m3-ultramem-32 by Google Cloud Platform

m3-ultramem-32 is a 32 vCPUs, 976 GB RAM server offered by Google Cloud Platform with 32 vCPUs, 976 GiB of memory and 0 GB of storage. The pricing starts at 0.5728 USD per hour.
32 vCPU
976 GiB Memory
Spare SCore
30,212
(All-cores)
1,844
(Single-core)

Specifications

Server Metadata

Vendor ID
gcp
Server ID
1410032
Name
m3-ultramem-32
Description
32 vCPUs, 976 GB RAM
Family
m3
Hw Virt
Average Time To Start
23.5
Status
active
Observed At
2026-10-21T21:59:31.160862

Availability

REGION / IDSPOTONDEMAND
Ashburn (US) / us-east4
1.1252 USD/h 6.0713 USD/h
Council Bluffs (US) / us-central1
2.8202 USD/h 6.0912 USD/h
The Dalles (US) / us-west1
2.8202 USD/h 6.0912 USD/h
Moncks Corner (US) / us-east1
2.8202 USD/h 6.0912 USD/h
Mumbai (IN) / asia-south1
0.6305 USD/h 6.3407 USD/h
Delhi (IN) / asia-south2
0.5728 USD/h 6.3409 USD/h
Bangkok (TH) / asia-southeast3
2.8987 USD/h 6.4006 USD/h
Eemshaven (NL) / europe-west4
0.576 USD/h 6.4022 USD/h
Tel Aviv (IL) / me-west1
1.5037 USD/h 6.7003 USD/h
Hamina (FI) / europe-north1
0.6033 USD/h 6.707 USD/h
Toronto (CA) / northamerica-northeast2
0.6033 USD/h 6.707 USD/h
Montréal (CA) / northamerica-northeast1
0.6059 USD/h 6.707 USD/h
St. Ghislain (BE) / europe-west1
1.418 USD/h 6.707 USD/h
Johannesburg (ZA) / africa-south1
1.4527 USD/h 6.707 USD/h
Las Vegas (US) / us-west4
0.7142 USD/h 6.8605 USD/h
London (GB) / europe-west2
0.6247 USD/h 6.9354 USD/h
Changhua County (TW) / asia-east1
0.6403 USD/h 7.053 USD/h
Paris (FR) / europe-west9
0.6364 USD/h 7.0658 USD/h
Milan (IT) / europe-west8
0.6373 USD/h 7.0658 USD/h
Frankfurt (DE) / europe-west3
0.6639 USD/h 7.1787 USD/h
Madrid (ES) / europe-southwest1
0.6801 USD/h 7.1876 USD/h
Salt Lake City (US) / us-west3
1.2286 USD/h 7.3222 USD/h
Warsaw (PL) / europe-central2
1.9457 USD/h 7.37 USD/h
Doha (QA) / me-central1
1.3438 USD/h 7.4008 USD/h
Jurong West (SG) / asia-southeast1
0.6766 USD/h 7.5147 USD/h
Sydney (AU) / australia-southeast1
0.6857 USD/h 7.6113 USD/h
Tokyo (JP) / asia-northeast1
0.7035 USD/h 7.8223 USD/h
Osaka (JP) / asia-northeast2
0.7035 USD/h 7.8223 USD/h
Seoul (KR) / asia-northeast3
2.332 USD/h 7.8223 USD/h
Turin (IT) / europe-west12
0.7116 USD/h 7.8576 USD/h
Melbourne (AU) / australia-southeast2
1.5731 USD/h 7.9158 USD/h
Zurich (CH) / europe-west6
0.7181 USD/h 7.9799 USD/h
Santiago (CL) / southamerica-west1
0.7836 USD/h 8.7104 USD/h
Osasco (BR) / southamerica-east1
0.8735 USD/h 9.6722 USD/h
Dammam (SA) / me-central2
0.8838 USD/h 9.7459 USD/h

