1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
|
<!---
Licensed to the Apache Software Foundation (ASF) under one
or more contributor license agreements. See the NOTICE file
distributed with this work for additional information
regarding copyright ownership. The ASF licenses this file
to you under the Apache License, Version 2.0 (the
"License"); you may not use this file except in compliance
with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing,
software distributed under the License is distributed on an
"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
KIND, either express or implied. See the License for the
specific language governing permissions and limitations
under the License.
-->
Apache Arrow for Go
===================
[![GoDoc](https://godoc.org/github.com/apache/arrow/go/arrow?status.svg)](https://godoc.org/github.com/apache/arrow/go/arrow)
[Apache Arrow][arrow] is a cross-language development platform for in-memory
data. It specifies a standardized language-independent columnar memory format
for flat and hierarchical data, organized for efficient analytic operations on
modern hardware. It also provides computational libraries and zero-copy
streaming messaging and inter-process communication.
Reference Counting
------------------
The library makes use of reference counting so that it can track when memory
buffers are no longer used. This allows Arrow to update resource accounting,
pool memory such and track overall memory usage as objects are created and
released. Types expose two methods to deal with this pattern. The `Retain`
method will increase the reference count by 1 and `Release` method will reduce
the count by 1. Once the reference count of an object is zero, any associated
object will be freed. `Retain` and `Release` are safe to call from multiple
goroutines.
### When to call `Retain` / `Release`?
* If you are passed an object and wish to take ownership of it, you must call
`Retain`. You must later pair this with a call to `Release` when you no
longer need the object. "Taking ownership" typically means you wish to
access the object outside the scope of the current function call.
* You own any object you create via functions whose name begins with `New` or
`Copy` or when receiving an object over a channel. Therefore you must call
`Release` once you no longer need the object.
* If you send an object over a channel, you must call `Retain` before sending
it as the receiver is assumed to own the object and will later call `Release`
when it no longer needs the object.
Performance
-----------
The arrow package makes extensive use of [c2goasm][] to leverage LLVM's
advanced optimizer and generate PLAN9 assembly functions from C/C++ code. The
arrow package can be compiled without these optimizations using the `noasm`
build tag. Alternatively, by configuring an environment variable, it is
possible to dynamically configure which architecture optimizations are used at
runtime. See the `cpu` package [README](arrow/internal/cpu/README.md) for a
description of this environment variable.
### Example Usage
The following benchmarks demonstrate summing an array of 8192 values using
various optimizations.
Disable no architecture optimizations (thus using AVX2):
```sh
$ INTEL_DISABLE_EXT=NONE go test -bench=8192 -run=. ./math
goos: darwin
goarch: amd64
pkg: github.com/apache/arrow/go/arrow/math
BenchmarkFloat64Funcs_Sum_8192-8 2000000 687 ns/op 95375.41 MB/s
BenchmarkInt64Funcs_Sum_8192-8 2000000 719 ns/op 91061.06 MB/s
BenchmarkUint64Funcs_Sum_8192-8 2000000 691 ns/op 94797.29 MB/s
PASS
ok github.com/apache/arrow/go/arrow/math 6.444s
```
**NOTE:** `NONE` is simply ignored, thus enabling optimizations for AVX2 and SSE4
----
Disable AVX2 architecture optimizations:
```sh
$ INTEL_DISABLE_EXT=AVX2 go test -bench=8192 -run=. ./math
goos: darwin
goarch: amd64
pkg: github.com/apache/arrow/go/arrow/math
BenchmarkFloat64Funcs_Sum_8192-8 1000000 1912 ns/op 34263.63 MB/s
BenchmarkInt64Funcs_Sum_8192-8 1000000 1392 ns/op 47065.57 MB/s
BenchmarkUint64Funcs_Sum_8192-8 1000000 1405 ns/op 46636.41 MB/s
PASS
ok github.com/apache/arrow/go/arrow/math 4.786s
```
----
Disable ALL architecture optimizations, thus using pure Go implementation:
```sh
$ INTEL_DISABLE_EXT=ALL go test -bench=8192 -run=. ./math
goos: darwin
goarch: amd64
pkg: github.com/apache/arrow/go/arrow/math
BenchmarkFloat64Funcs_Sum_8192-8 200000 10285 ns/op 6371.41 MB/s
BenchmarkInt64Funcs_Sum_8192-8 500000 3892 ns/op 16837.37 MB/s
BenchmarkUint64Funcs_Sum_8192-8 500000 3929 ns/op 16680.00 MB/s
PASS
ok github.com/apache/arrow/go/arrow/math 6.179s
```
[arrow]: https://arrow.apache.org
[c2goasm]: https://github.com/minio/c2goasm
|