• 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.
  • Furukawa Speaks! We discuss the announcement of the Nintendo Switch Successor and our June Direct Predictions on the new episode of the Famiboards Discussion Club! Check it out here!

StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)

Historically they rarely launched consoles with 3d Mario so I don’t expect it on launch even though it will be teased at the presentation

I expect Metroid and some surprises on launch
I think that's partially true. If we consider "launches with Mario", we have NES (US launch), SNES, N64, GB, DS, and Wii U. GCN had Luigi's Mansion, an internally developed flagship title in the Mario universe. They also can't launch with a Zelda (Wii, Nintendo Switch). The vast majority of Nintendo consoles have launched with either a Mario or a Zelda of some description.

While I think it would be a wiser market play to have a light launch and bring in Mario for the holiday season, I have to admit it looks like 3D Mario at launch is rather likely.
 
I honestly think that Raytacing actually looks best with simple models, colors and textures. Minecraft with Raytracing is literally the most graphically pleasing game that I have ever seen in my life. Same with Quake too.

I think ray-tracing can look really good in simple environments, but the implementation should depend on what the art style calls for. Animal Crossing makes liberal use of diffuse (almost lambertian) materials for the environment and even many object models, for instance, so while boosting the specularity on all of the materials in the environment just to showcase ray-traced specular reflections would look cool, it could also deviate from the style a bit. On the other hand, ray-traced diffuse interreflection would be well-suited to the style.

Note the key art below:

animal-crossing-new-horizons-key-artwork-january-2020-9000wide.jpg



It's vibrant, with simple, matte-like textures. The grass, trees, and characters, etc. all have a mostly diffuse, cozy look (with some specular sprinkles) that really contributes the game's aesthetic. Even the more reflective objects are primarily reflecting highlights from clear coats on top of diffusely colored surfaces, while their Fresnel reflections are more glossy than real life (likely artist-driven).

Keep in mind, the key art has the benefit of not needing to be rendered in real-time. If the artists wanted to feature specular reflections more prominently, they could have (and financial budget wouldn't have likely impeded this either). They didn't, likely because the simplicity is intentional and crucial to the style's aesthetic. Adding prominent specular reflections (ray-traced or otherwise) would introduce a level of visual complexity that wouldn't otherwise exist in this style.

That being said, as I mentioned before, there are other forms of ray-tracing (like global illumination, shadows) that would be well-suited to the style, but the point is that it needs to be used in a way that respects the style. The goal should be to enhance without overwhelming, maintaining the simple, charming look that makes Animal Crossing so appealing to casual gamers.
 
Although this is not explicitly about Nintendo, I found this rather intriguing.

(Translation below done by DeepL Translate.)
It's almost certainly PS4/5 games on ARM, I don't know what else they would need a "recent major architectures" JIT engineer for. The only question is if it's their own platform or someone else's... the wording is suggestive (mentioning "the new architecture system"), and it being a new PS device is way more likely than it being PS games on Switch.

It will be interesting to see if they manage to get adequate performance out of this. Using JITs to make modern games (ie, not emulating an old console) crossplatform for an official release is pretty new territory.
 
Last edited:
I think that's partially true. If we consider "launches with Mario", we have NES (US launch), SNES, N64, GB, DS, and Wii U. GCN had Luigi's Mansion, an internally developed flagship title in the Mario universe. They also can't launch with a Zelda (Wii, Nintendo Switch). The vast majority of Nintendo consoles have launched with either a Mario or a Zelda of some description.

While I think it would be a wiser market play to have a light launch and bring in Mario for the holiday season, I have to admit it looks like 3D Mario at launch is rather likely.
Add GBA. While it wasn't a new game, Super Mario Advance was still a Mario launch title.
 
I think ray-tracing can look really good in simple environments, but the implementation should depend on what the art style calls for. Animal Crossing makes liberal use of diffuse (almost lambertian) materials for the environment and even many object models, for instance, so while boosting the specularity on all of the materials in the environment just to showcase ray-traced specular reflections would look cool, it could also deviate from the style a bit. On the other hand, ray-traced diffuse interreflection would be well-suited to the style.

Note the key art below:

animal-crossing-new-horizons-key-artwork-january-2020-9000wide.jpg



It's vibrant, with simple, matte-like textures. The grass, trees, and characters, etc. all have a mostly diffuse, cozy look (with some specular sprinkles) that really contributes the game's aesthetic. Even the more reflective objects are primarily reflecting highlights from clear coats on top of diffusely colored surfaces, while their Fresnel reflections are more glossy than real life (likely artist-driven).

