Карта изначально не определяйка . Уже с шитым биосом. После реболла гпу определяеться. Код 43.
Mods/mats не проходит. Где то здесь читал что человек тоже столкнулся с такой проблеммой что без родного флэша код 43.
Вот лог . Кстати очень похож на неверно прошитый биос. Хотя пару вариантов попробывал. Слитое что было тоже приложу.

MODS arguments :

MODS start: Mon May 13 16:46:14 2024

[000001.459029593] Done parsing args
[000001.471068139] Command Line : gputest.jse -oqa -old_gold -skip_azalia_init -skip_rm_state_init -print_wallclock_time -notest -verbose

[000001.471080098] CPU
[000001.471129951] Arch : x86_64
[000001.471136968] Name : Intel(R) Celeron(R) CPU G1840 @ 2.80GHz
[000001.471166351] Cores : 2
[000001.471173289]
[000001.471245475] Version
[000001.471254236] MODS : 455.127
[000001.471305139]
[000001.471313221] System
[000001.471363535] OperatingSystem: Linux (x86_64)
[000001.471434605] Kernel : 5.9.1-gentoo-x86_64
[000001.471481217] KernelDriver : 4.00
[000001.471607347] SBIOS Version : 4.6.5
[000001.471816135] SBIOS Date : 08/15/2017
[000001.471894896] HostName : tinylinux
[000001.471982777] Available RAM : 3648/3968 MB (Free/Size)
[000001.472320768] NUMA Node 0 RAM: 3648/3968 MB (Free/Size)
[000001.472414266] Sysmem page size is 4 KB
[000001.472495833] Sys-uuid : 03000200-0400-0500-0006-000700080009
[000001.472733880] Failed to Read Hdisk Serial num, this requires root permissions to run
[000001.472759178] HDD-Serno :
[000001.472839341]
[000001.473083727] Turning off register exclusion regions.
[000001.474561077] >boards.jse
[000001.474567561] Preprocessing boards.jse.
[000001.474587215] Successfully opened /home/455.127/boards.jse
[000001.476174421] Compiling boards.jse.
[000001.476803312] Compiled boards.jse, pScript = 0x0xba92740.
[000001.476806598] Executing boards.jse.
[000001.477637890] <boards.jse
[000001.483979794] CreateNvLinkLibInterface : Creating NvLink library interface NvLinkLibIfUser
[000001.483991330] CreateAllDevInterfaces : Created NvLink device interface NvSwitchDevIf
[000001.484064501] CreateAllDevInterfaces : Created NvLink device interface NvGpuDevIf
[000001.484068654] CreateAllDevInterfaces : Created NvLink device interface IbmNpuDevIf
[000001.484120017] Warning: Can't find Pci device with class code 0x68000, index 0
[000001.484124406] Find NvSwitch ports
[000001.484146734] -------Finish NvSwitch detection-----
[000001.484581003] Physical Graphics controller at 0000:01:00.0
[000001.484662714] reg virtual base = 0x7f8b20ffc000
[000001.486790439] IOVA enabled
[000001.486794855] FindChip : Unable to find ChipId 0x000001ad, searching by PciDeviceId
[000001.489635210] Detected 47 sysmem address bits for chip id 0xc02 at 0000:01:00.0, IOVA enabled
[000001.490747717] mods [1]: set dma mask 47 for dev 0000:01:00.0
[000001.493867972] BAR0 base = 0x0_F6000000,
size = 0x01000000
[000001.493870368] BAR1 base = 0x0_E0000000,
size = 0x10000000
[000001.493935458] BAR2 base = 0x0_F0000000,
size = 0x02000000
[000001.493938266] Vendor ID = 0x10DE
[000001.493972064] Device ID = 0x2208
[000001.493976218] Chip ID = 0x000001AD
[000001.494009855] [0000:01:00.0] cap id 0x5 found at offs 0x68
[000001.494019866] [0000:01:00.0] cap id 0x11 not found
[000001.494023685] FindChip : Unable to find ChipId 0x000001ad, searching by PciDeviceId
[000001.496731552] Register manual file ga102_ref.txt not found
[000001.496744555] Warning: Can't find Pci device with class code 0x30000, index 1
[000001.496805698] Warning: Can't find Pci device with class code 0x30200, index 0
[000001.496824641] Warning: Can't find Pci device
[000001.496827392] Find ibmnpu ports
[000001.496862138] -------Finish ibmnpu detection-----
[000001.496865733] Found Test Devices :
[000001.496899066] GPU 0 [01:00.0]
[000001.496905928] CreateAll : Created topology manager implementation TopologyManagerProtobuf
[000001.496908103] CreateAll : Created topology manager implementation TopologyManagerProtobufLR
[000001.496909832] CreateAll : Created topology manager implementation TopologyManagerAuto
[000001.496910979] Initialize : Topology file not specified, skipping.
