Testy budżetowego serwera 1U AMD Epyc dla 4 dysków NVMe gen 4.0 (Czy BPN-SAS3-815TQ-N4 obsłuży dyski NVMe gen 4.0?)

| Nowości Supermicro • Testy • Nowości
Facebook Twitter Schowek

Testy budżetowego serwera 1U AMD dla 4 dysków NVMe gen 4.0 (Czy BPN-SAS3-815TQ-N4 obsłuży dyski NVMe gen 4.0?)

Zbudowaliśmy dla Klienta serwer 1U ITC-RR3R wyposażony w procesor AMD Epyc Rome 7542 32C/64T 2.9G 128MB i sprawdziliśmy czy backplane BPN-SAS3-815TQ-N4, w który wyposażona jest obudowa CSE-813MF2TQC4-R407CB obsłuży dyski NVMe Gen 4.0.

Supermicro 1U server NVMe ITC

Konfiguracja:

  1. Obudowa Supermicro CSE-813MF2TQC4-R407CB 1U, Hot-Swap 4xHDD SAS3/SATA3/NVMe, 400W Redundant
  2. Płyta główna Supermicro MBD-H12SSL-NT ATX, obsługująca jeden procesor AMD Epyc 7003/7002, do 2TB pamięci RAM w 8 modułach. Posiadająca dwa sloty M.2 (22x80, 22x110) oraz 2 porty SlimSAS x8, obsługujace 2 dyski NVMe każdy. Obsługa sieci zapewniana jest przez kartę sieciową Broadcom BCM57416 2x10GBase-T. Za zarzadzanie odpowiedzialny jest moduł ASPEED AST2500
  3. Procesor PSE-ROM7542-0075 AMD Epyc Rome 7542 32C/64T 2.9G 128MB
  4. Radiator Supermicro SNK-P0062P 1U pasywny, dedykowany socketowi SP3
  5. Pamieć RAM 512GB DDR4 ECC Registered (8 x 64GB MEM-DR464L-SL03-ER32)
  6. Dysk HDS-SUN0-MZQL2960HCJRA7 Samsung PM9A3 960GB NVMe PCIe4x4 U.2 7mm 1DWPD 5YR SED Opal
  7. Dysk HDS-TUN0-KCD6XLUL960G Kioxia CD6-R 960GB NVMe PCIe4x4 2.5" 15mm SIE 1DWPD 5YR
  8. Dysk M.2 HDS-MMN-MTFDKBA480TFR1BC - Micron 7450 PRO 480GB NVMe PCIe 4.0 M.2 22x80mm 3D TLC
  9. Dedykowany Riser Card RSC-S-6G4 1U Left Standard wyposażony w slot PCI-E 4.0 x16
  10. Kable CBL-SAST-0817 do połączenia płyty głównej z backplane do obsługi dysków NVMe
  11. Licencja SFT-OOB-LIC, rozszerzająca możliwości IPMI.

Pierwsze uruchomienie

  1. Podłączyliśmy serwer do prądu oraz sieci. Zalogowaliśmy się do IPMI. IPMI wykryło dysk.
  2. Uruchomiliśmy system operacyjny Kubuntu 23.10 live USB, jednak lspci nie pokazało żadnych urządzeń, które moglibyśmy uznać za nasze dyski NVMe.
  3. Uruchomiliśmy ponownie serwer i w BIOS zmieniliśmu opcję NVMe Firmware Source z Vendor na AMI Native Support
    • BIOS → Advanced → PCIe/PCI/PNP Configuration → NVMe Firmware Source Select [AMI Native Support].
  4. Uruchomiliśmy ponownie i wydaliśmy polecenie nvme oraz lspci: 
    • # nvme list -v

