Konstruktori arvuti - Asrock Taichi X570 + AMD Ryzen 5500

Allikas: Imre kasutab arvutit
Mine navigeerimisribaleMine otsikasti

Sissejuhatus

TODO

Riistvara

Arvutikomplekti komponentide koosseis

  • emaplaat - Asrock Taichi X570
  • protsessor - AMD Ryzen 5500
  • mälu - 4 x 32 GB
  • toiteplokk -
  • videokaart - Nvidia quadro k600 (lspci: 'NVIDIA Corporation GK107GL [Quadro K600]')
  • nvme seade - 2 x 'Samsung SSD 990 PRO with Heatsink 4TB'
  • nvme seade - 2 x 'micron 7450 pro u.3'

pcie nvme adapteri kasutamine

nvme u.3 micron 7450 pro plokkseade, iseloomulikud kontaktid (andmed + toide)

salvestuskomplekt

  • kaks nvme ketast
  • supermicro AOC-SLG4-4E4T pcie nvme adapter
  • ühenduskaablid (toide tuleb tavalise nn molex küljest)

andmekaabli kontaktid, kinnitub adapteri kaardi külge ja ühendab ketta

kus

  • valge lapik kontakt ja tema vastus kaardil on 'slimsas' ehk 'sff-8654 8i' - tavaliselt x8 pcie laiune, st teenindab ära kaks nvme ketast (antud juhul nii kasutatakse)
  • valge lapik kontakt ei ole MCIO (Mini Cool Edge IO) - tavaliselt x8 pcie laiune, st teenindab ära kaks nvme ketast (antud juhul nii kasutatakse)

arvuti bios/setup keskkond, tuleb väljendada, et pci liides töötab bifurcation režiimis - paistab, et antud juhul saab kasutada kahte x4 kanalit ja ka füüsilisi kettaid on parasjagu kaks

muud bifurcation valikud

Sissevaada arvuti korpusse, nö kõige kangemas pesas on videokaart, siis võrgukaart ja kõnealune storage adapter

Tulemusena on arvutis operatsioonisüsteemi jaoks olemas kaks füüsilist ketast

root@pve-svc-01:~# lspci | grep Micron
01:00.0 Non-Volatile memory controller: Micron Technology Inc 7450 PRO NVMe SSD (rev 01)
10:00.0 Non-Volatile memory controller: Micron Technology Inc 7450 PRO NVMe SSD (rev 01)

202604 - Mellanox kaardi lisamise trauma

root@pve-svc-01:~# zpool status
  pool: rpool
 state: DEGRADED
status: One or more devices has been removed by the administrator.
	Sufficient replicas exist for the pool to continue functioning in a
	degraded state.
action: Online the device using zpool online' or replace the device with
	'zpool replace'.
  scan: scrub repaired 0B in 00:00:16 with 0 errors on Sun Apr 12 00:24:17 2026
config:

	NAME                                                 STATE     READ WRITE CKSUM
	rpool                                                DEGRADED     0     0     0
	  mirror-0                                           DEGRADED     0     0     0
	    nvme-eui.000000000000000a00a075244cbac2ea-part3  ONLINE       0     0     0
	    nvme-eui.000000000000000200a075254da4607e-part3  REMOVED      0     0     0

errors: No known data errors

  pool: zp_svc
 state: DEGRADED
status: One or more devices has been removed by the administrator.
	Sufficient replicas exist for the pool to continue functioning in a
	degraded state.
action: Online the device using zpool online' or replace the device with
	'zpool replace'.
  scan: scrub repaired 0B in 00:35:07 with 0 errors on Sun Apr 12 00:59:09 2026
config:

	NAME                                                STATE     READ WRITE CKSUM
	zp_svc                                              DEGRADED     0     0     0
	  mirror-0                                          DEGRADED     0     0     0
	    nvme-Micron_7450_MTFDKCC15T3TFR_24484CBAC2EA_2  ONLINE       0     0     0
	    nvme-Micron_7450_MTFDKCC15T3TFR_25024DA4607E_2  REMOVED      0     0     0

