[アップデート] Intel Xeon 6を搭載したM8i、M8i-flexインスタンスが追加されました

[アップデート] Intel Xeon 6を搭載したM8i、M8i-flexインスタンスが追加されました

ワークロードに応じて適切なインスタンスタイプを選択しよう
2025.08.31

最新世代のIntel Xeonプロセッサーを搭載したEC2インスタンスを使用したい

こんにちは、のんピ(@non____97)です。

皆さんは最新世代のIntel Xeonプロセッサーを搭載したEC2インスタンスを使用したいなと思ったことはありますか? 私はあります。

インスタンスファミリー一覧を見ていると、C8gやM8gはあるのにC8iやM8iはないのかと気になった方もいらっしゃるのではないでしょうか。

今回、最新世代のIntel Xeon 6(Granite Rapids)プロセッサーを搭載したM8i及びM8i-flexのインスタンスが使用できるようになりました。

https://aws.amazon.com/about-aws/whats-new/2025/08/amazon-ec2-m8i-and-m8i-flex-instances-generally-available/

前世代と比較して最大15%以上コストパフォーマンスが向上しているようです。また、メモリ帯域幅に至っては2.5倍と大幅に増えています。

Amazon EC2 M8i and M8i-flex deliver the highest performance and fastest memory bandwidth among comparable Intel processors in the cloud. The M8i and M8i-flex instances offer up to 15% better price-performance, and 2.5x more memory bandwidth compared to previous generation Intel-based instances. They deliver 20% better performance than M7i and M7i-flex instances, with even higher gains for specific workloads. They are up to 30% faster for PostgreSQL databases, up to 60% faster for NGINX web applications, and up to 40% faster for AI deep learning recommendation models compared to M7i and M7i-flex.

Amazon EC2 M8i and M8i-flex Instances

使用できるリージョンは

  • us-east-1 : バージニア北部
  • us-east-2 : オハイオ
  • us-west-2 : オレゴン
  • eu-south-2 : スペイン

と日本のリージョンは現時点は含まれていません。

気になったので色々確認してみます。

dmesg

まずはdmesgです。

インスタンスタイプ以外は大きな違いはみられませんでした。

細かいところだと以下に差異が見られました。

  • m8i.largeではkvm-guest: PV spinlocks disabled, no host support、m8i-flex.largeではkvm-guest: PV spinlocks enabled
  • m8i.largeではPerformance Events: AnyThread deprecated, generic architected perfmon, full-width counters, Intel PMU driver.、m8i-flex.largeではPerformance Events: unsupported p6 CPU model 173 no PMU driver, software events only.

