tensorflow-aarch64 GPU version

[qiuweibo]

Hi,

Thanks so much for your work. But this is tensorfllow 1.4.0 aarch64 CPU version. However, normally we need to use tensorflow GPU version on ARM.

Surprisingly, I cannot find any Tensorflow 1.4.0 aarch64 GPU version. Have you seen any?

Thanks in advance.

Best regards,
Weibo Qiu.

[Haoyu Zhang]

很抱歉现在才回复你，关于gpu版的tf，你可以参考 nvidia 的预编译版本的。例如： https://developer.download.nvidia.cn/compute/redist/jp/v42/tensorflow-gpu/ 希望可以帮到你😉。 qiuweibo <[e-mail]> 于2019年8月2日周五 下午7:15写道：

…

Hi, Thanks so much for your work. But this is tensorfllow 1.4.0 aarch64 CPU version. However, normally we need to use tensorflow GPU version on ARM. Surprisingly, I cannot find any Tensorflow 1.4.0 aarch64 GPU version. Have you seen any? Thanks in advance. Best regards, Weibo Qiu. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#5?email_source=notifications&email_token=ADTVLVG3H37YJNXPU25UUVDQCQJNLA5CNFSM4II42NH2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HDBFAQQ>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADTVLVE4QWVSONB5JSZOD2TQCQJNLANCNFSM4II42NHQ> .

[qiuweibo]

很抱歉现在才回复你，关于gpu版的tf，你可以参考 nvidia 的预编译版本的。例如：
https://developer.download.nvidia.cn/compute/redist/jp/v42/tensorflow-gpu/
希望可以帮到你😉。

qiuweibo [e-mail] 于2019年8月2日周五 下午7:15写道：
…
Hi,

Thanks so much for your work. But this is tensorfllow 1.4.0 aarch64 CPU
version. However, normally we need to use tensorflow GPU version on ARM.

Surprisingly, I cannot find any Tensorflow 1.4.0 aarch64 GPU version. Have
you seen any?

Thanks in advance.

Best regards,
Weibo Qiu.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#5?email_source=notifications&email_token=ADTVLVG3H37YJNXPU25UUVDQCQJNLA5CNFSM4II42NH2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HDBFAQQ>,
or mute the thread
https://github.com/notifications/unsubscribe-auth/ADTVLVE4QWVSONB5JSZOD2TQCQJNLANCNFSM4II42NHQ
.

Hello!

我已经把毕业论文交上了，哈哈。

到最后我也没找到Tensorflow 1.4.0 aarch64 GPU的版本。你发的链接里nvidia预编译的tf版本都是1.9以上，我那个项目用不了，最后我就跑了cpu的版本分析了一下，就是非常慢。。

[Haoyu Zhang]

那个链接里的是 tf 1.14 左右的，适用于 jetson nano 这块开发板的，应该也适用于 aarch 64 架构的其他板子，如日后有需要可以试试看。

…

On Sun, Aug 25, 2019, 00:14 qiuweibo ***@***.***> wrote: 很抱歉现在才回复你，关于gpu版的tf，你可以参考 nvidia 的预编译版本的。例如： https://developer.download.nvidia.cn/compute/redist/jp/v42/tensorflow-gpu/ 希望可以帮到你😉。 qiuweibo ***@***.*** 于2019年8月2日周五 下午7:15写道： … Hi, Thanks so much for your work. But this is tensorfllow 1.4.0 aarch64 CPU version. However, normally we need to use tensorflow GPU version on ARM. Surprisingly, I cannot find any Tensorflow 1.4.0 aarch64 GPU version. Have you seen any? Thanks in advance. Best regards, Weibo Qiu. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#5 <#5> ?email_source=notifications&email_token=ADTVLVG3H37YJNXPU25UUVDQCQJNLA5CNFSM4II42NH2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HDBFAQQ>, or mute the thread https://github.com/notifications/unsubscribe-auth/ADTVLVE4QWVSONB5JSZOD2TQCQJNLANCNFSM4II42NHQ . Hello! 我已经把毕业论文交上了，哈哈。 到最后我也没找到Tensorflow 1.4.0 aarch64 GPU的版本。你发的链接里nvidia预编译的tf版本都是1.9以上，我那个项目用不了，最后我就跑了cpu的版本分析了一下，就是非常慢。。 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5?email_source=notifications&email_token=ADTVLVGAUOH6JPEIOUETA5DQGFM53A5CNFSM4II42NH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5CDBVQ#issuecomment-524562646>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADTVLVG4545MM7O6JSUBA3LQGFM53ANCNFSM4II42NHQ> .

[qiuweibo]

那个链接里的是 tf 1.14 左右的，适用于 jetson nano 这块开发板的，应该也适用于 aarch 64
架构的其他板子，如日后有需要可以试试看。
…

嗯，那个链接的前几级目录里有tf1.9开始的版本，但我都用不了。
我的板子是Jetson TX2， 也是aarch64，但是我的项目只能用tf1.4.0，所以最后就没找到gpu的

[cuih]

pytorch 和 tf都找不到GPU版本的aarch64平台支持（除了Jetson板子），明明CUDA已经可以安装在aarch64平台上了，请问是为啥呢？是pytorch-gpu/tensorflow-gpu的支持问题还是英伟达官方需要提供什么支持呢