errors: No known data errors
root@pve-svc-01:~# lsscsi -s
[N:0:1:1]    disk    Samsung SSD 990 PRO with Heatsink 4TB__1   /dev/nvme0n1  4.00TB
sdev_scandir_sort: left parse failed: nvme1n3
sdev_scandir_sort: left parse failed: nvme1n2
[N:1:0:2]    disk    Micron_7450_MTFDKCC15T3TFR__2              /dev/nvme1n2       -
[N:1:0:3]    disk    Micron_7450_MTFDKCC15T3TFR__3              /dev/nvme1n3       -
[N:1:0:1]    disk    Micron_7450_MTFDKCC15T3TFR__1              /dev/nvme1n1       -
[N:2:1:1]    disk    Samsung SSD 990 PRO with Heatsink 4TB__1   /dev/nvme2n1  4.00TB
sdev_scandir_sort: left parse failed: nvme3n1
sdev_scandir_sort: left parse failed: nvme3n3
[N:3:0:3]    disk    Micron_7450_MTFDKCC15T3TFR__3              /dev/nvme3n3  7.45TB
[N:3:0:1]    disk    Micron_7450_MTFDKCC15T3TFR__1              /dev/nvme3n1   214GB
[N:3:0:2]    disk    Micron_7450_MTFDKCC15T3TFR__2              /dev/nvme3n2  7.69TB
root@pve-svc-01:~# ./check_pcie.sh
Scanning PCIe Bus for active errors...
======================================
DEVICE: 00:01.2 PCI bridge: Advanced Micro Devices, Inc. [AMD] Renoir/Cezanne PCIe GPP Bridge (prog-if 00 [Normal decode])
  [!!!] Physical Lane Issues: 		LaneErrStat: LaneErr at lane: 0 1 2 3
--------------------------------------
DEVICE: 01:00.0 Non-Volatile memory controller: Micron Technology Inc 7450 PRO NVMe SSD (rev 01) (prog-if 02 [NVM Express])
  [!] Correctable: 		CESta:	RxErr+ BadTLP+ BadDLLP+ Rollover- Timeout+ AdvNonFatalErr+
  [!!] Uncorrectable: 		UESta:	DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq+ ACSViol-
  [!!!] Physical Lane Issues: 		LaneErrStat: LaneErr at lane: 0 1 2 3
  [W] Link Speed Downgraded: 8GT/s (Capable: 16GT/s)
--------------------------------------
DEVICE: 05:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller S4LV008[Pascal] (prog-if 02 [NVM Express])
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+
--------------------------------------
DEVICE: 06:00.0 PCI bridge: ASMedia Technology Inc. ASM1184e 4-Port PCIe x1 Gen2 Packet Switch (prog-if 00 [Normal decode])
  [!] Correctable: 		CESta:	RxErr+ BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+
--------------------------------------
DEVICE: 08:00.0 Network controller: Intel Corporation Wi-Fi 6 AX200 (rev 1a)
  [!] Correctable: 		CESta:	RxErr+ BadTLP- BadDLLP- Rollover- Timeout+ AdvNonFatalErr+
--------------------------------------
DEVICE: 0c:00.0 Ethernet controller: Intel Corporation 82574L Gigabit Network Connection
  [!] Correctable: 		CESta:	RxErr+ BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr-
  [W] Link Speed Downgraded: 		LnkSta:	Speed 2.5GT/s, Width x1 (Capable: 		LnkCap:	Port #4, Speed 2.5GT/s, Width x1, ASPM L0s L1, Exit Latency L0s <128ns, L1 <64us)
--------------------------------------
DEVICE: 10:00.0 Ethernet controller: Mellanox Technologies MT2894 Family [ConnectX-6 Lx]
  [!] Correctable: 		CESta:	RxErr- BadTLP+ BadDLLP+ Rollover- Timeout+ AdvNonFatalErr+
  [!!!] Physical Lane Issues: 		LaneErrStat: LaneErr at lane: 0 1 2 3 4 5 6 7
  [W] Link Speed Downgraded: 8GT/s (Capable: 16GT/s)
--------------------------------------
DEVICE: 10:00.1 Ethernet controller: Mellanox Technologies MT2894 Family [ConnectX-6 Lx]
  [!] Correctable: 		CESta:	RxErr- BadTLP+ BadDLLP+ Rollover- Timeout+ AdvNonFatalErr+
  [W] Link Speed Downgraded: 8GT/s (Capable: 16GT/s)
--------------------------------------
DEVICE: 11:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller S4LV008[Pascal] (prog-if 02 [NVM Express])
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+
  [W] Link Speed Downgraded: 8GT/s (Capable: 16GT/s)
--------------------------------------
DEVICE: 12:00.0 Non-Volatile memory controller: Micron Technology Inc 7450 PRO NVMe SSD (rev 01) (prog-if 02 [NVM Express])
  [!] Correctable: 		CESta:	RxErr+ BadTLP+ BadDLLP+ Rollover- Timeout+ AdvNonFatalErr+
  [!!!] Physical Lane Issues: 		LaneErrStat: LaneErr at lane: 0 1 2 3
  [W] Link Speed Downgraded: 8GT/s (Capable: 16GT/s)
--------------------------------------

202605 - ebastabiilsuse parandamine

gemini.google vestluses tekkis hüpotees, et kasutuses olev protsessor Ryzen 5 5500' on üsna tagasihoidlik seade

