Talk:Tensor Processing Unit: Difference between revisions

Content deleted Content added
TPU - Tensor Processing Unit: - updated to my previous entry
 
(21 intermediate revisions by 19 users not shown)
Line 1:
{{Talk header}}
{{WikiProjectBannerShell|{{WikiProject Computer science |class=Start |importance=Mid}}{{WikiProject Computing |class=Start |importance=Mid |science=y |science-importance=Mid |hardware=y |hardware-importance=Mid}}{{WikiProject Google |class=Start |importance=Mid}}}}
{{WikiProject banner shell|class=C|
{{findsourcesnotice}}
{{WikiProject Computer science |importance=Mid}}
{{WikiProjectBannerShell|{{WikiProject Computer science |class=Start |importance=Mid}}{{WikiProject Computing |class=Start |importance=Mid |science=y |science-importance=Mid |hardware=y |hardware-importance=Mid}}{{WikiProject Google |class=Start |importance=Mid}}}}
{{WikiProject Google |importance=Mid}}
}}
 
== adding Nvidia specific solutions? ==
Line 9 ⟶ 13:
 
== Rename to "Google TPU" ==
<div class="boilerplate" style="background-color: #efe; margin: 2em 0 0 0; padding: 0 10px 0 10px; border: 1px dotted #aaa;"><!-- Template:RM top -->
:''The following is a closed discussion of a [[WP:requested moves|requested move]]. <span style="color:red">'''Please do not modify it.'''</span> Subsequent comments should be made in a new section on the talk page. Editors desiring to contest the closing decision should consider a [[Wikipedia:move review|move review]]. No further edits should be made to this section. ''
 
The result of the move request was: '''not moved'''. <small>([[Wikipedia:Requested moves/Closing instructions#Non-admin closure|non-admin closure]])</small> [[User:Steel1943|<span style="color: #2F4F4F;">'''''Steel1943'''''</span>]] ([[User talk:Steel1943|talk]]) 20:34, 28 December 2017 (UTC)
{{requested move/dated|Google TPU}}
----
 
[[:Tensor processing unit]] → {{no redirect|Google TPU}} – The page describes mainly Google TPUs. I think that's why the company name should be put into a title. [[Special:Contributions/2.92.113.239|2.92.113.239]] ([[User talk:2.92.113.239|talk]]) 17:31, 17 December 2017 (UTC)
Line 17 ⟶ 24:
::{{u|scope creep}} and {{ping|JFG}} FYI, there is already a highly related general article: [[AI accelerator]]. Therefore I'm not sure this is the place for expansion/generalization, unless there is a lot of ''Tensor'' specific content. [[User:Dbsseven|Dbsseven]] ([[User talk:Dbsseven|talk]]) 17:25, 18 December 2017 (UTC)
*'''Oppose''' - I'm not sure adding "Google" aids the article in any way. Right now there isn't a need for further specificity in the title (IMO). [[User:Dbsseven|Dbsseven]] ([[User talk:Dbsseven|talk]]) 17:25, 18 December 2017 (UTC)
 
*'''Favour''' - Google does not have the right to appropriate the word "[[Tensor]]", a term specific to mathematics and physics, just as no vendor has the right to appropriate the term "[[CPU]]" or "[[GPU]]" for one of its products. TPUs are simply a new type of processor specializing in implementing tensor mathematics. Also, other vendors are already manufacturing TPUs, such as [[nVidia]]'s [[Volta (microarchitecture)|Volta]] microarchitecture chip, a combination GPU and TPU, 27,648 of which are currently powering [[Summit (supercomputer)|Summit]], the [[TOP500#Top_10_ranking|World's most powerful supercomputer]] at [https://www.olcf.ornl.gov/olcf-resources/compute-systems/summit/ Oak Ridge National Laboratory]. See: [https://www.nvidia.com/en-us/data-center/tensorcore/ Tensor Cores in NVIDIA's Tesla V100 GPU], a direct competitor of Google's TPU. [https://www.extremetech.com/extreme/269008-google-announces-8x-faster-tpu-3-0-for-ai-machine-learning This] media article offers an example of the TPU acronym used generically. Therefore this page ought to instead refer to generic TPUs and onclude references to other pages describing specific instances of TPUs offered by different vendors. (NOTE: The request to close this discussion is extemporaneous given important arguments such as this one have not been made.) [[Special:Contributions/181.88.207.203|181.88.207.203]] ([[User talk:181.88.207.203|talk]]) 05:06, 27 June 2018 (UTC)
 