Keep in mind, the key art has the benefit of not needing to be rendered in real-time. If the artists wanted to feature specular reflections more prominently, they could have (and financial budget wouldn't have likely impeded this either). They didn't, likely because the simplicity is intentional and crucial to the style's aesthetic. Adding prominent specular reflections (ray-traced or otherwise) would introduce a level of visual complexity that wouldn't otherwise exist in this style.

That being said, as I mentioned before, there are other forms of ray-tracing (like global illumination, shadows) that would be well-suited to the style, but the point is that it needs to be used in a way that respects the style. The goal should be to enhance without overwhelming, maintaining the simple, charming look that makes Animal Crossing so appealing to casual gamers.
Ironically: I find this to look rather flat.
I would compare it to the high key style of sitcoms. (Obviously just in concept)

And if this would be the style of a new AC... I would actually think that it's a regression.
 
What are the odds of Switch 2 launching with Mario Kart instead of 3d Mario?
Isn‘t Mario Kart the best selling Switch game?

Whether or not it's at launch or not, players will definitely not have to wait long for the next Mario Kart. It's literally the most popular Mario thing Nintendo has. More than the 2D games, 3D games...heck, Mario Kart is literally the most successful and popular Switch game by a HUGE margin. They'll definitely make sure it comes out in the first year of Switch 2.
 
Ironically: I find this to look rather flat.
I would compare it to the high key style of sitcoms. (Obviously just in concept)

And if this would be the style of a new AC... I would actually think that it's a regression.

I think it's perfectly fair and valid for you to not like the art style. I don't even think it's necessarily wrong for Nintendo to change the art style, but if they do, I do think a new style would still adhere to certain visual accessibility principles so as not to alienate casual gamers.
 
I’m sure that these questions have been answered before, but this thread is so long that it’s not easy to keep track of all of it.

1) Assuming that the successor of the Switch will have a 1080p screen, what is the common theory on how Nintendo will handle BC with Switch games in portable mode?

2) At some point (maybe a few years ago..), we were estimating that the CPU of the successor could run close to the speeds of the original Switch (1.02GHz) which would still be at least 6x the performance of the underclocked TX1 and significantly better than the Jaguar CPUs from the previous generation. With the latest information and how unlikely it is for the successor to be on a 8nm node, however, it seems like it much better chance for the CPU to be clocked much higher than that. What is the current reasonable expectation for the CPU’s clockspeed?
 
I think it's perfectly fair and valid for you to not like the art style. I don't even think it's necessarily wrong for Nintendo to change the art style, but if they do, I do think a new style would still adhere to certain visual accessibility principles so as not to alienate casual gamers.
Oh for sure. (And good analysis of the key art! Always nice to read technical opinions from you)

What I do think is more universal then my opinion: key art has a different goal then general in game graphics, so it should not inherently be seen as an ideal variant or the goal how a game should look.
 
I think ray-tracing can look really good in simple environments, but the implementation should depend on what the art style calls for. Animal Crossing makes liberal use of diffuse (almost lambertian) materials for the environment and even many object models, for instance, so while boosting the specularity on all of the materials in the environment just to showcase ray-traced specular reflections would look cool, it could also deviate from the style a bit. On the other hand, ray-traced diffuse interreflection would be well-suited to the style.

Note the key art below:

animal-crossing-new-horizons-key-artwork-january-2020-9000wide.jpg



It's vibrant, with simple, matte-like textures. The grass, trees, and characters, etc. all have a mostly diffuse, cozy look (with some specular sprinkles) that really contributes the game's aesthetic. Even the more reflective objects are primarily reflecting highlights from clear coats on top of diffusely colored surfaces, while their Fresnel reflections are more glossy than real life (likely artist-driven).

Keep in mind, the key art has the benefit of not needing to be rendered in real-time. If the artists wanted to feature specular reflections more prominently, they could have (and financial budget wouldn't have likely impeded this either). They didn't, likely because the simplicity is intentional and crucial to the style's aesthetic. Adding prominent specular reflections (ray-traced or otherwise) would introduce a level of visual complexity that wouldn't otherwise exist in this style.

