-
Hey everyone, staff have documented a list of banned content and subject matter that we feel are not consistent with site values, and don't make sense to host discussion of on Famiboards. This list (and the relevant reasoning per item) is viewable here.
StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)
- Thread starter Dakhil
- Start date
YugamiSekai
T239 vs. Picross 4K
- Pronouns
- He/Him
This kind of tech along with MicroLED and Graphene batteries always get me excited about how well a potential Switch 3 or 4 could perform.
- Pronouns
- He/Him
I wouldn't get too excited about graphene any time soon, they still need to figure out cost efficient mass production which has been the number 1 issue with graphene for decades.
This chip is really cool though (pun intended), I'm curious how cheap it can be at a mass production scale.
KCannon
Octorok
- Pronouns
- He/Him
RUMOR: Massive Smash Ultimate leak is allegedly an emotional roller coaster [UPDATES IN OP]
Updates are at the bottom of this post Definitely worthy of its own thread. There have been many "leaks" in the past, but this one is incredibly well made if it's actually fake. It's getting a ton of attention. A guy (accidentally?) leaked the entire roster through the final character art...
www.resetera.com
Just read the first post to get an idea of the absolute madness.
- Pronouns
- He/Him
Serif
𝕽𝖊𝖓𝖊𝖌𝖆𝖉𝖊 𝕬𝖓𝖌𝖊𝖑
I see GA10F on the Nvidia Drive OS Vulkan SC sample page published 08/11/2022,
But not on the 'latest' Drive OS 6.0 Linux SDK docs dated 11/30/2022 that is linked from the Nvidia Drive OS page.
The latest version only has gv11b and ga10b on that same 'running Vulcan samples' page. Also, ga10b/TEGRA234 is the only one listed in the NvGPU device tree.
From the 8/11 archived version:
From the 11/30 'latest' version:
So from my end, I couldn't find anything Drake related on the latest Linux SDK docs, though I could have missed something.
But not on the 'latest' Drive OS 6.0 Linux SDK docs dated 11/30/2022 that is linked from the Nvidia Drive OS page.
The latest version only has gv11b and ga10b on that same 'running Vulcan samples' page. Also, ga10b/TEGRA234 is the only one listed in the NvGPU device tree.
From the 8/11 archived version:
From the 11/30 'latest' version:
So from my end, I couldn't find anything Drake related on the latest Linux SDK docs, though I could have missed something.
mariodk18
Install Base Forum Namer
What's the deal with alpha particles/shaders/etc.? Why are they so taxing and/or so bandwidth heavy? I remember reading somewhere that the 360 had some special sauce that made it better at handling them, which I guess is why Bayo 1 ran better in some cases on 360 compared to Wii U (and maybe even Switch). Is there anything in Drake that could help with that other than just moar bandwidth?
- Pronouns
- He/him
Explain
LiC
Member
No.
Good catch. It's entirely possible that it was just a documentation error. GA10B and GA10F both support Vulkan, and share a lot of code and development resources, so it's believable that somebody putting together documentation would mistakenly cite support for both of them even if the product itself was only meant for GA10B.
- Pronouns
- He/Him
This is the type of joke I'd make and I really appreciate that you made it
Like he said, "somehow."
So you’re saying that drifting will now be considered a feature?
- Pronouns
- He/him
Explain better
- Pronouns
- He/Him
But then it wouldn't compare to the story of the Star Wars film he's referencing
Serif
𝕽𝖊𝖓𝖊𝖌𝖆𝖉𝖊 𝕬𝖓𝖌𝖊𝖑
Secrets only the Sith knew
ertaboy356b
Resident Troll
Didn't watch the whole video, how does it handle dust accummulation?
LiC
Member
Just for the sake of completeness: The original 6.0.3 release of Drive OS (June 2022) does not have the page about Vulkan SC samples. The 6.0.4 release (August 2022) adds it, and that's the one place where GA10F appears. The 6.0.5 release (November 2022) revises that page to remove the mention of GA10F, as @Serif found. Other pages that mention supported GPUs by name, such as this one, have never referenced GA10F in any release, and T239 has never been referenced on pages that reference T234.
