NCNN

NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent.

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

Project Site

github.com

Source Repository

github.com

Test Created

18 September 2020

Last Updated

1 August 2023

Test Maintainer

Michael Larabel

Test Type

System

Average Install Time

1 Minute, 28 Seconds

Average Run Time

20 Minutes, 48 Seconds

Test Dependencies

CMake + C/C++ Compiler Toolchain + Vulkan

Accolades

70k+ Downloads

Supported Platforms

* 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.
*** Test profile page view reporting began March 2021.
Data updated weekly as of 23 November 2024.

Revision History

pts/ncnn-1.5.0 [View Source] Tue, 01 Aug 2023 11:49:36 GMT
Update against latest upstream.

pts/ncnn-1.4.0 [View Source] Sat, 13 Aug 2022 09:59:41 GMT
Update against NCNN 20220729 upstream.

pts/ncnn-1.3.0 [View Source] Tue, 27 Jul 2021 16:34:42 GMT
Update against NCNN 2021-07-20 upstream, fix possible Vulkan build issue by including glslang source tree.

pts/ncnn-1.2.0 [View Source] Fri, 18 Jun 2021 08:11:33 GMT
Update against NCNN 20210525 release.

pts/ncnn-1.1.0 [View Source] Fri, 18 Dec 2020 08:06:41 GMT
Update against new upstream NCNN 20201218.

pts/ncnn-1.0.3 [View Source] Fri, 25 Sep 2020 06:36:39 GMT
Drop int8 tests per https://github.com/phoronix-test-suite/test-profiles/pull/167

pts/ncnn-1.0.2 [View Source] Thu, 24 Sep 2020 12:52:47 GMT
Expose Vulkan GPU support.

pts/ncnn-1.0.1 [View Source] Fri, 18 Sep 2020 12:28:10 GMT
Increase the run count.

pts/ncnn-1.0.0 [View Source] Fri, 18 Sep 2020 11:58:15 GMT
Initial commit of Tencent NCNN.

Suites Using This Test

Machine Learning

HPC - High Performance Computing

NVIDIA GPU Compute

Vulkan Compute

Performance Metrics

Analyze Test Configuration:

NCNN 20230517

Target: CPU - Model: googlenet

OpenBenchmarking.org metrics for this test profile configuration based on 284 public results since 1 August 2023 with the latest data as of 10 November 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

ms (Average)

97th

94th

92nd

90th

87th

Mid-Tier

75th

> 8

Intel Core i5-14500

75th

AMD Ryzen 9 7950X 16-Core

73rd

9 ^{+/- 1}

Apple M2

72nd

AMD Ryzen 7 PRO 6850U

68th

Intel Core i7-5960X

65th

Intel Core i9-10980XE

65th

Intel Xeon E-2388G

63rd

AMD Ryzen 7 PRO 5850U

57th

Intel Xeon E-2336

53rd

AMD Ryzen 7 3800XT 8-Core

51st

Median

50th

AMD Ryzen 9 5900HX

49th

12 ^{+/- 1}

Intel Core i7-1165G7

47th

14 ^{+/- 2}

AMD Ryzen 9 3900XT 12-Core

45th

AMD Ryzen Threadripper 3970X 32-Core

43rd

AMD EPYC 7763 64-Core

42nd

Intel Core i7-8565U

37th

AMD Ryzen 5 5500U

35th

2 x Intel Xeon Platinum 8380

34th

17 ^{+/- 1}

Intel Core i7-1065G7

34th

17 ^{+/- 1}

AMD Ryzen 7 4700U

31st

AMD EPYC 7F32 8-Core

29th

2 x Intel Xeon Platinum 8490H

28th

19 ^{+/- 1}

Low-Tier

25th

> 21

AMD EPYC 9384X 32-Core

25th

AMD Ryzen 5 4500U

24th

AMD Ryzen Threadripper PRO 5965WX 24-Cores

21st

25 ^{+/- 4}

2 x AMD EPYC 9124 16-Core

17th

AMD EPYC 9754 128-Core

14th

AMD Ryzen Threadripper 3990X 64-Core

13th

2 x AMD EPYC 9334 32-Core

12th

33 ^{+/- 1}

2 x AMD EPYC 9754 128-Core

8th

56 ^{+/- 1}

rv64imafdcvsu

3rd

262 ^{+/- 1}

Based on OpenBenchmarking.org data, the selected test / test configuration (NCNN 20230517 - Target: CPU - Model: googlenet) has an average run-time of 26 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.

Based on public OpenBenchmarking.org results, the selected test / test configuration has an average standard deviation of 1.2%.

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

SSE2 (SSE2)

Used by default on supported hardware.

MOVD PSHUFD PSHUFLW MOVDQU MOVDQA SUBSD MOVAPD MINSD MAXSD ADDSD CVTSI2SD DIVSD PUNPCKLQDQ CVTSS2SD MULSD MULPD PSRLDQ CVTSD2SS CVTTPS2DQ CVTDQ2PS PUNPCKHQDQ PMULUDQ PADDQ SHUFPD COMISD PSHUFHW

Advanced Vector Extensions (AVX)

Used by default on supported hardware.
Found on Intel processors since Sandy Bridge (2011).
Found on AMD processors since Bulldozer (2011).

VZEROUPPER VBROADCASTSS VPERM2F128 VINSERTF128 VEXTRACTF128 VPERMILPS VMASKMOVPS VBROADCASTF128 VBROADCASTSD

Advanced Vector Extensions 2 (AVX2)

Used by default on supported hardware.
Found on Intel processors since Haswell (2013).
Found on AMD processors since Excavator (2016).

VPBROADCASTD VINSERTI128 VPERMD VEXTRACTI128 VPERMQ VPERM2I128 VPBROADCASTW VPBROADCASTB VPBROADCASTQ VGATHERDPS VPGATHERQQ VPSRLVQ

Advanced Vector Extensions 512 (AVX512)

Used by default on supported hardware.

(ZMM REGISTER USE)

FMA (FMA)

Used by default on supported hardware.
Found on Intel processors since Haswell (2013).
Found on AMD processors since Bulldozer (2011).

VFMADD132PS VFMADD132SS VFNMADD231PS VFMADD231PS VFMADD213PS VFNMADD132PS VFNMADD132SS VFNMADD213SS VFMSUB132SS VFNMADD231SS VFMADD231SS VFMSUB132PS VFMSUB231SS VFMADD213SS VFMADD132PD VFMADD132SD VFNMSUB132SS VFNMSUB231SS

AVX Vector Neural Network Instructions (AVX-VNNI)

Requires passing a supported compiler/build flag (verified with targets: tigerlake, cascadelake).

VPDPWSSD

The test / benchmark does honor compiler flag changes.

Last automated analysis: 18 September 2023

This test profile binary relies on the shared libraries libgomp.so.1, libm.so.6, libmvec.so.1, libc.so.6.

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)

RISC-V 64-bit

riscv64

SiFive RISC-V, rv64imafdcvsu

Loongson LoongArch 64-bit

loongarch64

(Many Processors)

ARMv7 32-bit

armv7l

ARMv7 Cortex-A72 4-Core

ARMv8 64-bit

arm64

Apple M2 Pro