J. Roeleveld wrote:
On December 8, 2018 6:23:04 PM UTC, Dale <rdalek1967@gmail.com> wrote:
Alexander Puchmayr wrote:
Am Donnerstag, 6. Dezember 2018, 10:27:31 CET schrieb Dale:
Howdy, I mentioned in other threads that I'm doing some upgrades to my system. My first question is about a CPU upgrade. I currently have this for my CPU, from cpuinfo: AMD Phenom(tm) II X4 955 Processor I've bought but not yet installed a FX-8350 CPU. I have this in my make.conf file: CFLAGS="-march=native -O2 -pipe"
Compiling the whole system with -march=native might lead to troubles, especially when doing a CPU change. This option means that gcc is determining the type of CPU automatically and adjusts the instruction set used to exactly this CPU. Although, in your case, it is highly likely that your new CPU understands all commands from the old, but I wouldn't bet on it. Its possible that your existing software encounters problems like "illegal instruction" or the like. Very bad if your compiler crashes after CPU replacement, then you cannot emerge anything. I highly recommend using CFLAGS="-O2 -pipe" and nothing more, the performance difference is, if measurable at all, negligible.
USE_CPU="fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt nodeid_msr hw_pstate npt lbrv svm_lock nrip_save"
As someone else in this thread already mentioned, USE_CPU is not used. What you're looking for is CPU_FLAGS_X86=..., which defines what cpu-specific options will be enabled for packages supporting it and where it makes sense. See package cpuid2cpuflags for details. Regards Alex
It seems the holiday shopping is slowing down delivery.  My fan was supposed to be here today but didn't arrive.  Since I got time, I'll change the CFLAGS for at least the @system stuff, that should get me booted for sure.  While the native setting makes things easier for normal use, I can see the point of not using it when changing CPUs.  That is one reason for this thread.  The CPUs are different and may require some changes during the swap.  Is there a easy way to see what if any changes will be made?  I did a emerge -UDNa @system but it's not showing any change.  Does it require a emerge -e @system to force the change?  Or is it not changing anything? Thanks much.  Better safe than sorry.  ;-) Dale :-)  :-) 

A CFLAGS change requires a rebuild of all packages done with gcc. I am not aware of a simple way of only doing those, so a "emerge --empty @world" will be needed.

--
Joost
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.


Based on the output, that's what I was thinking. Emerge picks up on other USE changes but it seems it only grabs the CFLAGS during the compile/configure phase for each package. Would this change the kernel image as well or would it remain the same?  I may build a new kernel just to be sure.

One good thing about this, I can compare the times with current CPU and new CPU later and get a rough idea of speed increases.  ;-) 

Pardon me while I generate some heat.  o_O

Dale

:-)  :-)