l40s-90 by OVHcloud

l40s-90 is a Cloud GPU (15 vCPUs, 90 GiB RAM, 400 GB NVMe, 1xL40S 48 GiB VRAM) server offered by OVHcloud with 15 vCPUs, 90 GiB of memory and 400 GB of storage. The pricing starts at 1.5641 USD per hour.
15 vCPU
90 GiB Memory
400 GB Storage
1 GPU
Spare SCore
45,529
(All-cores)
3,043
(Single-core)

Specifications

Server Metadata

Vendor ID
ovh
Name
l40s-90
Description
Cloud GPU (15 vCPUs, 90 GiB RAM, 400 GB NVMe, 1xL40S 48 GiB VRAM)
Family
Cloud GPU
Hw Virt
Average Time To Start
140
Status
active
Observed At
2026-11-01T14:10:48.217173

Availability

REGION / IDSPOTONDEMAND
GRA11 (FR) / GRA11
- 1.5641 USD/h

Processor

vCPUs
15
Hypervisor
KVM
CPU Allocation
Dedicated
CPU Cores
15
CPU Speed
2.75 GHz
CPU Architecture
x86_64
CPU Manufacturer
AMD
CPU Family
EPYC
CPU Model
9124
CPU L1D Cache
64 KiB
CPU L1D Cache Total
960 KiB
CPU L1I Cache
64 KiB
CPU L1I Cache Total
960 KiB
CPU L2 Cache
512 KiB
CPU L2 Cache Total
8 MiB
CPU L3 Cache
16 MiB
CPU L3 Cache Total
240 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, 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, svm, cr8_legacy, abm, sse4a, misalignsse, 3dnowprefetch, osvw, perfctr_core, ssbd, ibrs, ibpb, stibp, vmmcall, fsgsbase, tsc_adjust, bmi1, avx2, smep, bmi2, erms, invpcid, avx512f, avx512dq, rdseed, adx, smap, avx512ifma, clflushopt, clwb, avx512cd, sha_ni, avx512bw, avx512vl, xsaveopt, xsavec, xgetbv1, xsaves, avx512_bf16, clzero, xsaveerptr, wbnoinvd, arat, npt, lbrv, nrip_save, tsc_scale, vmcb_clean, pausefilter, pfthreshold, v_vmsave_vmload, vgif, avx512vbmi, umip, pku, ospke, avx512_vbmi2, gfni, vaes, vpclmulqdq, avx512_vnni, avx512_bitalg, avx512_vpopcntdq, la57, rdpid, overflow_recov, succor, fsrm, arch_capabilities
Ecpus
15
Scalability
100

System Resources and Accelerators

MEMORY
Memory Amount
90 GiB
Memory Amount Actual
88 GiB
ACCELERATORS
Accelerator Count
1
Accelerator Memory Min
48 GiB
Accelerator Memory Total
48 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: 590.48.01, BIOS version: 95.02.66.00.02, Clock rate: 2520)
STORAGE
Storage Size
400 GB
Storage Type
nvme ssd
Storages
  • 400 GB nvme ssd
NETWORK
Network Speed Baseline
8 Gbps
Network Speed Max
8 Gbps
Inbound Traffic
0 GB/month
Outbound Traffic
0 GB/month
IPv4
1

CPU and System Topology

Server Description

A dedicated AMD EPYC and NVIDIA L40S GPU server delivering top-tier LLM inference performance and balanced resource density for accelerated workloads.

GPU AcceleratedCompute OptimizedStorage & Database

The OVHcloud l40s-90 is a KVM-based Cloud GPU server designed for high-performance accelerated computing. It features an x86_64 AMD EPYC 9124 processor with 15 dedicated physical cores running at 2.75 GHz, paired with 90.0 GB of system memory and 400 GB of local NVMe SSD storage. The server is equipped with a single NVIDIA L40S GPU based on the Ada Lovelace architecture, providing 48 GB of VRAM. In benchmark testing, the l40s-90 achieves top-tier performance in LLM inference, prompt processing, and text generation across various model sizes, alongside strong single-core CPU performance. With an 8 Gbps baseline network bandwidth, this configuration is highly efficient for machine learning inference, ray tracing, and GPU-accelerated data processing. It offers a cost-effective alternative to multi-GPU nodes by matching a high-capacity GPU with a balanced, dedicated CPU and memory allocation.

Economics

Average Price per Region

Prices per Zone

Lowest Prices

Performance

Workload Profiles

0.83Score
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) 1,226,740 ops/sec 1,918,343 ops/sec 0.63 50.00% 50.00% -21.00%
Redis RPS (pipeline=16, SET) 12,975,621 ops/sec 11,704,734 ops/sec 1.1 20.00% 20.00% +1.90%
PassMark Memory Mark (composite) 2,331 2,416 0.96 10.00% 10.00% -0.41%
Memory bandwidth (read, 16 MB ~ L3) 118,954 MB/sec 108,466 MB/sec 1.1 10.00% 10.00% +0.96%
PassMark single-thread CPU 2,854 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
22.6 GB/s
Node GPU-to-host bandwidth
26.4 GB/s
Node host-to-GPU bandwidth
26.8 GB/s
Single GPU slot
Per-GPU GPU-to-host bandwidth
26.4 GB/s
Per-GPU host-to-GPU bandwidth
26.8 GB/s
Per-GPU host-device latency
678 ns
Benchmark Environment
Driver Version
 535.274.02
CUDA Version (driver)
 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: 1,941