[000001.496911986] Initialize : Topology file not specified, skipping.
[000001.496913792] Find bridge ports
[000001.496921154] [0000:00:01.0] bridge sec=01 sub=01 vendor=8086 devid=0c01 uniq_link_poll=true
[000001.496931233] [0000:00:1c.0] bridge sec=02 sub=02 vendor=8086 devid=8c10 uniq_link_poll=false
[000001.496941016] [0000:00:1c.2] bridge sec=03 sub=03 vendor=8086 devid=8c10 uniq_link_poll=false
[000001.496944576] Warning: Can't find Pci device with class code 0x60400, index 3
[000001.496947819] Find siblings at 0000:00:xx.x
[000001.496949100] Sibling found: domain=0000 sec=01 sub=01
[000001.496950432] Sibling found: domain=0000 sec=02 sub=02
[000001.496951533] Sibling found: domain=0000 sec=03 sub=03
[000001.496952697] Found chipset controller's port at 0000:00:01.0
[000001.496958761] [0000:00:01.0] cap id 0x10 found at offs 0xa0
[000001.496965258] [0000:00:01.0] extended cap id 0x1 found at offs 0x01c0
[000001.496972205] [0000:00:1c.0] cap id 0x10 found at offs 0x40
[000001.496981070] [0000:00:1c.0] extended cap id 0x1 not found
[000001.496987523] [0000:00:1c.2] cap id 0x10 found at offs 0x40
[000001.496996832] [0000:00:1c.2] extended cap id 0x1 not found
[000001.496999677] Root check: 0000:00:01.0 IsRoot=true
[000001.497001381] Connecting ports for Chipset 0000:00:01.0:
[000001.497006681] DP0 0000:00:01.0 connects to test device GPU 0 [01:00.0]
[000001.497010017] ------Finish bridge detection-----
[000001.498106414] InitializeAllDevices : RM not initialized, skipping GPU nvlink device initialization
[000001.498592899] GPU 0 [01:00.0] dev.sub 0.0
[000001.498730619] ----------------------------------------
[000001.498954826] DevInst : 0
[000001.499326646] PCI Location : 0x00, 0x01, 0x00, 0x00
[000001.499592711] GPU DID : 0x2208
WARNING: [000001.518759892] ECID Vendor code is invalid and likely unfused.
[000001.518856521] ECID : N/A-W?V422???0W-32_x32_y32
[000001.519011090] Device Id : GA102
[000001.521978963] Revision : 20
[000001.523329069] NV Base : 0xf6000000
[000001.523487454] FB Base : 0xe0000000
[000001.523643328] IRQ : 11
[000001.523838258] MSI-X not supported on GPU 0 [01:00.0], skipping
[000001.523842130] Detected supported interrupt types: legacy MSI
ERROR: [000001.536511493] Couldn't clear GPU's interrupt state; int0 == 0xbadf1020.
[000001.536587329] First error recorded: 608 Interrupt request line stuck asserted
[000001.536589754] JavaScript stack trace:
[000001.536641021] #0: Initialize() in <native>:0
[000001.536645817] #1: InitializeGpuTests() in gputest.jse:341
[000001.536677203] #2: gputest_main() in gputest.jse:2199
[000001.536680434] #3: main() in gputest.jse:2136
[000001.536709607] Error 000000000608 : Gpu.Initialize Interrupt request line stuck asserted [0.064 seconds]
[000001.536798213] Error 000000000608 : Global.PrintGpuInitError Interrupt request line stuck asserted [0.000 seconds]
[000001.536873677] Error 000000000608 : Global.InitializeGpuTests Interrupt request line stuck asserted [0.077 seconds]
[000001.537617951]
[000001.537724609] Primary GPU not detected, skipping DisableUserInterface.
[000001.537818358] Warning: Can't find Pci device with class code 0x40300, index 2
[000001.537835285] Bypassing unsupported vendor: 0x8086Device already uninitialized
[000001.537889855] GA102 Azalia: DeviceId: 0x1aef.
[000001.537895793] Found 1 controllers (DevId 127)
[000001.538547423] Device already uninitialized
[000001.538736676] Beginning GPU shut down
[000001.538767908] Shutdown : GPU devices not initialized, skipping GPU nvlink shutdown
[000001.538989092] PexDevMgr not initialized. Bridge device shutdowns will likely do nothing.
[000001.538995305] PexDevice not initialized or already shutdown. Nothing to do.
[000001.539027811] Completed GPU shut down
[000001.539308147]
Error Code = 000000000608 (Interrupt request line stuck asserted)
[000001.539476558]
[000001.539499852]
####### #### ######## ###
####### ###### ######## ###
## ## ## ## ###
## ## ## ## ###
####### ######## ## ###
####### ######## ## ###
## ## ## ## ###
## ## ## ######## ########
## ## ## ######## ########