      Subsystem Subsystem-NQN Controllers

      ---------------- ------------------------------------------------------------------------------------------------ ----------------

      nvme-subsys2 nqn.2016-08.com.micron:nvme:nvm-subsystem-sn-24034667BF24 nvme2

      nvme-subsys1 nqn.1994-11.com.samsung:nvme:PM9A3:2.5-inch:S64FNG0X103815 nvme1

      nvme-subsys0 nqn.2019-10.com.kioxia:KCD6XLUL960G:53U0A029T5M8 nvme0

      Device SN MN FR TxPort Address Slot Subsystem Namespaces

      -------- -------------------- ---------------------------------------- -------- ------ -------------- ------ ------------ ----------------

      nvme2 24034667BF24 Micron_7450_MTFDKBA480TFR E2MU200 pcie 0000:03:00.0 nvme-subsys2 nvme2n1

      nvme1 S64FNG0X103815 SAMSUNG MZQL2960HCJR-00A07 GDC5A02Q pcie 0000:02:00.0 nvme-subsys1 nvme1n1

      nvme0 53U0A029T5M8 KCD6XLUL960G 0107 pcie 0000:01:00.0 nvme-subsys0 nvme0n1

      Device Generic NSID Usage Format Controllers

      ------------ ------------ ---------- -------------------------- ---------------- ----------------

      /dev/nvme2n1 /dev/ng2n1 0x1 14,95 GB / 480,10 GB 512 B + 0 B nvme2

      /dev/nvme1n1 /dev/ng1n1 0x1 0,00 B / 960,20 GB 512 B + 0 B nvme1

      /dev/nvme0n1 /dev/ng0n1 0x1 0,00 B / 960,20 GB 512 B + 0 B nvme0
    • # lspci -vv -nn -s 0000:03:00.0|grep Lnk

       LnkCap: Port #0, Speed 16GT/s, Width x4, ASPM L0s L1, Exit Latency L0s <256ns, L1 unlimited

       LnkCtl: ASPM Disabled; RCB 64 bytes, Disabled- CommClk+

       LnkSta: Speed 16GT/s, Width x4

       LnkCap2: Supported Link Speeds: 2.5-16GT/s, Crosslink- Retimer+ 2Retimers+ DRS-

       LnkCtl2: Target Link Speed: 16GT/s, EnterCompliance- SpeedDis-

       LnkSta2: Current De-emphasis Level: -3.5dB, EqualizationComplete+ EqualizationPhase1+

       LnkCtl3: LnkEquIntrruptEn- PerformEqu-
    • # lspci -vv -nn -s 0000:02:00.0|grep Lnk

       LnkCap: Port #0, Speed 16GT/s, Width x4, ASPM not supported

       LnkCtl: ASPM Disabled; RCB 64 bytes, Disabled- CommClk+

       LnkSta: Speed 16GT/s, Width x4

       LnkCap2: Supported Link Speeds: 2.5-16GT/s, Crosslink- Retimer+ 2Retimers+ DRS-

       LnkCtl2: Target Link Speed: 16GT/s, EnterCompliance- SpeedDis-

       LnkSta2: Current De-emphasis Level: -3.5dB, EqualizationComplete+ EqualizationPhase1+

       LnkCtl3: LnkEquIntrruptEn- PerformEqu-
    • # lspci -vv -nn -s 0000:01:00.0|grep Lnk

       LnkCap: Port #0, Speed 16GT/s, Width x4, ASPM not supported

       LnkCtl: ASPM Disabled; RCB 64 bytes, Disabled- CommClk+

       LnkSta: Speed 16GT/s, Width x4

       LnkCap2: Supported Link Speeds: 2.5-16GT/s, Crosslink- Retimer+ 2Retimers+ DRS-

       LnkCtl2: Target Link Speed: 16GT/s, EnterCompliance- SpeedDis-

       LnkSta2: Current De-emphasis Level: -3.5dB, EqualizationComplete+ EqualizationPhase1+

       LnkCtl3: LnkEquIntrruptEn- PerformEqu-
  5. Aby przeprowadzić testy, zainstalowaliśmy Kubuntu na dysku NVMe M.2

Testy

Użyliśmy polecenia fio aby przeprowadzić testy, najpierw odczyt sekwencyjny, następnie odczyt losowy.

