fleers commented 2 years ago

make

cc  -Wall -Wfatal-errors -pthread -c corefreqd.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqd.o
cc  -Wall -Wfatal-errors -c corefreqm.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqm.o
cc  -Wall -Wfatal-errors corefreqd.c corefreqm.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqd -lpthread -lm -lrt
cc  -Wall -Wfatal-errors -c corefreq-cli.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli.o
cc  -Wall -Wfatal-errors -c corefreq-ui.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-ui.o
cc  -Wall -Wfatal-errors -c corefreq-cli-rsc.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli-rsc.o
cc  -Wall -Wfatal-errors -c corefreq-cli-json.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-cli-json.o
cc  -Wall -Wfatal-errors -c corefreq-cli-extra.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-cli-extra.o
cc  -Wall -Wfatal-errors \
  corefreq-cli.c corefreq-ui.c corefreq-cli-rsc.c \
  corefreq-cli-json.c corefreq-cli-extra.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli -lm -lrt
make -j1 -C /lib/modules/4.18.0-240.el8.x86_64/build M=/tmp/CoreFreq modules
make[1]: Entering directory '/usr/src/kernels/4.18.0-240.el8.x86_64'
  CC [M]  /tmp/CoreFreq/corefreqk.o
/tmp/CoreFreq/corefreqk.c:358:26: error: initialization of ‘int (*)(struct cpufreq_policy_data *)’ from incompatible pointer type ‘int (*)(struct cpufreq_policy *)’ [-Werror=incompatible-pointer-types]
  /*MANDATORY*/ .verify = CoreFreqK_Policy_Verify,
                          ^~~~~~~~~~~~~~~~~~~~~~~
compilation terminated due to -Wfatal-errors.
cc1: some warnings being treated as errors
make[2]: *** [scripts/Makefile.build:322: /tmp/CoreFreq/corefreqk.o] Error 1
make[1]: *** [Makefile:1544: _module_/tmp/CoreFreq] Error 2
make[1]: Leaving directory '/usr/src/kernels/4.18.0-240.el8.x86_64'
make: *** [Makefile:76: all] Error 2

cat /etc/redhat-release

CentOS Linux release 8.3.2011

uname -a

Linux gb 4.18.0-240.el8.x86_64 #1 SMP Fri Sep 25 19:48:47 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

lscpu

Architecture:        x86_64
CPU op-mode(s):      32-bit, 64-bit
Byte Order:          Little Endian
CPU(s):              128
On-line CPU(s) list: 0-127
Thread(s) per core:  1
Core(s) per socket:  64
Socket(s):           2
NUMA node(s):        32
Vendor ID:           AuthenticAMD
CPU family:          23
Model:               49
Model name:          AMD EPYC 7742 64-Core Processor
Stepping:            0
CPU MHz:             3072.430
CPU max MHz:         2250.0000
CPU min MHz:         1500.0000
BogoMIPS:            4500.17
Virtualization:      AMD-V
L1d cache:           32K
L1i cache:           32K
L2 cache:            512K
L3 cache:            16384K
NUMA node0 CPU(s):   0-3
NUMA node1 CPU(s):   4-7
NUMA node2 CPU(s):   8-11
NUMA node3 CPU(s):   12-15
NUMA node4 CPU(s):   16-19
NUMA node5 CPU(s):   20-23
NUMA node6 CPU(s):   24-27
NUMA node7 CPU(s):   28-31
NUMA node8 CPU(s):   32-35
NUMA node9 CPU(s):   36-39
NUMA node10 CPU(s):  40-43
NUMA node11 CPU(s):  44-47
NUMA node12 CPU(s):  48-51
NUMA node13 CPU(s):  52-55
NUMA node14 CPU(s):  56-59
NUMA node15 CPU(s):  60-63
NUMA node16 CPU(s):  64-67
NUMA node17 CPU(s):  68-71
NUMA node18 CPU(s):  72-75
NUMA node19 CPU(s):  76-79
NUMA node20 CPU(s):  80-83
NUMA node21 CPU(s):  84-87
NUMA node22 CPU(s):  88-91
NUMA node23 CPU(s):  92-95
NUMA node24 CPU(s):  96-99
NUMA node25 CPU(s):  100-103
NUMA node26 CPU(s):  104-107
NUMA node27 CPU(s):  108-111
NUMA node28 CPU(s):  112-115
NUMA node29 CPU(s):  116-119
NUMA node30 CPU(s):  120-123
NUMA node31 CPU(s):  124-127
Flags:               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 constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 hw_pstate sme ssbd mba sev ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr wbnoinvd arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif umip rdpid overflow_recov succor smca

cyring commented 2 years ago

Hello,

Can you print the following please:

zgrep CONFIG_CPU_FREQ /proc/config.gz

cyring commented 2 years ago

Forget the above request. I have virtualized CentOS 8 and encountered the same issue.

