MetaX-Tech Developer Forum
  • 沐曦开发者
search
Sign in

jianjungki

  • Members
  • Joined 2025年8月21日
  • message 帖子
  • forum 主题
  • favorite 关注者
  • favorite_border Follows
  • person_outline 详细信息

jianjungki has posted 6 messages.

  • See post chevron_right
    jianjungki
    Members
    metax现在安装了驱动不认卡 已解决 2025年8月25日 14:11

    lscpu

    Architecture:            x86_64
      CPU op-mode(s):        32-bit, 64-bit
      Address sizes:         52 bits physical, 57 bits virtual
      Byte Order:            Little Endian
    CPU(s):                  128
      On-line CPU(s) list:   0-127
    Vendor ID:               GenuineIntel
      Model name:            INTEL(R) XEON(R) GOLD 6530
        CPU family:          6
        Model:               207
        Thread(s) per core:  2
        Core(s) per socket:  32
        Socket(s):           2
        Stepping:            2
        CPU max MHz:         4000.0000
        CPU min MHz:         800.0000
        BogoMIPS:            4200.00
    
    Virtualization features: 
      Virtualization:        VT-x
    Caches (sum of all):     
      L1d:                   3 MiB (64 instances)
      L1i:                   2 MiB (64 instances)
      L2:                    128 MiB (64 instances)
      L3:                    320 MiB (2 instances)
    NUMA:                    
      NUMA node(s):          4
      NUMA node0 CPU(s):     0-15,64-79
      NUMA node1 CPU(s):     16-31,80-95
      NUMA node2 CPU(s):     32-47,96-111
      NUMA node3 CPU(s):     48-63,112-127
    Vulnerabilities:         
      Itlb multihit:         Not affected
      L1tf:                  Not affected
      Mds:                   Not affected
      Meltdown:              Not affected
      Spec store bypass:     Mitigation; Speculative Store Bypass disabled via prctl and seccomp
      Spectre v1:            Mitigation; usercopy/swapgs barriers and __user pointer sanitization
      Spectre v2:            Mitigation; Enhanced IBRS, IBPB conditional, RSB filling
      Srbds:                 Not affected
      Tsx async abort:       Not affected
    

    dmidecode -t baseboard

    # dmidecode 3.3
    Getting SMBIOS data from sysfs.
    SMBIOS 3.6.0 present.
    # SMBIOS implementations newer than version 3.5.0 are not
    # fully supported by this version of dmidecode.
    
    Handle 0x0002, DMI type 2, 15 bytes
    Base Board Information
            Manufacturer: OEM
            Product Name: T3DGQ
            Version: VER.B
            Serial Number: 22210020THW01030016212503100222
            Asset Tag: N/A
            Features:
                    Board is a hosting board
                    Board is replaceable
            Location In Chassis: Not Specified
            Chassis Handle: 0x0003
            Type: Motherboard
            Contained Object Handles: 0
    
    Handle 0x0045, DMI type 41, 11 bytes
    Onboard Device
            Reference Designation: Onboard VGA
            Type: Video
            Status: Enabled
            Type Instance: 1
            Bus Address: 0000:02:00.0
    
    Handle 0x0046, DMI type 41, 11 bytes
    Onboard Device
            Reference Designation: Onboard SATA Controller
            Type: SATA Controller
            Status: Enabled
            Type Instance: 1
            Bus Address: 0000:00:18.0
    
    Handle 0x0047, DMI type 41, 11 bytes
    Onboard Device
            Reference Designation: Onboard SATA Controller
            Type: SATA Controller
            Status: Enabled
            Type Instance: 2
            Bus Address: 0000:00:19.0
    
    Handle 0x0048, DMI type 41, 11 bytes
    Onboard Device
            Reference Designation: Onboard ETHERNET Controller
            Type: Ethernet
            Status: Enabled
            Type Instance: 1
            Bus Address: 0000:5a:00.0
    
