Citation needed. I don't recall anyone having an issue with the instruction set. The performance of the CPU certainly would have been an issue for some, but that's independent of the ISA (just as any issues developers have porting games to Switch will be independent of the ISA).
Nintendo Switch Dev Kit Stats Leaked? Cortex A57, 4GB RAM, 32GB Storage, Multi-Touch.
- Thread starter Robert at Zeboyd Games
- Start date
- Status
Citation needed. I don't recall anyone having an issue with the instruction set. The performance of the CPU certainly would have been an issue for some, but that's independent of the ISA (just as any issues developers have porting games to Switch will be independent of the ISA).
DeuceGamer
Member
I could see it getting a Call of Duty. Even Wii U got two COD games. There was also a fanbase for the games on Wii.
Obviously not a given and not something I would expect but I could see it.
Candescence
Member
High sales will drive ports even if technical challenges exist. The Wii U didn't get ports because it didn't sell for a variety of reasons, but the Wii sure as hell got blockbusters such as pretty much every single CoD since Modern Warfare (and even got a port of the original Modern Warfare that was a rather impressive technical feat for the studio that worked on it). If the Switch inherits the audiences of both the Wii U and the over 50 million-selling 3DS (in spite of its problems), that's a hefty potential install base that can't be ignored, and the Switch is sure as hell likely going to be closer to the Xbone than Wii U in terms of power than the power gap between the Wii and 360. There's also the lesser example of the PS3 being a nightmare to develop for and initially selling kinda poorly for a good chunk of its lifespan and yet still getting ports, if partly because Japan simply doesn't buy Xboxes.
If Nintendo continues to knock it out of the park with marketing and it works dividends for them? Third parties will flock to the Switch and throw multiplats onto it, technical challenges be damned.
Also, modern engines are a lot more scalable. If the Switch has a good enough CPU (and chances are it will, since the PS4/Xbone's Jaguar cores are garbage), scaling down the GPU side of things is the only main issue, and probably not that difficult.
frankie_baby
Member
I expect they'll put out at least one COD game on it, don't quite understand why they never bothered doing any for the 3ds though
th3sicknness
Member
I hope that is true because looking at the sales numbers Nintendo posted isnt encouraging. Lets hope people come back to the platform.
Wii 100 Million sales / DS 154 Million sales vs Wii U 13 Million sales / 3Ds 60 million sales.... Everything has been trending down. I see what Nintendo was thinking though.. You take 13+ 60 and you get a base that is larger than the PS4 or the X1..
///PATRIOT
Banned
TEAM FAKE
RampagingSoul
Member
Did the DS CODs (Modern Warfare, Black Ops I believe) even sell that well? That could have been a factor.
Louis Cyphre
Banned
Those ports quickly dried up on the Wii. Outside of a few titles I don't recall sales to be all that good and some publishers like Electronic Arts stopped making games on it. The online services were not comparable and most publishers are not going to make special editions just to cater the uniqueness of Nintendo's hardware.
So if the system is yet another weaker console will the handheld gamer pick up the slack? That's iffy if those games are now $60 a pop.
The issue with PPC is that many common frameworks and tools didn't support it and developers weren't going to do all that port work to make it run. UE4 was a particularly bad problem because in additional to western adoption, Japanese devs are making more and more use of it but it doesn't run on Wii U. In-house engines like Frostbite had no reason to support Wii U's architecture. The difference with Switch is that all of these do target ARM-based mobile devices and so devs don't have to port the whole toolchain, just the assets and maybe a few other things.
I expect Nintendo partners with one of the many cross-platform app development framework providers to provide easy ways to port Android apps to Switch.
I expect Nintendo partners with one of the many cross-platform app development framework providers to provide easy ways to port Android apps to Switch.
th3sicknness
Member
^^
Hopefully Nintendo and Nvidia are working on something to make this process easier than it was on PPC. I am hoping maybe the Vulkan API can be used as well.
Herb Alpert
Banned
Will there be twich on switch?
sorry...
Candescence
Member
To be fair, the Wii U and 3DS were both plagued by numerous problems. The 3DS had a gimmick that was introduced in a time when 3D viewing was already set to lose public interest, and too expensive, causing Nintendo to do a panic price drop months after. What helped it recover was said price drop alongside a few big titles like Mario 3D Land, and a fairly steady supply of good titles. The Wii U was poorly marketed and also expensive compared to the tech it was using, making it a dead starter. It's mostly due to Nintendo releasing some really popular titles that helped it cross the 10+ million line in the first place.
The Switch, on the other hand, has been marketed damn well perfectly and the reveal trailer has gotten over 13 million views worldwide with an insanely positive like/dislike ratio. It's obvious that Nintendo has learned from one of its biggest mistakes it made with the Wii U and 3DS. If they stay on form and provide an appealing launch lineup, I could see this selling a lot. And if they successfully market the Switch as an essential companion device for smart devices? They could actually make the dominance of mobile work for them. This will be especially true in Japan, where regular home consoles are dying, Japan will eat this thing up easily, especially the moment the inevitable Monster Hunter title for it is announced.
https://developer.nvidia.com/develop4shield
Nvidia already did a lot of this work with the Shield. That can be expanded upon with the Switch.
