ROCM-SMI(1)

📦
version 10.0 · 3 versions
, in debian trixie, ubuntu questing, ubuntu resolute, ubuntu stonking

🎀 ROCM-SMI 💅

okay bestie, gather round, because this is rocm-smi — the manual page for it, specifically, in section 1 of the User Commands manual. it's a tool to monitor AMD accelerators and GPUs. yes, that's the whole NAME section, and yes it's iconic. ✨

✨ SYNOPSIS ✨

usage: rocm-smi [-h] [-V] [-d DEVICE [DEVICE ...]] [--alldevices]
       [--showhw] [-a] [-i] [-v] [-e [EVENT ...]]

              [--showdriverversion] [--showtempgraph] [--showfwinfo [BLOCK
              ...]] [--showmclkrange] [--showmemvendor] [--showsclkrange]
              [--showproductname] [--showserial] [--showuniqueid]
              [--showvoltagerange] [--showbus] [--showpagesinfo]
              [--showpendingpages] [--showretiredpages]
              [--showunreservablepages] [-f] [-P] [-t] [-u] [--showmemuse]
              [--showvoltage] [-b] [-c] [-g] [-l] [-M] [-m] [-o] [-p] [-S] [-s]
              [--showmeminfo TYPE [TYPE ...]] [--showpids [VERBOSE]]
              [--showpidgpus [SHOWPIDGPUS ...]]  [--showreplaycount]
              [--showrasinfo [SHOWRASINFO ...]] [--showvc] [--showxgmierr]
              [--showtopo] [--showtopoaccess] [--showtopoweight]
              [--showtopohops] [--showtopotype] [--showtoponuma]
              [--showenergycounter] [--shownodesbw] [--showcomputepartition]
              [--showmemorypartition] [-r] [--resetfans] [--resetprofile]
              [--resetpoweroverdrive] [--resetxgmierr] [--resetperfdeterminism]
              [--resetcomputepartition] [--resetmemorypartition] [--setclock
              TYPE LEVEL] [--setsclk LEVEL [LEVEL ...]]  [--setmclk LEVEL
              [LEVEL ...]] [--setpcie LEVEL [LEVEL ...]] [--setslevel SCLKLEVEL
              SCLK SVOLT] [--setmlevel MCLKLEVEL MCLK MVOLT] [--setvc POINT
              SCLK SVOLT] [--setsrange SCLKMIN SCLKMAX] [--setextremum min|max
              sclk|mclk CLK] [--setmrange MCLKMIN MCLKMAX] [--setfan LEVEL]
              [--setperflevel LEVEL] [--setoverdrive %] [--setmemoverdrive %]
              [--setpoweroverdrive WATTS] [--setprofile SETPROFILE]
              [--setperfdeterminism SCLK] [--setcomputepartition
              {CPX,SPX,DPX,TPX,QPX,cpx,spx,dpx,tpx,qpx}] [--setmemorypartition
              {NPS1,NPS2,NPS4,NPS8,nps1,nps2,nps4,nps8}] [--rasenable BLOCK
              ERRTYPE] [--rasdisable BLOCK ERRTYPE] [--rasinject BLOCK]
              [--gpureset] [--load FILE | --save FILE] [--autorespond RESPONSE]
              [--loglevel LEVEL] [--json] [--csv]

💖 DESCRIPTION 💖

so the way this works is: that whole wall of flags up there is the description. rocm-smi takes all of them. and then, right under the usage text, the tool announces itself like the main character it is:

AMD ROCm System Management Interface | ROCM-SMI version: 2.2.0

that's the vibe check and it passed. 🫶

🌸 OPTIONS 🌸

options:

  • -h, --help — shows this help message and then exits. the "wait what does this do again" button, we love her.
  • --gpureset — resets the specified GPU. heads up bestie: exactly one GPU must be specified.
  • --load FILE — loads Clock, Fan, Performance and Profile settings from FILE.
  • --save FILE — saves Clock, Fan, Performance and Profile settings to FILE.
  • -V, --version — shows version information.
  • -d DEVICE [DEVICE ...], --device DEVICE [DEVICE ...] — run the command on the device(s) you specify.

