a3-ultragpu-8g-nolssd by Google Cloud Platform

a3-ultragpu-8g-nolssd is a Accelerator Optimized: 8 NVIDIA H200 GPU, 224 vCPUs, 2952GB RAM, No LocalSSD server offered by Google Cloud Platform with 224 vCPUs, 2952 GiB of memory and 0 GB of storage. The pricing starts at 42.6169 USD per hour.
224 vCPU
2952 GiB Memory
8 GPU
Spare SCore
180,185
(All-cores)
1,838
(Single-core)

Specifications #

Server Metadata #

Vendor ID
gcp
Server ID
1722004
Name
a3-ultragpu-8g-nolssd
Description
Accelerator Optimized: 8 NVIDIA H200 GPU, 224 vCPUs, 2952GB RAM, No LocalSSD
Family
GPU Machines
Series
a3
Hw Virt
Average Time To Start
264
Status
active
Observed At
2026-11-17T04:14:20.755614

Availability #

Region / IDSpotOndemand
Eemshaven (NL) / europe-west4
55.0923 USD/h
-
Dallas (US) / us-south1
58.9646 USD/h
-
Ashburn (US) / us-east4
42.6169 USD/h
83.4918 USD/h
Council Bluffs (US) / us-central1
50.0854 USD/h
83.4918 USD/h
The Dalles (US) / us-west1
50.0854 USD/h
83.4918 USD/h
St. Ghislain (BE) / europe-west1
55.1555 USD/h
91.9472 USD/h
Mumbai (IN) / asia-south1
56.6741 USD/h
100.1957 USD/h

Processor #

vCPUs
224
CPU Allocation
Dedicated
CPU Cores
112
CPU Speed
2.1 GHz
CPU Architecture
x86_64
CPU Manufacturer
Intel
CPU Family
Xeon
CPU Model
8581C
CPU L1D Cache
48 KiB
CPU L1D Cache Total
5 MiB
CPU L1I Cache
32 KiB
CPU L1I Cache Total
4 MiB
CPU L2 Cache
2 MiB
CPU L2 Cache Total
224 MiB
CPU L3 Cache
260 MiB
CPU L3 Cache Total
520 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, monitor, vmx, ssse3, fma, cx16, pdcm, pcid, sse4_1, sse4_2, x2apic, movbe, popcnt, aes, xsave, avx, f16c, rdrand, hypervisor, lahf_lm, abm, 3dnowprefetch, ssbd, ibrs, ibpb, stibp, ibrs_enhanced, tpr_shadow, flexpriority, ept, vpid, ept_ad, fsgsbase, tsc_adjust, bmi1, hle, 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, wbnoinvd, arat, vnmi, 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
98
Scalability
43.75

System Resources and Accelerators #

Memory
Memory Amount
3 TB
Memory Amount Actual
3 TB
Accelerators
Accelerator Count
8
Accelerator Memory Min
141 GiB
Accelerator Memory Total
1 TB
Accelerator Type
GPU
Accelerator Manufacturer
NVIDIA
Accelerator Family
Hopper
Accelerator Model
H200
Accelerators
  • NVIDIA Hopper NVIDIA H200 (Memory amount: 143771, Firmware version: 590.48.01, BIOS version: 96.00.A5.00.03, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H200 (Memory amount: 143771, Firmware version: 590.48.01, BIOS version: 96.00.A5.00.03, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H200 (Memory amount: 143771, Firmware version: 590.48.01, BIOS version: 96.00.A5.00.03, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H200 (Memory amount: 143771, Firmware version: 590.48.01, BIOS version: 96.00.A5.00.03, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H200 (Memory amount: 143771, Firmware version: 590.48.01, BIOS version: 96.00.A5.00.03, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H200 (Memory amount: 143771, Firmware version: 590.48.01, BIOS version: 96.00.A5.00.03, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H200 (Memory amount: 143771, Firmware version: 590.48.01, BIOS version: 96.00.A5.00.03, Clock rate: 1980)
  • NVIDIA Hopper NVIDIA H200 (Memory amount: 143771, Firmware version: 590.48.01, BIOS version: 96.00.A5.00.03, Clock rate: 1980)
Storage
Storage Size
0 GB
Storages
    Compatible Storage IDs
    30014, 30017, 30012, 30016
    Network
    Inbound Traffic
    0 GB/month
    Outbound Traffic
    0 GB/month
    IPv4
    0

    CPU and System Topology #

    Server Description #

    An accelerator-optimized platform featuring eight high-performance GPUs and massive system memory for demanding artificial intelligence and parallel computing workloads.

