abseil-cpp

117 Commits

9 Branches

22 Tags

14 MiB

Tag: Branch: Tree: e7ca23acac

Commit Graph

Author	SHA1	Message	Date
Abseil Team	4fd9a1ec50	Export of internal Abseil changes -- 03700706d80f0939e2b5b8c02a326f045b643730 by Abseil Team <absl-team@google.com>: Reduced latency and code-size of some InlinedVector methods: 1. Simpler fast path for push_back/emplace_back. 2. Do not inline slow path of push-back/emplace_back. 3. Simplify resize implementation. Performance: A simple benchmark that does the following per iteration: ``` push_back on an InlinedVector<int64> push_back on an InlinedVector<bool> ``` Sees iteration time go from 4.3ns to 2.8ns and code size shrink from 1129 bytes to 175 bytes. PiperOrigin-RevId: 343335635 -- 16f74277a9e8bf228c164b053da8b8098f76de62 by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 343332753 -- 886b6d5d0244783d309e34f03c21710f411e3cb3 by Abseil Team <absl-team@google.com>: Optimize `Status::Status`: When creating a status, we currently create an empty struct first, then assign fields. This is suboptimal: https://screenshot.googleplex.com/5HqDuFBKUEqrVgy. Relevant Benchmarks: ``` BM_StatusCopyError_Deep/threads:1 26.9ns ±13% 21.2ns ±16% -21.46% (p=0.000 n=15+15) BM_StatusCopyError_Deep/threads:2 32.0ns ±30% 25.6ns ±37% -20.17% (p=0.004 n=15+14) BM_StatusCopyError_Deep/threads:4 37.4ns ±84% 30.6ns ±58% -18.26% (p=0.029 n=15+15) BM_StatusCopyError_Deep/threads:8 47.2ns ±33% 33.5ns ±56% -28.91% (p=0.000 n=15+14) ``` PiperOrigin-RevId: 343303312 -- 2f9d945654292e8e52cad410fa41dae794cff42c by Abseil Team <absl-team@google.com>: Set SOVERSION for the installed libraries PiperOrigin-RevId: 343287682 -- 600bbfffe91cfbdc60b43cdad5619258298d0b0d by Abseil Team <absl-team@google.com>: Fix a typo in a comment (than -> that) PiperOrigin-RevId: 343187724 -- 310c82cd97b3f1f0d1ee93a0ee2b0aee828b2a93 by Abseil Team <absl-team@google.com>: Simplify unaligned memory access functions. The #ifdef to produce calls to __sanitizer_unaligned_load16 etc were needed in past versions of this code, when we were lying to the compiler about the alignment of the loads/stores, by using a reinterpret_cast. However, a year ago, absl switched to simply use memcpy. Sanitizers support this correctly by default, nothing extra is required. PiperOrigin-RevId: 343159883 -- bdf6fcf99180c371fda6ba8af82fd44656e372fa by Gennadiy Rozental <rogeeff@google.com>: Migrate usage flags to global variables instead of modeling them as Abseil Flags. Also introduce new semantic for --help=substring command line argument. PiperOrigin-RevId: 343019883 GitOrigin-RevId: 03700706d80f0939e2b5b8c02a326f045b643730 Change-Id: I4ad40dfa9606f8b8bfb2d91fd09e327105311bfb	4 years ago
Abseil Team	c6b3f2cf58	Export of internal Abseil changes -- 0c8282d75798c77733eee6167870bcc6acc0bfc1 by Evan Brown <ezb@google.com>: Provide mutable access to the key in node handles using std::launder when compiled with C++17 or later. Also, document why we can't provide mutable access to the key without C++17. Note: we use Policy::mutable_key() because btree already uses Policy::key() internally to get const key access, and we want to avoid calling std::launder unless we need mutable access to the key. PiperOrigin-RevId: 326519000 -- 8018d0c3044400f0a731b0d2d00b606742c98818 by Xiaoyi Zhang <zhangxy@google.com>: Move `Status` internal symbols from the public header into an internal header file. PiperOrigin-RevId: 326471847 -- 87a7644864ba7c003b0611898aaba1b71c840376 by Abseil Team <absl-team@google.com>: Avoid a costly divide (the division accounts for 10% of the time spent in the function). When the division is signed, the compiler has to generate a div. When it is unsigned, it can generate a shift: https://godbolt.org/z/vGfTv4. As per the test above the div, we know that the value is unsigned. PiperOrigin-RevId: 326453275 GitOrigin-RevId: 0c8282d75798c77733eee6167870bcc6acc0bfc1 Change-Id: I0a953558358055ab3dc6a533d8930698509b1195	4 years ago
Abseil Team	768eb2ca28	Export of internal Abseil changes -- f012012ef78234a6a4585321b67d7b7c92ebc266 by Laramie Leavitt <lar@google.com>: Slight restructuring of absl/random/internal randen implementation. Convert round-keys.inc into randen_round_keys.cc file. Consistently use a 128-bit pointer type for internal method parameters. This allows simpler pointer arithmetic in C++ & permits removal of some constants and casts. Remove some redundancy in comments & constexpr variables. Specifically, all references to Randen algorithm parameters use RandenTraits; duplication in RandenSlow removed. PiperOrigin-RevId: 312190313 -- dc8b42e054046741e9ed65335bfdface997c6063 by Abseil Team <absl-team@google.com>: Internal change. PiperOrigin-RevId: 312167304 -- f13d248fafaf206492c1362c3574031aea3abaf7 by Matthew Brown <matthewbr@google.com>: Cleanup StrFormat extensions a little. PiperOrigin-RevId: 312166336 -- 9d9117589667afe2332bb7ad42bc967ca7c54502 by Derek Mauro <dmauro@google.com>: Internal change PiperOrigin-RevId: 312105213 -- 9a12b9b3aa0e59b8ee6cf9408ed0029045543a9b by Abseil Team <absl-team@google.com>: Complete IGNORE_TYPE macro renaming. PiperOrigin-RevId: 311999699 -- 64756f20d61021d999bd0d4c15e9ad3857382f57 by Gennadiy Rozental <rogeeff@google.com>: Switch to fixed bytes specific default value. This fixes the Abseil Flags for big endian platforms. PiperOrigin-RevId: 311844448 -- bdbe6b5b29791dbc3816ada1828458b3010ff1e9 by Laramie Leavitt <lar@google.com>: Change many distribution tests to use pcg_engine as a deterministic source of entropy. It's reasonable to test that the BitGen itself has good entropy, however when testing the cross product of all random distributions x all the architecture variations x all submitted changes results in a large number of tests. In order to account for these failures while still using good entropy requires that our allowed sigma need to account for all of these independent tests. Our current sigma values are too restrictive, and we see a lot of failures, so we have to either relax the sigma values or convert some of the statistical tests to use deterministic values. This changelist does the latter. PiperOrigin-RevId: 311840096 GitOrigin-RevId: f012012ef78234a6a4585321b67d7b7c92ebc266 Change-Id: Ic84886f38ff30d7d72c126e9b63c9a61eb729a1a	5 years ago

