Llama.cpp

Llama.cpp is a port of Facebook's LLaMA model in C/C++ developed by Georgi Gerganov. Llama.cpp allows the inference of LLaMA and other supported models in C/C++. For CPU inference Llama.cpp supports AVX2/AVX-512, ARM NEON, and other modern ISAs along with features like OpenBLAS usage.

To run this test with the Phoronix Test Suite, the basic command is: phoronix-test-suite benchmark llama-cpp.

Project Site

github.com

Source Repository

github.com

Test Created

10 January 2024

Last Updated

23 November 2024

Test Maintainer

Michael Larabel 

Test Type

System

Average Install Time

29 Seconds

Average Run Time

17 Minutes, 28 Seconds

Test Dependencies

C/C++ Compiler Toolchain + BLAS (Basic Linear Algebra Sub-Routine)

Accolades

10k+ Downloads + Recently Updated Test Profile

Supported Platforms


Public Result Uploads *Reported Installs **Reported Test Completions **Test Profile Page ViewsOpenBenchmarking.orgEventsLlama.cpp Popularity Statisticspts/llama-cpp2024.012024.022024.032024.042024.052024.062024.072024.082024.092024.102024.114K8K12K16K20K
* Uploading of benchmark result data to OpenBenchmarking.org is always optional (opt-in) via the Phoronix Test Suite for users wishing to share their results publicly.
** Data based on those opting to upload their test results to OpenBenchmarking.org and users enabling the opt-in anonymous statistics reporting while running benchmarks from an Internet-connected platform.
Data updated weekly as of 23 November 2024.

Revision History

pts/llama-cpp-2.0.0   [View Source]   Sat, 23 Nov 2024 17:33:55 GMT
Update against Llama.cpp upstream, switch to llama-bench executor.

pts/llama-cpp-1.1.0   [View Source]   Sun, 02 Jun 2024 10:36:25 GMT
Update against Llama.cpp upstream, switch to Llama 3 model.

pts/llama-cpp-1.0.0   [View Source]   Wed, 10 Jan 2024 18:02:20 GMT
Initial commit of llama.cpp CPU benchmark.

Suites Using This Test

Machine Learning

HPC - High Performance Computing

Large Language Models


Performance Metrics

Analyze Test Configuration:

Llama.cpp b3067

Model: Meta-Llama-3-8B-Instruct-Q8_0.gguf

OpenBenchmarking.org metrics for this test profile configuration based on 95 public results since 2 June 2024 with the latest data as of 11 October 2024.

Below is an overview of the generalized performance for components where there is sufficient statistically significant data based upon user-uploaded results. It is important to keep in mind particularly in the Linux/open-source space there can be vastly different OS configurations, with this overview intended to offer just general guidance as to the performance expectations.

Component
Percentile Rank
# Compatible Public Results
Tokens Per Second (Average)
93rd
4
17.71 +/- 0.04
Mid-Tier
75th
< 9.69
71st
6
9.50 +/- 0.02
61st
3
8.75 +/- 0.73
60th
5
8.54 +/- 0.29
Median
50th
8.29
48th
7
8.12 +/- 0.11
26th
3
7.11 +/- 0.01
Low-Tier
25th
< 7.11
20th
3
6.22 +/- 0.02
15th
3
5.76 +/- 0.01
12th
3
5.48 +/- 0.57
OpenBenchmarking.orgDistribution Of Public Results - Model: Meta-Llama-3-8B-Instruct-Q8_0.gguf95 Results Range From 0 To 23 Tokens Per Second24681012141618202224816243240

Based on OpenBenchmarking.org data, the selected test / test configuration (Llama.cpp b3067 - Model: Meta-Llama-3-8B-Instruct-Q8_0.gguf) has an average run-time of 4 minutes. By default this test profile is set to run at least 3 times but may increase if the standard deviation exceeds pre-defined defaults or other calculations deem additional runs necessary for greater statistical accuracy of the result.

OpenBenchmarking.orgMinutesTime Required To Complete BenchmarkModel: Meta-Llama-3-8B-Instruct-Q8_0.ggufRun-Time714212835Min: 1 / Avg: 3.84 / Max: 34

Based on public OpenBenchmarking.org results, the selected test / test configuration has an average standard deviation of 0.1%.

OpenBenchmarking.orgPercent, Fewer Is BetterAverage Deviation Between RunsModel: Meta-Llama-3-8B-Instruct-Q8_0.ggufDeviation246810Min: 0 / Avg: 0.08 / Max: 2

Notable Instruction Set Usage

Notable instruction set extensions supported by this test, based on an automatic analysis by the Phoronix Test Suite / OpenBenchmarking.org analytics engine.

Instruction Set
Support
Instructions Detected
Used by default on supported hardware.
Found on Intel processors since Sandy Bridge (2011).
Found on AMD processors since Bulldozer (2011).

 
VZEROUPPER VEXTRACTF128 VPERMILPS VBROADCASTSS VPERM2F128 VINSERTF128 VBROADCASTSD VMASKMOVPS
Used by default on supported hardware.
Found on Intel processors since Haswell (2013).
Found on AMD processors since Excavator (2016).

 
VPBROADCASTQ VEXTRACTI128 VPBROADCASTD VPERMQ VGATHERQPS VPBROADCASTW VINSERTI128 VPBROADCASTB VPSLLVD VPERMD VPSRLVD VPERM2I128
Advanced Vector Extensions 512 (AVX512)
Used by default on supported hardware.
 
(ZMM REGISTER USE)
AVX Vector Neural Network Instructions (AVX-VNNI)
Used by default on supported hardware.
 
VPDPBUSD
FMA (FMA)
Used by default on supported hardware.
Found on Intel processors since Haswell (2013).
Found on AMD processors since Bulldozer (2011).

 
VFMADD231PS VFMADD132PS VFNMADD231PS VFMADD213PS VFNMADD231SS VFMADD132SS VFNMADD213SS VFMADD132SD VFMADD231SS VFMADD213SS
The test / benchmark does honor compiler flag changes.
Last automated analysis: 6 June 2024

This test profile binary relies on the shared libraries libopenblas.so.0, libm.so.6, libc.so.6, libgfortran.so.5, libquadmath.so.0.

Tested CPU Architectures

This benchmark has been successfully tested on the below mentioned architectures. The CPU architectures listed is where successful OpenBenchmarking.org result uploads occurred, namely for helping to determine if a given test is compatible with various alternative CPU architectures.

CPU Architecture
Kernel Identifier
Verified On
Intel / AMD x86 64-bit
x86_64
(Many Processors)
ARMv8 64-bit
aarch64
ARMv8 Cortex-A76 4-Core, ARMv8 Neoverse-V1

Recent Test Results

OpenBenchmarking.org Results Compare

1 System - 323 Benchmark Results

AMD Ryzen Threadripper 7960X 24-Cores - Gigabyte TRX50 AERO D - AMD Device 14a4

Ubuntu 24.04 - 6.8.0-48-generic - GNOME Shell 46.0

2 Systems - 541 Benchmark Results

ARMv8 Cortex-A76 - Raspberry Pi 5 Model B Rev 1.0 - Broadcom BCM2712

Ubuntu 24.04 - 6.8.0-1013-raspi - KDE Plasma 5.27.11

2 Systems - 525 Benchmark Results

ARMv8 Cortex-A76 - Raspberry Pi 5 Model B Rev 1.0 - Broadcom BCM2712

Ubuntu 24.04 - 6.8.0-1013-raspi - KDE Plasma 5.27.11

1 System - 342 Benchmark Results

Intel Core i9-12900K - ASUS PRIME Z790-V AX - Intel Raptor Lake-S PCH

Ubuntu 24.04 - 6.8.0-47-generic - GNOME Shell 46.0

1 System - 1 Benchmark Result

AMD Ryzen 7 7800X3D 8-Core - 32GB - 0GB Virtual Disk + 9GB Virtual Disk + 1100GB Virtual Disk

Ubuntu 24.04 - 5.15.153.1-microsoft-standard-WSL2 - 4.6 Mesa 24.0.9-0ubuntu0.1

1 System - 334 Benchmark Results

Intel Core i3-12100 - ASRock H610M-HDV/M.2 R2.0 - Intel Device 7aa7

Ubuntu 20.04 - 5.15.0-89-generic - GNOME Shell 3.36.9

1 System - 102 Benchmark Results

AMD Ryzen 9 9950X 16-Core - ASUS ROG STRIX X670E-E GAMING WIFI - AMD Device 14d8

Ubuntu 24.04 - 6.10.0-phx - GNOME Shell 46.0

1 System - 80 Benchmark Results

AMD Ryzen 7 PRO 7840U - LENOVO ThinkPad T14s Gen 4 21F80041GE - AMD Device 14e8

Gentoo 2.15 - 6.10.3-gentoo-dist - GNOME Shell 45.5

1 System - 80 Benchmark Results

AMD Ryzen 7 PRO 7840U - LENOVO ThinkPad T14s Gen 4 21F80041GE - AMD Device 14e8

Gentoo 2.15 - 6.10.2-gentoo-dist - GNOME Shell 45.5

1 System - 1 Benchmark Result

Intel Core i7-8550U - LENOVO ThinkPad T480 20L5000AMZ - Intel Xeon E3-1200 v6

Ubuntu 24.04 - 6.8.0-38-generic - GNOME Shell 46.0

Most Popular Test Results

Find More Test Results