    Google Cloud Platform a3-ultragpu-8g-nolssd is an accelerator-optimized server designed for intensive parallel computing and machine learning workloads. The hardware configuration features 8 NVIDIA H200 Hopper GPUs with 1128 GB of total VRAM, paired with 224 vCPUs from an Intel Xeon 8581C processor and 2952.0 GB of system memory. It does not include local SSD storage. Benchmarks show top-tier multi-core CPU performance, exceptional memory bandwidth, and high-throughput capabilities in database operations and software compilation. The processor supports advanced instruction sets like AVX-512 and Intel AMX to accelerate matrix operations. This server is optimized for large-scale AI model training, LLM inference, and complex scientific simulations where GPU acceleration and massive memory capacity are critical, representing a high-density resource profile for specialized enterprise workloads.

    Economics #

    Average Price per Region #

    Prices per Zone #

    Lowest Prices #

    Performance #

    Workload Profiles #

    7.7Score
    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) 30,271,633 ops/sec 1,933,252 ops/sec 16 50.00% 50.00% +300.00%
    Redis RPS (pipeline=16, SET) 154,487,015 ops/sec 11,814,424 ops/sec 12 20.00% 20.00% +64.00%
    PassMark Memory Mark (composite) 2,204 2,408 0.92 10.00% 10.00% -0.83%
    Memory bandwidth (read, 16 MB ~ L3) 513,010 MB/sec 108,674 MB/sec 4.6 10.00% 10.00% +16.00%
    PassMark single-thread CPU 2,810 Mops/s 2,365 Mops/s 1.2 10.00% 10.00% +1.80%

    GPU Bandwidth and Latency #

    Node (all GPUs)
    Node baseline directional bandwidth
    155.1 GB/s
    Node GPU-to-host bandwidth
    224.9 GB/s
    Node host-to-GPU bandwidth
    222.8 GB/s
    Peer-to-peer
    GPU-to-GPU P2P duplex bandwidth
    383.4 GB/s
    GPU-to-GPU P2P gather bandwidth
    398.3 GB/s
    GPU-to-GPU P2P latency
    1,270 ns
    GPU-to-GPU P2P bandwidth
    388.1 GB/s
    Single GPU slot
    Per-GPU GPU-to-host bandwidth
    55.3 GB/s
    Per-GPU host-to-GPU bandwidth
    55.4 GB/s
    Per-GPU host-device latency
    1,185 ns
    Benchmark Environment
    Driver Version
    535.230.02
    CUDA Version (driver)
    12.2
    nvbandwidth Version
    v0.5
    GPU Count
    8
    P2P Supported
    Yes
    Efficiency
    SM/CE copy efficiency
    0.93

    Memory Bandwidth#

    Compression #

    OpenSSL #

    Geekbench Single-Core #

    Score: 1,959

    Geekbench Multi-Core #

    Score: 19,083

    Passmark CPU Scores #

    BENCHMARKSCORE
    Mark
    125,594
    Compression
    3,082,097
    Encryption
    188,989
    Extended Instructions
    229,315
    Floating Point Maths
    609,446
    Integer Maths
    860,934
    Physics
    23,762
    Prime Numbers
    1,615
    Single Threaded
    2,810
    String Sorting
    399,516

    Passmark Memory Scores #

    BENCHMARKSCORE
    Memory Mark
    2,204
    Database Operations
    33,038
    Memory Latency
    65
    Memory Read Cached
    24,180
    Memory Read Uncached
    7,386
    Memory Write
    7,225