They also have very good relationships with a lot of the big devs and big middleware companies.
DerekOfTheDykes
Member
I have a strong suspicion that you will still see a majority of handheld gamers accept higher prices, due to the perceived increase in value of Switch games over the 3DS. Even if all they're getting is HD, prettier versions of franchises they were playing on 3DS, they'll probably accept $60 games.
I mean, recent example, I sure as hell didn't buy World of FF for the Vita, even though it was the same game at $20 cheaper. I gladly spent that extra money on a prettier version. And Switch may not be PS4 quality, but for the handheld gamers mentioned it's going to be a pretty massive jump for them.
th3sicknness
Member
There are a lot of things that will matter a lot outside of their sizzle reel.
Battery
Storage
Region Locking
Internet infrastructure (purchases stored by account instead of device)
OS
Achievements
Party Chat
They have a lot of things they MUST include in order for this to be a success. Because if this thing lasts 3 hours max and doesnt have anything from the above it is going to get slotted again into the severe limitations of the 3ds and Wii U and it wont be able to break the narrative that Nintendo just doesnt understand the market.
Didn't Eurogamer and Laura Gale leak pretty much the entire concept?
Cid Highwind
Banned
But how many bits does it have
TyMiles2012
Member
That's what I'm hoping. Even though consoles always had little to no overhead, Vulkan and DX12 will really help with optimization because it will be able to spread the workload across all the ARM and CUDA cores easily, while also offloading CPU computations to the GPU (Nintendo seems to enjoy doing that, especially with Xenoblade X and Breath of The Wild). Also, this could be the first non-Microsoft console to technically support DirectX at all since newer Tegra chips support it. If Microsoft lets Nintendo have the DirectX12 functionality enabled for devs, that will make a LOT of developers happy who don't want to mess with Vulkan, but of course, Vulkan will be very available and it's just as good.
EDIT: Looking at that Shield developer page above, looks like it'll just be Vulkan, though that was for Android. Who knows what'll happen, but I hope the NVN API is built from Vulkan's base (It's Open Source, right?)
Herb Alpert
Banned
A lot.
64 bits when running on battery, 128 when docked.
Nevar. Vulkan would be more than enough.
Gotdatmoney
Member
The majority of handheld gamers are younger and the majority of the 3DS sales and handheld sales in general aren't driven by your average gaffer. $80 handheld games is absolutely gonna hurt the bottom line.
TyMiles2012
Member
True. Though what makes me most curious is the actual performance increase. This new exclusive API will probably benefit the Switch greatly, but look at DOOM for PC. Nvidia cards didn't see a HUGE improvement between OGL 4.5 and Vulkan, but AMD cards, even after OGL 4.5 support got patched in (AMD users were gimped at launch only supporting OGL 4.3 with the 390X getting outclassed by the 970), saw a HUGE jump.
A Super Ultra amount.
OpenGL and Vulkan mainly exists in order to make 3D rendering calls portable between OS and hardware that operates within a very modular setups (PC and smartphone/tablet, embedded devices, etc.) Consoles usually don't have to worry about that as the hardware configuration is static. The NVN API should allow the Switch to easily fit within the current model of development process most developer are currently working with. Probably the first Nintendo platform able to do so.
Schnozberry
Member
NVN is likely a Nintendo customized version of NVAPI, a low level software development kit that leverages specific Nvidia hardware features. On Windows in integrates with DirectX to give developers deeper access to Nvidia's hardware. I'm sure it serves a similar function here with Nintendo's version of the Vulkan.
Yeah, i don't know why he's saying single precision in the article.
Schnozberry
Member
Pairing A72 cores with A57s would be very unconventional, to say the least. They are essentially iterations of one another. If they were going with A72, they would probably just replace the A57s and go with a quad A72 design. It would be the least disruptive to developers who already have access to the dev kits.
I somehow thought that it would be the main API.
dampflokfreund
Banned
So Switch will likely have 50 GB/s RAM with a decent amount of e/dRAM, eh?
That would be enough.
That would be enough.
Schnozberry
Member
Looking back at Nvidia's press release for the Switch:
It sounds like you might be right.
At this point in time, everything about the Switch's final hardware specs are speculative since the leaks are probably based on Tegra development boards.
I suspect NVN is more likely to be Vulkan-based. Nintendo joined the Khronos Vulkan working group (something they never did for OpenGL despite using it as a secondary API choice on multiple consoles), and the two Ns could stand for Nintendo and Nvidia, with the V standing for Vulkan.
Schnozberry
Member
An increased GPU L2 or added L3 could still be useful even with a 128 bit bus. Of course if you're talking about a non-cache memory pool, then I wouldn't expect that regardless of the main memory bandwidth.
Schnozberry
Member
How large of an L3 cache would you think necessary? If they went with the same cache configuration as Parker, L1 and L2 are both 2MB.