[000001.539506023]
[000001.577138106] mods [1]: driver closed
[000001.577441772] Max queue depth = 27
[000001.577468971] Tasker Yields: 4
[000001.577481193] Tasker Suspends: 151
[000001.577491864] Tasker Detaches: 47
[000001.578319267] Device already uninitialized
[000001.578342234] Device already uninitialized
[000001.582846923] MODS end : Mon May 13 16:46:15 2024 [1.583 seconds (00:00:01.583 h:m:s)]
---------- Добавлено спустя 3 минуты 9 секунд: ----------
P.S Все сопротивления в допуске и напряжения поднимаються. PCI-E тоже +- одинаковый. Любые замечания приветсвуются )

попробуйте посвежее модс. 455.219 455.229
---------- Добавлено спустя 2 минуты 24 секунды: ----------
также если не проходит матс,не пройдет и модс. Ну и тестятся GA-102 через training stаtus - на проблемные каналы. Есть видео у хорошо известных блоггеров

И эта версия прекрасно тестит проблемные ga102 .
Взял 520 модс , лог
MODS arguments :

MODS start: Tue May 14 05:34:52 2024

Command Line : gputest.js -skip_rm_state_init

CPU
Arch : x86_64
Name : Intel(R) Celeron(R) CPU G1840 @ 2.80GHz
Cores : 2

Version
MODS : 520.112

System
OperatingSystem: Linux (x86_64)
Kernel : 5.9.1-gentoo-x86_64
KernelDriver : 4.00
SBIOS Version : 4.6.5
SBIOS Date : 08/15/2017
HostName : tinylinux
Available RAM : 3527/3968 MB (Free/Size)
Hypervisor : No
Sys-uuid : 03000200-0400-0500-0006-000700080009
HDD-Serno :

GPU 0 [01:00.0] dev.sub 0.0
----------------------------------------
DevInst : 0
PCI Location : 0x00, 0x01, 0x00, 0x00
GPU DID : 0x2208
WARNING: ECID Vendor code is invalid and likely unfused.
ECID : N/A-W???0W-32_x32_y32
Device Id : GA102
Revision : 20
NV Base : 0xf6000000
FB Base : 0xe0000000
IRQ : 11
ERROR: Couldn't clear GPU's interrupt state; int0 == 0xbadf1020.
Error 000000000608 : Gpu.Initialize Interrupt request line stuck asserted [0.065 seconds]
Error 000000000608 : Global.PrintGpuInitError Interrupt request line stuck asserted [0.000 seconds]
Error 000000000608 : Global.InitializeGpuTests Interrupt request line stuck asserted [0.078 seconds]
PexDevMgr not initialized. Bridge device shutdowns will likely do nothing.

Error Code = 000000000608 (Interrupt request line stuck asserted)


####### #### ######## ###
####### ###### ######## ###
## ## ## ## ###
## ## ## ## ###
####### ######## ## ###
####### ######## ## ###
## ## ## ## ###
## ## ## ######## ########
## ## ## ######## ########


MODS end : Tue May 14 05:34:53 2024 [1.891 seconds (00:00:01.891 h:m:s)]



ECID : N/A-W???0W-32_x32_y32 меня вот эта строчка смущает .Кто с таким сталкивался. Я обычно такое видел при гпу полутрупе .
---------- Добавлено спустя 35 минут 48 секунд: ----------
ODS arguments :

MODS start: Tue May 14 06:09:59 2024

Command Line : gputest.js -skip_rm_state_init -oqa -skip_azalia_init

CPU
Arch : x86_64
Name : Intel(R) Celeron(R) CPU G1840 @ 2.80GHz
Cores : 2

Version
MODS : 455.127

System
OperatingSystem: Linux (x86_64)
Kernel : 5.9.1-gentoo-x86_64
KernelDriver : 4.00
SBIOS Version : 4.6.5
SBIOS Date : 08/15/2017
HostName : tinylinux
Available RAM : 3646/3968 MB (Free/Size)
Sys-uuid : 03000200-0400-0500-0006-000700080009
HDD-Serno :

GPU 0 [01:00.0] dev.sub 0.0
----------------------------------------
DevInst : 0
PCI Location : 0x00, 0x01, 0x00, 0x00
GPU DID : 0x2216
PDI : 0xef398921eb982d08
Raw ECID : 0x008052800000005d984f1dc4
Raw ECID (GHS) : 0x4705d984f1c0000000a028100
ECID : SAMSUNG-SNPXF7-10_x5_y4
Device Id : GA102
Revision : a1
Sub Revision : 1
NV Base : 0xf6000000
FB Base : 0xe0000000
IRQ : 11