If one believes that this removed reference is a sign that GA10F is meant to be included in Drive OS (rather than just an error in the documentation), it should be clear that at the very least it's not supported right now. IMO that makes it implausible that the reference coincided with the release of any GA10F/T239 product to Drive customers, even if we entertain the notion that that's on the table at all.
The timing probably isn't a coincidence. The release of 6.0.4 happened just after the release of Jetson Linux 35.1, which is the L4T source that revealed all the T239 commits we've been discussing since then. And I think it's a very natural explanation that, when creating a new page on Drive OS GPU support, someone found internal information on GA10B and GA10F in their recent release, and wrote them both into the documentation. But to me that just looks like a mistake which was corrected in the following version.
We also don't know how T239 development is continuing in the Linux source. As @oldpuck corrected me on the other day, those commits don't become public when they happen, only when Nvidia pulls it all together and does a public source release (the most recent being 35.1). Based on what was in that last release, I would say they aren't done, since there are many classes that only exist for T23x/T234 with no version to distinguish T239, which are probably still in the mentioned private repo, and the overlay file still lists it as something that is being separated out from the main source. That also works against the idea that Linux development wrapped up in August and they started shipping a product afterwards.
If one believes that this removed reference is a sign that GA10F is meant to be included in Drive OS (rather than just an error in the documentation), it should be clear that at the very least it's not supported right now. IMO that makes it implausible that the reference coincided with the release of any GA10F/T239 product to Drive customers, even if we entertain the notion that that's on the table at all.
The timing probably isn't a coincidence. The release of 6.0.4 happened just after the release of Jetson Linux 35.1, which is the L4T source that revealed all the T239 commits we've been discussing since then. And I think it's a very natural explanation that, when creating a new page on Drive OS GPU support, someone found internal information on GA10B and GA10F in their recent release, and wrote them both into the documentation. But to me that just looks like a mistake which was corrected in the following version.
We also don't know how T239 development is continuing in the Linux source. As @oldpuck corrected me on the other day, those commits don't become public when they happen, only when Nvidia pulls it all together and does a public source release (the most recent being 35.1). Based on what was in that last release, I would say they aren't done, since there are many classes that only exist for T23x/T234 with no version to distinguish T239, which are probably still in the mentioned private repo, and the overlay file still lists it as something that is being separated out from the main source. That also works against the idea that Linux development wrapped up in August and they started shipping a product afterwards.
Magic-Man
Perseus Jackson
- Pronouns
- He/Him
Drake car tires be like
Raccoon
Fox Brigade
- Pronouns
- He/Him
somehow, bloomberg was wrong
somehow, natedrake's switch was shelved
somehow, it ended up in a fucking car (except not really probably)
- Pronouns
- He/him
- Pronouns
- He/Him
This was actually me earlier today when looking for an address I wrote on a sticky note a few months ago
- Pronouns
- He/Him
Honestly if it turned out after all this Drake is just for a car I’d probably die of laughter, the most underwhelming ending possible. Though it would also be kind of exciting since it means we wouldn’t have much idea what the Switch 2 would be specs wise.
Look over there
Bob-omb
The claim is that there's enough pressure such that you can cover the intake with IP68 rated water/dustproof material and it'll still work fine.
Here's an accompanying PC World article: https://www.pcworld.com/article/1388332/new-airjet-chips-can-double-a-laptops-performance.html
Coincidentally, recently I've been thinking of alternative options for Sony/MS/some other company to pivot to for dedicated gaming devices in a more energy-consumption-aware future.
The Switch form factor checks off: Handheld, Tabletop, and hooked-up-to-a-TV/display.
Big ol set top boxes cover just the TV/display usage.
To extend from the PC-in-a-smaller-box description... how about the Laptop form factor? You can still plug it into a display, but then you pick up the option of Tabletop usage. And you have room to go up to... 28-35 watts. That should be enough carve out an interesting segment. While I'm thinking about this, I am eyeing Intel...
(admittedly, I kinda wanted to hold off on mentioning this until after the 26th, when presumably Intel says something about Lunar Lake).
