GPU-20xCPU-256GB-2xL40S by UpCloud

GPU-20xCPU-256GB-2xL40S is a GPU (20 vCPUs, 256 GiB RAM, 2x L40S) server offered by UpCloud with 20 vCPUs, 256 GiB of memory and 0 GB of storage. The pricing starts at 2.6863 USD per hour.
20 vCPU
256 GiB Memory
2 GPU
Spare SCore
83,934
(All-cores)
4,182
(Single-core)

Specifications

Server Metadata

Vendor ID
upcloud
Name
GPU-20xCPU-256GB-2xL40S
Description
GPU (20 vCPUs, 256 GiB RAM, 2x L40S)
Family
GPU
Hw Virt
Average Time To Start
86
Status
active
Observed At
2026-10-23T12:29:25.982040

Availability

REGION / IDSPOTONDEMAND
Helsinki #2 (FI) / fi-hel2
2.6863 USD/h 3.5761 USD/h

Processor

vCPUs
20
Hypervisor
KVM
CPU Allocation
Shared
CPU Cores
20
CPU Speed
2 GHz
CPU Architecture
x86_64
CPU Manufacturer
AMD
CPU Family
EPYC
CPU Model
9575F
CPU L1D Cache
64 KiB
CPU L1D Cache Total
1 MiB
CPU L1I Cache
64 KiB
CPU L1I Cache Total
1 MiB
CPU L2 Cache
512 KiB
CPU L2 Cache Total
10 MiB
CPU L3 Cache
16 MiB
CPU L3 Cache Total
16 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, ht, syscall, nx, mmxext, fxsr_opt, pdpe1gb, rdtscp, lm, rep_good, nopl, cpuid, extd_apicid, tsc_known_freq, pni, pclmulqdq, ssse3, fma, cx16, pcid, sse4_1, sse4_2, x2apic, movbe, popcnt, tsc_deadline_timer, aes, xsave, avx, f16c, rdrand, hypervisor, lahf_lm, cmp_legacy, cr8_legacy, abm, sse4a, misalignsse, 3dnowprefetch, osvw, perfctr_core, ssbd, ibrs, ibpb, stibp, vmmcall, fsgsbase, tsc_adjust, bmi1, avx2, smep, bmi2, invpcid, avx512f, avx512dq, rdseed, adx, smap, avx512ifma, clflushopt, clwb, avx512cd, sha_ni, avx512bw, avx512vl, xsaveopt, xsavec, xgetbv1, xsaves, avx_vnni, avx512_bf16, clzero, xsaveerptr, wbnoinvd, arat, avx512vbmi, umip, pku, ospke, avx512_vbmi2, gfni, vaes, vpclmulqdq, avx512_vnni, avx512_bitalg, avx512_vpopcntdq, la57, rdpid, movdiri, movdir64b, avx512_vp2intersect, arch_capabilities
Ecpus
20.1
Scalability
100.5

System Resources and Accelerators

MEMORY
Memory Amount
256 GiB
Memory Amount Actual
256 GiB
ACCELERATORS
Accelerator Count
2
Accelerator Memory Min
48 GiB
Accelerator Memory Total
96 GiB
Accelerator Type
GPU
Accelerator Manufacturer
NVIDIA
Accelerator Family
Ada Lovelace
Accelerator Model
L40S
Accelerators
  • NVIDIA Ada Lovelace NVIDIA L40S (Memory amount: 46068, Firmware version: 595.84, BIOS version: 95.02.66.00.02, Clock rate: 2520)
  • NVIDIA Ada Lovelace NVIDIA L40S (Memory amount: 46068, Firmware version: 595.84, BIOS version: 95.02.66.00.02, Clock rate: 2520)
STORAGE
Storage Size
0 GB
Storages
    NETWORK
    Inbound Traffic
    0 GB/month
    Outbound Traffic
    36864 GB/month
    IPv4
    1

    CPU and System Topology

    Server Description

    A dual-GPU accelerated instance featuring AMD EPYC processing and 256 GB of memory for demanding machine learning and data science workloads.