    Handle 0x0049, DMI type 41, 11 bytes
    Onboard Device
            Reference Designation: Onboard ETHERNET Controller
            Type: Ethernet
            Status: Enabled
            Type Instance: 2
            Bus Address: 0000:5b:00.0
    
  • See post chevron_right
    jianjungki
    Members
    metax现在安装了驱动不认卡 已解决 2025年8月25日 10:42

    把内核降低到 5.15.0-25-generic 安装成功了

    但是发现mx-smi 的持久化模式好像不能用
    mx-smi -L
    mx-smi version: 2.2.6
    GPU#0 MXN260 0000:27:00.0 Available (UUID: GPU-3017405f-8bb9-e80a-d0c1-c05985c378ce)
    GPU#1 MXN260 0000:38:00.0 Available (UUID: GPU-c946e025-1f82-e717-d37a-2087991651c1)
    GPU#2 MXN260 0000:a8:00.0 Available (UUID: GPU-1ec42d0e-ae59-e0ee-526c-dfe202e320c2)
    GPU#3 MXN260 0000:b8:00.0 Available (UUID: GPU-be6c05da-a6a3-cd1f-b2d4-e73e5abdff99)

    mx-smi --set-persistence-mode 1 -i 3
    mx-smi version: 2.2.6
    GPU#3 set-persistence-mode Sysfs write error: Resource temporarily unavailable

  • See post chevron_right
    jianjungki
    Members
    metax现在安装了驱动不认卡 已解决 2025年8月21日 16:18

    metax-driver-3.0.0.5-deb-x86_64.run

    安装的是这个驱动

  • See post chevron_right
    jianjungki
    Members
    metax现在安装了驱动不认卡 已解决 2025年8月21日 15:16
    @shuai_chen has written:

    尊敬的开发者您好,请强制刷新固件developer.metax-tech.com/api/client/document/preview/549/C500_mxsmiManual_CN.html#ahs0dv2z2csa1

    mx-smi --force-vbios-flash mxvbios-1.26.1.0-840-N260.bin -i all

    mx-smi version: 2.2.6
    GPU#0 force-vbios-flash Operation not support in target device: Please load metax
    GPU#1 force-vbios-flash Operation not support in target device: Please load metax
    GPU#2 force-vbios-flash Operation not support in target device: Please load metax
    GPU#3 force-vbios-flash Operation not support in target device: Please load metax

  • See post chevron_right
    jianjungki
    Members
    metax现在安装了驱动不认卡 已解决 2025年8月21日 15:13

    mx-smi --force-vbios-flash mxvbios-1.26.1.0-840-N260.bin -i all

    mx-smi version: 2.2.6
    GPU#0 force-vbios-flash Operation not support in target device: Please load metax
    GPU#1 force-vbios-flash Operation not support in target device: Please load metax
    GPU#2 force-vbios-flash Operation not support in target device: Please load metax
    GPU#3 force-vbios-flash Operation not support in target device: Please load metax