A fix is available in the development branch Here is the result into the VM

2021-09-26-113123_1024x794_scrot

When you will fetch the develop branch, be sure :

not to mix source code with any previous downloads
CoreFreq' Daemon and Driver are fully unloaded from memory before starting up again

This fix is part of CoreFreq version 1.87.4

Waiting for your confirmation before merging fix into master

fleers commented 2 years ago

Thanks. The compilation error has changed to this:

# make
cc  -Wall -Wfatal-errors -pthread -c corefreqd.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqd.o
cc  -Wall -Wfatal-errors -c corefreqm.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqm.o
cc  -Wall -Wfatal-errors corefreqd.c corefreqm.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqd -lpthread -lm -lrt
cc  -Wall -Wfatal-errors -c corefreq-cli.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli.o
cc  -Wall -Wfatal-errors -c corefreq-ui.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-ui.o
cc  -Wall -Wfatal-errors -c corefreq-cli-rsc.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli-rsc.o
cc  -Wall -Wfatal-errors -c corefreq-cli-json.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-cli-json.o
cc  -Wall -Wfatal-errors -c corefreq-cli-extra.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-cli-extra.o
cc  -Wall -Wfatal-errors \
  corefreq-cli.c corefreq-ui.c corefreq-cli-rsc.c \
  corefreq-cli-json.c corefreq-cli-extra.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli -lm -lrt
make -j1 -C /lib/modules/4.18.0-240.el8.x86_64/build M=/tmp/CoreFreq modules
make[1]: Entering directory '/usr/src/kernels/4.18.0-240.el8.x86_64'
  CC [M]  /tmp/CoreFreq/corefreqk.o
/tmp/CoreFreq/corefreqk.c:361:17: error: initialization of ‘int (*)(int)’ from incompatible pointer type ‘int (*)(struct cpufreq_policy *, int)’ [-Werror=incompatible-pointer-types]
    .set_boost = CoreFreqK_SetBoost
                 ^~~~~~~~~~~~~~~~~~
compilation terminated due to -Wfatal-errors.
cc1: some warnings being treated as errors
make[2]: *** [scripts/Makefile.build:322: /tmp/CoreFreq/corefreqk.o] Error 1
make[1]: *** [Makefile:1544: _module_/tmp/CoreFreq] Error 2
make[1]: Leaving directory '/usr/src/kernels/4.18.0-240.el8.x86_64'
make: *** [Makefile:76: all] Error 2

cyring commented 2 years ago

Can you replace the following lines with the code changed for each;

https://github.com/cyring/CoreFreq/blob/6e1d7d414017a24cce69edb24c5586324d0c1fa4/corefreqk.c#L15624

#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0) && (REDHAT != 8)

https://github.com/cyring/CoreFreq/blob/6e1d7d414017a24cce69edb24c5586324d0c1fa4/corefreqk.h#L5362

#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0) && (REDHAT != 8)

Next fully rebuild make clean all

cyring commented 2 years ago

EDIT My Minor is 4, your is 3 Try the bellow lines instead.

https://github.com/cyring/CoreFreq/blob/6e1d7d414017a24cce69edb24c5586324d0c1fa4/corefreqk.c#L15624

#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0) \
  || ((REDHAT == 8) && (RHEL_MINOR > 3))

https://github.com/cyring/CoreFreq/blob/6e1d7d414017a24cce69edb24c5586324d0c1fa4/corefreqk.h#L5362

#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0) \
  || ((REDHAT == 8) && (RHEL_MINOR > 3))

fleers commented 2 years ago

# diff -u corefreqk.c.orig corefreqk.c
--- corefreqk.c.orig    2021-09-26 17:42:38.405889600 -0600
+++ corefreqk.c 2021-09-26 17:44:38.486764252 -0600
@@ -15621,7 +15621,7 @@
     }
 }