Display Options:

  • --alldevices — no description given in the source, it's just out here living its truth 💀
  • --showhw — shows Hardware details.
  • -a, --showallinfo — shows Temperature, Fan and Clock values.

Topology:

  • -i, --showid — shows DEVICE IDs.
  • -v, --showvbios — shows the VBIOS version.
  • -e [EVENT ...], --showevents [EVENT ...] — shows the event list.
  • --showdriverversion — shows the kernel driver version.
  • --showtempgraph — shows the Temperature Graph.
  • --showfwinfo [BLOCK ...] — shows FW information.
  • --showmclkrange — shows the mclk range.
  • --showmemvendor — shows the GPU memory vendor.
  • --showsclkrange — shows the sclk range.
  • --showproductname — shows product details.
  • --showserial — shows the GPU's Serial Number.
  • --showuniqueid — shows the GPU's Unique ID.
  • --showvoltagerange — shows the voltage range.
  • --showbus — shows the PCI bus number.

Pages information:

  • --showpagesinfo — shows retired, pending and unreservable pages. the whole trio ✨
  • --showpendingpages — shows pending retired pages.
  • --showretiredpages — shows retired pages.
  • --showunreservablepages — shows unreservable pages.
  • -f, --showfan — shows the current fan speed.
  • -P, --showpower — shows the current average or instant socket graphics package power consumption.
  • -t, --showtemp — shows the current temperature.
  • -u, --showuse — shows the current GPU use.
  • --showmemuse — shows the current GPU memory used.
  • --showvoltage — shows the current GPU voltage.
  • -b, --showbw — shows estimated PCIe use.
  • -c, --showclocks — shows the current clock frequencies.
  • -g, --showgpuclocks — shows the current GPU clock frequencies.
  • -l, --showprofile — shows Compute Profile attributes.
  • -M, --showmaxpower — shows the maximum graphics package power this GPU will consume.
  • -m, --showmemoverdrive — shows the current GPU Memory Clock OverDrive level.
  • -o, --showoverdrive — shows the current GPU Clock OverDrive level.
  • -p, --showperflevel — shows the current DPM Performance Level.
  • -S, --showclkvolt — shows supported GPU and Memory Clocks and Voltages.
  • -s, --showclkfrq — shows supported GPU and Memory Clock.
  • --showmeminfo TYPE [TYPE ...] — shows Memory usage information for the given block(s) TYPE.
  • --showpids [VERBOSE] — shows current running KFD PIDs (pass details to VERBOSE if you want the detailed version, nosey bestie 👀).
  • --showpidgpus [SHOWPIDGPUS ...] — shows GPUs used by the specified KFD PIDs (all of them if no arg is given).
  • --showreplaycount — shows PCIe Replay Count.
  • --showrasinfo [SHOWRASINFO ...] — shows RAS enablement information and error counts for the specified block(s) (all if no arg is given).
  • --showvc — shows the voltage curve.
  • --showxgmierr — shows XGMI error information since the last read.
  • --showtopo — shows hardware topology information.
  • --showtopoaccess — shows the link accessibility between GPUs.
  • --showtopoweight — shows the relative weight between GPUs.
  • --showtopohops — shows the number of hops between GPUs.
  • --showtopotype — shows the link type between GPUs.
  • --showtoponuma — shows the numa nodes.
  • --showenergycounter — Energy accumulator that stores amount of energy consumed.
  • --shownodesbw — shows the numa nodes.
  • --showcomputepartition — shows current compute partitioning.
  • --showmemorypartition — shows current memory partition.