  • toetab kuni pcie v. 3.0
  • algselt laptop kasutusjuhule orienteeritud
  • mitte-paindlik pcie ressursside esitamises (nt iommu abil grupeerimine)
  • materjalivaliku osas mitte kõige kvaliteetsem (signaalide läbikostvus, müra jms)

ja hüpotees, et protsessor 'Ryzen 9 5900XT' võiks anda palju parema tulemuse

  • toetab kuni pcie v. 4.0
  • pigem serveri laadne, ja downgrade'itud desktop kasutusjuhuks
  • pcie ressursside paindlik korraldamine (nt bifurcation osakonda ilmus juurde mõni valik (nt 4 x4)
  • kvaliteetsemad materjalid, võimsam signaal pcie peal jne

ning tõesti, tulemus on looduses palju parem

root@pve-svc-01:~# ./check_pcie.sh
Scanning PCIe Bus for active errors...
======================================
DEVICE: 00:01.3 PCI bridge: Advanced Micro Devices, Inc. [AMD] Starship/Matisse GPP Bridge (prog-if 00 [Normal decode])
  [!!!] Physical Lane Issues: 		LaneErrStat: LaneErr at lane: 0 1 2 3
--------------------------------------
DEVICE: 04:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller S4LV008[Pascal] (prog-if 02 [NVM Express])
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 05:00.0 PCI bridge: ASMedia Technology Inc. ASM1184e 4-Port PCIe x1 Gen2 Packet Switch (prog-if 00 [Normal decode])
  [!] Correctable: 		CESta:	RxErr+ BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 08:00.0 Ethernet controller: Intel Corporation I211 Gigabit Network Connection (rev 03)
  [!] Correctable: 		CESta:	RxErr+ BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr- CorrIntErr- HeaderOF-
  [W] Link Speed Downgraded: 		LnkSta:	Speed 2.5GT/s, Width x1 (Capable: 		LnkCap:	Port #5, Speed 2.5GT/s, Width x1, ASPM L0s L1, Exit Latency L0s <2us, L1 <16us)
--------------------------------------
DEVICE: 0a:00.0 Ethernet controller: Intel Corporation 82574L Gigabit Network Connection
  [!] Correctable: 		CESta:	RxErr+ BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr- CorrIntErr- HeaderOF-
  [W] Link Speed Downgraded: 		LnkSta:	Speed 2.5GT/s, Width x1 (Capable: 		LnkCap:	Port #4, Speed 2.5GT/s, Width x1, ASPM L0s L1, Exit Latency L0s <128ns, L1 <64us)
--------------------------------------
DEVICE: 0e:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller S4LV008[Pascal] (prog-if 02 [NVM Express])
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 0f:00.0 Non-Volatile memory controller: Micron Technology Inc 7450 PRO NVMe SSD (rev 01) (prog-if 02 [NVM Express])
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 10:00.0 Non-Volatile memory controller: Micron Technology Inc 7450 PRO NVMe SSD (rev 01) (prog-if 02 [NVM Express])
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 11:00.0 Ethernet controller: Mellanox Technologies MT2894 Family [ConnectX-6 Lx]
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 11:00.1 Ethernet controller: Mellanox Technologies MT2894 Family [ConnectX-6 Lx]
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 13:00.0 Non-Essential Instrumentation [1300]: Advanced Micro Devices, Inc. [AMD] Starship/Matisse Reserved SPP
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 13:00.1 Encryption controller: Advanced Micro Devices, Inc. [AMD] Starship/Matisse Cryptographic Coprocessor PSPCPP
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 13:00.3 USB controller: Advanced Micro Devices, Inc. [AMD] Matisse USB 3.0 Host Controller (prog-if 30 [XHCI])
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------
DEVICE: 13:00.4 Audio device: Advanced Micro Devices, Inc. [AMD] Starship/Matisse HD Audio Controller (prog-if 00 [HDA compatible])
  [!] Correctable: 		CESta:	RxErr- BadTLP- BadDLLP- Rollover- Timeout- AdvNonFatalErr+ CorrIntErr- HeaderOF-
--------------------------------------

kus

  • 8 GT/s downgrade'isid ei ole (nt nvme x4 pcie seadmed töötavad oma loomuliku x4 ehk 16 GT/s kiiruse lane'iga)
  • LaneErr esineb ainult 00:01.3 juures, paistab, et seal on üks seade ja oluline seade, gemini arvates on see lingi kokkuleppimise ühekordne nähtus, eriti arvestades, et süsteem säilitab kiirust 16 GT/s, näis
# lspci -PP -s  0e:00.0
00:01.3/0e:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller S4LV008[Pascal]

20260530 - bifurcation märkused

SuperMicro nvme retimer kaardi kasutamist aitab

  • bios juures bifruction valikute vahel katsetamine (4 x4, 2 x8 jne)
  • füüsiliste kaablite ühendamine ringi nt ühe Y kaabli küljes on kaks seadete; kummagi Y kaabli küljes on üks seade
  • pcie kaardi passimine ühte ja teisse pcie füüsilisse pessa (tõenäoliselt on kõige parem kaarti paigutada nn video kaardi pessa)
  • videokaardina võiks kasutada midagi võimalikult tagasihoidlikku, nt pcie v. 2.0 nvidia midagi 56 euro eest; ja nö madalas pcie pesas

Kasulikud lisamaterjalid

  • TODO