Geekbench Multi-Core

Score: 11,588

Passmark CPU Scores

BENCHMARKSCORE
Mark
 34,459
Compression
 384,470
Encryption
 19,427
Extended Instructions
 36,664
Floating Point Maths
 78,812
Integer Maths
 89,932
Physics
 4,583
Prime Numbers
 286
Single Threaded
 2,854
String Sorting
 56,562

Passmark Memory Scores

BENCHMARKSCORE
Memory Mark
 2,331
Database Operations
 11,733
Memory Latency
 99
Memory Read Cached
 27,260
Memory Read Uncached
 24,828
Memory Write
 23,680

Stress-ng Stressors

CPU integerSingleMultiScaling
atomic
 719.92 557.56 0.77×
bitops
 159,682 2,371,935 14.85×
branch
 213,185 3,156,986 14.81×
funccall
 617.2 9,158 14.84×
funcret
 20,852 316,108 15.16×
hash
 162,439 2,405,463 14.81×
intmath
 66,114 987,302 14.93×
nop
 624,121 9,241,691 14.81×
prime
 475,438 7,054,100 14.84×
rotate
 86,862 1,297,925 14.94×
str
 1,081,219 16,087,452 14.88×
wcs
 688,425 10,318,585 14.99×
CryptographySingleMultiScaling
crypt
 3,709 54,720 14.75×
CompressionSingleMultiScaling
jpeg
 818.48 11,883 14.52×
zlib
 70.73 853.78 12.07×
HypervisorSingleMultiScaling
clock
 25,312 145,573 5.75×
clone
 193.8 1,711 8.83×
context
 2,839 42,300 14.90×
cpu-sched
 69.07 943.59 13.66×
eventfd
 37,100 487,957 13.15×
fault
 9,720 89,254 9.18×
fork
 1,515 22,846 15.08×
getrandom
 50,102 748,515 14.94×
hrtimers
 565,166 1,484,851 2.63×
mprotect
 400,928 2,750,342 6.86×
mtx
 36,112,702 1,625,084,229 45.00×
munmap
 1,067 19,410 18.20×
null
 331,594 4,970,334 14.99×
pipe
 222,663 5,004,208 22.47×
priv-instr
 409,591 4,891,775 11.94×
resched
 4,646,436 2,530,217 0.54×
switch
 79,413 1,899,448 23.92×
timermix
 139,577 1,860,678 13.33×
tlb-shootdown
 12,493 60,035 4.81×
tsc
 780,785 11,603,723 14.86×
usersyscall
 314,536 4,074,960 12.96×
vdso
 82,140,304 1,219,320,838 14.84×
vfork
 13,827 105,288 7.61×
x86syscall
 3,814,969 56,643,084 14.85×
yield
 8,717,490 8,734,481 1.00×
zero
 2,658,822 39,454,966 14.84×
CPU float/SIMDSingleMultiScaling
eigen
 16,495 241,192 14.62×
expmath
 3,113 46,238 14.85×
fma
 917,190 13,616,357 14.85×
fp
 251.22 3,730 14.85×
fractal
 7.4 109.03 14.73×
hyperbolic
 3,807 56,111 14.74×
logmath
 4,527 67,483 14.91×
matrix
 4,916 73,016 14.85×
matrix-3d
 855.92 4,309 5.03×
monte-carlo
 161.75 2,425 14.99×
mpfr
 1,690 25,100 14.85×
powmath
 1,329 19,751 14.87×
trig
 1,865 27,683 14.84×
veccmp
 19,012 281,005 14.78×
vecfp
 1,805 26,938 14.92×
vecmath
 4,013 59,527 14.83×
vecshuf
 636.44 9,450 14.85×
vecwide
 28,527 422,809 14.82×
AlgorithmsSingleMultiScaling
bitonicsort
 10.2 154.19 15.12×
bsearch
 303.89 4,547 14.96×
bubblesort
 0.8 12.2 15.25×
fibsearch
 376.82 5,607 14.88×
heapsort
 10.8 159.79 14.80×
hsearch
 11,079 168,734 15.23×
insertionsort
 12.6 184.2 14.62×
malloc
 3,172,592 22,772,539 7.18×
mcontend
 53.62 1,875 34.96×
mergesort
 33.6 506.58 15.08×
qsort
 21.8 323.19 14.83×
skiplist
 7,006 103,814 14.82×
tree
 1.8 24.79 13.77×
MemorySingleMultiScaling
cache
 359,490 530,371 1.48×
cachehammer
 123,395 2,413,910 19.56×
cacheline
 43.15 563.42 13.06×
l1cache
 417.07 6,416 15.38×
memcpy
 444.22 6,619 14.90×
memrate
 3.53 9.07 2.57×
memthrash
 170.14 633.38 3.72×
ptr-chase
 13,586,383 152,452,701 11.22×
spinmem
 232.58 776.29 3.34×
stream
 253.57 10,577 41.71×
vm
 58,250 293,009 5.03×

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
rtx5000-28 428 GiB1
t1-le-45 845 GiB1
rtx5000-56 856 GiB2
t2-le-45 1545 GiB1
a100-180 15180 GiB1
t1-le-90 1690 GiB2
rtx5000-84 1684 GiB3

Similar Servers

INSTANCEVENDORvCPUsMEMORYAccelerators

l40s-90 FAQs