  • See post chevron_right
    jianjungki
    Members
    metax现在安装了驱动不认卡 已解决 2025年8月21日 14:57
    modinfo metax
    filename:       /lib/modules/6.8.0-49-generic/extra/metax.ko
    build_date:     2025-07-24 10:39:32
    build_user:     jenkins_sw_bot
    git_commit:     6d6350ab
    c500_vbios_min_version:       1.12.0.0
    vbios_min_version:  2.0.0.0
    license:        GPL and additional rights
    description:    MetaX GPU Kernel Mode Driver
    version:        2.16.0
    import_ns:      "DMA_BUF"
    import_ns:      DMA_BUF
    srcversion:     185AB48B00B0DEC332058FD
    alias:          pci:v00009999d0000401Esv*sd*bc*sc*i*
    alias:          pci:v00009999d00004010sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004021sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004020sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004040sv*sd*bc*sc*i*
    alias:          pci:v00009999d00000011sv*sd*bc*sc*i*
    alias:          pci:v00009999d00000A18sv*sd*bc*sc*i*
    alias:          pci:v00009999d00000A01sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004080sv*sd*bc*sc*i*
    alias:          pci:v00009999d00000A19sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004019sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004038sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004081sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004004sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004011sv*sd*bc*sc*i*
    alias:          pci:v00009999d0000409Bsv*sd*bc*sc*i*
    alias:          pci:v00009999d0000401Csv*sd*bc*sc*i*
    alias:          pci:v00009999d00004028sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004058sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004018sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004000sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004098sv*sd*bc*sc*i*
    alias:          pci:v00009999d00000A03sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004083sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004001sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004006sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004099sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004039sv*sd*bc*sc*i*
    alias:          pci:v00009999d00004029sv*sd*bc*sc*i*
    depends:        ttm
    retpoline:      Y
    name:           metax
    vermagic:       6.8.0-49-generic SMP preempt mod_unload modversions 
    parm:           batches_of_task:how many tasks metax_scheduler processes before appending an EOP? (default is 10) (uint)
    parm:           mb_stress:mailbox stress test(0 disable, nozero indicates ping count)
    parm:           xtt_userptr_limit:set system memory limit for (xtt + userptr) (default is 1)
    parm:           vram_free_sync_watermark:watermark of vram free sync, x/8 of total visible vram (default is 6/8) (ullong)
    parm:           log_flags:log flags
    parm:           log_level:log level(0: NONE 1: EMERG 2: ALERT 3: CRIT 4: ERR 5: WARN 6: NOTICE 7: INFO 8: DBG)
    parm:           sys_hugetlb:system memory map hugepage(maybe 2M) PTE to gpu, default is false (bool)
    parm:           xvm_page_size:VM page size (0 = 4K, 2 = 16K, 4 = 64K, 9 = 2M), default is 9 (int)
    parm:           xcore_page_size:XCORE ATUL1 page size (9 to 16, 9 = 2M, ... 16 = 256M), default is 14, deprecated (int)
    parm:           vf_only:driver only supports vf device, default is 0 (bool)
    parm:           vf_message_disable:disable message between vf and pf, default is false (bool)
    parm:           vf_access_timeout:disable vf access timeout in ms, default is 0 (int)
    parm:           flr_reinit:specify vf device to do flr reinitialization, default is 1 (int)
    parm:           flr_kill:enable vf flr kill bound process feature, default is true (bool)
    parm:           bad_pages:bad pages test, "ip:addr_type:err_code:addr" with comma for multiple entries (string)
    parm:           use_vpud:choose vpud or g2d, default is g2d (bool)
    parm:           vf_indirect_proxy_disable:disable indirect register access proxy, default is 0 (bool)
    parm:           ras_sync:sync ras table form SMP, default is 1 (bool)
    parm:           big_timeout:enable big timeout mode for EMU verification, default is false (bool)
    parm:           atu_inv_by_smp:invalidate ATU by SMP, default is 0 (bool)
    parm:           pasid_table_set_by_smp:update INT xpid-pasid mapping table by SMP, default is 1 (bool)