That being said, as I mentioned before, there are other forms of ray-tracing (like global illumination, shadows) that would be well-suited to the style, but the point is that it needs to be used in a way that respects the style. The goal should be to enhance without overwhelming, maintaining the simple, charming look that makes Animal Crossing so appealing to casual gamers.
This post reminds me of how much I dislike the key art for New Horizons. The proportions look way off, the random quasi-realistic textures look out of place, the composition is a mess, it looks saturated as hell, and I particularly dislike the overly realistic clouds and random N64-looking hill in the background on the right. Your technical analysis is great, but man the key art just bugs me.
 
2) At some point (maybe a few years ago..), we were estimating that the CPU of the successor could run close to the speeds of the original Switch (1.02GHz) which would still be at least 6x the performance of the underclocked TX1 and significantly better than the Jaguar CPUs from the previous generation. With the latest information and how unlikely it is for the successor to be on a 8nm node, however, it seems like it much better chance for the CPU to be clocked much higher than that. What is the current reasonable expectation for the CPU’s clockspeed?

I found the following post fairly useful with that:

I base myself on this slide from ARM:
images

1x A78 on 5nm @3.0GHz = 1W

1x A77 on 7nm @2.6GHz = 1W
thus, 1x A77 on 7nm @2.3GHz < 1W
which means 1x A78 on 5nm @2.1GHz < 0.5W.

Power curve tend to be roughly quadratic, so 1.5GHz would in theory be around 0.25W.

I expect the OS core be clocked low and them to give 2~3W for the CPU, which should be something like 1.5~2.1GHz according to my napkin math.

Thraktor also calculated the wattage of several clocks for Orin (8nm) and guesstimated Drake to be ~1.6GHz for a similar wattage as OG Switch CPU.

There are many caveats, so take these numbers with a good bit of salt.
 
This post reminds me of how much I dislike the key art for New Horizons. The proportions look way off, the random quasi-realistic textures look out of place, the composition is a mess, it looks saturated as hell, and I particularly dislike the overly realistic clouds and random N64-looking hill in the background on the right. Your technical analysis is great, but man the key art just bugs me.
the saturated look in combination with the composition ind the "just throw everything in there so that we can show multiple aspects off" ... yeah, i really don't like it. Random screenshots look way better imho.
 
Whether or not it's at launch or not, players will definitely not have to wait long for the next Mario Kart. It's literally the most popular Mario thing Nintendo has. More than the 2D games, 3D games...heck, Mario Kart is literally the most successful and popular Switch game by a HUGE margin. They'll definitely make sure it comes out in the first year of Switch 2.
You know, that is a pretty good point. Racing games were a big technical showcase for Sony and Microsoft this generation, and certainly for all 3 in the past. And it would give Nintendo a chance to show off what Switch 2 can do with reflections and ray tracing. My only concern is...the MK team has been working on Mario Kart Tour and the Booster Pass for Mario Kart 8 Deluxe, so who are they getting to work on Mario Kart 9? I was always disappointed we never got a true 9 with different mechanics, because I played MK8 to death on Wii U and the Switch version is something I only pull out when friends are over.
I’m sure that these questions have been answered before, but this thread is so long that it’s not easy to keep track of all of it.

1) Assuming that the successor of the Switch will have a 1080p screen, what is the common theory on how Nintendo will handle BC with Switch games in portable mode?

2) At some point (maybe a few years ago..), we were estimating that the CPU of the successor could run close to the speeds of the original Switch (1.02GHz) which would still be at least 6x the performance of the underclocked TX1 and significantly better than the Jaguar CPUs from the previous generation. With the latest information and how unlikely it is for the successor to be on a 8nm node, however, it seems like it much better chance for the CPU to be clocked much higher than that. What is the current reasonable expectation for the CPU’s clockspeed?
If Switch 2 has a 1080p screen, in BC mode, Switch 1 games would run at their original 720p resolution, unless the developers went back and patched them/ We know they're going to have to make some sort of software solution just to run backwards compatibility, so I don't see older games suddenly being boosted up to docked-level visuals.

I did a quick search and found this YouTube video:
Where Doctre81 claims to have found a resume on LinkedIn detailing an engineer's work on an Nvidia project for 5nm (Nvidia 4N is a 5nm node) and they listed clock rate as 2.653GHz. I've never heard of Doctre81 before, and don't know if he's a legit source or not, but the resume document shared is pretty technical and looks like a real document.
 
I used that power calculator to determine that the T239 gpu for sec8n should be 6.3W at 408MHz, which is quite absurd.
 