Processor

vCPUs
32
CPU Allocation
Dedicated
CPU Cores
16
CPU Speed
2.6 GHz
CPU Architecture
x86_64
CPU Manufacturer
Intel
CPU Family
Xeon
CPU L1D Cache
48 KiB
CPU L1D Cache Total
768 KiB
CPU L1I Cache
32 KiB
CPU L1I Cache Total
512 KiB
CPU L2 Cache
1 MiB
CPU L2 Cache Total
20 MiB
CPU L3 Cache
54 MiB
CPU L3 Cache Total
54 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, avx512f, avx512dq, rdseed, adx, smap, avx512ifma, clflushopt, clwb, avx512cd, sha_ni, avx512bw, avx512vl, xsaveopt, xsavec, xgetbv1, xsaves, arat, avx512vbmi, umip, avx512_vbmi2, gfni, vaes, vpclmulqdq, avx512_vnni, avx512_bitalg, avx512_vpopcntdq, rdpid, fsrm, md_clear, arch_capabilities
Ecpus
16.4
Scalability
51.25

System Resources and Accelerators

MEMORY
Memory Amount
976 GiB
Memory Amount Actual
976 GiB
ACCELERATORS
Accelerator Count
0
Accelerator Memory Min
0 MiB
Accelerator Memory Total
0 MiB
Accelerators
    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

      A memory-optimized server featuring massive RAM capacity per vCPU, designed for large in-memory databases and memory-intensive enterprise workloads.

      Memory Optimized

      The Google Cloud Platform m3-ultramem-32 is an x86_64 memory-optimized server configured with 32 vCPUs and 976 GB of RAM, yielding a high memory-to-core ratio of 30.5 GB per vCPU. Powered by a 2.6 GHz Intel Xeon processor with 16 physical cores, this instance lacks local storage and GPU accelerators. Benchmark data indicates average performance across standard compute and memory tasks, with a PassMark CPU score of 35890.88 and Redis SET operations reaching 30.48 million operations per second. While it handles small-scale language model text generation efficiently, performance drops significantly on larger models, such as a poor 2.45 tokens per second for 70B model text generation. This hardware profile is qualitatively cost-effective for memory-bound applications, making it highly suitable for large in-memory databases, caching layers, and memory-intensive enterprise workloads that do not require local NVMe storage or GPU acceleration.

      Economics

      Average Price per Region

      Prices per Zone

      Lowest Prices

      Performance

      Workload Profiles

      1.7Score
      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) 4,220,585 ops/sec 1,918,343 ops/sec 2.2 50.00% 50.00% +48.00%
      Redis RPS (pipeline=16, SET) 21,171,580 ops/sec 11,704,734 ops/sec 1.8 20.00% 20.00% +12.00%
      PassMark Memory Mark (composite) 2,710 2,416 1.1 10.00% 10.00% +0.96%
      Memory bandwidth (read, 16 MB ~ L3) 152,201 MB/sec 108,420 MB/sec 1.4 10.00% 10.00% +3.40%
      PassMark single-thread CPU 2,351 Mops/s 2,365 Mops/s 1 10.00% 10.00% 0.00%

      Memory Bandwidth

      Compression

      OpenSSL

      Geekbench Single-Core

      Score: 1,570

      Geekbench Multi-Core

      Score: 13,551

      Passmark CPU Scores

      BENCHMARKSCORE
      Mark
      		 35,891
      Compression
      		 478,087
      Encryption
      		 28,441
      Extended Instructions
      		 27,254
      Floating Point Maths
      		 66,947
      Integer Maths
      		 130,876
      Physics
      		 3,292
      Prime Numbers
      		 194
      Single Threaded
      		 2,351
      String Sorting
      		 67,005

      Passmark Memory Scores

      BENCHMARKSCORE
      Memory Mark
      		 2,710
      Database Operations
      		 15,229
      Memory Latency
      		 55
      Memory Read Cached
      		 23,438
      Memory Read Uncached
      		 11,857
      Memory Write
      		 10,963