m8i.large

$ sudo dmesg
[    0.000000] Linux version 6.1.144-170.251.amzn2023.x86_64 (mockbuild@ip-10-0-47-63) (gcc (GCC) 11.5.0 20240719 (Red Hat 11.5.0-5), GNU ld version 2.41-50.amzn2023.0.3) #1 SMP PREEMPT_DYNAMIC Tue Jul 15 13:53:09 UTC 2025
[    0.000000] Command line: BOOT_IMAGE=(hd0,gpt1)/boot/vmlinuz-6.1.144-170.251.amzn2023.x86_64 root=UUID=852bff28-b4ae-47f9-ae3b-21cb21646c50 ro console=tty0 console=ttyS0,115200n8 nvme_core.io_timeout=4294967295 rd.emergency=poweroff rd.shell=0 selinux=1 security=selinux quiet
[    0.000000] KASLR enabled
[    0.000000] x86/tme: enabled by BIOS
[    0.000000] x86/tme: Unknown policy is active: 0x2
[    0.000000] x86/mktme: No known encryption algorithm is supported: 0x0
[    0.000000] x86/mktme: disabled by BIOS
[    0.000000] x86/split lock detection: #DB: warning on user-space bus_locks
[    0.000000] BIOS-provided physical RAM map:
[    0.000000] BIOS-e820: [mem 0x0000000000000000-0x000000000009ffff] usable
.
.
(中略)
.
.
[    0.000000] BIOS-e820: [mem 0x0000000240100000-0x0000000242afffff] reserved
[    0.000000] NX (Execute Disable) protection: active
[    0.000000] efi: EFI v2.70 by EDK II
[    0.000000] efi: ACPI=0xbfb7e000 ACPI 2.0=0xbfb7e014 SMBIOS=0xbfa8b000 MEMATTR=0xbe217118
[    0.000000] SMBIOS 2.7 present.
[    0.000000] DMI: Amazon EC2 m8i.large/, BIOS 1.0 10/16/2017
[    0.000000] Hypervisor detected: KVM
[    0.000000] kvm-clock: Using msrs 4b564d01 and 4b564d00
[    0.000000] kvm-clock: using sched offset of 1507858603 cycles
[    0.000002] clocksource: kvm-clock: mask: 0xffffffffffffffff max_cycles: 0x1cd42e4dffb, max_idle_ns: 881590591483 ns
[    0.000004] tsc: Detected 2700.000 MHz processor
[    0.000095] e820: update [mem 0x00000000-0x00000fff] usable ==> reserved
[    0.000096] e820: remove [mem 0x000a0000-0x000fffff] usable
[    0.000100] last_pfn = 0x236000 max_arch_pfn = 0x400000000
[    0.000139] x86/PAT: Configuration [0-7]: WB  WC  UC- UC  WB  WP  UC- WT
[    0.000145] last_pfn = 0xbff7c max_arch_pfn = 0x400000000
[    0.004719] Using GB pages for direct mapping
[    0.004913] Secure boot disabled
[    0.004913] RAMDISK: [mem 0xbc8b6000-0xbd861fff]
[    0.004917] ACPI: Early table checksum verification disabled
[    0.004919] ACPI: RSDP 0x00000000BFB7E014 000024 (v02 AMAZON)
.
.
(中略)
.
.
[    0.013234] setup_percpu: NR_CPUS:8192 nr_cpumask_bits:2 nr_cpu_ids:2 nr_node_ids:1
.
.
(中略)
.
.
[    0.040087] rcu: Preemptible hierarchical RCU implementation.
[    0.040088] rcu:     RCU restricting CPUs from NR_CPUS=8192 to nr_cpu_ids=2.
[    0.040089]  Trampoline variant of Tasks RCU enabled.
[    0.040089]  Rude variant of Tasks RCU enabled.
[    0.040089]  Tracing variant of Tasks RCU enabled.
[    0.040090] rcu: RCU calculated value of scheduler-enlistment delay is 10 jiffies.
[    0.040090] rcu: Adjusting geometry for rcu_fanout_leaf=16, nr_cpu_ids=2
[    0.043304] NR_IRQS: 524544, nr_irqs: 440, preallocated irqs: 16
[    0.043493] rcu: srcu_init: Setting srcu_struct sizes based on contention.
[    0.043532] Console: colour dummy device 80x25
[    0.043542] printk: console [tty0] enabled
[    0.044472] printk: console [ttyS0] enabled
[    0.044479] ACPI: Core revision 20220331
[    0.044642] clocksource: hpet: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 30580167144 ns
[    0.044657] APIC: Switch to symmetric I/O mode setup
[    0.045031] x2apic enabled
[    0.045387] Switched APIC routing to physical x2apic.
[    0.045389] kvm-guest: setup PV IPIs
[    0.046879] clocksource: tsc-early: mask: 0xffffffffffffffff max_cycles: 0x26eb3e3cc63, max_idle_ns: 440795236593 ns
[    0.046882] Calibrating delay loop (skipped) preset value.. 5400.00 BogoMIPS (lpj=27000000)
[    0.046986] x86/cpu: User Mode Instruction Prevention (UMIP) activated
[    0.047122] process: using mwait in idle threads
[    0.047125] Last level iTLB entries: 4KB 0, 2MB 0, 4MB 0
[    0.047126] Last level dTLB entries: 4KB 0, 2MB 0, 4MB 0, 1GB 0
[    0.047131] Spectre V1 : Mitigation: usercopy/swapgs barriers and __user pointer sanitization
[    0.047135] Spectre V2 : Mitigation: Enhanced / Automatic IBRS
[    0.047136] Spectre V2 : mitigation: Enabling conditional Indirect Branch Prediction Barrier
[    0.047137] Speculative Store Bypass: Mitigation: Speculative Store Bypass disabled via prctl
[    0.047168] x86/fpu: Supporting XSAVE feature 0x001: 'x87 floating point registers'
[    0.047169] x86/fpu: Supporting XSAVE feature 0x002: 'SSE registers'
[    0.047170] x86/fpu: Supporting XSAVE feature 0x004: 'AVX registers'
[    0.047170] x86/fpu: Supporting XSAVE feature 0x020: 'AVX-512 opmask'
[    0.047171] x86/fpu: Supporting XSAVE feature 0x040: 'AVX-512 Hi256'
[    0.047171] x86/fpu: Supporting XSAVE feature 0x080: 'AVX-512 ZMM_Hi256'
[    0.047172] x86/fpu: Supporting XSAVE feature 0x200: 'Protection Keys User registers'
[    0.047173] x86/fpu: Supporting XSAVE feature 0x20000: 'AMX Tile config'
[    0.047173] x86/fpu: Supporting XSAVE feature 0x40000: 'AMX Tile data'
[    0.047174] x86/fpu: xstate_offset[2]:  576, xstate_sizes[2]:  256
[    0.047175] x86/fpu: xstate_offset[5]:  832, xstate_sizes[5]:   64
[    0.047175] x86/fpu: xstate_offset[6]:  896, xstate_sizes[6]:  512
[    0.047176] x86/fpu: xstate_offset[7]: 1408, xstate_sizes[7]: 1024
[    0.047177] x86/fpu: xstate_offset[9]: 2432, xstate_sizes[9]:    8
[    0.047177] x86/fpu: xstate_offset[17]: 2496, xstate_sizes[17]:   64
[    0.047178] x86/fpu: xstate_offset[18]: 2560, xstate_sizes[18]: 8192
[    0.047179] x86/fpu: Enabled xstate features 0x602e7, context size is 10752 bytes, using 'compacted' format.