Last edited:
Oh for sure. (And good analysis of the key art! Always nice to read technical opinions from you)

What I do think is more universal then my opinion: key art has a different goal then general in game graphics, so it should not inherently be seen as an ideal variant or the goal how a game should look.

Thanks for the feedback. I do agree that in-game graphics don't necessarily need to look like key art, but the reason I used the key art as an example is because it happens to share of lot of design principles as the in-game art and isn't constrained by real-time rendering, which helped to illustrate the point about intentionally avoiding the use of high specularity even when the hardware is plenty capable.
 
Or just more development of things like this
They specifically use new architecture (新しいアーキテクチャ), so obviously not x86 since Sony has been using it since PS4.
■職務内容/ Job Description(Roles & Responsibilities)
当社エンタテインメントゲーム機器向け開発されたプログラムを、異なるアーキテクチャをもつゲーム機器の上で実行可能にするためのシステムの開発を行います。具体的には PlayStation 向けに開発されている多数のゲームタイトルを、新しいアーキテクチャのシステム上で動作させることを実現させます。
このシステムは、問題発生時の不具合解析を容易にするための機構を備えることが求められています。
(Sorry for the OOT, let's go back discussing Nintendo console again)
 
They specifically use new architecture (新しいアーキテクチャ), so obviously not x86 since Sony has been using it since PS4.

(Sorry for the OOT, let's go back discussing Nintendo console again)
unless they've said something elsewhere specifically talking about meaning CPUs, its entirely possible that theyre looking for GPU instruction translation like we are thinking nintendo needs for switch games on switch2
 
Thanks for the feedback. I do agree that in-game graphics don't necessarily need to look like key art, but the reason I used the key art as an example is because it happens to share of lot of design principles as the in-game art and isn't constrained by real-time rendering, which helped to illustrate the point about intentionally avoiding the use of high specularity even when the hardware is plenty capable.
Yeah, that's a good way to look at it.

To much use of high specularity can give of a plastic flair (or prerendered).

Maybe an equivalent would be 48/60fps movies without motion blur. To sharp.
 
You know, that is a pretty good point. Racing games were a big technical showcase for Sony and Microsoft this generation, and certainly for all 3 in the past. And it would give Nintendo a chance to show off what Switch 2 can do with reflections and ray tracing. My only concern is...the MK team has been working on Mario Kart Tour and the Booster Pass for Mario Kart 8 Deluxe, so who are they getting to work on Mario Kart 9? I was always disappointed we never got a true 9 with different mechanics, because I played MK8 to death on Wii U and the Switch version is something I only pull out when friends are over.

If Switch 2 has a 1080p screen, in BC mode, Switch 1 games would run at their original 720p resolution, unless the developers went back and patched them/ We know they're going to have to make some sort of software solution just to run backwards compatibility, so I don't see older games suddenly being boosted up to docked-level visuals.

I did a quick search and found this YouTube video:
Where Doctre81 claims to have found a resume on LinkedIn detailing an engineer's work on an Nvidia project for 5nm (Nvidia 4N is a 5nm node) and they listed clock rate as 2.653GHz. I've never heard of Doctre81 before, and don't know if he's a legit source or not, but the resume document shared is pretty technical and looks like a real document.

Are we even allowed to link to Doctre anymore? He tried joining the forum and got banned for some pretty terrible takes and acquaintances.
 
I’m sure that these questions have been answered before, but this thread is so long that it’s not easy to keep track of all of it.

1) Assuming that the successor of the Switch will have a 1080p screen, what is the common theory on how Nintendo will handle BC with Switch games in portable mode?
One theory is that they will run the games in their docked graphic profile regardless if docked or not. Games that use touch screen have been found to have their touchscreen controls enabled in docked mode by people with emulators.
 
You know, that is a pretty good point. Racing games were a big technical showcase for Sony and Microsoft this generation, and certainly for all 3 in the past. And it would give Nintendo a chance to show off what Switch 2 can do with reflections and ray tracing. My only concern is...the MK team has been working on Mario Kart Tour and the Booster Pass for Mario Kart 8 Deluxe, so who are they getting to work on Mario Kart 9? I was always disappointed we never got a true 9 with different mechanics, because I played MK8 to death on Wii U and the Switch version is something I only pull out when friends are over.

If Switch 2 has a 1080p screen, in BC mode, Switch 1 games would run at their original 720p resolution, unless the developers went back and patched them/ We know they're going to have to make some sort of software solution just to run backwards compatibility, so I don't see older games suddenly being boosted up to docked-level visuals.