-#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0) || (REDHAT == 8)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0) || ((REDHAT == 8) && (RHEL_MINOR > 3))
 static int CoreFreqK_SetBoost(struct cpufreq_policy *policy, int state)
 {
        UNUSED(policy);

# diff -u corefreqk.h.orig corefreqk.h
--- corefreqk.h.orig    2021-09-26 17:42:51.556985407 -0600
+++ corefreqk.h 2021-09-26 17:45:04.314952338 -0600
@@ -5359,7 +5359,7 @@
 #endif
 static int CoreFreqK_SetPolicy(struct cpufreq_policy*) ;
 static int CoreFreqK_Bios_Limit(int, unsigned int*) ;
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0) || (REDHAT == 8)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0) || ((REDHAT == 8) && (RHEL_MINOR > 3))
 static int CoreFreqK_SetBoost(struct cpufreq_policy*, int) ;
 #else
 static int CoreFreqK_SetBoost(int) ;

# make clean && make
rm -f corefreqd corefreq-cli
make -j1 -C /lib/modules/4.18.0-240.el8.x86_64/build M=/tmp/CoreFreq clean
make[1]: Entering directory '/usr/src/kernels/4.18.0-240.el8.x86_64'
  CLEAN   /tmp/CoreFreq/.tmp_versions
make[1]: Leaving directory '/usr/src/kernels/4.18.0-240.el8.x86_64'
cc  -Wall -Wfatal-errors -pthread -c corefreqd.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqd.o
cc  -Wall -Wfatal-errors -c corefreqm.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqm.o
cc  -Wall -Wfatal-errors corefreqd.c corefreqm.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreqd -lpthread -lm -lrt
cc  -Wall -Wfatal-errors -c corefreq-cli.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli.o
cc  -Wall -Wfatal-errors -c corefreq-ui.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-ui.o
cc  -Wall -Wfatal-errors -c corefreq-cli-rsc.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli-rsc.o
cc  -Wall -Wfatal-errors -c corefreq-cli-json.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-cli-json.o
cc  -Wall -Wfatal-errors -c corefreq-cli-extra.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0 \
  -o corefreq-cli-extra.o
cc  -Wall -Wfatal-errors \
  corefreq-cli.c corefreq-ui.c corefreq-cli-rsc.c \
  corefreq-cli-json.c corefreq-cli-extra.c \
  -D CORE_COUNT=256 -D TASK_ORDER=5 -D MAX_FREQ_HZ=5250000000 -D UBENCH=0  \
  -o corefreq-cli -lm -lrt
make -j1 -C /lib/modules/4.18.0-240.el8.x86_64/build M=/tmp/CoreFreq modules
make[1]: Entering directory '/usr/src/kernels/4.18.0-240.el8.x86_64'
  CC [M]  /tmp/CoreFreq/corefreqk.o
/tmp/CoreFreq/corefreqk.c: In function ‘CoreFreqK_IdleDriver_Init’:
/tmp/CoreFreq/corefreqk.c:15260:18: error: assignment to ‘void (*)(struct cpuidle_device *, struct cpuidle_driver *, int)’ from incompatible pointer type ‘int (*)(struct cpuidle_device *, struct cpuidle_driver *, int)’ [-Werror=incompatible-pointer-types]
   ].enter_s2idle = CoreFreqK_Alt_S2_MWAIT_Handler;
                  ^
compilation terminated due to -Wfatal-errors.
cc1: some warnings being treated as errors
make[2]: *** [scripts/Makefile.build:322: /tmp/CoreFreq/corefreqk.o] Error 1
make[1]: *** [Makefile:1544: _module_/tmp/CoreFreq] Error 2
make[1]: Leaving directory '/usr/src/kernels/4.18.0-240.el8.x86_64'
make: *** [Makefile:76: all] Error 2

cyring commented 2 years ago

Can you post a link to the ISO of your current CentOS version ?

fleers commented 2 years ago

8.3 ISOs can be found here I'm using the dvd1 ISO

cyring commented 2 years ago

8.3 ISOs can be found here I'm using the dvd1 ISO

I'm downloading CentOS-8.3.2011-x86_64-minimal.iso because that server link is too slow. Let's hope I will find enough packages for kernel development ...

cyring commented 2 years ago

Notes for the CentOS-8.3.2011-x86_64-minimal.iso

dnf install perl-interpreter gcc make elfutils-libelf-devel

rpm -i https://vault.centos.org/8.3.2011/BaseOS/x86_64/os/Packages/kernel-devel-4.18.0-240.el8.x86_64.rpm

rpm -i https://vault.centos.org/8.3.2011/BaseOS/x86_64/os/Packages/kernel-headers-4.18.0-240.el8.x86_64.rpm

cyring commented 2 years ago

Please try latest commit in develop branch.

fleers commented 2 years ago

It builds successfully now! Thanks for the quick fix :-)

cyring commented 2 years ago

It builds successfully now! Thanks for the quick fix :-)

That's great !

