5 Commits

Author	SHA1	Message	Date
Owen Anderson	0b78ae86c5	Cleanup byte swapping utilities to generate optimal code on the platforms we care about. (#11394) ... Summary: While the use of memcpy as part of the byte swapping sequence looks funky, all major compilers recognize and optimize this pattern reliably, resulting in essentially optimal code generation. For example, decodeUInt32LE goes from this on iOS arm64: > ldrb w8, [x0, #3] > ldrb w9, [x0, #2] > bfi w8, w9, #8, #8 > ldrb w9, [x0, #1] > bfi w8, w9, #16, #8 > ldrb w9, [x0] > bfi w8, w9, #24, #8 > mov x0, x8 > ret To this: > ldr w8, [x0] > rev w0, w8 > ret Pull Request resolved: https://github.com/pytorch/pytorch/pull/11394 Reviewed By: SsnL Differential Revision: D9728659 Pulled By: resistor fbshipit-source-id: 9afbd4adfad1d1fb7b01f1179e6707ee21fa726f	2018-09-10 15:40:24 -07:00
Sam Gross	45f665d05c	Fix decodeUInt64BE ... Fixes #1658	2017-05-26 11:21:31 -07:00
Adam Paszke	67f94557ff	Expose torch.HalfTensor	2017-02-27 19:35:47 -05:00
Adam Paszke	0c9670ddf0	Allow remapping storages at load time and serialize data in little endian order	2016-10-04 12:54:55 -07:00
Sam Gross	1486d880b0	Add Storage.from_buffer ... The from_buffer is similar to numpy's frombuffer. It decodes a Python buffer object into a Storage object. For byte and char storages, it simply copies the bytes.	2016-09-07 15:32:33 -07:00