    GPU AcceleratedMemory Optimized

    UpCloud GPU-20xCPU-256GB-2xL40S is a GPU-accelerated server instance built on the KVM hypervisor. It features 20 vCPUs from an AMD EPYC 9575F processor, 256.0 GB of system memory, and two NVIDIA L40S GPUs with a total of 96 GB of VRAM. The instance does not include local storage, requiring external storage attachments. In benchmarks, the server delivers top-tier single-core CPU performance and strong multi-core capabilities, alongside high memory bandwidth. It excels in small-to-medium LLM inference, with prompt processing for a 135M model reaching 4555.72 tokens per second. This hardware profile provides a qualitatively cost-efficient balance of high-capacity VRAM and system memory relative to its CPU core count. It is well-suited for machine learning inference, image processing, data science, and memory-intensive database workloads.

    Economics

    Average Price per Region

    Prices per Zone

    Lowest Prices

    Performance

    Workload Profiles

    3.7Score
    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) 8,130,330 ops/sec 1,919,289 ops/sec 4.3 50.00% 50.00% +110.00%
    Redis RPS (pipeline=16, SET) 51,803,292 ops/sec 11,728,058 ops/sec 4.3 20.00% 20.00% +34.00%
    PassMark Memory Mark (composite) 2,843 2,416 1.2 10.00% 10.00% +1.80%
    Memory bandwidth (read, 16 MB ~ L3) 871,422 MB/sec 108,466 MB/sec 7.9 10.00% 10.00% +23.00%
    PassMark single-thread CPU 4,256 Mops/s 2,365 Mops/s 1.8 10.00% 10.00% +6.10%

    GPU Bandwidth and Latency

    Node (all GPUs)
    Node baseline directional bandwidth
    		45.3 GB/s
    Node GPU-to-host bandwidth
    		52.7 GB/s
    Node host-to-GPU bandwidth
    		53.7 GB/s
    Single GPU slot
    Per-GPU GPU-to-host bandwidth
    		26.4 GB/s
    Per-GPU host-to-GPU bandwidth
    		26.9 GB/s
    Per-GPU host-device latency
    		602 ns
    Benchmark Environment
    Driver Version
    		 595.84
    CUDA Runtime Version
    		 13.2
    nvbandwidth Version
    		 v0.8
    GPU Count
    		 2
    P2P Supported
    No
    Efficiency
    SM/CE copy efficiency
    		1.00

    Memory Bandwidth

    Compression

    OpenSSL

    Geekbench Single-Core

    Score: 2,478

    Geekbench Multi-Core

    Score: 20,060

    Passmark CPU Scores

    BENCHMARKSCORE
    Mark
    		 62,859
    Compression
    		 726,715
    Encryption
    		 35,460
    Extended Instructions
    		 64,317
    Floating Point Maths
    		 170,442
    Integer Maths
    		 170,350
    Physics
    		 11,003
    Prime Numbers
    		 670
    Single Threaded
    		 4,256
    String Sorting
    		 101,175

    Passmark Memory Scores

    BENCHMARKSCORE
    Memory Mark
    		 2,843
    Database Operations
    		 27,649
    Memory Latency
    		 88
    Memory Read Cached
    		 38,108
    Memory Read Uncached
    		 35,161
    Memory Write
    		 35,223