*'''Comment''' I added a [[WP:PAID]] disclosure request to the user, [[User talk:2.92.113.239]], because I think it is push to get some free advertising on the part of Google, re: this article. I'm not saying generalize it, I'm looking to add as much detail as possible. This is one processor of class of processors, that should be described by their architecture and api, not by name. I don't like the [[AI accelerator]] article, it is essentially a pamphlet offering free advertising. [[User:Dbsseven|Dbsseven]], Kudos to yourself, I see you have tried to smarten it up a bit, but the article is a first cut, and I think it is rank. Statement like this, (in this article) which are almost fancruft: ''Google stated the first generation TPU design was memory bandwidth limited'', instead of ''the first generation TPU design was memory bandwidth limited''.The first violates [[WP:NOTAVERTISING]]. The second doesn't. The point i'm trying is make is. It is new field, and lots of new disruptive designs are coming out, and everybody is trying to find what works, but processor design follows an ethos, a design language, they come of the uni's after decades of research, so I know for a fact there is other ultra high speed, low precision FPU processors out there. This article is about the TPU, but it should have been a architecture articles describing ultra high speed, high bandwidth, low precision FPU processors, with the TPU an example, amongst several others, including a good descritpion of the architecture. [[User:Scope creep|scope_creep]] ([[User talk:Scope creep|talk]]) 18:21, 18 December 2017 (UTC)
----
:''The above discussion is preserved as an archive of a [[Wikipedia:Requested moves|requested move]]. <span style="color:red">'''Please do not modify it.'''</span> Subsequent comments should be made in a new section on this talk page or in a [[Wikipedia:Move review|move review]]. No further edits should be made to this section.''</div><!-- Template:RM bottom -->
 
== Should this remain only dedicated to Google's TPU? ==
 
I'm not proposing we move or rename the article, but other companies are starting to sell chips they call "Tensor Processing Units" as well, and the term's been used in research for quite some time. {{As of|October 2018}}, is it better to incorporate encyclopedic coverage of other TPU content besides Google's? --[[User:Daviddwd|Daviddwd]] ([[User talk:Daviddwd|talk]]) 22:55, 4 October 2018 (UTC)
 
== TPU - Tensor Processing Unit ==
 
Again, TPU's are a type of processor dedicated to computing Tensors and are manufactured by multiple vendors, as is the case with CPUs and GPUs. TPUs are NOT specific to Google, thus this article needs to me made generic to incorporate info on TPUs from other vendors as well. Thanks. [[Special:Contributions/181.12.115.177|181.12.115.177]] ([[User talk:181.12.115.177|talk]]) 04:22, 4 April 2019 (UTC)
:The product name "Tensor Processing Unit" '''is''' specific to Google. However products that provide tensor processing are not specific to Google; see [[AI accelerator]]. A possible cause of confusion is the "tensor cores" used by nVidia, including [[Volta (microarchitecture)]]. This confusing, but to my knowlegde Google is the only company that uses the specific wording "Tensor processing unit ". [[User:Mrjulesd|<span style="color:orange;">--Jules</span>]]&nbsp;[[User talk:Mrjulesd|(Mrjulesd)]] 10:47, 4 April 2019 (UTC)
 
The article says "Google's TPUs are proprietary and are not commercially available". But they '''are''' commercially available. https://coral.withgoogle.com/docs/accelerator/get-started/ . I've updated the article. [[User:Sayitclearly|Sayitclearly]] ([[User talk:Sayitclearly|talk]]) 13:43, 3 September 2019 (UTC)
 
Again, Google is appropriating the word "Tensor" for its own use. This is inappropriate: Words are for everyone to use and "tensor", like "vector", is no exception. For the uninitiated, tensor describes a field of vectors and is not at all bound to the AI. To the point, tensors were invented by italian mathematician Ricci-Curbastro to aid Einstein express his famous General Relativity in a single equation in an epoch were the computer hadn't even been invented, much less AI. The other words are "Processor" and "Unit", also, I believe, of general use. So should companies also have rights to CPU - Code Processing Unit, FPU - Floating Point Unit and GPU - Graphics Processing Unit? Of course not! So why would TPU - Tensor Processing Unit be the exception? Please, let's recover TPU for general use. [[Special:Contributions/181.12.78.206|181.12.78.206]] ([[User talk:181.12.78.206|talk]]) 19:34, 14 December 2020 (UTC)
 
It would be useful for the article to, somewhere, include a link to [[Tensor (machine_learning)]], regardless of whether or not a TPU is Google-specific. I had never heard of a TPU (or a Tensor) until today, so I expected that, upon finding this article, it would make it clear what a Tensor is as well as what a TPU is. [[User:Gmporr|Gmporr]] ([[User talk:Gmporr|talk]]) 17:46, 24 March 2025 (UTC)
:I went ahead and made a change (made the article's first word "Tensor") a wikilink. But as it's now doubly highlighted (bold and typical wikilink purple highlight) it may not be readily clear to readers that it's a wikilink. So I'd value other editors' changes if there seems to be a better place in the article to include the link. [[User:Gmporr|Gmporr]] ([[User talk:Gmporr|talk]]) 18:30, 24 March 2025 (UTC)
 
== August 2021: Google Nest cameras will include TPUs ==
 
Worth adding that the 2021 refresh of Google Nest Cams (including the Nest Doorbell, no longer called Nest Hello) will include TPUs? the TPUs will allow immediate processing and alerting. (Also offline recording and storing, due to battery inclusion). This section by DAVIDBSPALDING, but using work computer, so can't sign in to Wiki. -- [[Special:Contributions/208.242.14.199|208.242.14.199]] ([[User talk:208.242.14.199|talk]]) 15:34, 12 August 2021 (UTC)
 
== Products ==
The table is cryptic.
Perhaps "Mib" and "W" could be use full names.
Bill
[[User:W102102|W102102]] ([[User talk:W102102|talk]]) 09:57, 13 August 2022 (UTC)