Serif
𝕽𝖊𝖓𝖊𝖌𝖆𝖉𝖊 𝕬𝖓𝖌𝖊𝖑
Even if T239 were used in a car which I doubt due to the information above, it's still most likely the Switch 2 chipset due to the NVN2 support, DLSS + RT, and file decompression engine intended for games.
- Pronouns
- He/Him
I was just about to say, this would be the most hilarious way to reconcile "for all these reasons, Drake must be in production or close to it" with "the 2022-23 Switch was shelved."
It would be absolutely wild (and for multiple reasons I don't expect it to be the case) but it would answer a couple questions at least.
Smash Kirby
Bob-omb
Erathix
Famiro
- Pronouns
- He/Him/They/Them
Erathix
Famiro
- Pronouns
- He/Him/They/Them
LiC
Member
T239 being used in automotive would be incredibly bizarre, and I'm pretty sure that's never going to happen, but it wouldn't change my expectations for Nintendo hardware. "The deal with Nintendo fell through, so Nvidia decided to repurpose their gaming SoC design for cars instead" is even less believable, and @oldpuck wasn't suggesting that. The development timeline also doesn't work. The Linux commits go back to April 2021 at least, while NVN2 was being worked on as of February 2022.* Nintendo's hardware couldn't have been cancelled any earlier than that, so does that mean it got cancelled in or after March, and Nvidia pivoted to making it part of Drive inside 5 months? Or does the 2021 Linux history mean they always planned to do it, in which case it's not evidence of anything with regards to Nintendo hardware?
* And when I say this, I really mean it:
Hidden content is only available for registered users. Sharing it outside of Famiboards is subject to moderation.
D
Deleted member 887
Guest
Very good catch
For the record, I was speculating wildly, and I agree, Drake Drive doesn't make a lot of sense. Having seen Serif's find, I think it's obviously ruled out
However, this explanation doesn't quite fly to me, which is why I brought it up in the first place. The interesting thing is not the documentation reference, but what the documentation says:
This isn't an issue of a technical writer including a ga10f reference. At some point, someone wrote a Vulkan SC backend for GA10F. Vulkan SC is the "Safety Critical" Vulkan variant for car (and plane and medical) applications. Much like Linux support for the FDE, Nvidia seems invested in making sure the entire Orin software stuck runs on Drake.
Hidden content is only available for registered users. Sharing it outside of Famiboards is subject to moderation.
D
Deleted member 887
Guest
Yeah, the only thing I was suggesting that this might change is creating a buffer in the timeline, just as his been suggested with Nvidia making a new shield.
The other possibility that I considered is "Drive Sim" - Nvidia lets Drive developers simulate cars in a cloud based environment. T239 would have some specific advantages over Orin Nano in that very narrow environment (server density specifically) but I think it's such a narrow use case I have difficulty imagining it.
LiC
Member
Yes.
Such as?
You're assuming that the pipeline cache data for GA10F must have actually existed for the author to write this. However, I see it as "the sample already contains the pipeline cache generated for <insert list of supported GPUs here>." And when they were filling in the list of supported GPUs, they included one that shouldn't have been included.
I guess if anyone's got friends who work with Nvidia's automotive software, you could ask them whether the 6.0.4 SDK's pipeline_cache.hpp file has three sets of data and includes GA10F, or just two.
Serif
𝕽𝖊𝖓𝖊𝖌𝖆𝖉𝖊 𝕬𝖓𝖌𝖊𝖑
I agree, I'm unsure if you're responding to a point I made or "Yes,and"-ing, I'm beginning to lose track of this conversation with the multiple posts about cars.
Hidden content is only available for registered users. Sharing it outside of Famiboards is subject to moderation.
Concernt
Optimism is non-negotiable
- Pronouns
- She/Her
None of that is true, thankfully.
LiC
Member
It was a "yes and."I agree, I'm unsure if you're responding to a point I made or "Yes,and"-ing, I'm beginning to lose track of this conversation with the multiple posts about cars.
Corim
Cappy
- Pronouns
- he/him
bloopland33
Like Like
Well, hopefully the podcast comes this (Wednesday) morning and we can mercifully be freed from the worst of this uncertainty.
I wouldn't expect that. I think it's more about 2023 games for Switch. He mentioned that there won't be any speculations in that video.