TEST 1

Na pierwszy ogień poszedł dysk Kioxia HDS-TUN0-KCD6XLUL960G Kioxia CD6-R 960GB NVMe PCIe4x4 2.5" 15mm SIE 1DWPD 5YR.

fio seq read (Kioxia KCD6XLUL960G)

fio --filename=/dev/nvme0n1 --rw=read --direct=1 --bs=1M --ioengine=libaio --runtime=60 --numjobs=1 --time_based --group_reporting --name=seq_read --iodepth=16
seq_read: (g=0): rw=read, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=16
fio-3.35
Starting 1 process
Jobs: 1 (f=1): [R(1)][100.0%][r=2852MiB/s][r=2851 IOPS][eta 00m:00s]
seq_read: (groupid=0, jobs=1): err= 0: pid=22195: Thu Mar 7 15:34:53 2024
 read: IOPS=2848, BW=2849MiB/s (2987MB/s)(167GiB/60006msec)
 slat (usec): min=61, max=991, avg=161.52, stdev=70.30
 clat (usec): min=2165, max=11294, avg=5453.75, stdev=103.34
 lat (usec): min=2291, max=12285, avg=5615.27, stdev=77.17
 clat percentiles (usec):
 | 1.00th=[ 5276], 5.00th=[ 5276], 10.00th=[ 5342], 20.00th=[ 5407],
 | 30.00th=[ 5407], 40.00th=[ 5407], 50.00th=[ 5473], 60.00th=[ 5473],
 | 70.00th=[ 5473], 80.00th=[ 5538], 90.00th=[ 5538], 95.00th=[ 5604],
 | 99.00th=[ 5669], 99.50th=[ 5669], 99.90th=[ 5669], 99.95th=[ 5669],
 | 99.99th=[ 6718]
 bw ( MiB/s): min= 2812, max= 2854, per=100.00%, avg=2851.14, stdev= 3.76, samples=119
 iops : min= 2812, max= 2854, avg=2851.14, stdev= 3.76, samples=119
 lat (msec) : 4=0.01%, 10=99.99%, 20=0.01%
 cpu : usr=0.74%, sys=54.06%, ctx=169635, majf=0, minf=4109
 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=100.0%, 32=0.0%, >=64=0.0%
 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
 issued rwts: total=170935,0,0,0 short=0,0,0,0 dropped=0,0,0,0
 latency : target=0, window=0, percentile=100.00%, depth=16
Run status group 0 (all jobs):
 READ: bw=2849MiB/s (2987MB/s), 2849MiB/s-2849MiB/s (2987MB/s-2987MB/s), io=167GiB (179GB), run=60006-60006msec
Disk stats (read/write):
 nvme0n1: ios=1364954/0, merge=0/0, ticks=7044197/0, in_queue=7044197, util=99.92%

Drugi dysk, który testowaliśmy to HDS-SUN0-MZQL2960HCJRA7 Samsung PM9A3 960GB NVMe PCIe4x4 U.2 7mm 1DWPD 5YR SED Opal

fio seq read (Samsung MZQL2960HCJRA7)

fio --filename=/dev/nvme1n1 --rw=read --direct=1 --bs=1M --ioengine=libaio --runtime=60 --numjobs=1 --time_based --group_reporting --name=seq_read --iodepth=16
seq_read: (g=0): rw=read, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=16
fio-3.35
Starting 1 process
Jobs: 1 (f=1): [R(1)][100.0%][r=6922MiB/s][r=6921 IOPS][eta 00m:00s]
seq_read: (groupid=0, jobs=1): err= 0: pid=22263: Thu Mar 7 15:36:19 2024
 read: IOPS=6913, BW=6913MiB/s (7249MB/s)(405GiB/60003msec)
 slat (usec): min=82, max=967, avg=112.97, stdev=15.55
 clat (usec): min=1730, max=11732, avg=2200.57, stdev=44.01
 lat (usec): min=1837, max=12700, avg=2313.54, stdev=45.45
 clat percentiles (usec):
 | 1.00th=[ 2147], 5.00th=[ 2180], 10.00th=[ 2180], 20.00th=[ 2180],
 | 30.00th=[ 2180], 40.00th=[ 2180], 50.00th=[ 2212], 60.00th=[ 2212],
 | 70.00th=[ 2212], 80.00th=[ 2212], 90.00th=[ 2212], 95.00th=[ 2212],
 | 99.00th=[ 2212], 99.50th=[ 2245], 99.90th=[ 2278], 99.95th=[ 2802],
 | 99.99th=[ 2933]
 bw ( MiB/s): min= 6604, max= 6924, per=100.00%, avg=6919.13, stdev=29.14, samples=119
 iops : min= 6604, max= 6924, avg=6919.13, stdev=29.14, samples=119
 lat (msec) : 2=0.05%, 4=99.94%, 10=0.01%, 20=0.01%
 cpu : usr=1.29%, sys=88.29%, ctx=373362, majf=0, minf=4110
 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=100.0%, 32=0.0%, >=64=0.0%
 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
 issued rwts: total=414807,0,0,0 short=0,0,0,0 dropped=0,0,0,0
 latency : target=0, window=0, percentile=100.00%, depth=16
Run status group 0 (all jobs):
 READ: bw=6913MiB/s (7249MB/s), 6913MiB/s-6913MiB/s (7249MB/s-7249MB/s), io=405GiB (435GB), run=60003-60003msec
Disk stats (read/write):
 nvme1n1: ios=3311288/0, merge=0/0, ticks=7132629/0, in_queue=7132629, util=99.92%

