Command Line Options - Command Line Options

Command Line Options - Command Line Options - 57368

uProf User Guide

Document ID

57368

Release Date

2025-06-09

Revision

5.1 English

Here is a list of all command line options.

Table 1. AMDuProfPcm Options
Option	Description
`-h`	Displays this help information on the console/terminal.
`-m <metric,...>`	Metrics to report. The supported metric groups and the corresponding metrics are Platform, OS, and Hypervisor specific. Run `./AMDuProfPcm -h` to get the list of supported metrics. The following metric groups are supported: Core ipc – reports metrics such as CEF, Utilization, CPI, and IPC fp – reports GFLOPS dc – advanced caching metrics such as DC refills by source (supported only on AMD “Zen 3”, AMD “Zen 4”, and AMD"Zen 5" processors) l1 – L1 cache related metrics (DC access and IC Fetch miss ratio) l2 – L2D and L2I cache related access/hit/miss metrics swpfdc – software prefetch data cache from various nodes and CCX (supported only on AMD “Zen 3”, AMD “Zen 4”, and AMD “Zen 5” processors) hwpfdc – hardware prefetch data cache from various nodes and CCX (supported only on AMD “Zen 3”, AMD “Zen 4”, and AMD “Zen 5” processors) pipeline_util – top-down metrics to visualize the bottlenecks in the CPU pipeline (supported only on AMD “Zen 4” and AMD “Zen 5” processors) avx_imix - SSE/AVX instruction mix percentage cache_miss - L1 Data and Insturction cache misses, L2 Data and Code Read misses tlb - translation lookaside buffer misses L3 l3 – L3 cache metrics like L3 Access, L3 Miss, and Average Miss latency DF Metrics memory – approximate memory read and write bandwidths in GB/s for all the channels pcie – PCIe bandwidth in GB/s (supported only on AMD “Zen 2”, AMD “Zen 4”, and AMD “Zen 5” processors) xgmi – approximate xGMI outbound databytes in GB/s for all the remote links dma – DMA bandwidth in GB/s (supported only on AMD “Zen 4” and AMD “Zen 5” processors) ccm_bw - Core complex inbound and outbound bandwidth in GB/s (supported only on AMD “Zen 4” processors and later) cxl - Compute Express Link bandwidth in GB/s (supported only on AMD “Zen 5” processors and later)
`-c <core\|ccx\|l3\|numa\|package>=<n>`	Collect from the specified core \| ccx \| ccd \| package. The default is 'core=0'. If 'ccx' or 'l3' is specified: Core events will be collected from all the cores of this ccx. l3 event will be collected for ccx this core belongs to. df events will be collected for the package this core belongs to. If 'package' is specified: Core events will be collected for all the cores of this package. l3 events will be collected for all the ccxs of this package. df events will be collected for this package. If 'numa' is specified: Core events will be collected from all the cores of this numa. l3 events will be collected for all the ccxs of this numa. df events will be collected for all the packages of this numa.
`-a`	Collect from all the cores. Note: Options `-c` and `-a` cannot be used together.
`-C`	Prints the cumulative data at the end of the profile duration. Else, all the samples will be reported as timeseries data.
`-A <system,package,ccd,ccx,core>`	Prints aggregated metrics at various component levels. The following granularities are supported: system – samples from all the cores in the system will be aggregated package – samples from all the cores in the package will be aggregated and reported for all the packages available in the system; applicable for multi-package systems. ccd – samples from all the cores in CCD will be aggregated and reported for all the CCDs. ccx – samples from all the cores in CCX will be aggregated and reported for all the CCXs. Note: CCX is only applicable to Core and L3 metrics. core – samples from all the logical cores on which samples are collected will be reported without aggregation. Note: Option `-a` should be used along with this option to collect samples from all the cores. Note: Comma separated list of components can be specified.
`-I <ms>`	Print the metrics at regular intervals. By default, it is enabled with an interval of 1000 ms.
`-i <config file>`	User defined XML config file that specifies Core\|L3\|DF counters to monitor. Refer sample files in the `<install-dir>/bin/Data/Config/` directory for the format. Note: Options `-i` and `-m` cannot be used together. If option `-i` is used, all the events mentioned in the user defined config file will be collected.
`-d <seconds>`	Profile duration to run.
`-t <multiplex interval in ms>`	The interval in which pmc count values will be read, the minimum is 100 ms.
`-o <output file>`	The output file name, it is in CSV format.
`-P <n>`	Sets precision of the metrics reported, the default value is 2.
`-q`	Hide CPU topology section in the output report.
`-r`	Force resets the MSRs.
`-k`	Prefixes 'pkg' in package level counters.
`-s`	Displays time stamp in the time series report.
`-l`	Lists the supported raw PMC events.
`-z <pmc-event>`	Prints the name, description, and available unit masks for the event.
`-w <dir>`	Specifies the working directory. The default will be the path of the launched application.
`-n`	Print cpu topology.
`-v`	Print version.
`-X`	Collect data using perf subsystem without root privileges. This option is enabled by default and will be deprecated soon. Note: This is only supported on Linux.
`-p <process ID>`	Specify the target process ID to monitor. Note: This is only supported with the option `-X` on Linux.
`-f <util:<n>>`	Filter the roofline data based on the utilization. For example, `-f util:90` will filter all data points with less than 90% utilization. Note: This is applicable only with the roofline command.
`--html`	Collect data and generate HTML report.
`--msr`	Collect data using MSR mode. Requires root privilege.
`--percentile`	Generate custom percentile in html report. Default is 95^th Percentile.
`--collect-xgmi`	Collect xgmi data.
`--collect-pcie`	Collect pcie data.
`--collect-power`	Collect power data.
`--per-core`	Prints metrics at core level (no aggregation).
`--per-die`	Prints aggregated metrics at die level.
`--per-socket`	Prints aggregated metrics at socket level.
`--report-roofline`	Report roofline data under the 'profile' command.
`--no-aggr`	Prints non-aggregated metrics data: Core metrics will be reported at core level. l3 metrics will be reported at ccx level. df metrics will be printed at package/aid level.
`--collect-clk`	Collect clock data.
`--collect-guest`	Count only the guest events. Note: This is applicable only in the host when hypervisor is enabled.
`--collect-host`	Count only the host events. (Default behavior is to collect host and guest data). Note: This is applicable only in the host when hypervisor is enabled.
`--read-smbios`	Read memory speed and total memory channels from SMBIOS. Applicable only for roofline command.
`--show-used-ccx`	Show only the CCXs where the target application ran in the heatmap's L3 Cache section in the html report. Not applicable with `--msr` option.
`-O <output dir path>`	Path to create the output directory.