Talk:Nvidia Drive
Appearance
![]() | This article has not yet been rated on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||||||||||||
|
Drive AGX Orin 2x Tegra Orin + 2x Next-Gen GPU
I don't see any cite for the release of such system with 2xOrin+ 2xdGPUs. Don't see anything available online, otherwise I would have added one. Thoughtbox (talk) 23:22, 24 June 2020 (UTC)
Generations
I notice that the Xavier and Pegasus systems has been assigned to the third generation. However, I have not seen a cite for this, or a cite for any of the PX systems being a particular generation. Given the very large difference in computational power and autonomous driving level between Xavier and Pegasus, I am personally very skeptical they would be assigned the same generation. Dbsseven (talk) 15:46, 10 October 2017 (UTC)
- @WikiEwout: I see you have twice changed Xavier to "Third" generation. However, there is no cite provided to support this. And the only references I can find only list Pegasus as "Third" generation.[1][2] I have revered these changes, per WP:BRD. Please provide a cite if relabeling, otherwise it is original research. Dbsseven (talk) 16:54, 1 December 2017 (UTC)
- Generations tags are typically done for SoC families (see also the GPU boards from Nvidia). Following that all Xavier boards should nicely fit into the "third generation" section. Actually even after some research i can not see a mentioning from Nvidia that puts the single-chip Xavier board to _any_ generation, not even the second. Maybe their marketing terms were not ready with the generation term when those board went public. (Yes, marketing is free to do anything, even the technically non-reasonable thing - and guess what, it really does take this freedom any now and then.) - My best/worst understanding so far would put all Xavier based boards into the very same category - and add the note that Nvidia never officially announced any generation level for that specific item until now. Logic can be sharp. Keeping article sufficiently structured and useable might enforce us to use this little editor's trick. --Alexander.stohr (talk) 17:44, 13 December 2017 (UTC)
- @Alexander.stohr: I see your point about organization and clarity. However, I am hesitant to go classifying products based on how we as editors believe they should be should be classified. This seems like original research to me, especially as there are cites that already state Pegasus is the "first" third generation product. However an independent organization, perhaps based on GPU microarchitecture, might be justified so that there is clear stratification. So rather than saying "first generation, second...", organize as Maxwell/Pascal/Xavier based. Dbsseven (talk) 17:58, 13 December 2017 (UTC)
- No problem with that. Maybe time will tell. The other way GPU vendors look on their SoCs is by the integrated type of core - for nvidia these are the CUDA enabled core blocks. So we would probably have a pairing of 1st/2nd/3rd gen = Maxwell/Pascal/Volta with semiconductors Erista/Parker/Xaviar (code names) or Tegra_X1/Tegra_X2/Xavier (assumed final product names) or T210/T186/<t.b.d.> (chip numbers) - all stated under the not yet found evidence in a valid external source. --Alexander.stohr (talk) 08:49, 14 December 2017 (UTC)
- @Alexander.stohr: I see your point about organization and clarity. However, I am hesitant to go classifying products based on how we as editors believe they should be should be classified. This seems like original research to me, especially as there are cites that already state Pegasus is the "first" third generation product. However an independent organization, perhaps based on GPU microarchitecture, might be justified so that there is clear stratification. So rather than saying "first generation, second...", organize as Maxwell/Pascal/Xavier based. Dbsseven (talk) 17:58, 13 December 2017 (UTC)