Author

SHA1

Message

Date

Abseil Team

4fd9a1ec50

Export of internal Abseil changes

--
03700706d80f0939e2b5b8c02a326f045b643730 by Abseil Team <absl-team@google.com>:

Reduced latency and code-size of some InlinedVector methods:

1. Simpler fast path for push_back/emplace_back.
2. Do not inline slow path of push-back/emplace_back.
3. Simplify resize implementation.

Performance:
A simple benchmark that does the following per iteration:

```
push_back on an InlinedVector<int64>
push_back on an InlinedVector<bool>
```

Sees iteration time go from 4.3ns to 2.8ns and code size shrink from 1129 bytes to 175 bytes.

PiperOrigin-RevId: 343335635

--
16f74277a9e8bf228c164b053da8b8098f76de62 by Derek Mauro <dmauro@google.com>:

Internal change

PiperOrigin-RevId: 343332753

--
886b6d5d0244783d309e34f03c21710f411e3cb3 by Abseil Team <absl-team@google.com>:

Optimize `Status::Status`: When creating a status, we currently create an empty struct first, then assign fields. This is suboptimal: https://screenshot.googleplex.com/5HqDuFBKUEqrVgy.

Relevant Benchmarks:
```
BM_StatusCopyError_Deep/threads:1              26.9ns ±13%   21.2ns ±16%  -21.46%  (p=0.000 n=15+15)
BM_StatusCopyError_Deep/threads:2              32.0ns ±30%   25.6ns ±37%  -20.17%  (p=0.004 n=15+14)
BM_StatusCopyError_Deep/threads:4              37.4ns ±84%   30.6ns ±58%  -18.26%  (p=0.029 n=15+15)
BM_StatusCopyError_Deep/threads:8              47.2ns ±33%   33.5ns ±56%  -28.91%  (p=0.000 n=15+14)
```

PiperOrigin-RevId: 343303312

--
2f9d945654292e8e52cad410fa41dae794cff42c by Abseil Team <absl-team@google.com>:

Set SOVERSION for the installed libraries

PiperOrigin-RevId: 343287682

--
600bbfffe91cfbdc60b43cdad5619258298d0b0d by Abseil Team <absl-team@google.com>:

Fix a typo in a comment (than -> that)

PiperOrigin-RevId: 343187724

--
310c82cd97b3f1f0d1ee93a0ee2b0aee828b2a93 by Abseil Team <absl-team@google.com>:

Simplify unaligned memory access functions.