    parm:           pci_err_irq:use IRQ method to detect PCI error, default is 0 (bool)
    parm:           p2p_cpu_check:check if CPU supports p2pdma, default is 1 (bool)
    parm:           pci_acs_redir_disable:disable pcie acs redirect, default is 0 (bool)
    parm:           bad_page_threshold:maximum bad pages threshold, default is 1024 (int)
    parm:           nonxcore:use to control if the ko will set xcore atu and ce ip (int)
    parm:           mb_polling:enable polling mode for VF mailbox, default is false (bool)
    parm:           stream_mb_enable:enable stream mailbox, default is false (bool)
    parm:           disable_metalink:disable metalink completely, default is false (bool)
    parm:           pri_mem_sz:set private memory size in Kbyte, default is 4K (uint)
    parm:           mas:enable metax advanced storage, default is 0 (bool)
    parm:           support_sku_mismatch:always support creating queue even if SKU mismatched, default is 0 (bool)
    parm:           warmreset_hv_wdog:enable hypervisor warmreset watchdog, default is false (bool)
    parm:           sdma_burst_length:sdma burst length(0xf0 = 64 bytes, 0xfa = 128 bytes, 0xff = 256 bytes), deprecated (int)
    parm:           gpu_sched_timeout:gpu sched timeout in ms, default is 10s (uint)
    parm:           runpm:enable runtime pm, default is 1 (int)
    parm:           ras_ce_mon_debug:enable ras ce mon debug, default is 0 (int)
    parm:           ras_ce_mon_on:enable ras ce mon switch, default is 0 (int)
    parm:           warmreset_timeout_scale:warmreset timeout scale, default is 1 (int)
    parm:           ib_reg_addr:ib reg function address (ulong)
    parm:           ib_unreg_addr:ib unreg function address (ulong)
    parm:           reserve_cap:reserve capacity for multi-process, default is 1 (uint)
    parm:           test_mini_mode:test mini mode, default is 0 (bool)
    parm:           runpm_lp2:enable runtime lp2 power management, default is 0 (int)
    parm:           runpm_link:enable runtime pcie/metalink speed change, default is 1 (int)
    parm:           dump_ras_ce:enable dump ras ce error, default is true (bool)
    parm:           ecc_bypass:ecc bypass param (bit[63:32] are AP bitmask, bit[31:0] are dpc/chain id) (ulong)
    parm:           disable_new_fuse_id:disable new fuse id, default is 0 (int)
    parm:           inner_metalink_5bit_tag:force inner metalink to use 5-bit tag field (default is true) (bool)
    parm:           enable_warmreset_select_gpu:allow selected GPU to do warmreset independently, default is false (bool)
    parm:           enable_tba_in_vram:enable tba memory in vram for xcore initialization, default is false (bool)
    parm:           fw_path:customized firmware image search path with a higher priority than default path (string)
    parm:           gpu_model:GPU device model (0 = native, 1 = sgpu), default is 0 (int)
    parm:           topo_df:df topo for mgpu, default is 0, 0 fixed topo, 1 df topo (int)
    parm:           disable_pci_cpl_timeout:disables the completion timeout mechanism of pcie, default is false (bool)
    parm:           metax_layout_enable:activate the metax/mxcd layout within /sys/class folder, default is false (bool)
    parm:           wipe_on_release:clear vram to 0 when vram is freed, default is true (bool)
    parm:           mailbox_timeout_scale:SMP mailbox timeout scale, default is 1
    parm:           mxcd_log_flags:mxcd log flags
    parm:           mxcd_log_level:log level (0:NONE, 1: EMERG, 2: ALERT, 3: CRIT, 4: ERR, 5: WARN, 6: NOTICE, 7: INFO, 8: DBG)
    parm:           umh_path:Driver exception UMH(user mode helper) path (charp)
    
    lspci | grep 9999
    27:00.0 Display controller: Device 9999:4083 (rev 01)
    38:00.0 Display controller: Device 9999:4083 (rev 01)
    a8:00.0 Display controller: Device 9999:4083 (rev 01)
    b8:00.0 Display controller: Device 9999:4083 (rev 01)
    

    lsmod | egrep "(metax|mxgvm)"
    root@amax:/lib/firmware/metax/mxc500#

    显示空白,什么都没有,按理来说需要加载metax

    mx-smi -u mxvbios-1.26.1.0-840-N260.bin -t 600
    mx-smi version: 2.2.6
    Hint: -u only support upgrading vbios for all devices.
    No available devices were discovered

    安装固件也显示没有设备

    uname -m && uname -r && lsb_release -a

    x86_64
    6.8.0-49-generic
    No LSB modules are available.
    Distributor ID: Ubuntu
    Description: Ubuntu 22.04.5 LTS
    Release: 22.04
    Codename: jammy

  • 沐曦开发者论坛
powered by misago