ddcpuid is a x86 processor information tool and library.
$ ddcpuid # Summary is the default mode
Name: AuthenticAMD AMD Ryzen 9 5950X 16-Core Processor
Identifier: Family 0x19 Model 0x21 Stepping 0x0
Cores: 16 cores, 32 threads
Max. Memory: 256 TiB physical, 256 TiB virtual
Baseline: x86-64-v3
Features: core-performance-boost htt
Extensions: x87/fpu +f16c mmx extmmx amd64/x86-64 +lahf64 amd-v/vmx +svm=v1 aes-ni adx sha bmi1 bmi2
SSE: sse sse2 sse3 ssse3 sse4.1 sse4.2 sse4a fma
AVX: avx avx2
AMX:
Mitigations: ibpb ibrs ibrs_pref stibp stibp_on ssbd
Cache L1-D: 16x 32 KiB, 512 KiB total, si
Cache L1-I: 16x 32 KiB, 512 KiB total, si
Cache L2-U: 16x 512 KiB, 8 MiB total, si ci
Cache L3-U: 2x 32 MiB, 64 MiB total, si nwbv- Can be used as a stand-alone tool or as a library. DUB compatible.
- Fully supports DMD, GDC, and LDC compilers.
- BetterC compatible, and used by default for the application.
- Library does not rely on external functions (e.g., C runtime, Druntime, OS).
- Surpasses CPU-Z, Intel's Go CPUID module, and Druntime's
core.cpuidmodule in terms of x86-related information. - Currently featuring 240 CPUID bits documented and counting!
Want to better understand x86 and their technologies? There's the ddcpuid Manual (PDF)!
Officially supports these vendors:
"GenuineIntel"- Intel Corporation"AuthenticAMD"- Advanced Micro Devices Inc."KVMKVMKVM\0\0\0"- Linux built-in Kernel Virtual Machine"Microsoft Hv"- Microsoft Hyper-V interface"VBoxVBoxVBox"- VirtualBox Hyper-V interface"\0\0\0\0\0\0\0\0\0\0\0\0"- VirtualBox minimal interface
NOTE: Features may be influenced by the virtual environment.
$ ddcpuid
Name: AuthenticAMD AMD Ryzen 9 5950X 16-Core Processor
Identifier: Family 0x19 Model 0x21 Stepping 0x0
Cores: 4 cores, 4 threads
Max. Memory: 256 TiB physical, 256 TiB virtual
Baseline: x86-64
Features: htt
Extensions: x87/fpu mmx extmmx amd64/x86-64 +lahf64 amd-v/vmx +svm=v1 aes-ni
SSE: sse sse2 sse3 ssse3 sse4.1 sse4.2 sse4a
AVX: avx avx2
AMX:
Mitigations:
ParaVirt.: Hyper-V
Cache L1-D: 2x 32 KiB, 64 KiB total, si
Cache L1-I: 2x 32 KiB, 64 KiB total, si
Cache L2-U: 2x 512 KiB, 1 MiB total, si ci
Cache L3-U: 1x 32 MiB, 32 MiB total, si nwbvWhile displayed in the default operating mode, the --baseline option
can be useful in compiltation scripts.
WARNING: This depends on the capabilities of the processor regardless of the operating system. Results may vary.
WARNING: The compilation baseline for DMD and LDC is the Pentium Pro (i686). No guaranties are given for detecting i486 and i586 family processors.
$ ddcpuid --baseline
x86-64-v3
| Machine | Values |
|---|---|
| 32-bit | i386, i486, i586, or i686 |
| 64-bit | x86-64, x86-64-v2, x86-64-v3, or x86-64-v4 |
$ ddcpuid --raw
| Leaf | Sub-leaf | EAX | EBX | ECX | EDX |
|----------|----------|----------|----------|----------|----------|
| 0 | 0 | 10 | 68747541 | 444d4163 | 69746e65 |
| 1 | 0 | a20f10 | 1d200800 | 7ed8320b | 178bfbff |
| 2 | 0 | 0 | 0 | 0 | 0 |
| 3 | 0 | 0 | 0 | 0 | 0 |
| 4 | 0 | 0 | 0 | 0 | 0 |
| 5 | 0 | 40 | 40 | 3 | 11 |
| 6 | 0 | 4 | 0 | 1 | 0 |
| 7 | 0 | 0 | 219c97a9 | 40069c | 10 |
| 8 | 0 | 0 | 0 | 0 | 0 |
| 9 | 0 | 0 | 0 | 0 | 0 |
| a | 0 | 0 | 0 | 0 | 0 |
| b | 0 | 1 | 2 | 100 | 1d |
| c | 0 | 0 | 0 | 0 | 0 |
| d | 0 | 207 | 988 | 988 | 0 |
| e | 0 | 0 | 0 | 0 | 0 |
| f | 0 | 0 | ff | 0 | 2 |
| 10 | 0 | 0 | 2 | 0 | 0 |
| 80000000 | 0 | 80000023 | 68747541 | 444d4163 | 69746e65 |
| 80000001 | 0 | a20f10 | 20000000 | 75c237ff | 2fd3fbff |