    Stress-ng Stressors

    CPU integerSingleMultiScaling
    atomic
    		 488.42 426.71 0.87×
    bitops
    		 227,407 4,171,096 18.34×
    branch
    		 170,345 3,388,637 19.89×
    funccall
    		 775.81 14,204 18.31×
    funcret
    		 24,862 472,022 18.99×
    hash
    		 209,780 3,889,357 18.54×
    intmath
    		 80,359 1,473,611 18.34×
    nop
    		 394,612 7,086,405 17.96×
    prime
    		 578,838 10,570,973 18.26×
    rotate
    		 132,415 2,416,583 18.25×
    str
    		 1,418,752 23,467,940 16.54×
    wcs
    		 929,696 15,180,940 16.33×
    CryptographySingleMultiScaling
    crypt
    		 4,905 83,665 17.06×
    CompressionSingleMultiScaling
    jpeg
    		 1,176 22,548 19.18×
    zlib
    		 72.8 1,380 18.95×
    HypervisorSingleMultiScaling
    clock
    		 78,732 279,709 3.55×
    clone
    		 449.35 6,012 13.38×
    context
    		 8,672 159,827 18.43×
    cpu-sched
    		 61.28 1,292 21.09×
    eventfd
    		 109,094 1,054,235 9.66×
    fault
    		 33,411 191,325 5.73×
    fork
    		 5,595 61,541 11.00×
    getrandom
    		 91,393 1,470,060 16.09×
    hrtimers
    		 546,606 4,781,595 8.75×
    mprotect
    		 670,770 8,257,345 12.31×
    mtx
    		 33,518,112 1,542,010,515 46.01×
    munmap
    		 3,871 50,711 13.10×
    null
    		 1,330,848 24,453,365 18.37×
    pipe
    		 1,648,690 13,829,754 8.39×
    priv-instr
    		 1,301,472 15,017,657 11.54×
    resched
    		 13,013,094 9,358,552 0.72×
    switch
    		 272,595 7,410,676 27.19×
    timermix
    		 319,694 5,610,226 17.55×
    tlb-shootdown
    		 20,888 133,637 6.40×
    tsc
    		 858,199 17,169,693 20.01×
    usersyscall
    		 1,041,229 10,939,492 10.51×
    vdso
    		 97,891,260 1,876,276,433 19.17×
    vfork
    		 42,993 200,451 4.66×
    x86syscall
    		 18,186,806 348,291,822 19.15×
    yield
    		 28,864,184 32,784,629 1.14×
    zero
    		 7,967,977 138,052,075 17.33×
    CPU float/SIMDSingleMultiScaling
    eigen
    		 17,004 321,282 18.89×
    expmath
    		 3,802 71,487 18.80×
    fma
    		 1,547,093 27,708,907 17.91×
    fp
    		 210.47 3,758 17.85×
    fractal
    		 8.8 167.62 19.05×
    hyperbolic
    		 5,207 100,778 19.35×
    logmath
    		 5,544 101,907 18.38×
    matrix
    		 8,779 128,490 14.64×
    matrix-3d
    		 1,347 10,833 8.05×
    monte-carlo
    		 421.18 7,433 17.65×
    mpfr
    		 2,228 38,940 17.48×
    powmath
    		 1,743 32,682 18.75×
    trig
    		 2,219 39,679 17.88×
    veccmp
    		 29,054 502,238 17.29×
    vecfp
    		 2,458 43,920 17.87×
    vecmath
    		 4,604 87,810 19.07×
    vecshuf
    		 910.9 15,618 17.15×
    vecwide
    		 38,468 688,440 17.90×
    AlgorithmsSingleMultiScaling
    bitonicsort
    		 12.2 199.6 16.36×
    bsearch
    		 360.59 6,661 18.47×
    bubblesort
    		 1.4 23.8 17.00×
    fibsearch
    		 413.03 7,389 17.89×
    heapsort
    		 12.6 242.99 19.28×
    hsearch
    		 12,583 213,972 17.01×
    insertionsort
    		 14.8 254 17.16×
    malloc
    		 3,718,007 41,573,771 11.18×
    mcontend
    		 131.22 3,273 24.94×
    mergesort
    		 38.2 685.78 17.95×
    qsort
    		 25.4 444.39 17.50×
    skiplist
    		 8,047 143,791 17.87×
    tree
    		 2 21.38 10.69×
    MemorySingleMultiScaling
    cache
    		 1,152,053 1,632,824 1.42×
    cachehammer
    		 262,967 4,699,083 17.87×
    cacheline
    		 88.66 1,044 11.77×
    l1cache
    		 679.63 11,735 17.27×
    memcpy
    		 595.38 11,255 18.90×
    memrate
    		 0.99 12.68 12.81×
    memthrash
    		 178.69 1,077 6.03×
    ptr-chase
    		 12,736,481 147,702,570 11.60×
    spinmem
    		 420.98 1,737 4.13×
    stream
    		 357.54 30,211 84.50×
    vm
    		 234,912 294,223 1.25×

    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

    PostgreSQL heavy read-only throughput

    Alternatives

    Servers of the Same Family

    INSTANCEvCPUsMEMORYAccelerators
    GPU-8xCPU-64GB-1xL40S 864 GiB1
    GPU-12xCPU-128GB-1xL40S 12128 GiB1
    GPU-12xCPU-128GB-2xL40S 12128 GiB2
    GPU-16xCPU-80GB-1xRTXPRO6000 1680 GiB1
    GPU-16xCPU-192GB-1xL40S 16192 GiB1
    GPU-16xCPU-192GB-2xL40S 16192 GiB2
    GPU-16xCPU-192GB-3xL40S 16192 GiB3

    Similar Servers

    INSTANCEVENDORvCPUsMEMORYAccelerators

    GPU-20xCPU-256GB-2xL40S FAQs