      Stress-ng Stressors

      CPU integerSingleMultiScaling
      atomic
      		 267.86 279.65 1.04×
      bitops
      		 127,062 2,069,275 16.29×
      branch
      		 133,665 3,394,206 25.39×
      funccall
      		 618.4 10,091 16.32×
      funcret
      		 11,475 258,015 22.48×
      hash
      		 146,924 3,408,004 23.20×
      intmath
      		 59,600 1,178,398 19.77×
      nop
      		 183,682 2,601,967 14.17×
      prime
      		 383,931 6,686,883 17.42×
      rotate
      		 95,198 1,620,339 17.02×
      str
      		 874,351 14,773,174 16.90×
      wcs
      		 598,136 8,448,796 14.13×
      CryptographySingleMultiScaling
      crypt
      		 2,728 40,959 15.01×
      CompressionSingleMultiScaling
      jpeg
      		 710.01 13,995 19.71×
      zlib
      		 52.24 1,354 25.92×
      HypervisorSingleMultiScaling
      clock
      		 42,868 446,760 10.42×
      clone
      		 478.88 3,940 8.23×
      context
      		 4,616 111,171 24.09×
      cpu-sched
      		 66.54 2,082 31.29×
      eventfd
      		 59,699 1,115,436 18.68×
      fault
      		 17,969 294,236 16.37×
      fork
      		 5,908 86,200 14.59×
      getrandom
      		 48,047 879,992 18.32×
      hrtimers
      		 512,258 4,564,524 8.91×
      mprotect
      		 1,363,598 8,421,292 6.18×
      mtx
      		 13,268,145 974,469,281 73.44×
      munmap
      		 4,195 64,784 15.44×
      null
      		 724,236 14,563,376 20.11×
      pipe
      		 919,301 20,251,865 22.03×
      priv-instr
      		 741,871 14,235,607 19.19×
      resched
      		 11,473,442 3,230,140 0.28×
      switch
      		 91,000 7,589,025 83.40×
      timermix
      		 207,457 5,348,780 25.78×
      tlb-shootdown
      		 12,517 59,133 4.72×
      tsc
      		 747,983 17,648,131 23.59×
      usersyscall
      		 622,763 12,036,092 19.33×
      vdso
      		 106,914,583 1,977,494,992 18.50×
      vfork
      		 37,223 281,492 7.56×
      x86syscall
      		 8,330,746 184,380,612 22.13×
      yield
      		 24,583,290 28,972,023 1.18×
      zero
      		 4,374,417 80,230,371 18.34×
      CPU float/SIMDSingleMultiScaling
      eigen
      		 14,431 297,506 20.62×
      expmath
      		 2,808 65,666 23.39×
      fma
      		 622,144 17,295,020 27.80×
      fp
      		 187.2 3,210 17.15×
      fractal
      		 4.8 114.42 23.84×
      hyperbolic
      		 2,988 70,791 23.69×
      logmath
      		 4,723 109,209 23.12×
      matrix
      		 3,409 82,950 24.34×
      matrix-3d
      		 776.99 5,068 6.52×
      monte-carlo
      		 247.43 5,471 22.11×
      mpfr
      		 1,375 23,489 17.08×
      powmath
      		 1,642 39,134 23.84×
      trig
      		 1,917 38,938 20.32×
      veccmp
      		 19,802 305,651 15.44×
      vecfp
      		 1,486 29,000 19.51×
      vecmath
      		 3,145 72,754 23.13×
      vecshuf
      		 647.88 11,247 17.36×
      vecwide
      		 31,063 573,494 18.46×
      AlgorithmsSingleMultiScaling
      bitonicsort
      		 7.8 128 16.41×
      bsearch
      		 241.66 5,255 21.74×
      bubblesort
      		 1 12.8 12.80×
      fibsearch
      		 280.33 5,532 19.74×
      heapsort
      		 9.2 172.8 18.78×
      hsearch
      		 6,917 127,312 18.41×
      insertionsort
      		 10.2 96 9.41×
      malloc
      		 2,705,103 35,833,105 13.25×
      mcontend
      		 116.12 6,644 57.21×
      mergesort
      		 26.2 582.59 22.24×
      qsort
      		 20.2 335.59 16.61×
      skiplist
      		 5,292 123,799 23.39×
      tree
      		 1.6 25.37 15.86×
      MemorySingleMultiScaling
      cache
      		 413,885 771,697 1.86×
      cachehammer
      		 374,795 9,601,053 25.62×
      cacheline
      		 90.95 3,763 41.37×
      l1cache
      		 895.46 5,872 6.56×
      memcpy
      		 394.95 6,229 15.77×
      memrate
      		 1.78 12.15 6.83×
      memthrash
      		 231.89 990.41 4.27×
      ptr-chase
      		 35,885,952 306,814,525 8.55×
      spinmem
      		 226.59 1,343 5.93×
      stream
      		 36.41 8,695 238.81×
      vm
      		 113,943 297,473 2.61×

      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
      m3-megamem-64 64976 GiB0
      m3-ultramem-64 641952 GiB0
      m3-megamem-128 1281952 GiB0
      m3-ultramem-128 1283904 GiB0

      Similar Servers

      INSTANCEVENDORvCPUsMEMORYAccelerators

      m3-ultramem-32 FAQs