[    0.056881] Freeing SMP alternatives memory: 36K
[    0.056881] pid_max: default: 32768 minimum: 301
[    0.056881] LSM: Security Framework initializing
[    0.056881] Yama: becoming mindful.
[    0.056881] SELinux:  Initializing.
[    0.056881] LSM support for eBPF active
[    0.056881] Mount-cache hash table entries: 16384 (order: 5, 131072 bytes, linear)
[    0.056881] Mountpoint-cache hash table entries: 16384 (order: 5, 131072 bytes, linear)
[    0.056881] smpboot: CPU0: Intel(R) Xeon(R) 6975P-C (family: 0x6, model: 0xad, stepping: 0x1)
[    0.056881] cblist_init_generic: Setting adjustable number of callback queues.
[    0.056881] cblist_init_generic: Setting shift to 1 and lim to 1.
[    0.056881] cblist_init_generic: Setting adjustable number of callback queues.
[    0.056881] cblist_init_generic: Setting shift to 1 and lim to 1.
[    0.056881] cblist_init_generic: Setting adjustable number of callback queues.
[    0.056881] cblist_init_generic: Setting shift to 1 and lim to 1.
[    0.056881] Performance Events:  AnyThread deprecated, generic architected perfmon, full-width counters, Intel PMU driver.
[    0.056881] ... version:                5
[    0.056881] ... bit width:              48
[    0.056881] ... generic registers:      8
[    0.056881] ... value mask:             0000ffffffffffff
[    0.056881] ... max period:             00007fffffffffff
[    0.056881] ... fixed-purpose events:   2
[    0.056881] ... event mask:             00000005000000ff
[    0.056881] signal: max sigframe size: 11952
[    0.056881] Estimated ratio of average max frequency by base frequency (times 1024): 1479
[    0.056881] rcu: Hierarchical SRCU implementation.
[    0.056881] rcu:     Max phase no-delay instances is 1000.
[    0.056881] smp: Bringing up secondary CPUs ...
[    0.056881] x86: Booting SMP configuration:
[    0.056881] .... node  #0, CPUs:      #1
[    0.056881] smp: Brought up 1 node, 2 CPUs
[    0.056881] smpboot: Max logical packages: 1
[    0.056881] smpboot: Total of 2 processors activated (10800.00 BogoMIPS)
[    0.056881] node 0 deferred pages initialised in 0ms
[    0.056881] devtmpfs: initialized
[    0.056881] x86/mm: Memory block size: 128MB
[    0.056881] ACPI: PM: Registering ACPI NVS region [mem 0xbfb7f000-0xbfbfefff] (524288 bytes)
[    0.056881] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[    0.056881] futex hash table entries: 512 (order: 3, 32768 bytes, linear)
[    0.056951] NET: Registered PF_NETLINK/PF_ROUTE protocol family
[    0.057010] DMA: preallocated 1024 KiB GFP_KERNEL pool for atomic allocations
[    0.057014] DMA: preallocated 1024 KiB GFP_KERNEL|GFP_DMA pool for atomic allocations
[    0.057019] DMA: preallocated 1024 KiB GFP_KERNEL|GFP_DMA32 pool for atomic allocations
[    0.057025] audit: initializing netlink subsys (disabled)
[    0.057029] audit: type=2000 audit(1756560848.034:1): state=initialized audit_enabled=0 res=1
.
.
(中略)
.
.
[    0.212362] pcieport 0000:26:00.3: pciehp: Slot #37 AttnBtn+ PwrCtrl+ MRL- AttnInd+ PwrInd+ HotPlug+ Surprise- Interlock- NoCompl+ IbPresDis- LLActRep+
[    0.212840] shpchp: Standard Hot Plug PCI Controller Driver version: 0.4
[    0.213039] Serial: 8250/16550 driver, 4 ports, IRQ sharing disabled
[    0.213237] 00:04: ttyS0 at I/O 0x3f8 (irq = 4, base_baud = 115200) is a 16550A
[    0.213953] AMD-Vi: AMD IOMMUv2 functionality not available on this system - This is not a bug.
[    0.214431] nvme nvme0: pci function 0000:00:04.0
[    0.215876] rtc_cmos 00:00: RTC can wake from S4
[    0.216808] rtc_cmos 00:00: registered as rtc0
[    0.216957] rtc_cmos 00:00: setting system clock to 2025-08-30T13:34:06 UTC (1756560846)
[    0.217014] rtc_cmos 00:00: alarms up to one day, 114 bytes nvram
[    0.217083] vmclock AMZNC10C:00: vmclock0: registered miscdev
[    0.217092] intel_pstate: CPU model not supported
[    0.217110] pstore: Registered efi as persistent store backend
[    0.217119] hid: raw HID events driver (C) Jiri Kosina
[    0.222528] Freeing initrd memory: 16048K
[    0.226434] NET: Registered PF_INET6 protocol family
[    0.229918] Segment Routing with IPv6
[    0.229926] In-situ OAM (IOAM) with IPv6
[    0.229974] NET: Registered PF_PACKET protocol family
[    0.230416] IPI shorthand broadcast: enabled
[    0.230422] AVX2 version of gcm_enc/dec engaged.
[    0.230424] nvme nvme0: 2/0/0 default/read/poll queues
[    0.230455] AES CTR mode by8 optimization enabled
[    0.230497] sched_clock: Marking stable (226779790, 3382772)->(246652068, -16489506)
[    0.230593] registered taskstats version 1
[    0.230599] Loading compiled-in X.509 certificates
[    0.235788] Loaded X.509 cert 'Amazon.com: Amazon Linux Kernel Signing Key: 9a007247b45e883e4aae0fcf7a6612b12f5fcb39'
[    0.235842] zswap: loaded using pool lzo/zbud
[    0.235904] Key type .fscrypt registered
[    0.235905] Key type fscrypt-provisioning registered
[    0.236011] pstore: Using crash dump compression: deflate
[    0.236176] ima: No TPM chip found, activating TPM-bypass!
[    0.236182] ima: Allocated hash algorithm: sha256
[    0.236189] ima: No architecture policies found
[    0.237591]  nvme0n1: p1 p127 p128
[    0.314172] clk: Disabling unused clocks
[    0.318672] Freeing unused decrypted memory: 2036K
[    0.318979] Freeing unused kernel image (initmem) memory: 2448K
[    0.318980] Write protecting the kernel read-only data: 26624k
[    0.319296] Freeing unused kernel image (text/rodata gap) memory: 2036K
[    0.319465] Freeing unused kernel image (rodata/data gap) memory: 1336K
[    0.319469] Run /init as init process
.
.
(以下略)
.
.