Przy okazji przetestowaliśmy nasz dysk systemowy HDS-MMN-MTFDKBA480TFR1BC - Micron 7450 PRO 480GB NVMe PCIe 4.0 M.2 22x80mm 3D TLC

fio seq read (Micron MTFDKBA480TFR1BC)

fio --filename=/dev/nvme2n1 --rw=read --direct=1 --bs=1M --ioengine=libaio --runtime=60 --numjobs=1 --time_based --group_reporting --name=seq_read --iodepth=16
seq_read: (g=0): rw=read, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=16
fio-3.35
Starting 1 process
Jobs: 1 (f=1): [R(1)][100.0%][r=5042MiB/s][r=5042 IOPS][eta 00m:00s]
seq_read: (groupid=0, jobs=1): err= 0: pid=22330: Thu Mar 7 15:37:59 2024
 read: IOPS=5000, BW=5001MiB/s (5244MB/s)(293GiB/60002msec)
 slat (usec): min=64, max=848, avg=102.37, stdev=20.28
 clat (usec): min=967, max=11332, avg=3096.30, stdev=225.10
 lat (usec): min=1062, max=12181, avg=3198.67, stdev=224.67
 clat percentiles (usec):
 | 1.00th=[ 2933], 5.00th=[ 2999], 10.00th=[ 3032], 20.00th=[ 3032],
 | 30.00th=[ 3064], 40.00th=[ 3064], 50.00th=[ 3064], 60.00th=[ 3097],
 | 70.00th=[ 3097], 80.00th=[ 3097], 90.00th=[ 3130], 95.00th=[ 3163],
 | 99.00th=[ 4113], 99.50th=[ 4948], 99.90th=[ 5800], 99.95th=[ 6128],
 | 99.99th=[ 7439]
 bw ( MiB/s): min= 3970, max= 5044, per=100.00%, avg=5005.58, stdev=158.93, samples=119
 iops : min= 3970, max= 5044, avg=5005.58, stdev=158.93, samples=119
 lat (usec) : 1000=0.01%
 lat (msec) : 2=0.03%, 4=98.86%, 10=1.11%, 20=0.01%
 cpu : usr=0.65%, sys=58.17%, ctx=276012, majf=0, minf=4108
 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=100.0%, 32=0.0%, >=64=0.0%
 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
 issued rwts: total=300053,0,0,0 short=0,0,0,0 dropped=0,0,0,0
 latency : target=0, window=0, percentile=100.00%, depth=16
Run status group 0 (all jobs):
 READ: bw=5001MiB/s (5244MB/s), 5001MiB/s-5001MiB/s (5244MB/s-5244MB/s), io=293GiB (315GB), run=60002-60002msec
Disk stats (read/write):
 nvme2n1: ios=2396174/3, merge=0/1, ticks=7200202/0, in_queue=7200202, util=99.92%

TEST 2

Następnie przetestowaliśmy dyski pod kątem odczytu losowego. Kolejność testów taka sama jak przy teśc

fio random read (Kioxia KCD6XLUL960G)

