SPEC CPU2017 Flag Description - Platform settings

Operating System Tuning Parameters

ulimit

This sets user limits of system-wide resources and can set the stack size to n kbytes, or unlimited to allow the stack size to grow without limit. Some common ulimit commands may include:

• ulimit -s [n | unlimited]: Set the stack size to n kbytes, or unlimited to allow the stack size to grow without limit.
• ulimit -l (number): Set the maximum size that can be locked into memory.

Kernel parameters

The following Linux Kernel parameters were set for better optimize performance.

• dirty_ratio: This setting is the absolute maximum amount of system memory that can be filled with dirty pages before everything must get committed to disk.The command like "sysctl -w vm.dirty_ratio=8" is used to set.
• drop caches: Free the file system page cache. The command like "echo 1> /proc/sys/vm/drop_caches" is used to free up the filesystem page cache.
• randome va space: This setting can disable address space layout randomization (ASLR). The command like "sysctl -w kernel.randomize_va_space=0" is used to reduce run-to-run variability.
• swappiness: The swappiness value can range from 1 to 100. A value of 100 will cause the kernel to swap out inactive processes frequently in favor of file system performance, resulting in large disk cache sizes. A value of 1 tells the kernel to only swap processes to disk if absolutely necessary. This can be set through a command like "sysctl -w vm.swappiness=1"
• Transparent Hugepages (THP): THP is an abstraction layer that automates most aspects of creating, managing and using huge pages. Huge pages increase the memory page size from 4 KB to 2 MB. The command like "echo always > /sys/kernel/mm/transparent_hugepage/enabled" and "echo always > /sys/kernel/mm/transparent_hugepage/defrag" is used to enable THP
• zone reclaim mode: Zone reclaim allows the reclaiming of pages from a zone if the number of free pages falls below a watermark even if other zones still have enough pages available. Reclaiming a page can be more beneficial than taking the performance penalties that are associated with allocating a page on a remote zone, especially for NUMA machines. To tell the kernel to free local node memory rather than grabbing free memory from remote nodes, use a command like "sysctl -w vm.zone_reclaim_mode=1"

Firmware / BIOS / Microcode Settings

Determinism Slider

This BIOS switch allows 2 options: "Power", and "Performance". This feature is for the determinism to control performance. "Perforomance" setting uses default values for deterministic performance control. "Power" setting provides predicable performance across all processors of the same type. "Power" setting maximizes performance withing the power limits defined by "TDP Limit". Default is "Performance".

Fan Control

This BIOS switch allows 2 options: "Auto" and "Full". The default setting is "Auto", which allows the system to control the fan speed according to the system temperature. If "Full" is selected, the system runs fans at 100% speed and it mayimprove the system performance. But it increases the power consumption of the system.

Interleaving Region Size

This BIOS switch allows 3 options: "1K Region Size", and "2K Region Size", and "Auto". This feature can select interleaving region size between a 1K or 2K region size. Default is "Auto".

L1 Stride Prefetcher

This BIOS switch allows 2 options: "Disabled", "Enabled, and "Auto". This feature allows enabling or disabling of L1 Stride Prefetcher, which uses the L1 cache memory access history of individual instructions to fetch additional lines when each access is a constant distance from the previous. Default is "Auto".

L1 Region Prefetcher

This BIOS switch allows 2 options: "Disabled", "Enabled, and "Auto". This feature allows enabling or disabling of L1 Region Prefetcher, which uses the L1 cache memory access history to fetch additional lines when the data access for a given instruction that tends to be followed by a consistent patter of other access within a localized region. Default is "Auto".

NUMA nodes per socket

This BIOS switch allows 4 options: "NPS0", "NPS1", "NPS2", and "NPS4". This feature specifies the number of desired NUMA nodes per populated socket in the system:

• NPS0: All physical processor is a NUMA node and memory accesses are interleaved across all memory channels of two sockets together.
• NPS1: Each physical processor is a NUMA node, and memory accesses are interleaved across all memory channels directly connected to the physical processor.
• NPS2: Each physical processor is two NUMA nodes, and memory accesses are interleaved across 4 memory channels.
• NPS4: Each physical processor is four NUMA nodes, and memory accesses are interleaved across 2 memory channels.

Default is "NPS1".

Probe Filter Organization

This BIOS switch allows 3 options: "Auto", "Dedicated", and "Shared". This feature specifies whether multiple memory channels will share ("Shared") probe filter storage or not ("Dedicated"). For memory sizes of 16 TB or larger, this feature is automatically set to "Shared". "Auto" selects the optimized setting in the configuration. Default is "Dedicated".

Package Power Limit Control

This BIOS switch allows 2 options: "Manual" and "Auto". This feature enables or disables user to specify "Package Power Limit" "Auto" uses the fused values. "Manual" enables user to configure customized values in "Package Power Limit" switch. Default is "Auto".

Package Power Limit

This BIOS switch specifies the maximum power that each CPU package may consume in the system. The actual power limit is the maximum of the Package Power Limit and "TDP Limit".

Power Profile Selection

This BIOS switch allows 3 options: "High Performance Mode", "Efficiency Mode", and "Maximum IO Performance Mode". This feature select DF Pstate based on each profile policy. Default is "Efficiency Mode".

SMT Control

This BIOS switch allows 2 options: "Enabled" and "Disabled". This feature allows enabling or disabling of symmetric multithreading on processors. When enabled ("Enabled"), each physical processor core operates as two logical processor cores. When disabled ("Disabled"), each physical core operates as only one logical processor core. "Enabled" can improve overall performance for applications that benefit from a higher processor core count. Default is "Enabled".

SVM Mode

This BIOS switch allows 2 options: "Enabled" and "Disabled". This feature allows enabling or disabling CPU Virtualization feature based on AMD Secure Virtual Machine architecture. Default is "Enabled".

TDP Control

This BIOS switch allows 2 options: "Manual" and "Auto". This feature enables or disables user to specify "TDP Limit". "Auto" uses the fused values. "Manual" enables user to configure customized values in "TDP Limit" switch. Default is "Auto".

TDP Limit

This BIOS switch specifies the maximum power that the CPU will consume, up to the platform power limit. Valid values vary by CPU model. If value outside the valid range is set, the CPU will automatically adjust the value so that it does fall within the valid range. When increasing TDP Limit, additional power will only be consumed up to the Package Power Limit, which may be less than the TDP Limit setting.

Model	Minimum TDP	Maximum TDP
EPYC 9845	320	400
EPYC 9745	320	400
EPYC 9655	320	400
EPYC 9575F	320	400
EPYC 9555	320	400
EPYC 9475F	320	400
EPYC 9455	240	300
EPYC 9375F	320	400
EPYC 9355	240	300
EPYC 9275F	320	400
EPYC 9255	200	240
EPYC 9175F	320	400
EPYC 9135	200	240
EPYC 9115	125	155
EPYC 9015	120	155