m8i-flex.large

$ sudo dmesg
[    0.000000] Linux version 6.1.144-170.251.amzn2023.x86_64 (mockbuild@ip-10-0-47-63) (gcc (GCC) 11.5.0 20240719 (Red Hat 11.5.0-5), GNU ld version 2.41-50.amzn2023.0.3) #1 SMP PREEMPT_DYNAMIC Tue Jul 15 13:53:09 UTC 2025
[    0.000000] Command line: BOOT_IMAGE=(hd0,gpt1)/boot/vmlinuz-6.1.144-170.251.amzn2023.x86_64 root=UUID=852bff28-b4ae-47f9-ae3b-21cb21646c50 ro console=tty0 console=ttyS0,115200n8 nvme_core.io_timeout=4294967295 rd.emergency=poweroff rd.shell=0 selinux=1 security=selinux quiet
[    0.000000] KASLR enabled
[    0.000000] x86/tme: enabled by BIOS
[    0.000000] x86/tme: Unknown policy is active: 0x2
[    0.000000] x86/mktme: No known encryption algorithm is supported: 0x0
[    0.000000] x86/mktme: disabled by BIOS
[    0.000000] x86/split lock detection: #DB: warning on user-space bus_locks
[    0.000000] BIOS-provided physical RAM map:
[    0.000000] BIOS-e820: [mem 0x0000000000000000-0x000000000009ffff] usable
.
.
(中略)
.
.
[    0.000000] BIOS-e820: [mem 0x0000000240100000-0x0000000242afffff] reserved
[    0.000000] NX (Execute Disable) protection: active
[    0.000000] efi: EFI v2.70 by EDK II
[    0.000000] efi: ACPI=0xbfb7e000 ACPI 2.0=0xbfb7e014 SMBIOS=0xbfa8b000 MEMATTR=0xbe217118
[    0.000000] SMBIOS 2.7 present.
[    0.000000] DMI: Amazon EC2 m8i-flex.large/, BIOS 1.0 10/16/2017
[    0.000000] Hypervisor detected: KVM
[    0.000000] kvm-clock: Using msrs 4b564d01 and 4b564d00
[    0.000001] kvm-clock: using sched offset of 1638729323 cycles
[    0.000002] clocksource: kvm-clock: mask: 0xffffffffffffffff max_cycles: 0x1cd42e4dffb, max_idle_ns: 881590591483 ns
[    0.000004] tsc: Detected 2700.000 MHz processor
[    0.000097] e820: update [mem 0x00000000-0x00000fff] usable ==> reserved
[    0.000099] e820: remove [mem 0x000a0000-0x000fffff] usable
[    0.000102] last_pfn = 0x236000 max_arch_pfn = 0x400000000
[    0.000140] x86/PAT: Configuration [0-7]: WB  WC  UC- UC  WB  WP  UC- WT
[    0.000146] last_pfn = 0xbff7c max_arch_pfn = 0x400000000
[    0.004760] Using GB pages for direct mapping
[    0.004985] Secure boot disabled
[    0.004985] RAMDISK: [mem 0xbc8b6000-0xbd861fff]
[    0.004989] ACPI: Early table checksum verification disabled
[    0.004990] ACPI: RSDP 0x00000000BFB7E014 000024 (v02 AMAZON)
.
.
(中略)
.
.
[    0.013916] setup_percpu: NR_CPUS:8192 nr_cpumask_bits:2 nr_cpu_ids:2 nr_node_ids:1
.
.
(中略)
.
.
[    0.044868] rcu: Preemptible hierarchical RCU implementation.
[    0.044869] rcu:     RCU restricting CPUs from NR_CPUS=8192 to nr_cpu_ids=2.
[    0.044871]  Trampoline variant of Tasks RCU enabled.
[    0.044871]  Rude variant of Tasks RCU enabled.
[    0.044871]  Tracing variant of Tasks RCU enabled.
[    0.044872] rcu: RCU calculated value of scheduler-enlistment delay is 10 jiffies.
[    0.044873] rcu: Adjusting geometry for rcu_fanout_leaf=16, nr_cpu_ids=2
[    0.048140] NR_IRQS: 524544, nr_irqs: 440, preallocated irqs: 16
[    0.048336] rcu: srcu_init: Setting srcu_struct sizes based on contention.
[    0.048385] Console: colour dummy device 80x25
[    0.048396] printk: console [tty0] enabled
[    0.049355] printk: console [ttyS0] enabled
[    0.049364] ACPI: Core revision 20220331
[    0.049538] clocksource: hpet: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 30580167144 ns
[    0.049555] APIC: Switch to symmetric I/O mode setup
[    0.049929] x2apic enabled
[    0.050283] Switched APIC routing to physical x2apic.
[    0.050285] kvm-guest: setup PV IPIs
[    0.051760] clocksource: tsc-early: mask: 0xffffffffffffffff max_cycles: 0x26eb3e3cc63, max_idle_ns: 440795236593 ns
[    0.051764] Calibrating delay loop (skipped) preset value.. 5400.00 BogoMIPS (lpj=27000000)
[    0.051864] x86/cpu: User Mode Instruction Prevention (UMIP) activated
[    0.051992] Last level iTLB entries: 4KB 0, 2MB 0, 4MB 0
[    0.051993] Last level dTLB entries: 4KB 0, 2MB 0, 4MB 0, 1GB 0
[    0.051999] Spectre V1 : Mitigation: usercopy/swapgs barriers and __user pointer sanitization
[    0.052003] Spectre V2 : Mitigation: Enhanced / Automatic IBRS
[    0.052005] Spectre V2 : mitigation: Enabling conditional Indirect Branch Prediction Barrier
[    0.052006] Speculative Store Bypass: Mitigation: Speculative Store Bypass disabled via prctl
[    0.052037] x86/fpu: Supporting XSAVE feature 0x001: 'x87 floating point registers'
[    0.052038] x86/fpu: Supporting XSAVE feature 0x002: 'SSE registers'
[    0.052038] x86/fpu: Supporting XSAVE feature 0x004: 'AVX registers'
[    0.052039] x86/fpu: Supporting XSAVE feature 0x020: 'AVX-512 opmask'
[    0.052039] x86/fpu: Supporting XSAVE feature 0x040: 'AVX-512 Hi256'
[    0.052040] x86/fpu: Supporting XSAVE feature 0x080: 'AVX-512 ZMM_Hi256'
[    0.052040] x86/fpu: Supporting XSAVE feature 0x200: 'Protection Keys User registers'
[    0.052041] x86/fpu: Supporting XSAVE feature 0x20000: 'AMX Tile config'
[    0.052041] x86/fpu: Supporting XSAVE feature 0x40000: 'AMX Tile data'
[    0.052042] x86/fpu: xstate_offset[2]:  576, xstate_sizes[2]:  256
[    0.052043] x86/fpu: xstate_offset[5]:  832, xstate_sizes[5]:   64
[    0.052043] x86/fpu: xstate_offset[6]:  896, xstate_sizes[6]:  512
[    0.052044] x86/fpu: xstate_offset[7]: 1408, xstate_sizes[7]: 1024
[    0.052044] x86/fpu: xstate_offset[9]: 2432, xstate_sizes[9]:    8
[    0.052045] x86/fpu: xstate_offset[17]: 2496, xstate_sizes[17]:   64
[    0.052046] x86/fpu: xstate_offset[18]: 2560, xstate_sizes[18]: 8192
[    0.052046] x86/fpu: Enabled xstate features 0x602e7, context size is 10752 bytes, using 'compacted' format.
[    0.061762] Freeing SMP alternatives memory: 36K
[    0.061762] pid_max: default: 32768 minimum: 301
[    0.061762] LSM: Security Framework initializing
[    0.061762] Yama: becoming mindful.
[    0.061762] SELinux:  Initializing.
[    0.061762] LSM support for eBPF active
[    0.061762] Mount-cache hash table entries: 16384 (order: 5, 131072 bytes, linear)
[    0.061762] Mountpoint-cache hash table entries: 16384 (order: 5, 131072 bytes, linear)
[    0.061762] smpboot: CPU0: Intel(R) Xeon(R) 6975P-C (family: 0x6, model: 0xad, stepping: 0x1)
[    0.061762] cblist_init_generic: Setting adjustable number of callback queues.
[    0.061762] cblist_init_generic: Setting shift to 1 and lim to 1.
[    0.061762] cblist_init_generic: Setting adjustable number of callback queues.
[    0.061762] cblist_init_generic: Setting shift to 1 and lim to 1.
[    0.061762] cblist_init_generic: Setting adjustable number of callback queues.
[    0.061762] cblist_init_generic: Setting shift to 1 and lim to 1.
[    0.061762] Performance Events: unsupported p6 CPU model 173 no PMU driver, software events only.
[    0.061762] signal: max sigframe size: 11952
[    0.061762] rcu: Hierarchical SRCU implementation.
[    0.061762] rcu:     Max phase no-delay instances is 1000.
[    0.061762] smp: Bringing up secondary CPUs ...
[    0.061762] x86: Booting SMP configuration:
[    0.061762] .... node  #0, CPUs:      #1
[    0.061762] smp: Brought up 1 node, 2 CPUs
[    0.061762] smpboot: Max logical packages: 1
[    0.061762] smpboot: Total of 2 processors activated (10800.00 BogoMIPS)
[    0.061784] node 0 deferred pages initialised in 10ms
[    0.062106] devtmpfs: initialized
[    0.062106] x86/mm: Memory block size: 128MB
[    0.062118] ACPI: PM: Registering ACPI NVS region [mem 0xbfb7f000-0xbfbfefff] (524288 bytes)
[    0.062118] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[    0.062118] futex hash table entries: 512 (order: 3, 32768 bytes, linear)
[    0.062118] NET: Registered PF_NETLINK/PF_ROUTE protocol family
[    0.062118] DMA: preallocated 1024 KiB GFP_KERNEL pool for atomic allocations
[    0.062118] DMA: preallocated 1024 KiB GFP_KERNEL|GFP_DMA pool for atomic allocations
[    0.062118] DMA: preallocated 1024 KiB GFP_KERNEL|GFP_DMA32 pool for atomic allocations
[    0.062118] audit: initializing netlink subsys (disabled)
[    0.062118] audit: type=2000 audit(1756599393.260:1): state=initialized audit_enabled=0 res=1
.
.
(以下略)
.
.