I did a quick search and found this YouTube video:
Where Doctre81 claims to have found a resume on LinkedIn detailing an engineer's work on an Nvidia project for 5nm (Nvidia 4N is a 5nm node) and they listed clock rate as 2.653GHz. I've never heard of Doctre81 before, and don't know if he's a legit source or not, but the resume document shared is pretty technical and looks like a real document.


Looking at the comments section for that video, a very interesting comment my attention. It's a comment by the user "syferz" who says:

The chip was leaked in a hack 18 months ago, it's "GA10F" (Geforce Ampere 10F) Orin is GA10B, it does have some Ada features, and it's OFA engine (Frame Generation stuff) is the same as Orin and different from both Ampere and Ada, it has dense pixel flow, so it's possible to use frame generation with Switch 2, but we don't know the speed of the OFA engine, considering Orin is for cars, I'd expect it to be much faster than Ampere's normal OFA. We will see.

The bit that gets my attention is about the OFA for Frame Generation....this is surprising to me because I assumed Frame Generation was completely off the table because, to my knowledge, Frame Generation is only good for taking already very high fps (e.g. 60fps) and taking them to silly fast (e.g. 120fps). I could be completely wrong on that, so please correct me if I am.

For the Switch 2 which is likely to be around 30/60fps for each title, it seems like it would just be a waste of real-estate. But according to this user, it's in there?

Is it possible for Switch 2 to have Frame Generation after all? perhaps a completely bespoke kind of Frame Generation developed by Nvidia specifically for the Switch 2 that allows for it to work effectively at lower frame-rates? so it's not just DLSS boosting the fps, but also the FG?

Curious what people's thoughts are.
 
I could only see them add it if they think they can get some of the CoD fans to buy into Switch 2, a type of gamer that usually is not that into Nintendo consoles.

I won't go as far as to say they shouldn't touch the graphics, but just pointing out that it cannot faithfully adhere to the art style by chasing as much detail as the hardware will allow. Even implementing ray tracing is something the AC developers might not be too keen to do. I do think something like using polygons instead of normal maps to represent detail is something that could be a genuine improvement without compromising the art style.

Something I think a lot of core gamers don't realize about Nintendo is that visual accessibility is something that they consider in their art design. Psychologically, simple, vibrant colors and basic detail tend to be more inviting to casual gamers. It's the same logic used to create content for young children. AC is made for a very broad audience, and the simple art style is one of the ways they have been able to achieve that. Having specular reflections everywhere and high frequency detail everywhere might feel too off-putting for many casual gamers, as these people already avoid games with complex graphics.
First they need to upper the frame rate to 60 FPS, instead of NH stuttering 30 FPS.

In-House the lightning is already way better than the lightning outdoors. I think outdoors has only one light source (sun or moon) and all other light sources only have a colored spot around them. Indoors lights have actual influence on the surroundings but it gets weird when you put lights to close or more than three lights sources in a room, the lightning gets sometimes very unrealistic/uncanny and ugly.

So implementing "real light" with ray tracing would be a big step up.
 
Looking at the comments section for that video, a very interesting comment my attention. It's a comment by the user "syferz" who says:

The chip was leaked in a hack 18 months ago, it's "GA10F" (Geforce Ampere 10F) Orin is GA10B, it does have some Ada features, and it's OFA engine (Frame Generation stuff) is the same as Orin and different from both Ampere and Ada, it has dense pixel flow, so it's possible to use frame generation with Switch 2, but we don't know the speed of the OFA engine, considering Orin is for cars, I'd expect it to be much faster than Ampere's normal OFA. We will see.

The bit that gets my attention is about the OFA for Frame Generation....this is surprising to me because I assumed Frame Generation was completely off the table because, to my knowledge, Frame Generation is only good for taking already very high fps (e.g. 60fps) and taking them to silly fast (e.g. 120fps). I could be completely wrong on that, so please correct me if I am.

For the Switch 2 which is likely to be around 30/60fps for each title, it seems like it would just be a waste of real-estate. But according to this user, it's in there?

Is it possible for Switch 2 to have Frame Generation after all? perhaps a completely bespoke kind of Frame Generation developed by Nvidia specifically for the Switch 2 that allows for it to work effectively at lower frame-rates? so it's not just DLSS boosting the fps, but also the FG?