Chipset
VID : FFFF (Unknown)
Chipset DID : FFFF (Unknown)
Chipset ASPM : L0s/L1
Chipset LTR : Enabled

Error 000000000021 : CheckClocks (test 10) script failed to execute [0.000 seconds]
Error 000000000021 : Global.RequireGoldenValues script failed to execute [0.004 seconds]
Error 000000000021 : Global.LoadGoldenValuesHelper script failed to execute [0.004 seconds]
Error 000000000021 : Global.LoadGoldenValues script failed to execute [0.004 seconds]
Error 000000000021 : Global.InitializeGpuTests script failed to execute [0.281 seconds]
gputest.jse:32: TypeError: subdev.Perf has no properties

Error Code = 000000010021 (script failed to execute)

Это вот с рабочей 3080 нсмотря на то что модс не запустился матс следом проходит

mats version 455.127. Testing GA102 with 5 MB of memory starting with 0 MB.

Memory Errors on
Read Error Count: 0
Write Error Count: 0
Unknown Error Count: 0

=== MEMORY ERRORS BY SUBPARTITION ===
SUBPART READ ERRORS WRITE ERRORS UNKNOWN ERRS
------- ----------- ------------ ------------
FBIOB0 0 0 0
FBIOB1 0 0 0
FBIOC0 0 0 0
FBIOC1 0 0 0
FBIOD0 0 0 0
FBIOD1 0 0 0
FBIOE0 0 0 0
FBIOE1 0 0 0
FBIOF0 0 0 0
FBIOF1 0 0 0

Failing Bits:
None

В чем прикол тестировать 3080ti версией для 4090? Вам же уже дали совет, попробуйте посвежее чем 455.127, от 455.178 и свежее желательно, 455.219-455.229.

"Прикол"в том что и 127 прекрасно все тестит. Тут гпу не инцализируеться а вы про версии. 520 свежее некуда)))
Лог 219 пжл.
MODS arguments :

MODS start: Tue May 14 07:54:34 2024

Command Line : gputest.js -skip_azalia_init -skip_rm_state_init -oqa

CPU
Arch : x86_64
Name : Intel(R) Celeron(R) CPU G1840 @ 2.80GHz
Cores : 2

Version
MODS : 455.219

System
OperatingSystem: Linux (x86_64)
Kernel : 5.9.1-gentoo-x86_64
KernelDriver : 4.00
SBIOS Version : 4.6.5
SBIOS Date : 08/15/2017
HostName : tinylinux
Available RAM : 3645/3968 MB (Free/Size)
Sys-uuid : 03000200-0400-0500-0006-000700080009
HDD-Serno :

GPU 0 [01:00.0] dev.sub 0.0
----------------------------------------
DevInst : 0
PCI Location : 0x00, 0x01, 0x00, 0x00
GPU DID : 0x2208
WARNING: ECID Vendor code is invalid and likely unfused.
ECID : N/A-W?V422???0W-32_x32_y32
Device Id : GA102
Revision : 20
NV Base : 0xf6000000
FB Base : 0xe0000000
IRQ : 11
ERROR: Couldn't clear GPU's interrupt state; int0 == 0xbadf1020.
Error 000000000608 : Gpu.Initialize Interrupt request line stuck asserted [0.062 seconds]
Error 000000000608 : Global.PrintGpuInitError Interrupt request line stuck asserted [0.000 seconds]
Error 000000000608 : Global.InitializeGpuTests Interrupt request line stuck asserted [0.076 seconds]

PexDevMgr not initialized. Bridge device shutdowns will likely do nothing.

Error Code = 000000000608 (Interrupt request line stuck asserted)
---------- Добавлено спустя 2 минуты 10 секунд: ----------
Причем если попытаться прошить биос через nvflash он тоже не шьет. Ругаеться на неизвестного вендора. Хотя GPU-Z Вендора видит . Ничего непонятно.

Посмотрел бы я, как вы будете прекрасно тестировать какую-нибудь LHR карту верcией 455.127. Про версии вам было сказано к тому, что каждая мажорная версия под свое поколение, а минорная под конкретную модель, поэтому бездумно брать "свежее некуда" и тестировать карты предыдущего поколения или наоборот более новые - полнейшая глупость. Хотя если вам чисто MATS погонять, то да почти любая версия подойдет, и то не всегда.

Это вообще не одно и тоже. Тут речь об ECID, а не о том, что там видит GPU-Z. Возможно GPU и правда битый, раз MODS не может распознать на чьих мощностях был изготовлен GPU - TSMC или SAMSUNG. Логи MODS то хотя бы читайте. И поймете все неуместность упоминая GPU-Z и что он там видит.

Ко всему прочему ревизия не как A1 определяется, а как 20.

Вот за конкретику спасибо!