| 80000002 | 0 | 20444d41 | 657a7952 | 2039206e | 30353935 |
| 80000003 | 0 | 36312058 | 726f432d | 72502065 | 7365636f |
| 80000004 | 0 | 20726f73 | 20202020 | 20202020 | 202020 |
| 80000005 | 0 | ff40ff40 | ff40ff40 | 20080140 | 20080140 |
| 80000006 | 0 | 48002200 | 68004200 | 2006140 | 2009140 |
| 80000007 | 0 | 0 | 3b | 0 | 6799 |
| 80000008 | 0 | 3030 | 111ef657 | 501f | 10000 |
| 80000009 | 0 | 0 | 0 | 0 | 0 |
| 8000000a | 0 | 1 | 8000 | 0 | 101bbcff |
| 8000000b | 0 | 0 | 0 | 0 | 0 |
| 8000000c | 0 | 0 | 0 | 0 | 0 |
| 8000000d | 0 | 0 | 0 | 0 | 0 |
| 8000000e | 0 | 0 | 0 | 0 | 0 |
| 8000000f | 0 | 0 | 0 | 0 | 0 |
| 80000010 | 0 | 0 | 0 | 0 | 0 |
| 80000011 | 0 | 0 | 0 | 0 | 0 |
| 80000012 | 0 | 0 | 0 | 0 | 0 |
| 80000013 | 0 | 0 | 0 | 0 | 0 |
| 80000014 | 0 | 0 | 0 | 0 | 0 |
| 80000015 | 0 | 0 | 0 | 0 | 0 |
| 80000016 | 0 | 0 | 0 | 0 | 0 |
| 80000017 | 0 | 0 | 0 | 0 | 0 |
| 80000018 | 0 | 0 | 0 | 0 | 0 |
| 80000019 | 0 | f040f040 | f0400000 | 0 | 0 |
| 8000001a | 0 | 6 | 0 | 0 | 0 |
| 8000001b | 0 | 3ff | 0 | 0 | 0 |
| 8000001c | 0 | 0 | 0 | 0 | 0 |
| 8000001d | 0 | 4121 | 1c0003f | 3f | 0 |
| 8000001e | 0 | 1d | 10e | 0 | 0 |
| 8000001f | 0 | 1780f | 173 | 1fd | 1 |
| 80000020 | 0 | 0 | 2 | 0 | 0 |
| 80000021 | 0 | 4d | 0 | 0 | 0 |
| 80000022 | 0 | 0 | 0 | 0 | 0 |
| 80000023 | 0 | 0 | 0 | 0 | 0 |
For more information in the fashion of Linux's /proc/cpuinfo,
use the -l or --list switches. Warning, this really outputs a lot of
information!
This mode really includes all processor information, including paravirtualization features and Hyper-V's extremely long list of feature bits, if detected.
The best way to compile ddcpuid is by using DUB.
Compilers supported:
- DMD >= 2.068.0 (best supported).
- For earlier versions (tested on dmd 2.067.1), see manual compilation.
- GDC >= 7.0.0 (good optimizations, see GDC Issues).
- LDC >= 1.0.0 (best optimizations, see LDC Issues).
- For 0.17.1, see how to perform a manual compilation.
Using dub(1) is rather straightforward.
Recommended builds for releases:
- DMD:
dub build -b release-nobounds --compiler=dmd - GDC:
dub build -b release-nobounds-gdc --compiler=gdc- For GDC <=10:
dub build -b release-nobounds --compiler=gdc
- For GDC <=10:
- LDC:
dub build -b release-nobounds --compiler=ldc2
For more information how to use DUB, visit this page.
The Makefile relies on GNU Make (gmake/gnumake).
Available variables:
DC: D compiler, defaults todmdPREFIX: installation path prefix, defaults to/usr/local
Available actions:
- debug (default)
- release
- install
- uninstall
Examples:
make: Produce a debug buildmake release DC=ldc: Produce a release build with LDC
Since ddcpuid only consists of two source files, both being in the src
folder, it's still pretty simple to perform a compilation by hand.
Here's an example that works on any compiler:
dmd src/ddcpuid.d src/main.d -ofddcpuid
ldc2 src/ddcpuid.d src/main.d -ofddcpuid
gdc src/ddcpuid.d src/main.d -oddcpuid
If you want an optimized build:
dmd -betterC -release -O -boundscheck=off src/ddcpuid.d src/main.d -oddcpuid
ldc2 -betterC -release -O -boundscheck=off src/ddcpuid.d src/main.d -oddcpuid
gdc -fno-druntime -release -O -fbounds-check=off src/ddcpuid.d src/main.d -oddcpuid
You get the idea.
Versions earlier than 11 will not compile using -fno-druntime due to linking
issues: undefined reference to '__gdc_personality_v0'.
On Windows, LDC versions 1.13 and 1.14 do not include
legacy_stdio_definitions.lib when linking, making it impossible to compile
the project using -betterC.