This is because AMD does not have 2 functional units (I have never seen an AMD with two functional units).

There seems to be no user-level way to get these variables and most kernels just return "0"for all the LEVELn system configuration variables:

% getconf -a | grep CACHE
LEVEL1_ICACHE_SIZE                 0
LEVEL1_ICACHE_ASSOC                0
LEVEL1_ICACHE_LINESIZE             64
LEVEL1_DCACHE_SIZE                 0
LEVEL1_DCACHE_ASSOC                0
LEVEL1_DCACHE_LINESIZE             64
...

I've been doing some research and it looks like your best bet is to use the uarch of the chip itself and approximate/best-guess by using a static LUT of known values for each uarch, much like pytorch does by using the technical reference manuals:

https://github.com/pytorch/cpuinfo/blob/ed8b86a253800bafdb7b25c5c399f91bff9cb1f3/src/arm/cache.c#L950-L965

Output:

Name:              AMD FX(tm)-6300 Six-Core Processor 
Microarchitecture: Piledriver
Technology:        32nm
Max Frequency:     3.500 GHz
Cores:             3 cores (6 threads)
AVX:               AVX
FMA:               FMA3,FMA4
L1i Size:          64KB (192KB Total)
L1d Size:          16KB (96KB Total)
L2 Size:           2MB (6MB Total)
L3 Size:           8MB
Peak Performance:  168.00 GFLOP/s

This processor definitely has 6 physical cores:

cpufetch $ lscpu | grep CPU
CPU op-mode(s):                  32-bit, 64-bit
CPU(s):                          6

Instead of using ASCII characters, print it using spaces with a background color. Something like this:

RED="\e[48;2;255;0;0m"
RESET="\e[0m"
BLUE="\e[48;2;0;0;200m"
echo -e $RED' '$BLUE' '$RESET

In some enviroments like VMs, /sys/devices/system/cpu/cpu0/cpufreq dir may not exist and it will report incorrect values because not contemplating this issue

I noticed it in a Xeon Gold 6248

rPiZeroW.

compiles fine, get a "Segmentation fault" message. nothing else returns. just single line 2 words.

[ERROR]: Can't read frequency information from cpuid (needed level is 0x00000016, max is 0x0000000D)
[ERROR]: Found unified cache at level 4 (expected == 2 or 3)
Please, create a new issue with this error message and the output of 'cpufetch --debug' in https://github.com/Dr-Noob/cpufetch/issues

I gather the first error is a known issue but I couldn't find anything on the second one.

CPU: Crystal Well Core i7 2.8GHz
It is an MSI Laptop if that matters, please tell what else you need to debug this issue.

For fun I tried cpufetch in PS, cmd and WSL2 debian all returned the same errors (both), so at least the code is consistent.

lscpu from WSL2 Debian
oscar@MSI-Laptop:~$ lscpu
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 8
On-line CPU(s) list: 0-7
Thread(s) per core: 2
Core(s) per socket: 4
Socket(s): 1
Vendor ID: GenuineIntel
CPU family: 6
Model: 70
Model name: Intel(R) Core(TM) i7-4980HQ CPU @ 2.80GHz
Stepping: 1
CPU MHz: 2801.000
CPU max MHz: 2801.0000
BogoMIPS: 5602.00
Virtualization: VT-x
Hypervisor vendor: Windows Subsystem for Linux
Virtualization type: container
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave osxsave avx f16c rdrand lahf_lm abm fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid ibrs ibpb stibp ssbd

Hi,

Thanks for making this nice fetching tool.

When cpufetch (Release 0.7) is executed, I get the error:

[ERROR]: Can't read cache information from cpuid (needed extended level is -2147483619, max is -2147483621)

Replace 'x86_64' string by a real check of the arch

All of the ascii art would benefit from a big rework. Key ideas are:

• Replace ascii art (using the same characters for all the art) with ascii art with different characters which can enable to do better and more complex shapes
• Each ascii art should have its own with (but mantaining the height), instead of a common one. This would allow a better fit for each art
• Fancy style should be removed. Drawing with rectangles may be a good choice for some arts, but not for all of them
• Default color scheme should be RGB, instead the system color. This would allow more realistic arts. Add option --system-color to use the system colors instead. Remove --color x
• (Optional) Specify RGB color for each pixel in art, instead of setting a single color for all pixels. This would be very nice, since it could achieve a better feel of the art using fading colors.

Attempting to build on an ARM MacBook Pro shows the below result:

$ make
gcc -Wall -Wextra -Werror -pedantic -fstack-protector-all -pedantic -std=c99 -DARCH_ARM -Wno-unused-parameter -Wfloat-equal -Wshadow -Wpointer-arith src/common/main.c src/common/cpu.c src/common/udev.c src/common/printer.c src/common/args.c src/common/global.c src/arm/midr.c src/arm/uarch.c src/arm/soc.c src/arm/udev.c -o cpufetch
src/arm/midr.c:7:10: fatal error: 'sys/auxv.h' file not found
#include <sys/auxv.h>
         ^~~~~~~~~~~~
1 error generated.
make: *** [cpufetch] Error 1

uname -a returns this:

Darwin MacBook-Pro.local 20.3.0 Darwin Kernel Version 20.3.0: Thu Jan 21 00:06:51 PST 2021; root:xnu-7195.81.3~1/RELEASE_ARM64_T8101 arm64

sujeet@sujeet-pc:~/MyStuff/cpufetch$ make
gcc -Wall -Wextra -Werror -pedantic -fstack-protector-all -pedantic -std=c99 -DARCH_ARM -Wno-unused-parameter -Wfloat-equal -Wshadow -Wpointer-arith src/common/main.c src/common/cpu.c src/common/udev.c src/common/printer.c src/common/args.c src/common/global.c src/arm/midr.c src/arm/uarch.c src/arm/soc.c src/arm/udev.c -o cpufetch
src/arm/midr.c:8:10: fatal error: asm/hwcap.h: No such file or directory
#include <asm/hwcap.h>
^~~~~~~~~~~~~
compilation terminated.
make: *** [Makefile:45: cpufetch] Error 1
how do I deal with this

This is the output of the command:

╭─eduardo@charles ~/cpufetch ‹master›
╰─$ cpufetch
open: No such file or directory
[ERROR]: Could not open '/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq'
Please, create a new issue with this error message and the output of 'cpufetch --debug' in https://github.com/Dr-Noob/cpufetch/issues
[ERROR]: SMT level was not found when querying topology

                                                          Name:              Pentium(R) Dual-Core CPU E5700 @ 3.00GHz
                                                          Microarchitecture: Penryn
                                                          Technology:        45nm
                                                          Max Frequency:     Unknown
                                                          Cores:             0 cores
                                                          AVX:               No
                                                          FMA:               No
                                                          L1i Size:          32KB
                                                          L1d Size:          32KB
                                                          L2 Size:           2MB
                                                          Peak Performance:  Unknown

╭─eduardo@charles ~/cpufetch ‹master›
╰─$ cpufetch --debug
open: No such file or directory
[ERROR]: Could not open '/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq'
Please, create a new issue with this error message and the output of 'cpufetch --debug' in https://github.com/Dr-Noob/cpufetch/issues
[ERROR]: SMT level was not found when querying topology
cpufetch v0.96 (x86_64 build)
Pentium(R) Dual-Core CPU E5700 @ 3.00GHz

• Max standard level: 0x0000000D
• Max extended level: 0x80000008
• CPUID dump: 0x0001067A
  ╭─eduardo@charles ~/cpufetch ‹master›
  ╰─$

Instead, the topology should be queried properly. For example, on a AMD Ryzen 2600 (see https://en.wikichip.org/wiki/amd/ryzen_5/2600), cpufetch reports 8MB of L3 which is wrong, because it has two L3...

why -std=c99 on CXXFLAGS ? It's invalid for c++

When installing through the git repository, I am unable to make the package.

Support for VMs; print VM name and other info.
Instead of print the logo in the usual way, print it with random colors, like:

for i = 0; i < len; i++
  color = rand

This color idea can also be added as --color wild

Is there a Way to compile cpufetch in 32Bit Linux?
I am using Kubuntu 18.04 LTS (32Bit)
Thanks,

For example, in a Ryzen 5 2600, cpufetch reports 3.4 GHz, but the turbo is 3.9 GHz (see https://en.wikichip.org/wiki/amd/ryzen_5/2600)

It does not seem like it can be fetched from cpuid. At least, dmidecode is able to obtain such information.

Currently, the port to Windows is impossible due to the udev dependency. Removing it could make cpufetch a better tool, because it would not depend on any operating system. This is probably the most complex feature to add. Removing this dependency, it should be easy to port cpufetch to Windows

Please add a feature to detect and display the microarchitecture level (x86-64 v2, etc.)

This might be helpful if you decide to implement this: https://unix.stackexchange.com/questions/631217/how-do-i-check-if-my-cpu-supports-x86-64-v2

cpufetch v0.67 says:
L1i: 2 x 64 KB = 128 KB
L1d: 4 x 16 KB = 64 KB

lscpu 2.36 says:
L1i: 2 x 64 KB = 128 KB
L1d: 2 x 16 KB = 32 KB

cpuworld agrees with cpufetch, but it looks strange that lscpu (which uses the kernel functionality) is wrong...

For example, on a Xeon Gold, the logo could appear in yellow, and some other cool tweaks.

For a future project, developing a cpufetch like tool for GPU could be interesting.

For NVIDIA GPUs, I know that CUDA can give you any kind of details. For radeon and integrated GPUs (Intel, for example), I don't know how to approach it.

At the moment, if cpuid (0x16) does not support it, cpufetch isn't able to do so. To know how we can get this, look at the Linux kernel source code.

If using certain terminal colors, output may not be visible(like white background in Ubuntu)

Print how many NUMA nodes are present

We could know it by checking the microarchitecture, since 2 VPUs are a thing since Haswell microarchitecture (according to https://github.com/Dr-Noob/FLOPS)

When I type either cpufetch or cpufetch --debug the program segfaults with the same message

Please, create a new issue with this error message and the output of 'cpufetch --debug' in https://github.com/Dr-Noob/cpufetch/issues
zsh: segmentation fault (core dumped)  cpufetch --debug

Here is the output of inxi -Fxz concerning system info and cpu

           base: Arch Linux
Machine:   Type: Laptop System: LENOVO product: 81N3 v: Lenovo IdeaPad S145-15AST serial: <filter>
           Mobo: LENOVO model: LNVNB161216 v: SDK0R32866WIN serial: <filter> UEFI: LENOVO v: AYCN18WW date: 06/05/2019
CPU:       Info: Dual Core model: AMD A6-9225 RADEON R4 5 COMPUTE CORES 2C+3G bits: 64 type: MCP arch: Excavator rev: 0
           L2 cache: 1024 KiB
           flags: avx avx2 lm nx pae sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3 svm bogomips: 10384
           Speed: 2595 MHz min/max: 1300/2600 MHz boost: enabled Core speeds (MHz): 1: 2595 2: 2595

The ASCII logo could be improved significantly if it would show the specific model, along with the manufacturer. For example, showing the Intel trademark, and Core below it, for Intel Core family. The same applies for Xeon, AMD Ryzen, etc. Adding colours to this different families could also be interesting.

If there is more than one L3 cache, just shows the size of one of them insted of adding the value of each one

doing cpufetch --debug it shown this:

open: No such file or directory
[ERROR]: Could not open '/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq'
Please, create a new issue with this error message and the output of 'cpufetch --debug' in https://github.com/Dr-Noob/cpufetch/issues
cpufetch v0.96 (x86_64 build)
AMD Ryzen 3 3200G with Radeon Vega Graphics

• Max standard level: 0x0000000D
• Max extended level: 0x8000001F
• CPUID dump: 0x00810F81

$ ./cpufetch --debug
open: No such file or directory
[ERROR]: Could not open '/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq'
Please, create a new issue with this error message and the output of 'cpufetch --debug' in https://github.com/Dr-Noob/cpufetch/issues
cpufetch v0.94 (x86_64 build)
Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz
- Max standart level: 0x0000000D
- Max extended level: 0x80000008

also, typo: standart standard

For example, something like this:

Maybe with an approach similar to this one

In the case of Kryo 260 / 280 (and more), MIDR is not enough for detect specific model, but using SoC we could do so.

Just like XKB(D) XKB(I)

This is challenging, though

The mentioned phone has a Snapdragon 730G processor but the output shows a SD 675 and a completely different architecture.

$./cpufetch
[ERROR]: Invalid L2 size: 4MB
Please, create a new issue with this error message and the output of 'cpufetch --debug' in https://github.com/Dr-Noob/cpufetch/issues
Segmentation fault (core dumped)

Git version: 4900c10
openSUSE Leap 15.2
Kernel 5.3.18
AMD A10-8700P Radeon R6, 10 Compute Cores 4C+6G

cpufetch --debug
[ERROR]: Invalid L1i size: 96KB
Please, create a new issue with this error message and the output of 'cpufetch --debug' in https://github.com/Dr-Noob/cpufetch/issues

cpufetch --debug gives the same error, nothing new.