Curious what people's thoughts are.
Frame Generation is only good for taking already very high fps (e.g. 60fps) and taking them to silly fast (e.g. 120fps).
That's an AMD thing, their less sophisticated frame generation tech starts to completely fall apart below 60 fps . Nvidia's ML-based frame generation is still usable at a minimum of 30 fps, albeit with some artifacting, but I'd argue AMD's FSR frame generation looks quite similar with a 60fps base framerate. Artifacting wildly depends on the game and it's implementation, however. Third person games with relatively slower camera movement work best, while FPS games suffer the most. In other words, fast motion is bad.

My friend has a RTX 4080, but even so, some areas in Dragons Dogma 2 (especially the city) can cause his framerate to drop down into the 40s. His R7 7700 is the bottleneck here. So he modded DLSS framegen into the game and based on what I've seen it's a night and day difference. Artifacts, if any, are almost imperceptible while framerate is comfortably in the 70s at the minimum. The 1440p input resolution might be helping with this, but I think it's still viable at lower resolutions.

If any games are to use frame gen on the Switch 2, I'd suggest they run at a base framerate of 40-45 and then go up from there. Starting at 30 fps means you'll still fall short of 60, which defeats the purpose. There will be less artifacting too. Since it's a console, devs can fine-tune and tweak parameters and reduce artifacts even further, and if Nvidia's Reflex tech makes it into the Switch 2 then they can leverage it to reduce the latency as well. There's potentially a lot of amazing tech at play here.
 
Last edited:
It's a big mystery why they would keep the Orin OFA, if its indeed is larger than the regular Ampere one. Since the use cases for it are at best limited.
If it's a fully custom chip they could've removed it entirely, I think. That tells me they have plans for it.
 
I'd love to see bigger maps for the next Animal Crossing, better terraforming capabilities... let some of the old limitations be a thing of the past, within reason of course.

And maybe stop having breakable tools... once and for all. Take a look at Fae Farm for all its QoL features.
 
Looking at the comments section for that video, a very interesting comment my attention. It's a comment by the user "syferz" who says:

The chip was leaked in a hack 18 months ago, it's "GA10F" (Geforce Ampere 10F) Orin is GA10B, it does have some Ada features, and it's OFA engine (Frame Generation stuff) is the same as Orin and different from both Ampere and Ada, it has dense pixel flow, so it's possible to use frame generation with Switch 2, but we don't know the speed of the OFA engine, considering Orin is for cars, I'd expect it to be much faster than Ampere's normal OFA. We will see.

The bit that gets my attention is about the OFA for Frame Generation....this is surprising to me because I assumed Frame Generation was completely off the table because, to my knowledge, Frame Generation is only good for taking already very high fps (e.g. 60fps) and taking them to silly fast (e.g. 120fps). I could be completely wrong on that, so please correct me if I am.

For the Switch 2 which is likely to be around 30/60fps for each title, it seems like it would just be a waste of real-estate. But according to this user, it's in there?

Is it possible for Switch 2 to have Frame Generation after all? perhaps a completely bespoke kind of Frame Generation developed by Nvidia specifically for the Switch 2 that allows for it to work effectively at lower frame-rates? so it's not just DLSS boosting the fps, but also the FG?

Curious what people's thoughts are.
Ask syferz yourself. @Z0m3le :p

FG is possible, but I don't expect it to be used at all. The current parlance is that you need a high and stable enough frame rate to keep input latency in check. And FG isn't free already
 
There’s mentions of FG in nvn2? Making sure I interpreted correctly
No there isn't, but the version of nvn2 we have access to is 2+ years old.

What I meant is that even if FG is technically possible, I don't think they would bother with an nvn2 implementation unless they felt it was worthwhile actually using.
 
but the resume document shared is pretty technical and looks like a real document.

I was able to find the resume in question on LinkedIn, among other LinkedIn profiles also found by folks here.

The resume doc file is indeed real and still there when I checked few days ago. We just cannot explicitly tie 5nm project to T239, that’s the problem.
 
No there isn't, but the version of nvn2 we have access to is 2+ years old.

What I meant is that even if FG is technically possible, I don't think they would bother with an nvn2 implementation unless they felt it was worthwhile actually using.

Ah gotcha. So the usual question remains, why is there OFA in T239?