Set options:

  • --setclock TYPE LEVEL — sets Clock Frequency Level(s) for the specified clock. (requires manual Perf level.)
  • --setsclk LEVEL [LEVEL ...] — sets GPU Clock Frequency Level(s). (requires manual Perf level.)
  • --setmclk LEVEL [LEVEL ...] — sets GPU Memory Clock Frequency Level(s). (requires manual Perf level.)
  • --setpcie LEVEL [LEVEL ...] — sets PCIE Clock Frequency Level(s). (requires manual Perf level.)
  • --setslevel SCLKLEVEL SCLK SVOLT — changes the GPU Clock frequency (MHz) and Voltage (mV) for a specific Level.
  • --setmlevel MCLKLEVEL MCLK MVOLT — changes the GPU Memory clock frequency (MHz) and Voltage for (mV) a specific Level.
  • --setvc POINT SCLK SVOLT — changes the SCLK Voltage Curve (MHz mV) for a specific point.
  • --setsrange SCLKMIN SCLKMAX — sets the min and max SCLK speed.
  • --setextremum min|max sclk|mclk CLK — sets the min/max of SCLK/MCLK speed.
  • --setmrange MCLKMIN MCLKMAX — sets the min and max MCLK speed.
  • --setfan LEVEL — sets the GPU Fan Speed (Level or %).
  • --setperflevel LEVEL — sets the Performance Level.
  • --setoverdrive % — sets the GPU OverDrive level. (requires manual|high Perf level.)
  • --setmemoverdrive % — sets the GPU Memory Overclock OverDrive level. (requires manual|high Perf level.)
  • --setpoweroverdrive WATTS — sets the maximum GPU power using Power OverDrive, in Watts.
  • --setprofile SETPROFILE — specify a Power Profile level (#), or a quoted string of CUSTOM Profile attributes like "# # # #...". (requires manual Perf level.)
  • --setperfdeterminism SCLK — sets a clock frequency limit to get minimal performance variation.
  • --setcomputepartition {CPX,SPX,DPX,TPX,QPX,cpx,spx,dpx,tpx,qpx} — sets the compute partition.
  • --setmemorypartition {NPS1,NPS2,NPS4,NPS8,nps1,nps2,nps4,nps8} — sets the memory partition.
  • --rasenable BLOCK ERRTYPE — enables RAS for the specified block and error type.
  • --rasdisable BLOCK ERRTYPE — disables RAS for the specified block and error type.
  • --rasinject BLOCK — injects RAS poison for the specified block. ONLY WORKS ON UNSECURED BOARDS — like, do not try this on a locked-down one, it will simply not eat. 💀

Reset options:

  • -r, --resetclocks — resets clocks and OverDrive to default.
  • --resetfans — resets fans to automatic (driver) control.
  • --resetprofile — resets the Power Profile back to default.
  • --resetpoweroverdrive — sets the maximum GPU power back to the device default state.
  • --resetxgmierr — resets the XGMI error count.
  • --resetperfdeterminism — disables performance determinism.
  • --resetcomputepartition — resets to the boot compute partition state.
  • --resetmemorypartition — resets to the boot memory partition state.

Auto-response options:

  • --autorespond RESPONSE — the response to automatically provide for all prompts. NOT RECOMMENDED, and we mean it — you're basically telling the tool to answer for you and it will take you at your word. 🙈

Output options:

  • --loglevel LEVEL — how much output gets printed for what the program is doing. one of debug/info/warning/error/critical.
  • --json — print output in JSON format.
  • --csv — print output in CSV format.

🦋 SEE ALSO 🦋

The full documentation for rocm-smi is maintained as a Texinfo manual. If the info and rocm-smi programs are properly installed at your site, then this command

info rocm-smi

should give you access to the complete manual. iykyk 🧚

okay that's the whole thing, you're fully caught up bestie. this was rocm-smi 2.2.0, from November 2024, section ROCM-SMI(1) — go forth and monitor those accelerators. slay, periodt, and don't forget to reset your fans. 💜🌈

📦 where this page comes from