Z0m3le
Bob-omb
Linux development never wraps up until the product is no longer supported. It has physical traits in the kernal code that would suggest that silicon is finished, not software development, drivers for that is only possible after final silicon is complete anyways, and we've seen signs of that. Engineering samples showed up as early as April 2022 via the Kernal updates. Rich from digital foundry also believes the Linux commits are indications that silicon has finalized and begun initial production from that September 5th commits. That is pretty much the last update we have on the hardware, just because final silicon has completed, doesn't mean they are in full model production of Switch "2" right now, but it strongly suggests that physical retail T239 chips exist and could begin production at any time.
I was looking through history of SoCs and how long they waited to come to market after being finalized, it's usually within 12 months, there was one exception I found though, at least suggested, and that is Wii U's MCM, it seems it was ready for production in summer 2011, remembering back to early Wii U information, there was official talks from Nintendo about a delay to Wii U because they couldn't get the wireless Gamepad to work as seemlessly as they wanted, and it took nearly a year to get it working, being a component that stalled the entire system. This does make some sense, as Wii U hardware was "seen" at E3 2011, the gamepads were wired to the console, but if that was real hardware, it would mean retail MCM chips were inside of them. This is the only modern console that I could find where the chip was custom and was ready over a year before product launch, but information is not straight forward and engineering samples are often done 6 months or more before final silicon.
I think people need to be patient with Drake hardware announcement, there is really only 2 legit windows for a new more powerful Switch announcement using Drake to be made, one is in 3 weeks during the earnings report for Q3 on Feb 7th... If nothing is announced by Feb 10th, Zelda launch with Drake just isn't happening and people will need to move on. The next window is around E3 for a holiday release, this could drag into July, but that is the last time we would hear about it this year. Switch 1 using TX1 needs to remain a strong seller, and cutting holiday sales by announcing Switch "2" this year for a next year launch, is just going to frustrate potential customers and stall Switch sales through the holidays. It would make more sense to wait until this time frame next year for an announcement, even with a Spring launch for hardware, though there is no need to rush a launch next spring if they miss Zelda, it only makes sense to launch in spring because of Zelda's release, a Spring release limits initial hardware sales to about 3M, this is simply the amount of space the retail chain has for a hardware launch in the Spring, a holiday launch benefits greatly from end caps and pallets placed on retail floors, as well as heavy foot traffic, keeping inventory moving, allowing for ~5M sales in the holidays, this 2M boost out of the gate is actually why Switch took so long to catch PS4 sales when launch aligned.
So keep an eye out for early February announcement, and then move on to E3 if it doesn't happen, after that, you can stop worrying about a hardware launch this year, as the window has closed for 2023 at that point.
Last edited:
Lant_War
Understand, Understand
- Pronouns
- She/Her
Speculation on Nintendo hardware always finds a way to leave people disappointed even when the information seems basically rock-solid
SiG
Chain Chomp
Except it was. People were so sure back in the day it would be another AMD powered device.
In fact, people were so sure it was that oblong looking thing that was all screen and two analogue sticks based on patents.
I don't get that either. It's a speculation thread. Anything stated here shouldn't be taken as being set on stone.
TheGreatMightyPoo
Like Like
I’ve changed my stance on the possible price of Drake.
I’m thinking 45,000$ now minimum.
I’m thinking 45,000$ now minimum.
Z0m3le
Bob-omb
The handheld and console teams merged at the new building that opened on Jan 2013, just a couple months after Wii U's launch. We knew they would be merging their platforms into an account platform in 2015 when Iwata talked about brothers in a family of systems and how they should emulate iOS/Android's software development behavior... Semi-Accurate had the leak about Tegra powering Nintendo NX in Spring 2016, and Eurogamer had the devkit leak in July 2016, then the October Reddit leaker who laid out everything before the Switch reveal on October 20th 2016... but yeah there were still people sure about AMD powering the next console the entire time, someone named SMD64 still believes that happened and the entire world is living a lie, but you must remember to think for yourself, don't believe the Switch lies.
I have still refused to watch that movie, mostly because I read that quote online. And Im a SW fan and I liked ep 8.
Last edited:
