Skip to content

Commit 5a678ae

Browse files
aminyaaminya
authored
Merge pull request #30 from JuliaMath/AcuteBenchmark
using AcuteBenchmark
2 parents 4eab004 + b0e348f commit 5a678ae

File tree

8 files changed

+333
-148
lines changed

8 files changed

+333
-148
lines changed

‎.github/workflows/CompatHelper.yml‎

Lines changed: 27 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,27 @@
1+
name: CompatHelper
2+
3+
on:
4+
schedule:
5+
- cron: '00 * * * *'
6+
issues:
7+
types: [opened, reopened]
8+
9+
jobs:
10+
build:
11+
runs-on: ${{ matrix.os }}
12+
strategy:
13+
matrix:
14+
julia-version: [1.3.1]
15+
julia-arch: [x86]
16+
os: [ubuntu-latest]
17+
steps:
18+
- uses: julia-actions/setup-julia@latest
19+
with:
20+
version: ${{ matrix.julia-version }}
21+
- name: Install dependencies
22+
run: julia -e 'using Pkg; Pkg.add(Pkg.PackageSpec(name = "CompatHelper", url = "https://github.com/bcbi/CompatHelper.jl.git"))'
23+
- name: CompatHelper.main
24+
env:
25+
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
26+
JULIA_DEBUG: CompatHelper
27+
run: julia -e 'using CompatHelper; CompatHelper.main()'

‎.gitmodules‎

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,3 @@
1+
[submodule "benchmark-results"]
2+
path = benchmark-results
3+
url = C:/Users/yahyaaba/Documents/GitHub/AcuteBenchmark-Results

‎README.md‎

Lines changed: 84 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -3,7 +3,7 @@
33
![](https://github.com/JuliaMath/VML.jl/workflows/julia%20nightly/badge.svg)
44

55
This package provides bindings to the Intel MKL [Vector Mathematics Functions](https://software.intel.com/en-us/node/521751).
6-
This is often substantially faster than broadcasting Julia's built-in functions, especially when applying a transcendental function over a large array.
6+
This is often substantially faster than broadcasting Julia's built-in functions, especially when applying a transcendental function over a large array.
77
Until Julia 0.6 the package was registered as `VML.jl`.
88

99
Similar packages are [Yeppp.jl](https://github.com/JuliaMath/Yeppp.jl), which wraps the open source Yeppp library, and [AppleAccelerate.jl](https://github.com/JuliaMath/AppleAccelerate.jl), which provides access to macOS's Accelerate framework.
@@ -37,8 +37,8 @@ In the event that MKL was not installed properly you will get an error when firs
3737

3838
## Using IntelVectorMath
3939
After loading `IntelVectorMath`, you have the supported function listed below, for example `IntelVectorMath.sin(rand(100))`. These should provide a significant speed-up over broadcasting the Base functions.
40-
Since the package name is quite long, an alias `IVM` is also exported to allow `IVM.sin(rand(100))` after `using` the package.
41-
If you `import` the package, you can add this alias via `const IVM = IntelVectorMath`. Equally, you can replace `IVM` with another alias of your choice.
40+
Since the package name is quite long, an alias `IVM` is also exported to allow `IVM.sin(rand(100))` after `using` the package.
41+
If you `import` the package, you can add this alias via `const IVM = IntelVectorMath`. Equally, you can replace `IVM` with another alias of your choice.
4242

4343
#### Example
4444
```julia
@@ -100,16 +100,87 @@ regarding these options is available on
100100
[Intel's website](http://software.intel.com/sites/products/documentation/hpc/mkl/IntelVectorMath/vmldata.htm).
101101

102102
## Performance
103-
(These results are currently outdated and will be updated in due course)
104-
![IntelVectorMath Performance Comparison](/benchmark/performance.png)
105-
106-
![IntelVectorMath Complex Performance Comparison](/benchmark/performance_complex.png)
107-
108-
Tests were performed on an Intel(R) Core(TM) i7-3930K CPU. Error bars
109-
are 95% confidence intervals based on 25 repetitions of each test with
110-
a 1,000,000 element vector. The dashed line indicates equivalent
111-
performance for IntelVectorMath versus the implementations in Base. Both Base and
112-
IntelVectorMath use only a single core when performing these benchmarks.
103+
Summary of Results:
104+
105+
Relative speed of IntelVectorMath/Base
106+
107+
![IntelVectorMath Performance Comparison](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set6-relative.png)
108+
109+
![IntelVectorMath Complex Performance Comparison](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set6-relative.png)
110+
111+
Full Results:
112+
113+
<details>
114+
<summary>Real Functions - Full Benchmark Results</summary>
115+
116+
![Dimension set 1](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set1-relative.png)
117+
![Dimension set 2](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set2-relative.png)
118+
![Dimension set 3](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set3-relative.png)
119+
![Dimension set 4](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set4-relative.png)
120+
![Dimension set 5](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set5-relative.png)
121+
![Dimension set 6](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set6-relative.png)
122+
![Dimension set 7](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set7-relative.png)
123+
![Dimension set 8](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set8-relative.png)
124+
![Dimension set 9](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set9-relative.png)
125+
![Dimension set 10](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/bar/bench-dims-set10-relative.png)
126+
127+
</details>
128+
129+
<details>
130+
<summary>Complex Functions - Full Benchmark Results</summary>
131+
132+
![Dimension set 1](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set1-relative.png)
133+
![Dimension set 2](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set2-relative.png)
134+
![Dimension set 3](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set3-relative.png)
135+
![Dimension set 4](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set4-relative.png)
136+
![Dimension set 5](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set5-relative.png)
137+
![Dimension set 6](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set6-relative.png)
138+
![Dimension set 7](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set7-relative.png)
139+
![Dimension set 8](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set8-relative.png)
140+
![Dimension set 9](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set9-relative.png)
141+
![Dimension set 10](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Complex/bar/bench-dims-set10-relative.png)
142+
143+
</details>
144+
145+
<details>
146+
<summary>Real Functions - Performance over dimensions</summary>
147+
148+
![abs](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-abs-Type-Float64.png)
149+
![abs2](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-abs2-Type-Float64.png)
150+
![acos](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-acos-Type-Float64.png)
151+
![acosh](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-acosh-Type-Float64.png)
152+
![asin](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-asin-Type-Float64.png)
153+
![asinh](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-asinh-Type-Float64.png)
154+
![atan](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-atan-Type-Float64.png)
155+
![atanh](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-atanh-Type-Float64.png)
156+
![cbrt](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-cbrt-Type-Float64.png)
157+
![ceil](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-ceil-Type-Float64.png)
158+
![cis](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-cis-Type-Float64.png)
159+
![cos](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-cos-Type-Float64.png)
160+
![cosh](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-cosh-Type-Float64.png)
161+
![erf](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-erf-Type-Float64.png)
162+
![erfc](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-erfc-Type-Float64.png)
163+
![erfcinv](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-erfcinv-Type-Float64.png)
164+
![erfcinv](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-erfcinv-Type-Float64.png)
165+
![exp](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-exp-Type-Float64.png)
166+
![expm1](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-expm1-Type-Float64.png)
167+
![floor](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-floor-Type-Float64.png)
168+
![gamma](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-gamma-Type-Float64.png)
169+
![hypot](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-hypot-Type-Float64.png)
170+
![log](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-log-Type-Float64.png)
171+
![round](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-round-Type-Float64.png)
172+
![sin](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-sin-Type-Float64.png)
173+
![sinh](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-sinh-Type-Float64.png)
174+
![sqrt](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-sqrt-Type-Float64.png)
175+
![tan](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-tan-Type-Float64.png)
176+
![tanh](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-tanh-Type-Float64.png)
177+
![trunc](https://raw.githubusercontent.com/aminya/AcuteBenchmark-Results/master/IntelVectorMath/Real/dimplot/bench-trunc-Type-Float64.png)
178+
179+
</details>
180+
181+
<br/>
182+
183+
Tests were performed on an Intel(R) Core(TM) i5-8250U @ 1.6 [GHz] 1800 Mhz. The dashed line indicates equivalent performance for IntelVectorMath versus the implementations in Base.
113184

114185
## Supported functions
115186

‎benchmark-results‎

Submodule benchmark-results added at 06b87cb

‎benchmark/Project.toml‎

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,2 @@
1+
[deps]
2+
AcuteBenchmark = "8f64fb4b-d933-4fa5-a855-09390d771cf3"

0 commit comments

Comments
 (0)