In future commit, I have to fix build combinations among CentOS/RHEL 8 Minor 3 vs 4; do no regression testings with CentOS 7; and probably other Redhat flavours. So I'll come back to you for a final pre-release testing.

In Wiki-Support there is a list of processors supported by CoreFreq: https://github.com/cyring/CoreFreq/wiki/CPU-support Feel free to add a page with your screenshots and Cli output, especially corefreq-cli -s Your data will help me a lot to find the remaining work. Thank you.

cyring commented 2 years ago

Hello,

In develop branch is now available a fix to compile with CentOS RHEL 8.3 and 8.4 (and still CentOS 7)

Can you give it a try ?

Your screenshots and CLI outputs are also welcomed.

fleers commented 2 years ago

Hello, here's cli output on that system - link

cyring commented 2 years ago

Hello, here's cli output on that system -

Issues found:

Architecture codename should be Rome

[Zen2/Castle Peak]

HSMP does not give PPT

PPT [ 4 W]

cyring commented 2 years ago

Issues found:

Architecture codename should be Rome

[Zen2/Castle Peak]

HSMP does not give PPT

PPT [ 4 W]

According to the CPU support page in Wiki, bellow processors have the same CPUID signature

EPYC 7742 CPUID converted to hexa model 31

CPU family:          23
Model:               49
Model name:          AMD EPYC 7742 64-Core Processor
Stepping:            0

EPYC 7502P has a signature 8F_31 Stepping 0
Threadripper 3970X has a signature 8F_31 Stepping 0

Consequently HSMP was not trigerred b/c brand AMD EPYC 7502P 32-Core Processor was out of the Castle Peak list. https://github.com/cyring/CoreFreq/blob/5127cc858078fbac640a443bfa9d4dabd552c5e0/corefreqk.h#L4551

I have no evidence that 8F_30 Stepping 0 specifies EPYC/Rome https://github.com/cyring/CoreFreq/blob/5127cc858078fbac640a443bfa9d4dabd552c5e0/corefreqk.h#L1484

cyring commented 2 years ago

Hello,

Can you test the latest develop branch ? Info: Rome and Castle Peak are fixed to the same CPUID signature.

And provide in another page the output of:

corefreq-cli -s -m -c 1 -i 1 -C 1

Thank you.

fleers commented 2 years ago

Here you go

cyring commented 2 years ago

Here you go

Thank you for this test. Not the results I was expecting. Architecture string have been changed to Zen2/EPYC/Rome

Do you confirm source code has been fetched from the develop branch ?

CLI and Daemon stopped
Driver unload
Full rebuild with make clean all
Start CoreFreq again

EDIT: at the present time, git log should displayed the following latest commits

commit 3c2acf35b5f81db6649ee94c28a5cb593309fcbf (HEAD -> develop, origin/develop)
Author: CyrIng <labs@cyring.fr>
Date:   Thu Oct 14 01:19:03 2021 +0200

    [AMD/UMC] tWTW[DLR] associated comment fix.

commit 2a961733eb0891b7262b7e74d1b0e65eeaa4b8c4
Author: CyrIng <labs@cyring.fr>
Date:   Wed Oct 13 03:45:50 2021 +0200

    [AMD/Zen2/MCM] Merged Rome and Castle Peak

commit c3fe57fce655b44767d1fc8419ed7284796d4ca0
Author: CyrIng <labs@cyring.fr>
Date:   Tue Oct 12 07:43:32 2021 +0200

    Misc source code re-styling.

commit 34f5aacfbcd4eee17824173b83592d33a0a991f7
Author: CyrIng <labs@cyring.fr>
Date:   Sun Oct 10 20:53:23 2021 +0200

    GCC & Clang: source code review.

