,299 214,598 16.14×
    vecfp
    		 1,151 22,700 19.72×
    vecmath
    		 1,863 46,396 24.91×
    vecshuf
    		 481.66 8,242 17.11×
    vecwide
    		 22,000 381,715 17.35×
    AlgorithmsSingleMultiScaling
    bitonicsort
    		 6.2 111.8 18.03×
    bsearch
    		 206.18 3,685 17.87×
    bubblesort
    		 0.6 6.4 10.67×
    fibsearch
    		 208.52 3,823 18.33×
    heapsort
    		 7.4 140.8 19.03×
    hsearch
    		 4,164 81,799 19.64×
    insertionsort
    		 7 108.8 15.54×
    malloc
    		 1,984,649 33,046,178 16.65×
    mcontend
    		 106.33 6,299 59.24×
    mergesort
    		 19.8 397.19 20.06×
    qsort
    		 12.6 225.19 17.87×
    skiplist
    		 4,162 93,585 22.48×
    tree
    		 1.2 18.98 15.82×
    MemorySingleMultiScaling
    cache
    		 321,861 632,243 1.96×
    cachehammer
    		 366,979 9,312,358 25.38×
    cacheline
    		 33.97 2,204 64.89×
    l1cache
    		 193.94 3,327 17.16×
    memcpy
    		 296.71 4,938 16.64×
    memrate
    		 1.17 11.62 9.93×
    memthrash
    		 146.84 785.88 5.35×
    ptr-chase
    		 26,868,442 302,396,749 11.25×
    spinmem
    		 195.19 1,155 5.92×
    stream
    		 26.96 8,018 297.39×
    vm
    		 52,608 297,094 5.65×

    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
    g2-standard-4 416 GiB1
    g2-standard-8 832 GiB1
    g2-standard-12 1248 GiB1
    g2-standard-16 1664 GiB1
    g2-standard-24 2496 GiB2
    g2-standard-48 48192 GiB4
    g2-standard-96 96384 GiB8

    Similar Servers

    INSTANCEVENDORvCPUsMEMORYAccelerators

    g2-standard-32 FAQs