lscpu

lscpuです。

Intel Xeon 6975P-Cを使っています。m8i.largeとm8i-flex.largeには特に違いは見られませんでした。

また、以下記事からm7i.largeとm7i-flex.largeと比較すると、それぞれL3キャッシュが105 MiBから480 MiBに大幅に増えていました。

https://dev.classmethod.jp/articles/ec2-instances-m7i-m7i-flex/

m8i.large

$ lscpu
Architecture:                x86_64
  CPU op-mode(s):            32-bit, 64-bit
  Address sizes:             48 bits physical, 48 bits virtual
  Byte Order:                Little Endian
CPU(s):                      2
  On-line CPU(s) list:       0,1
Vendor ID:                   GenuineIntel
  Model name:                Intel(R) Xeon(R) 6975P-C
    CPU family:              6
    Model:                   173
    Thread(s) per core:      2
    Core(s) per socket:      1
    Socket(s):               1
    Stepping:                1
    BogoMIPS:                5400.00
    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 c
                             onstant_tsc arch_perfmon rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pdcm
                              pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault
                             invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid avx512f avx512dq rdseed adx sma
                             p avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves avx_vnni avx512_bf16 wbnoinvd ida arat a
                             vx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq rdpid bus_lock_detec
                             t cldemote movdiri movdir64b md_clear serialize amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization features:
  Hypervisor vendor:         KVM
  Virtualization type:       full
Caches (sum of all):
  L1d:                       48 KiB (1 instance)
  L1i:                       64 KiB (1 instance)
  L2:                        2 MiB (1 instance)
  L3:                        480 MiB (1 instance)
NUMA:
  NUMA node(s):              1
  NUMA node0 CPU(s):         0,1
Vulnerabilities:
  Gather data sampling:      Not affected
  Indirect target selection: Not affected
  Itlb multihit:             Not affected
  L1tf:                      Not affected
  Mds:                       Not affected
  Meltdown:                  Not affected
  Mmio stale data:           Not affected
  Reg file data sampling:    Not affected
  Retbleed:                  Not affected
  Spec rstack overflow:      Not affected
  Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
  Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization
  Spectre v2:                Mitigation; Enhanced / Automatic IBRS; IBPB conditional; PBRSB-eIBRS Not affected; BHI BHI_DIS_S
  Srbds:                     Not affected
  Tsa:                       Not affected
  Tsx async abort:           Not affected

m8i-flex.large

$ lscpu
Architecture:                x86_64
  CPU op-mode(s):            32-bit, 64-bit
  Address sizes:             48 bits physical, 48 bits virtual
  Byte Order:                Little Endian
CPU(s):                      2
  On-line CPU(s) list:       0,1
Vendor ID:                   GenuineIntel
  Model name:                Intel(R) Xeon(R) 6975P-C
    CPU family:              6
    Model:                   173
    Thread(s) per core:      2
    Core(s) per socket:      1
    Socket(s):               1
    Stepping:                1
    BogoMIPS:                5400.00
    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 pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand
                              hypervisor lahf_lm abm 3dnowprefetch cpuid_fault invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced 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 avx_vnni avx512_bf16 wbnoinvd ida arat avx
                             512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq rdpid bus_lock_detect cldemote movdiri movdir64b
                              md_clear serialize amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization features:
  Hypervisor vendor:         KVM
  Virtualization type:       full
Caches (sum of all):
  L1d:                       48 KiB (1 instance)
  L1i:                       64 KiB (1 instance)
  L2:                        2 MiB (1 instance)
  L3:                        480 MiB (1 instance)
NUMA:
  NUMA node(s):              1
  NUMA node0 CPU(s):         0,1
Vulnerabilities:
  Gather data sampling:      Not affected
  Indirect target selection: Not affected
  Itlb multihit:             Not affected
  L1tf:                      Not affected
  Mds:                       Not affected
  Meltdown:                  Not affected
  Mmio stale data:           Not affected
  Reg file data sampling:    Not affected
  Retbleed:                  Not affected
  Spec rstack overflow:      Not affected
  Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
  Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization
  Spectre v2:                Mitigation; Enhanced / Automatic IBRS; IBPB conditional; PBRSB-eIBRS Not affected; BHI BHI_DIS_S
  Srbds:                     Not affected
  Tsa:                       Not affected
  Tsx async abort:           Not affected

