This website requires Javascript to function properly. Please go to the setting of your web browser and enable Javascript for this website.

Ad by

最优利率和cashback可以申请特批，好信用好收入offer更好。请点链接扫码加微信咨询，Scotiabank -- Nick Zhang 6478812600。

Ad by

最优利率和cashback可以申请特批，好信用好收入offer更好。请点链接扫码加微信咨询，Scotiabank -- Nick Zhang 6478812600。

Introducing Accelerated PyTorch Training on Mac

xmlhttprequest(build5381)

In collaboration with the Metal engineering team at Apple, we are excited to announce support for GPU-accelerated PyTorch training on Mac. Until now, PyTorch training on Mac only leveraged the CPU, but with the upcoming PyTorch v1.12 release, developers and researchers can take advantage of Apple silicon GPUs for significantly faster model training. This unlocks the ability to perform machine learning workflows like prototyping and fine-tuning locally, right on Mac.

Full Story:
https://pytorch.org/blog/introducing-accelerated-pytorch-training-on-mac/

(#14575432@0)
Last Updated: 2022-5-20
This post has been archived. It cannot be replied.

Report

Replies, comments and Discussions:

工作学习 / 学科技术 / Introducing Accelerated PyTorch Training on Mac -xmlhttprequest(build5381); 2022-5-20 {867} (#14575432@0)

In collaboration with the Metal engineering team at Apple, we are excited to announce support for GPU-accelerated PyTorch training on Mac. Until now, PyTorch training on Mac only leveraged the CPU, but with the upcoming PyTorch v1.12 release, developers and researchers can take advantage of Apple silicon GPUs for significantly faster model training. This unlocks the ability to perform machine learning workflows like prototyping and fine-tuning locally, right on Mac.

Full Story:
https://pytorch.org/blog/introducing-accelerated-pytorch-training-on-mac/

在mac上训练机器学习性价比不好 -donhill(Nickle); 2022-5-21 {65} (#14578454@0)
一个配M1 ultra 的本大几千刀了。搞个intel 12900 + nv 3090 不香吗。

多一个选择总是好的，而且配 12900 的本儿也不便宜。 -xmlhttprequest(build5381); 2022-5-21 (#14578560@0)

colab免费 -iamflying(叶和花); 2022-5-21 (#14578823@0)

刚入手的时候可以用。免费的东西也就玩玩toy project.想做点严肃东西那它是真心不方便。数据没个像样地方存，好像我记得一个session限时12小时？不过Colab 有付费选项的。 -donhill(Nickle); 2022-5-21 (#14578985@0)

数据可以挂上google drive存。拿来学习其实还可以啦。 -moonhalf(石室矢诗); 2022-5-21 (#14579339@0)
个人学习够了。如果要打比赛，可以考虑用GCP。 -iamflying(叶和花); 2022-5-21 (#14579467@0)

愿意付费就好办了，AWS, Azure 。。还可以支持国货阿里云 lol -donhill(Nickle); 2022-5-21 (#14579478@0)

是啊。可是家里一直就用Mac，不想为了这个单买机器的话，多个选择也挺好 -moonhalf(石室矢诗); 2022-5-21 (#14579337@0)

Mac上也可以用TensoFlow PluggableDevice -iamflying(叶和花); 2022-5-21 (#14579472@0)

呃，你的mac是 M1 芯片吗，我总以为兼容性不好。当然，我穷，没用过。 -donhill(Nickle); 2022-5-21 (#14579482@0)

是的。MacBook Pro 14 -iamflying(叶和花); 2022-5-22 (#14579949@0)
一般开发可以。兼容性不是特别好。比如python 3.6这种“老”版本很难装上，装好了什么numpy 的库也没法安装。不过总体还可以，我python, Scalia, Java 开发，就遇到这个麻烦。哦还有就是docker出来的是arm64版的image，和用intel的服务器不一样。 -moonhalf(石室矢诗); 2022-5-22 (#14580006@0) +1

兼容性是有些问题。scipy花了一年才支持M1。 -iamflying(叶和花); 2022-5-23 (#14581736@0)
在M1上面装开发系统，你是给自己自虐么？编译器对cpu指令集有非常挑剔的要求，不是普通软件可以替代的。M1也就适合和M1适配的视频编辑软件还效果不错。 -zhengy4(_); 2022-5-24 (#14583467@0) +1

开发看具体情况吧。实际使用，Java，python，scala，Javascript都挺好，没啥“自虐”的地方。估计包括了大部分写代码的情况。你是用到什么，还是看到哪种开发觉得会“自虐”吗？ -moonhalf(石室矢诗); 2022-5-24 (#14583812@0) +1

内行👍 -moonhalf(石室矢诗); 2022-5-22 (#14580001@0)

@Las Vegas

Introducing Accelerated PyTorch Training on Mac

Replies, comments and Discussions:

More Topics