EDIT: Now using argument -n a new line can be inserted between sections

corefreq-cli -n -s -n -m -n -c 1 -i 1 -C 1

fleers commented 2 years ago

OK, here's an updated cli output

cyring commented 2 years ago

OK, here's an updated cli output

Thank you very much. Let me know if you wish to provide more EPYC tests ?

fleers commented 2 years ago

Let me know if you wish to provide more EPYC tests ?

Yes, happy to - let me know what you would like to have.

cyring commented 2 years ago

Let me know if you wish to provide more EPYC tests ?

Yes, happy to - let me know what you would like to have.

Thanks for your help.

Features state

Technologies
|- Data Cache Unit
   |- L1 Prefetcher                                                L1 HW   <OFF>
   |- L2 Prefetcher                                                L2 HW   <OFF>
|- System Management Mode                                       SMM-Lock   [ ON]
|- Simultaneous Multithreading                                       SMT   [OFF]
|- PowerNow!                                                         CnQ   [OFF]
|- Core C-States                                                     CCx   [ ON]
|- Core Performance Boost                                            CPB   < ON>
|- Watchdog Timer                                                    WDT   < ON>
|- Virtualization                                                    SVM   [OFF]
   |- I/O MMU                                                      AMD-V   [OFF]
   |- Version                                                     [         N/A]
   |- Hypervisor                                                           [OFF]
   |- Vendor ID                                                   [         N/A]

There is no reason to have the Data Cache Unit disabled, unless you decided so. Because I have an unsolved synchronization issue, I suspect those states not being ready yet when displayed.

In the UI go to the Technologies window:

L1 HW and L2 HW are both OFF
Press the star key [*] to refresh the CoreFreq Machine
Make a screenshot of the results

Fyi, here is mine: 2021-10-15-192242_644x550_scrot

SMT

I believe you have disable Simultaneous Multithreading but I would like to check if the CoreFreq topology is correct with Rome ?

Once SMT enabled in your BIOS:

Build with a count of 128 CPUs

make CORE_COUNT=128 NO_UPPER=1 clean all

Restart CoreFreq

rmmod corefreqk.ko
insmod corefreqk.ko
./corefreqd

Print the Topology and Processor information
```
corefreq-cli -n -m -n -s
```
Temperature
- Check the Temperature per cluster:
In the UI menu, select Tools then Turbo < Select CPU > To clearly see Temperatures, move the window by pressing [Shift] + [Left]
Press [Enter] on a CPU to start stressing it

Now check if value is raising per cluster ?

2021-10-15-194448_644x550_scrot

You can add other CPU from the same CCD cluster (see the Topology)
[F10] to stop the stressing loop.

Voltage

Voltage Core My 3950X is giving the same P-State Vcore whatever is the stressed CPU

Probably Rome may be capable of per cluster Vcore reporting ?

Here is the Voltage view running the Turbo Random CPU of Tools menu:

2021-10-15-195805_644x550_scrot

fleers commented 2 years ago

I had to switch to another machine - it's single socket.

Here's the cli output after SMT on

L1/L2 off and ON -

l1-l2 off

l1-l2 on

Temp. with turbo on 4 cores -

turbo temp

Voltage with random turbo -

voltage random

cyring commented 2 years ago

I had to switch to another machine - it's single socket.

Here's the cli output after SMT on

L1/L2 off and ON -

Temp. with turbo on 4 cores -

Voltage with random turbo -

Thanks, I will update the Support wiki page.

Some questions:

How does CoreFreq work with multiple sockets ?
Are you observing coherency issues with such platform ?
Are the Memory Controller window and corefreq-cli -M returning some data ?
Are the timings correct ? Same as values left in BIOS ?
We're L1 L2 first OFF because of a wrong CoreFreq reporting ?
Does the P-States altering work ? In the Processor window, go to ratios 1C or 2C to modify the frequency (for whole Processor) (if Core Performance Boost is ON, you have to disable it in the Technology window)

cyring commented 2 years ago

289 for testings before merge

If things OK feel free to close the issue.

cyring commented 2 years ago

RHEL are part of the new version. Closing the issue.

cyring / CoreFreq

[Solved] build on AMD Rome w/CentOS8 4.18.0-240 #279

make

cat /etc/redhat-release

uname -a

lscpu

Features state

SMT

Temperature

Voltage

289 for testings before merge