/proc/cpuinfo

/proc/cpuinfoです。

m8i.largeとm8i-flex.largeには特に違いは見られませんでした。

m8i.large

$ cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 173
model name      : Intel(R) Xeon(R) 6975P-C
stepping        : 1
microcode       : 0x10003c2
cpu MHz         : 2700.000
cache size      : 491520 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 1
apicid          : 0
initial apicid  : 0
fpu             : yes
fpu_exception   : yes
cpuid level     : 36
wp              : yes
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 arch_perfmon rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pdcm pcid sse4_1 sse4_2 x2apic movbe popcnttsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced 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 avx_vnni avx512_bf16 wbnoinvd ida arat avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq rdpid bus_lock_detect cldemote movdiri movdir64b md_clear serialize amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
bugs            : spectre_v1 spectre_v2 spec_store_bypass swapgs bhi
bogomips        : 5400.00
clflush size    : 64
cache_alignment : 64
address sizes   : 48 bits physical, 48 bits virtual
power management:

processor       : 1
vendor_id       : GenuineIntel
cpu family      : 6
model           : 173
model name      : Intel(R) Xeon(R) 6975P-C
stepping        : 1
microcode       : 0x10003c2
cpu MHz         : 2700.000
cache size      : 491520 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 1
apicid          : 1
initial apicid  : 1
fpu             : yes
fpu_exception   : yes
cpuid level     : 36
wp              : yes
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 arch_perfmon rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq monitor ssse3 fma cx16 pdcm pcid sse4_1 sse4_2 x2apic movbe popcnttsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced 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 avx_vnni avx512_bf16 wbnoinvd ida arat avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq rdpid bus_lock_detect cldemote movdiri movdir64b md_clear serialize amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
bugs            : spectre_v1 spectre_v2 spec_store_bypass swapgs bhi
bogomips        : 5400.00
clflush size    : 64
cache_alignment : 64
address sizes   : 48 bits physical, 48 bits virtual
power management:

m8i-flex.large

$ cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 173
model name      : Intel(R) Xeon(R) 6975P-C
stepping        : 1
microcode       : 0x10003c2
cpu MHz         : 2700.000
cache size      : 491520 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 1
apicid          : 0
initial apicid  : 0
fpu             : yes
fpu_exception   : yes
cpuid level     : 36
wp              : yes
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 pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_niavx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves avx_vnni avx512_bf16 wbnoinvd ida arat avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq rdpid bus_lock_detect cldemote movdiri movdir64b md_clear serialize amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
bugs            : spectre_v1 spectre_v2 spec_store_bypass swapgs bhi
bogomips        : 5400.00
clflush size    : 64
cache_alignment : 64
address sizes   : 48 bits physical, 48 bits virtual
power management:

processor       : 1
vendor_id       : GenuineIntel
cpu family      : 6
model           : 173
model name      : Intel(R) Xeon(R) 6975P-C
stepping        : 1
microcode       : 0x10003c2
cpu MHz         : 2700.000
cache size      : 491520 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 1
apicid          : 1
initial apicid  : 1
fpu             : yes
fpu_exception   : yes
cpuid level     : 36
wp              : yes
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 pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_niavx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves avx_vnni avx512_bf16 wbnoinvd ida arat avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq rdpid bus_lock_detect cldemote movdiri movdir64b md_clear serialize amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
bugs            : spectre_v1 spectre_v2 spec_store_bypass swapgs bhi
bogomips        : 5400.00
clflush size    : 64
cache_alignment : 64
address sizes   : 48 bits physical, 48 bits virtual
power management:

dmidecode -t memory

dmidecode -t memoryです。

DDR5-7200のメモリのようですね。

m8i.largeとm8i-flex.largeには特に違いは見られませんでした。

m8i.large

$ sudo dmidecode -t memory
# dmidecode 3.6
Getting SMBIOS data from sysfs.
SMBIOS 2.7 present.

Handle 0x0008, DMI type 16, 23 bytes
Physical Memory Array
        Location: System Board Or Motherboard
        Use: System Memory
        Error Correction Type: Unknown
        Maximum Capacity: 8 GB
        Error Information Handle: Not Provided
        Number Of Devices: 1

Handle 0x0009, DMI type 17, 34 bytes
Memory Device
        Array Handle: 0x0008
        Error Information Handle: Not Provided
        Total Width: 80 bits
        Data Width: 64 bits
        Size: 8 GB
        Form Factor: DIMM
        Set: None
        Locator: Not Specified
        Bank Locator: Not Specified
        Type: DDR5
        Type Detail: Unknown Fast-paged RAMBus Window DRAM
        Speed: 7200 MT/s
        Manufacturer: Not Specified
        Serial Number: Not Specified
        Asset Tag: Not Specified
        Part Number: Not Specified
        Rank: Unknown
        Configured Memory Speed: Unknown

m8i-flex.large

$ sudo dmidecode -t memory
# dmidecode 3.6
Getting SMBIOS data from sysfs.
SMBIOS 2.7 present.

Handle 0x0008, DMI type 16, 23 bytes
Physical Memory Array
        Location: System Board Or Motherboard
        Use: System Memory
        Error Correction Type: Unknown
        Maximum Capacity: 8 GB
        Error Information Handle: Not Provided
        Number Of Devices: 1

Handle 0x0009, DMI type 17, 34 bytes
Memory Device
        Array Handle: 0x0008
        Error Information Handle: Not Provided
        Total Width: 80 bits
        Data Width: 64 bits
        Size: 8 GB
        Form Factor: DIMM
        Set: None
        Locator: Not Specified
        Bank Locator: Not Specified
        Type: DDR5
        Type Detail: Unknown Fast-paged RAMBus Window DRAM
        Speed: 7200 MT/s
        Manufacturer: Not Specified
        Serial Number: Not Specified
        Asset Tag: Not Specified
        Part Number: Not Specified
        Rank: Unknown
        Configured Memory Speed: Unknown

dmidecode -t baseboard

dmidecode -t baseboardです。

m8i.largeとm8i-flex.largeには特に違いは見られませんでした。

m8i.large

$ sudo dmidecode -t baseboard
# dmidecode 3.6
Getting SMBIOS data from sysfs.
SMBIOS 2.7 present.