The #ifdef to produce calls to __sanitizer_unaligned_load16 etc were needed in past versions of this code, when we were lying to the compiler about the alignment of the loads/stores, by using a reinterpret_cast.

However, a year ago, absl switched to simply use memcpy. Sanitizers support this correctly by default, nothing extra is required.

PiperOrigin-RevId: 343159883

--
bdf6fcf99180c371fda6ba8af82fd44656e372fa by Gennadiy Rozental <rogeeff@google.com>:

Migrate usage flags to global variables instead of modeling them as Abseil Flags.

Also introduce new semantic for --help=substring command line argument.

PiperOrigin-RevId: 343019883
GitOrigin-RevId: 03700706d80f0939e2b5b8c02a326f045b643730
Change-Id: I4ad40dfa9606f8b8bfb2d91fd09e327105311bfb

4 years ago

Abseil Team

c6b3f2cf58

Export of internal Abseil changes

--
0c8282d75798c77733eee6167870bcc6acc0bfc1 by Evan Brown <ezb@google.com>:

Provide mutable access to the key in node handles using std::launder when compiled with C++17 or later.

Also, document why we can't provide mutable access to the key without C++17.

Note: we use Policy::mutable_key() because btree already uses Policy::key() internally to get const key access, and we want to avoid calling std::launder unless we need mutable access to the key.
PiperOrigin-RevId: 326519000

--
8018d0c3044400f0a731b0d2d00b606742c98818 by Xiaoyi Zhang <zhangxy@google.com>:

Move `Status` internal symbols from the public header into an internal header file.

PiperOrigin-RevId: 326471847

--
87a7644864ba7c003b0611898aaba1b71c840376 by Abseil Team <absl-team@google.com>:

Avoid a costly divide (the division accounts for 10% of the time spent in the function).

When the division is signed, the compiler has to generate a div. When it is unsigned, it can generate a shift: https://godbolt.org/z/vGfTv4. As per the test above the div, we know that the value is unsigned.

PiperOrigin-RevId: 326453275
GitOrigin-RevId: 0c8282d75798c77733eee6167870bcc6acc0bfc1
Change-Id: I0a953558358055ab3dc6a533d8930698509b1195

4 years ago

Abseil Team

768eb2ca28

Export of internal Abseil changes

--
f012012ef78234a6a4585321b67d7b7c92ebc266 by Laramie Leavitt <lar@google.com>:

Slight restructuring of absl/random/internal randen implementation.

Convert round-keys.inc into randen_round_keys.cc file.

Consistently use a 128-bit pointer type for internal method parameters. This allows simpler pointer arithmetic in C++ & permits removal of some constants and casts.

Remove some redundancy in comments & constexpr variables. Specifically, all references to Randen algorithm parameters use RandenTraits; duplication in RandenSlow removed.

PiperOrigin-RevId: 312190313

--
dc8b42e054046741e9ed65335bfdface997c6063 by Abseil Team <absl-team@google.com>:

Internal change.

PiperOrigin-RevId: 312167304

--
f13d248fafaf206492c1362c3574031aea3abaf7 by Matthew Brown <matthewbr@google.com>:

Cleanup StrFormat extensions a little.

PiperOrigin-RevId: 312166336

--
9d9117589667afe2332bb7ad42bc967ca7c54502 by Derek Mauro <dmauro@google.com>:

Internal change

PiperOrigin-RevId: 312105213

--
9a12b9b3aa0e59b8ee6cf9408ed0029045543a9b by Abseil Team <absl-team@google.com>:

Complete IGNORE_TYPE macro renaming.

PiperOrigin-RevId: 311999699

--
64756f20d61021d999bd0d4c15e9ad3857382f57 by Gennadiy Rozental <rogeeff@google.com>:

Switch to fixed bytes specific default value.

This fixes the Abseil Flags for big endian platforms.

PiperOrigin-RevId: 311844448

--
bdbe6b5b29791dbc3816ada1828458b3010ff1e9 by Laramie Leavitt <lar@google.com>:

Change many distribution tests to use pcg_engine as a deterministic source of entropy.

It's reasonable to test that the BitGen itself has good entropy, however when testing the cross product of all random distributions x all the architecture variations x all submitted changes results in a large number of tests. In order to account for these failures while still using good entropy requires that our allowed sigma need to account for all of these independent tests.

Our current sigma values are too restrictive, and we see a lot of failures, so we have to either relax the sigma values or convert some of the statistical tests to use deterministic values.

This changelist does the latter.

PiperOrigin-RevId: 311840096
GitOrigin-RevId: f012012ef78234a6a4585321b67d7b7c92ebc266
Change-Id: Ic84886f38ff30d7d72c126e9b63c9a61eb729a1a

5 years ago

2 Commits (e7ca23acac146b10edc4f752edd0bd28b0f14ea3)