37 lines
1.8 KiB
Plaintext
37 lines
1.8 KiB
Plaintext
#
|
|
# NVreg_UsePageAttributeTable=1 (Default 0) - Activating the better memory
|
|
# management method (PAT). The PAT method creates a partition type table at a
|
|
# specific address mapped inside the register and utilizes the memory
|
|
# architecture and instruction set more efficiently and faster. If your system
|
|
# can support this feature, it should improve CPU performance.
|
|
#
|
|
# NVreg_InitializeSystemMemoryAllocations=0 (Default 1) - Disables clearing
|
|
# system memory allocation before using it for the GPU. Potentially improves
|
|
# performance, but at the cost of increased security risks. Write "options
|
|
# nvidia NVreg_InitializeSystemMemoryAllocations=1" in
|
|
# /etc/modprobe.d/nvidia.conf, if you want to return the default value. Note:
|
|
# It is possible to use more memory (?)
|
|
#
|
|
# NVreg_DynamicPowerManagement=0x02 - Enables the use of dynamic power
|
|
# management for Turing generation mobile cards, allowing the dGPU to be
|
|
# powered down during idle time.
|
|
#
|
|
# NVreg_RegistryDwords=RMIntrLockingMode=1 (default 0) - enables experimental
|
|
# switch for better frame-pacing this mainly improves it for high refresh rate
|
|
# monitors with VRR or VR headsets.
|
|
#
|
|
# Note: This only works for PRIME configurations if your dGPU is controlling an
|
|
# external monitor.
|
|
#
|
|
# For example: At 240Hz each frame is expected every 4ms. But if a 1ms
|
|
# task—say, in the kernel or on the GSP — runs when a frame is about to be
|
|
# displayed, it can delay the rendering. Instead of a neat sequence at T+4ms,
|
|
# T+8ms, T+12ms, the frames might appear at T+4ms, T+9ms, T+12ms, etc. This
|
|
# shows how even small delays can shift frame timing, potentially impacting
|
|
# smooth display output.
|
|
#
|
|
options nvidia NVreg_UsePageAttributeTable=1 \
|
|
NVreg_InitializeSystemMemoryAllocations=0 \
|
|
NVreg_DynamicPowerManagement=0x02 \
|
|
NVreg_RegistryDwords=RMIntrLockingMode=1
|