fio --filename=/dev/nvme0n1 --rw=randread --direct=1 --bs=1M --ioengine=libaio --runtime=60 --numjobs=1 --time_based --group_reporting --name=seq_read --iodepth=16
seq_read: (g=0): rw=randread, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=16
fio-3.35
Starting 1 process
Jobs: 1 (f=1): [r(1)][100.0%][r=2853MiB/s][r=2852 IOPS][eta 00m:00s]
seq_read: (groupid=0, jobs=1): err= 0: pid=22407: Thu Mar 7 15:41:23 2024
 read: IOPS=2848, BW=2849MiB/s (2987MB/s)(167GiB/60006msec)
 slat (usec): min=111, max=948, avg=224.10, stdev=36.10
 clat (usec): min=2371, max=11871, avg=5390.63, stdev=66.05
 lat (usec): min=2526, max=12819, avg=5614.72, stdev=57.78
 clat percentiles (usec):
 | 1.00th=[ 5276], 5.00th=[ 5342], 10.00th=[ 5342], 20.00th=[ 5342],
 | 30.00th=[ 5342], 40.00th=[ 5407], 50.00th=[ 5407], 60.00th=[ 5407],
 | 70.00th=[ 5407], 80.00th=[ 5407], 90.00th=[ 5473], 95.00th=[ 5473],
 | 99.00th=[ 5473], 99.50th=[ 5473], 99.90th=[ 5538], 99.95th=[ 5538],
 | 99.99th=[ 6783]
 bw ( MiB/s): min= 2808, max= 2854, per=100.00%, avg=2851.13, stdev= 4.24, samples=119
 iops : min= 2808, max= 2854, avg=2851.14, stdev= 4.24, samples=119
 lat (msec) : 4=0.01%, 10=99.99%, 20=0.01%
 cpu : usr=0.80%, sys=74.14%, ctx=168755, majf=0, minf=4106
 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=100.0%, 32=0.0%, >=64=0.0%
 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
 issued rwts: total=170938,0,0,0 short=0,0,0,0 dropped=0,0,0,0
 latency : target=0, window=0, percentile=100.00%, depth=16
Run status group 0 (all jobs):
 READ: bw=2849MiB/s (2987MB/s), 2849MiB/s-2849MiB/s (2987MB/s-2987MB/s), io=167GiB (179GB), run=60006-60006msec
Disk stats (read/write):
 nvme0n1: ios=1364996/0, merge=0/0, ticks=6979131/0, in_queue=6979131, util=99.91%

fio random read (Samsung MZQL2960HCJRA7)

fio --filename=/dev/nvme1n1 --rw=randread --direct=1 --bs=1M --ioengine=libaio --runtime=60 --numjobs=1 --time_based --group_reporting --name=seq_read --iodepth=16
seq_read: (g=0): rw=randread, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=16
fio-3.35
Starting 1 process
Jobs: 1 (f=1): [r(1)][100.0%][r=6922MiB/s][r=6921 IOPS][eta 00m:00s]
seq_read: (groupid=0, jobs=1): err= 0: pid=22522: Thu Mar 7 15:42:48 2024
 read: IOPS=6913, BW=6913MiB/s (7249MB/s)(405GiB/60003msec)
 slat (usec): min=86, max=958, avg=109.43, stdev=13.04
 clat (usec): min=1711, max=11925, avg=2204.04, stdev=43.59
 lat (usec): min=1853, max=12884, avg=2313.47, stdev=45.94
 clat percentiles (usec):
 | 1.00th=[ 2147], 5.00th=[ 2180], 10.00th=[ 2180], 20.00th=[ 2180],
 | 30.00th=[ 2212], 40.00th=[ 2212], 50.00th=[ 2212], 60.00th=[ 2212],
 | 70.00th=[ 2212], 80.00th=[ 2212], 90.00th=[ 2212], 95.00th=[ 2212],
 | 99.00th=[ 2212], 99.50th=[ 2245], 99.90th=[ 2376], 99.95th=[ 2802],
 | 99.99th=[ 2835]
 bw ( MiB/s): min= 6608, max= 6924, per=100.00%, avg=6919.16, stdev=28.78, samples=119
 iops : min= 6608, max= 6924, avg=6919.16, stdev=28.78, samples=119
 lat (msec) : 2=0.05%, 4=99.94%, 10=0.01%, 20=0.01%
 cpu : usr=1.50%, sys=86.22%, ctx=382785, majf=0, minf=4107
 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=100.0%, 32=0.0%, >=64=0.0%
 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
 issued rwts: total=414816,0,0,0 short=0,0,0,0 dropped=0,0,0,0
 latency : target=0, window=0, percentile=100.00%, depth=16
Run status group 0 (all jobs):
 READ: bw=6913MiB/s (7249MB/s), 6913MiB/s-6913MiB/s (7249MB/s-7249MB/s), io=405GiB (435GB), run=60003-60003msec
Disk stats (read/write):
 nvme1n1: ios=3311521/0, merge=0/0, ticks=7133786/0, in_queue=7133786, util=99.91%

