F5F Stay Refreshed Software Operating Systems No one has shared their experience with the Ryzen optimized OpenMandriva Lx version.

No one has shared their experience with the Ryzen optimized OpenMandriva Lx version.

No one has shared their experience with the Ryzen optimized OpenMandriva Lx version.

E
EdoubleO
Member
238
01-01-2023, 01:14 AM
#1
I haven't built a Ryzen system yet, but I'm considering assembling a new PC centered around Ryzen. Something about OpenMandriva Lx offers both standard and a Ryzen-specific version. Have others examined this? What performance gains do you think are possible by tailoring the whole distribution for Ryzen architecture? From what I understand, it's the only option like that, so it would be worth checking if it delivers similar improvements to older x86 Linux builds when switching from i386 to i586 or i686.
E
EdoubleO
01-01-2023, 01:14 AM #1

I haven't built a Ryzen system yet, but I'm considering assembling a new PC centered around Ryzen. Something about OpenMandriva Lx offers both standard and a Ryzen-specific version. Have others examined this? What performance gains do you think are possible by tailoring the whole distribution for Ryzen architecture? From what I understand, it's the only option like that, so it would be worth checking if it delivers similar improvements to older x86 Linux builds when switching from i386 to i586 or i686.

P
164
01-02-2023, 01:32 AM
#2
This approach can feel isolating since there aren’t many people around to share experiences. Previous CPU improvements have faded because storage capacity has increased and performance per instruction has risen. They usually optimize by using chip-specific features and dropping unused functions. Setup tends to be quicker, resulting in smaller components and potentially unique functionalities not found in broader versions. There might be some useful shortcuts for one type but not others. Don’t rely heavily on speed improvements that don’t depend on setup time.
P
PandaBearMan23
01-02-2023, 01:32 AM #2

This approach can feel isolating since there aren’t many people around to share experiences. Previous CPU improvements have faded because storage capacity has increased and performance per instruction has risen. They usually optimize by using chip-specific features and dropping unused functions. Setup tends to be quicker, resulting in smaller components and potentially unique functionalities not found in broader versions. There might be some useful shortcuts for one type but not others. Don’t rely heavily on speed improvements that don’t depend on setup time.

C
Commando__
Senior Member
744
01-02-2023, 09:07 AM
#3
I've never been married, so maybe that's the case? If I had, it'd likely be a joke from an old man.
C
Commando__
01-02-2023, 09:07 AM #3

I've never been married, so maybe that's the case? If I had, it'd likely be a joke from an old man.

R
Razlorus
Posting Freak
976
01-07-2023, 03:21 PM
#4
The variation is usually small unless you consider many factors. The primary improvements will stem from adjustments in the kernel and file system settings, though these also hinge on your intended use since each option involves trade-offs. For other aspects, optimizations within packages mainly rely on the developer’s decisions and the version provided by your distribution.
R
Razlorus
01-07-2023, 03:21 PM #4

The variation is usually small unless you consider many factors. The primary improvements will stem from adjustments in the kernel and file system settings, though these also hinge on your intended use since each option involves trade-offs. For other aspects, optimizations within packages mainly rely on the developer’s decisions and the version provided by your distribution.

T
147
01-09-2023, 03:00 PM
#5
My setup has always used march=znver1. I had to remove some CPU_FLAGS from bdver2 and recompile the base system before installing the hard drive on the new machine. Look for a suitable bench test, check its current flags, and adjust my package accordingly for a proper comparison.
T
TannerBannaner
01-09-2023, 03:00 PM #5

My setup has always used march=znver1. I had to remove some CPU_FLAGS from bdver2 and recompile the base system before installing the hard drive on the new machine. Look for a suitable bench test, check its current flags, and adjust my package accordingly for a proper comparison.