    Stress-ng Stressors #

    CPU integerSingleMultiScaling
    atomic
    219.02 45.75 0.21×
    bitops
    164,571 16,293,711 99.01×
    branch
    155,370 19,688,472 126.72×
    funccall
    839.61 87,715 104.47×
    funcret
    13,850 1,810,069 130.69×
    hash
    197,750 23,564,692 119.16×
    intmath
    70,928 7,021,667 99.00×
    nop
    293,565 19,350,922 65.92×
    prime
    524,064 49,671,090 94.78×
    rotate
    110,376 9,637,176 87.31×
    str
    1,010,690 108,299,524 107.15×
    wcs
    783,414 68,639,779 87.62×
    CryptographySingleMultiScaling
    crypt
    3,757 355,539 94.64×
    CompressionSingleMultiScaling
    jpeg
    939.05 89,921 95.76×
    zlib
    61.67 8,843 143.39×
    HypervisorSingleMultiScaling
    clock
    22,906 184,419 8.05×
    clone
    465.16 16,648 35.79×
    context
    5,723 782,328 136.71×
    cpu-sched
    58.57 13,612 232.40×
    eventfd
    169,837 2,965,871 17.46×
    fault
    35,234 180,714 5.13×
    fork
    3,978 57,820 14.53×
    getrandom
    65,763 7,807,999 118.73×
    hrtimers
    191,200 11,384,870 59.54×
    mprotect
    635,717 16,203,767 25.49×
    mtx
    19,005,009 4,980,110,070 262.04×
    munmap
    3,307 65,748 19.88×
    null
    1,000,596 128,815,306 128.74×
    pipe
    773,920 118,749,533 153.44×
    priv-instr
    922,849 8,686,320 9.41×
    resched
    10,344,056 29,524,940 2.85×
    switch
    311,354 15,280,991 49.08×
    timermix
    229,326 32,669,549 142.46×
    tlb-shootdown
    10,495 225,710 21.51×
    tsc
    646,216 86,978,562 134.60×
    usersyscall
    784,898 8,586,327 10.94×
    vdso
    99,727,135 12,769,742,311 128.05×
    vfork
    27,602 106,962 3.88×
    x86syscall
    11,090,573 1,532,591,023 138.19×
    yield
    216,201,107 228,650,952 1.06×
    zero
    7,217,363 941,842,723 130.50×
    CPU float/SIMDSingleMultiScaling
    eigen
    17,923 1,903,978 106.23×
    expmath
    3,914 465,024 118.80×
    fma
    1,450,480 150,262,228 103.59×
    fp
    198.2 27,566 139.08×
    fractal
    7.2 851.11 118.21×
    hyperbolic
    4,497 532,958 118.51×
    logmath
    5,759 750,185 130.27×
    matrix
    5,375 533,397 99.24×
    matrix-3d
    814.9 20,567 25.24×
    monte-carlo
    322.87 43,211 133.84×
    mpfr
    1,951 172,301 88.34×
    powmath
    2,181 271,382 124.45×
    trig
    2,977 304,532 102.28×
    veccmp
    27,737 2,483,868 89.55×
    vecfp
    1,704 179,289 105.25×
    vecmath
    4,353 500,136 114.88×
    vecshuf
    915.56 85,917 93.84×
    vecwide
    38,759 3,571,056 92.14×
    AlgorithmsSingleMultiScaling
    bitonicsort
    9.8 940.78 96.00×
    bsearch
    318.76 33,393 104.76×
    bubblesort
    1.2 89.6 74.67×
    fibsearch
    365.37 36,358 99.51×
    heapsort
    11.2 1,188 106.09×
    hsearch
    9,406 858,044 91.23×
    insertionsort
    14.2 1,030 72.53×
    malloc
    3,129,585 152,859,631 48.84×
    mcontend
    56.89 30,210 531.03×
    mergesort
    33.8 3,774 111.66×
    qsort
    24.4 2,597 106.45×
    skiplist
    6,054 784,685 129.61×
    tree
    1.8 178.82 99.34×
    MemorySingleMultiScaling
    cache
    194,814 8,135,293 41.76×
    cachehammer
    191,880 7,721,225 40.24×
    cacheline
    57.71 2,631 45.59×
    l1cache
    1,228 24,668 20.08×
    memcpy
    493.57 39,543 80.12×
    memrate
    0.12 7.52 62.67×
    memthrash
    480.14 1,023 2.13×
    ptr-chase
    31,193,364 1,695,388,295 54.35×
    spinmem
    254.39 9,225 36.26×
    stream
    2.66 16,901 6353.68×
    vm
    193,869 399,897 2.06×

    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 2081872 GiB8
    a3-edgegpu-8g-nolssd 2081872 GiB8

    Similar Servers #

    INSTANCEVENDORvCPUsMEMORYAccelerators

    a3-ultragpu-8g-nolssd FAQs #