Handle 0x0002, DMI type 2, 15 bytes
Base Board Information
        Manufacturer: Amazon EC2
        Product Name: Not Specified
        Version: Not Specified
        Serial Number: Not Specified
        Asset Tag: i-03158858ae2ecc6ba
        Features: None
        Location In Chassis: Not Specified
        Chassis Handle: 0x0003
        Type: Other
        Contained Object Handles: 0

m8i-flex.large

$ sudo dmidecode -t baseboard
# dmidecode 3.6
Getting SMBIOS data from sysfs.
SMBIOS 2.7 present.

Handle 0x0002, DMI type 2, 15 bytes
Base Board Information
        Manufacturer: Amazon EC2
        Product Name: Not Specified
        Version: Not Specified
        Serial Number: Not Specified
        Asset Tag: i-03158858ae2ecc6ba
        Features: None
        Location In Chassis: Not Specified
        Chassis Handle: 0x0003
        Type: Other
        Contained Object Handles: 0

CPUとメモリ以外の性能、上限

AWS公式ドキュメントに記載されている主だったCPUとメモリ以外の性能、上限を確認します。

ネットワークのベースライン/バースト帯域

ネットワークのベースライン/バースト帯域は以下のとおりです。単位はGbpsです。

インスタンスサイズ M8 M8i-flex M7 M7i-flex M6i M5
large 0.937 / 12.5 0.468 / 12.5 0.781 / 12.5 0.39 / 12.5 0.781 / 12.5 0.75 / 10.0
xlarge 1.875 / 12.5 0.937 / 12.5 1.562 / 12.5 0.781 / 12.5 1.562 / 12.5 1.25 / 10.0
2xlarge 3.75 / 15.0 1.875 / 15.0 3.125 / 12.5 1.562 / 12.5 3.125 / 12.5 2.5 / 10.0
4xlarge 7.5 / 15.0 3.75 / 15.0 6.25 / 12.5 3.125 / 12.5 6.25 / 12.5 5.0 / 10.0
8xlarge 15 7.5 / 15.0 12.5 6.25 / 12.5 12.5 10
12xlarge 22.5 11.25 / 22.5 18.75 9.375 / 18.75 18.75 12
16xlarge 30 15.0 / 30.0 25 12.5 / 25.0 25 20
24xlarge 40 - 37.5 - 37.5 25
32xlarge 50 - - - 50 -
48xlarge 75 - 50 - - -
96xlarge 100 - - - - -
metal - - - - 50 25
metal-24xl - - 37.5 - - -
metal-48xl 75 - 50 - - -
metal-96xl 100 - - - - -

参考 : Specifications for Amazon EC2 general purpose instances - Amazon EC2

M7 → M8、M7-flex → M8-flexで順当にスペックアップしていますね。

一方、M8とM8-flexはベースラインが大きく違います。CPUやメモリ使用量が少なくとも捌く必要があるネットワークトラフィック量が多いのであれば、M8を選択するモチベーションはあるでしょう。

ちなみに、M8およびM8-flexはネットワークまたはEBSの操作のいずれかを優先するベースライン帯域幅の重み付けを選択できます。詳細は以下AWS公式ドキュメントをご覧ください。

https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/configure-bandwidth-weighting.html

最大ENI数

最大ENI数は以下のとおりです。

インスタンスサイズ M8 M8i-flex M7 M7i-flex M6i M5
large 3 3 3 3 3 3
xlarge 4 4 4 4 4 4
2xlarge 4 4 4 4 4 4
4xlarge 8 8 8 8 8 8
8xlarge 10 10 8 8 8 8
12xlarge 12 12 8 8 8 8
16xlarge 16 16 15 15 15 15
24xlarge 16 - 15 - 15 15
32xlarge 24 - - - 15 -
48xlarge 24 - 15 - - -
96xlarge 24 - - - - -
metal - - - - 15 15
metal-24xl - - 15 - - -
metal-48xl 24 - 15 - - -
metal-96xl 24 - - - - -

12xlarge以上のサイズで割り当て可能な最大ENI数に違いが見られますね。

一つのENIあたりのIPアドレス数

一つのENIあたりのIPアドレス数は以下のとおりです。

インスタンスサイズ M8 M8i-flex M7 M7i-flex M6i M5
large 20 20 10 10 10 10
xlarge 30 30 15 15 15 15
2xlarge 30 30 15 15 15 15
4xlarge 50 50 30 30 30 30
8xlarge 50 50 30 30 30 30
12xlarge 50 50 30 30 30 30
16xlarge 64 64 50 50 50 50
24xlarge 64 - 50 - 50 50
32xlarge 64 - - - 50 -
48xlarge 64 - 50 - - -
96xlarge 64 - - - - -
metal - - - - 50 50
metal-24xl - 50 - - -
metal-48xl 64 - 50 - - -
metal-96xl 64 - - - - -

同一のインスタンスサイズで一つのENIに割り当て可能なIPアドレス数が倍に増えているものがありますね。

従来多数のIPアドレスを付与したいがために大きめのインスタンスサイズを選択せざるを得なかった場合はコスト削減できそうです。

EBSボリュームへのベースライン/バースト帯域

ネットワークのベースライン/バースト帯域は以下のとおりです。単位はMbpsです。

インスタンスサイズ M8 M8i-flex M7 M7i-flex M6i M5
large 650.00 / 10000.00 315.00 / 10000.00 650.00 / 10000.00 312.00 / 10000.00 650.00 / 10000.00 650.00 / 4750.00
xlarge 1250.00 / 10000.00 630.00 / 10000.00 1250.00 / 10000.00 625.00 / 10000.00 1250.00 / 10000.00 1150.00 / 4750.00
2xlarge 2500.00 / 10000.00 1250.00 / 10000.00 2500.00 / 10000.00 1250.00 / 10000.00 2500.00 / 10000.00 2300.00 / 4750.00
4xlarge 5000.00 / 10000.00 2500.00 / 10000.00 5000.00 / 10000.00 2500.00 / 10000.00 5000.00 / 10000.00 4750
8xlarge 10000 5000.00 / 10000.00 10000 5000.00 / 10000.00 10000 6800
12xlarge 15000 7500.00 / 15000.00 15000 7500.00 / 15000.00 15000 9500
16xlarge 20000 10000.00 / 20000.00 20000 10000.00 / 20000.00 20000 13600
24xlarge 30000 - 30000 - 30000 19000
32xlarge 40000 - - - 40000 -
48xlarge 60000 - 40000 - - -
96xlarge 80000 - - - - -
metal - - - - 40000 19000
metal-24xl - - 30000 - - -
metal-48xl 60000 - 40000 - - -
metal-96xl 80000 - - - - -