fio random read (Micron MTFDKBA480TFR1BC)

fio --filename=/dev/nvme2n1 --rw=randread --direct=1 --bs=1M --ioengine=libaio --runtime=60 --numjobs=1 --time_based --group_reporting --name=seq_read --iodepth=16
seq_read: (g=0): rw=randread, bs=(R) 1024KiB-1024KiB, (W) 1024KiB-1024KiB, (T) 1024KiB-1024KiB, ioengine=libaio, iodepth=16
fio-3.35
Starting 1 process
Jobs: 1 (f=1): [r(1)][100.0%][r=5043MiB/s][r=5043 IOPS][eta 00m:00s]
seq_read: (groupid=0, jobs=1): err= 0: pid=22590: Thu Mar 7 15:44:09 2024
 read: IOPS=5037, BW=5037MiB/s (5282MB/s)(295GiB/60002msec)
 slat (usec): min=66, max=920, avg=105.58, stdev=25.21
 clat (usec): min=1572, max=12070, avg=3069.73, stdev=110.75
 lat (usec): min=1687, max=12816, avg=3175.32, stdev=106.67
 clat percentiles (usec):
 | 1.00th=[ 2835], 5.00th=[ 2933], 10.00th=[ 2966], 20.00th=[ 2999],
 | 30.00th=[ 3032], 40.00th=[ 3064], 50.00th=[ 3064], 60.00th=[ 3097],
 | 70.00th=[ 3097], 80.00th=[ 3130], 90.00th=[ 3130], 95.00th=[ 3195],
 | 99.00th=[ 3490], 99.50th=[ 3589], 99.90th=[ 3785], 99.95th=[ 3851],
 | 99.99th=[ 4113]
 bw ( MiB/s): min= 4930, max= 5046, per=100.00%, avg=5041.65, stdev=10.39, samples=119
 iops : min= 4930, max= 5046, avg=5041.65, stdev=10.39, samples=119
 lat (msec) : 2=0.01%, 4=99.98%, 10=0.02%, 20=0.01%
 cpu : usr=0.81%, sys=59.64%, ctx=247929, majf=0, minf=4106
 IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=100.0%, 32=0.0%, >=64=0.0%
 submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
 complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
 issued rwts: total=302253,0,0,0 short=0,0,0,0 dropped=0,0,0,0
 latency : target=0, window=0, percentile=100.00%, depth=16
Run status group 0 (all jobs):
 READ: bw=5037MiB/s (5282MB/s), 5037MiB/s-5037MiB/s (5282MB/s-5282MB/s), io=295GiB (317GB), run=60002-60002msec
Disk stats (read/write):
 nvme2n1: ios=2413346/154, merge=0/29, ticks=7210715/18, in_queue=7210733, util=99.92% 

Więcej informacji o serwerze oraz przeprowadzonych testach i ich wynikach znajdziesz na naszym na kanale na YouTube:

1/5 Unboxing:

2/4 Pierwsze uruchomienie

3/5 Pobór energii w trybie bezczynności (IDLE) oraz przy maksymalnym obciążeniu (STRESS)

4/5 Testy wydajnościowe procesora AMD Epyc Rome 7542 32C/64T

5/5 Testy wydajności dysków NVMe

Powiązane strony:

  1. Supermicro BPN-SAS3-815TQ-N4 dla dysków SAS3/SATA3 i dysków NVMe (Jak ustawić zworki?)
  2. Test platformy Supermicro NVMe 1029U-TN10RT + Intel Virtual RAID on CPU (VROC)
  3. Kontroler Supermicro AOC-SLG3-2M2 wykrywa tylko jeden dysk NVMe, dlaczego? (PCIe Bifurcation)
  4. Oszczędności wynikające ze stosowania nowych CPU w instalacjach hybrydowych wykorzystujących odnawialne źródła energii (OZE)
  5. Aktualizacja firmware dysku NVMe za pomocą nvme-cli na przykładzie Samsung PM9A3
  6. Jak sformatować dysk NVMe? Jak usunąć bezpowrotnie dane z dysku NVMe?