This costs power draw too just leaving OFA in there even if it’s unused (if I understood oldpuck earlier, there’s power costs for running electricity thru the chip, including unused silicon unless removed by laser)

If OFA somehow is truly unused I would say that would make 4N more likely too (as if 4N wasn’t already likely). Based on playing with Jetson power estimator tool there was still optimization needed to bring power draw down to reasonable level because the power draw seems damn high on SEC8N
 
Historically they rarely launched consoles with 3d Mario so I don’t expect it on launch even though it will be teased at the presentation

I expect Metroid and some surprises on launch
Maybe Mario Kart then

Historically when it wasn’t Mario it was Zelda, now it wouldn’t be either in this case.

When a console has neither Mario nor Zelda at launch, historically it fails very violently. The GameCube suffered a lot, and let’s not even talk about the Wii U.

Besides it was never a will of Nintendo but something suffered because they launched the console too early each time. We know that Super Mario sunshine should have had more development time but there was already an absolute necessity to release the game.

The so-called delay that has led so many people on the Internet to roll over these days is a very good thing precisely because it will allow to have big games at launch and not only rely on Metroid Prime and its 3 million sales. And I say that as a Metroid fan.
 
Maybe Mario Kart then

Historically when it wasn’t Mario it was Zelda, now it wouldn’t be either in this case.

When a console has neither Mario nor Zelda at launch, historically it fails very violently. The GameCube suffered a lot, and let’s not even talk about the Wii U.
I mean, the Wii U technically had Mario at launch...
 
Super NES launched with Mario (iirc). N64 launched with Mario.

I don't think there's a pattern here, but the situation "If we can launch with Mario, we launch with Mario", imo.
 
Technically yea, but considering it was the first HD Mario game it was underwhelming at best. It felt like a level pack for the Wii game. If it had been even close to Wonders presentation I could have seen it making an impact.
Yeah, 3D World probably would have made a much better launch title, too.
 
Ah gotcha. So the usual question remains, why is there OFA in T239?

This costs power draw too just leaving OFA in there even if it’s unused (if I understood oldpuck earlier, there’s power costs for running electricity thru the chip, including unused silicon unless removed by laser)

If OFA somehow is truly unused I would say that would make 4N more likely too (as if 4N wasn’t already likely). Based on playing with Jetson power estimator tool there was still optimization needed to bring power draw down to reasonable level because the power draw seems damn high on SEC8N
Because it's something embedded into the Ampere design. Nvidia has been shipping OFA ever since Turing.

It's probably such a small piece of silicon that it's irrelevant whether it's to be used or not. It would be more expensive to rework the Ampere internal organization when you can just deactivate it or leave it be.

Lovelace just increased OFA performance to meet DLSS FrameGen demands.

Also, I'm pretty sure the T239 OFA is the same as Desktop Ampere rather than the more powerful Orin or Ada ones. So Nvidia FrameGen is ruled out. But that's something @Dakhil, @LiC or @oldpuck are more knowledgeable than me.
 
The n64 was really delayed a few months for mario 64.
I would say that mario 3d world may have a little less appeal, but I still think it's good
but not the best like odyssey or galaxy.
edit:Turns out mario 3d world isn't launch with wiiu together. I made a mistake
 
Last edited:
0
Because it's something embedded into the Ampere design. Nvidia has been shipping OFA ever since Turing.

Lovelace just increased OFA performance to meet DLSS FrameGen demands.

Also, I'm pretty sure the T239 OFA is the same as Desktop Ampere rather than the more powerful Orin or Ada ones. So Nvidia FrameGen is ruled out. But that's something @Dakhil, @LiC or @oldpuck are more knowledgeable than me.

I believe it has the same OFA as Orin, although I don't think that makes it the same as Ada's.

However, the OFA doesn't really matter when it comes to DLSS Frame Generation. It provides one of the inputs to the DLSS FG process, but the actual generation of the intermediate frame is done by the tensor cores. Based on everything I've seen from DLSS FG performance on bigger GPUs, it's simply not viable on a GPU as small as T239's. It would take longer to generate a frame with DLSS FG than it would to render one normally, which makes it pretty useless on a device like this.
 
I mean, the Wii U technically had Mario at launch...
Yes, but he post I was answering to specifically talks about 3D Mario.

And it’s a game that is all the less relevant in my opinion to talk about since New Super Mario Bros U had no novelty effect and brings nothing that could justify the purchase of a console.
 
Please read this new, consolidated staff post before posting.

Furthermore, according to this follow-up post, all off-topic chat will be moderated.
Last edited by a moderator:


Back
Top Bottom