Wikipedia:Articles for deletion/Cloud-native processor

Cloud-native processor

[Hide this box] New to Articles for deletion (AfD)? Read these primers!

(Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL)

This is a neologism that has little independent coverage or notability. It's not meaningfully different from just a GPCPU aside from in core count and who builds/buys them. From looking at the sourcing, it looks like a term being pushed by one vendor only (Ampere Computing) but without broader industry adoption. lizthegrey (talk) 18:30, 14 October 2023 (UTC)[reply]

Note: This discussion has been included in the deletion sorting lists for the following topics: Computing and Internet. lizthegrey (talk) 18:30, 14 October 2023 (UTC)[reply]
Keep All three major types of processors (Intel, ARM, AMD) are covered - especially now that I've just now fleshed out Intel's offering. Michaelmalak (talk) 20:16, 14 October 2023 (UTC)[reply]
Your recent addition is WP:SYNTH to change "for cloud-native workloads" into supporting the term "cloud-native processor". I don't dispute that processors can be suited to CN workloads, but the idea that a processor category of CNP exists is not broadly supported. lizthegrey (talk) 20:28, 14 October 2023 (UTC)[reply]
Also just now added a cite to a secondary source (book) Michaelmalak (talk) 20:35, 14 October 2023 (UTC)[reply]
I've read the original text for that cite, which doesn't seem to actually use the phrase "cloud-native processor", only states that "cloud computing environments use multi-core processors" (but this is hardly unique, everything is multi-core, even desktop/laptop processors). so again fails WP:SYNTH, and more severely than the intel example which at least says "for cloud-native workloads". lizthegrey (talk) 21:03, 14 October 2023 (UTC)[reply]

this doesn't make any sense, if a criterion for being a "cloud-native processor" is not using SMT (per lede), then Intel and AMD processors should be disqualified, thus making it an ARM and RISC only thing. And of the ARM vendors, only Ampere uses this phrasing; AWS/Annapurna does not for its Graviton line. https://www.google.com/search?q=%22cloud-native+processor%22+-ampere vs https://www.google.com/search?q=%22cloud-native+processor%22+ampere is pretty striking. Ampere specifically markets itself as the "first Cloud-Native Processor supplier", so this page is just marketing fluff for Ampere. lizthegrey (talk) 21:13, 14 October 2023 (UTC)[reply]

Delete Different lines of CPU models may be suitable or optimized for e.g. virtualization (feature flags for Intel VT-x or AMD SVM), gaming (cache size), high-frequency trading (clock speed), vectorized computations (microarchitecture), many-core workloads, power efficiency etc. etc., and marketed as such. That does not mean they are "X-native processors". It is just marketing. If there were truly a new type of processor that deserved the term "cloud-native processor" (as opposed to "cloud-native CPU", which would be a lower bar to justify, but even that is still not met IMO) it would have to be tangibly distinct in architecture and physical nature from conventional CPUs in the way that GPUs, TPUs, and QPUs are - not merely a usage trend or marketing term. The article (and references) fail to articulate what the specific cloud-native nature of the CPU is. The fact is that they are simply regular CPUs with many cores that are efficiency-optimized, and among the many types of workloads that can benefit from this, cloud-native applications may tend to be one because they are easily horizontally scaled. The explanation in the article, "allows for simultaneous connections in a cloud environment resulting in scalability" is highly tenuous and makes no sense at all. It is definitely not the case that cloud-native applications require extremely dense servers with high core counts, because that is antithetical to the distributed microservices architecture of cloud-native computing, where many lightweight containers are orchestrated and horizontally scaled across many redundant nodes. Rather, cloud-native architectures simply allow many-core CPUs to be exploited efficiently and scalably while eliminating the downside of the lower performance of individual cores. Cloud-native is not an easily-articulated or -understood concept, but it is easy to throw around as a buzzword, as in this case. Rotiro (talk) 06:05, 19 October 2023 (UTC)[reply]

Relisted to generate a more thorough discussion and clearer consensus.
Please add new comments below this notice. Thanks, Liz ^{Read! Talk!} 21:26, 21 October 2023 (UTC)[reply]

I think the best option is a partial merge to Cloud-native computing. This seems right per WP:NOPAGE, and also provides that article as context to this article's information. Additionally, I don't see a way this specific page can exist without being promotional. "Cloud-native computing" is still buzzwordy, but at least it vaguely encompasses some vaguely-agreed-upon industry expectations that we can attempt to use secondary sources to write an article about, and a paragraph noting that there are processors being designed for the purpose would probably fit there. —siro χ o 06:32, 22 October 2023 (UTC)[reply]
I was going to second your opinion initially, but couldn't find anything about "cloud-native computing" that would apply the term to CPUs in Google Scholar. All mentions in Google Books seem to refer to software as well. Do you have any sources that would support that the "vaguely-agreed-upon industry expectations" in this regard exist? Wonder if may have missed something. PaulT2022 (talk) 18:56, 30 October 2023 (UTC)[reply]
Delete as a poorly-sourced neologism. All sources echo Ampere Computing's marketing that appears to artificially differentiate processor with quantitative differences to competitors (at best), higher thread count per socket, by claiming there's an underlying qualitative difference that applies to cloud applications. The claim isn't supported by any engineering publications on hardware architecture. PaulT2022 (talk) 01:16, 23 October 2023 (UTC)[reply]
Keep. There will be no specific article for hardware dedicated to cloud applications. "Cloud-native computing" is not about the processors themselves. Merging would make it a grab bag. बिनोद थारू (talk) 05:05, 30 October 2023 (UTC)[reply]
For what it's worth, I'm not proposing to leave a redirect, I'm proposing to outright delete (or, second choice, redirect to CPU or similar) lizthegrey (talk) 12:25, 30 October 2023 (UTC)[reply]
Keep as it is redundant to merge an article when there is little focus on the topic of the article merged to it, and deleting the article would constantly cause new creations based on the topic. Equalwidth (talk) 10:34, 30 October 2023 (UTC)[reply]