TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Google AI Cloud TPUs

23 pointsby kundiisabout 8 years ago

2 comments

aub3bhatabout 8 years ago
180 teraflops per TPU seems great.<p>For reference latest Titan X offers 12 TFLOPs [1] and upcoming AMD card for Deep Learning [2] offers 13 . Though its not clear if TPU performance is calculated at fp16 or fp32[2]. The best GPUs currentLY available on AWS offer mere 2 TFLOPs per GPU [3].<p>[1] <a href="https:&#x2F;&#x2F;blogs.nvidia.com&#x2F;blog&#x2F;2017&#x2F;04&#x2F;06&#x2F;titan-xp&#x2F;" rel="nofollow">https:&#x2F;&#x2F;blogs.nvidia.com&#x2F;blog&#x2F;2017&#x2F;04&#x2F;06&#x2F;titan-xp&#x2F;</a><p>[2] <a href="http:&#x2F;&#x2F;pro.radeon.com&#x2F;en-us&#x2F;vega-frontier-edition&#x2F;" rel="nofollow">http:&#x2F;&#x2F;pro.radeon.com&#x2F;en-us&#x2F;vega-frontier-edition&#x2F;</a><p>[3] <a href="http:&#x2F;&#x2F;images.nvidia.com&#x2F;content&#x2F;pdf&#x2F;tesla&#x2F;NVIDIA-Kepler-GK110-GK210-Architecture-Whitepaper.pdf" rel="nofollow">http:&#x2F;&#x2F;images.nvidia.com&#x2F;content&#x2F;pdf&#x2F;tesla&#x2F;NVIDIA-Kepler-GK1...</a>
评论 #14361482 未加载
评论 #14363472 未加载
infogulchabout 8 years ago
&gt; To solve this problem, we’ve has designed an all-new ML accelerator from scratch<p>I feel like that should be &quot;we have designed&quot; or &quot;we&#x27;ve designed&quot;. It seems like someone was in the middle of rewriting it and only got halfway there.