抜粋 : Specifications for Amazon EC2 general purpose instances - Amazon EC2

M7 → M8は48xlarge以上で、M7-flex → M8-flexではlargeとxlargeで若干の増強されています。

128KiB IOの場合のEBSへのベースライン/最大スループット

128KiB IOの場合のEBSへのベースライン/最大スループットは以下のとおりです。単位はMBpsです。

インスタンスサイズ M8 M8i-flex M7 M7i-flex M6i M5
large 81.25 / 1250.00 39.38 / 1250.00 81.25 / 1250.00 39.06 / 1250.00 81.25 / 1250.00 81.25 / 1250.00
xlarge 156.25 / 1250.00 78.75 / 1250.00 156.25 / 1250.00 78.12 / 1250.00 156.25 / 1250.00 156.25 / 1250.00
2xlarge 312.50 / 1250.00 156.25 / 1250.00 312.50 / 1250.00 156.25 / 1250.00 312.50 / 1250.00 312.50 / 1250.00
4xlarge 625.00 / 1250.00 312.50 / 1250.00 625.00 / 1250.00 312.50 / 1250.00 625.00 / 1250.00 625.00 / 1250.00
8xlarge 1250 625.00 / 1250.00 1250 625.00 / 1250.00 1250 1250
12xlarge 1875 937.50 / 1875.00 1875 937.50 / 1875.00 1875 1875
16xlarge 2500 1250.00 / 2500.00 2500 1250.00 / 2500.00 2500 2500
24xlarge 3750 - 3750 - 3750 3750
32xlarge 5000 - - - 5000 5000
48xlarge 7500 - 5000 - - -
96xlarge 10000 - - - - -
metal - - - - 5000 2375
metal-24xl - - 3750 - - -
metal-48xl 7500 - 5000 - - -
metal-96xl 10000 - - - - -

M7 → M8は48xlarge以上で、M7-flex → M8-flexではlargeとxlargeで若干の増強されています。

16KiB IOの場合のEBSへのベースライン/最大IOPS

16KiB IOの場合のEBSへのベースライン/最大IOPSは以下のとおりです。

インスタンスサイズ M8 M8i-flex M7 M7i-flex M6i M5
large 3600.00 / 40000.00 2500.00 / 40000.00 3600.00 / 40000.00 2500.00 / 40000.00 3600.00 / 40000.00 3600.00 / 18750.00
xlarge 6000.00 / 40000.00 3600.00 / 40000.00 6000.00 / 40000.00 3600.00 / 40000.00 6000.00 / 40000.00 6000.00 / 18750.00
2xlarge 12000.00 / 40000.00 6000.00 / 40000.00 12000.00 / 40000.00 6000.00 / 40000.00 12000.00 / 40000.00 12000.00 / 18750.00
4xlarge 20000.00 / 40000.00 12000.00 / 40000.00 20000.00 / 40000.00 12000.00 / 40000.00 20000.00 / 40000.00 18750
8xlarge 40000 20000.00 / 40000.00 40000 20000.00 / 40000.00 40000 30000
12xlarge 60000 30000.00 / 60000.00 60000 30000.00 / 60000.00 60000 40000
16xlarge 80000 40000.00 / 80000.00 80000 40000.00 / 80000.00 80000 60000
24xlarge 120000 - 120000 - 120000 80000
32xlarge 160000 - - - 160000 -
48xlarge 240000 - 240000 - - -
96xlarge 480000 - - - - -
metal - - - - 160000 80000
metal-24xl - 120000 - - -
metal-48xl 240000 - 240000 - - -
metal-96xl 480000 - - - - -

M6i以降では特に変化はないですね。

料金

2025/8/30時点のバージニア北部リージョンのLinuxインスタンスのオンデマンド料金は以下のとおりです。

vCPU メモリ (GiB) インスタンスサイズ M8i M8i-flex M7i M7i-flex M6i M5
2 8 large 0.10584 0.10055 0.1008 0.09576 0.096 0.096
4 16 xlarge 0.21168 0.2011 0.2016 0.19152 0.192 0.192
8 32 2xlarge 0.42336 0.4022 0.4032 0.38304 0.384 0.384
16 64 4xlarge 0.84672 0.8044 0.8064 0.76608 0.768 0.768
32 128 8xlarge 1.69344 1.6088 1.6128 1.53216 1.536 1.536
48 192 12xlarge 2.54016 2.4132 2.4192 2.29824 2.304 2.304
64 256 16xlarge 3.38688 3.2176 3.2256 3.06432 3.072 3.072
96 384 24xlarge 5.08032 - 4.8384 - 4.608 4.608
128 512 32xlarge 6.77376 - - - 6.144 -
192 768 48xlarge 10.16064 - 9.6768 - - -
384 1536 96xlarge 20.32128 - - - - -
96 384 metal - - - - - 4.608
128 512 metal - - - - 6.144 -
96 384 metal-24xl - - 4.8384 - - -
192 768 metal-48xl 10.16064 - 9.6768 - - -
384 1536 metal-96xl 20.32128 - - - - -

抜粋 : EC2 On-Demand Instance Pricing

M7、M7i-flexと比較していずれも単純なvCPU数とメモリサイズでは料金は高くなっていますね。

ただし、CPU性能やメモリの帯域、L3キャッシュサイズなどといった単純なvCPU数やメモリサイズで測れない要素も多くあります。実際のワークロードで動かしながら選定する形が良いでしょう。

ワークロードに応じて適切なインスタンスタイプを選択しよう

Intel Xeon 6を搭載したM8i、M8i-flexインスタンスが追加されたアップデートを紹介しました。

カタログスペックでは測れない点も多くあります。実際のワークロードで検証をして、コストとパフォーマンスの最適な落としどこを探っていきましょう。

場合によっては最新世代のインスタンスファミリーを選択することで、インスタンスサイズを小さくすることも可能な場面もあるでしょう。

この記事が誰かの助けになれば幸いです。

以上、クラウド事業本部 コンサルティング部の のんピ(@non____97)でした!

この記事をシェアする

facebookのロゴhatenaのロゴtwitterのロゴ

© Classmethod, Inc. All rights reserved.