I see. But would 4xA72 deliver the same performance as 4xA57+2xDenver2, even if Denver isn't really gaming oriented? Or would it mainly be a less powerhungry setup?
Edit: so the A57 is in the Snapdragon 810 and inside the Helio X10 (which is in my phone) and the A72 is in the Snapdragon 650 and the Helio X20 and X25.
How likely is it for those A72 -should they find their way into the Switch - to get a sidekick like A53?
Edit2: Phones with 2xA72 + 4XA53 almost keep up with phones with 4xA57 + 4xA53.
Candescence
Member
Eh... I think you're being a tad overdramatic with all of those. Battery is absolutely important, no question, and so is storage (albeit to something of a lesser extent), but everything else is not something your average consumer cares much about, give or take. "OS" just means a decent OS, I suppose, and that's not really that difficult, it's not like the PS4 and Xbone have great OSes either. Achievements and party chat are nice but not essential. Only hardcores care about region locking.
And "internet infrastructure" is kind of a given, since Nintendo has been working closely with DeNA and transformed Club Nintendo into My Nintendo, clearly with more of an account-focused direction. It's easy to say "Nintendo gonna Nintendo", but there's enough signs to give them the benefit of the doubt, and they are improving things with each new system.
Really, most of those aren't things non-hardcore consumers will actively look for. As long as the battery and storage isn't a huge deal, they'll be focused more on the general concept and the games.
Spyder_Monkey
Banned
I'd say let the games speak for themselves before arguing over teraflops like you know what those words mean. The 3DS was below Gamecube level, yet outside of resolution, had games like RE Revelations, Kid Icarus Uprising, and Mario 3D Land. that looked on par, if not surpassing Wii games. It's not about raw power, it's about good dev tools, good engine support and what the devs can do to take advantage of the system as much as they can. Like remember how Sonic Boom looked and ran like ass on the Wii U? It wasn't because the Wii U couldn't handle a game like that, it was because the engine wasn't supported on the Wii U. Besides marketing, the biggest fault of the last console was because it was hard to make games for outside of Nintendo's bubble. If they can rectify that, then there's no reason why third parties can't be making games for it.
Schnozberry
Member
Tegra X1 uses in-kernel switching to activate the A53 cores in low power mode. It's not like the Helio or Exynos processors, with use heterogenous multiprocessing to allow all 8 cores to be active simultaneously at various clock speeds. Nintendo wouldn't be married to Nvidia's previous setup necessarily, but it makes the most sense in a gaming scenario. There's more optimization work required to take advantage of the A53 cores while gaming, because they are in order CPUs and have a different SIMD engine and instruction pipeline than A57 and A72.
I would almost completely rule out Denver at this point, as it would be an unsavory curve ball to drop a completely different architecture on developers in between the dev kits and final release.
On the wii many ports or spin off of AAA titles from the HD consoles were cash ins with little to no effort at all that sold very poorly most of the time, I'm afraid the "if the console sells well titles will come" it's quite up in the air. Obviously this time around porting would be less of a chore then porting an HD title on the wii, but still I'm not confident at all.
OG_Original Gamer
Member
Nintendo is going to have to sweeten the pot, in order to get developers on board to build a NS version of every game.
Well Nvidia already has an API named NV API. So the N could have just been added because it's Nintendo's version.
https://developer.nvidia.com/nvapi
th3sicknness
Member
I dont think im being over dramatic at all. When the Wii U launched (I bought day 1) you had to wait a PAINFULLY long amount of time to get the system turned on and to transition between different areas (even launching games). The quick start gamepad mode didnt exist. The OS is was so bland and outdated not to mention not including FOLDERS until the Wii U was dead. Nintendo needs to release a polished OS. The whole package matters.
th3sicknness
Member
You need to keep in mind AMD cards are garbage at Open GL.. So the gains you saw from Vulkan brought it up to par and in most cases exceed their Nvidia counterparts.Had AMD been good at Open GL the gains would not have been as drastic. The gains are mainly due to Async Compute working better on AMD Cards in addition to AMD focusing on future APIs like Direct X 12 and Vulkan. Nvidia cards dont suffer under Vulkan but they dont gain as much as AMD does because they are already at the peak of their potential.
Also Vulkan / Direct X 12 may not be the most efficient way to develop for the Switch. Keep in mind ps4 has its own API along with the X1 that code much more close to the metal. It seems that Nvidia is offering something similar so vulkan may not even be as prevalent as we think.
Father_Brain
Banned
Yeah, things like this are why I'm pretty certain I'm not buying Switch on day 1 even if the price and launch lineup are good. I'd rather not be a beta tester.
megabytecr
Member
A matter of taste, I was day 1 Wii U and yes the system had issues, you also forgot the hard resets, but to me having the system and games early overshadowed any issue. They have no margin for Error, so I beleive they will learn on past mistakes and even though having zero issues is almost impossible, they will have a smoother launch.
DeuceGamer
Member
I was there day one as well, and honestly I didn't think it was that big of an issue navigating menu's. Maybe I'm just more tolerant of it, or I played longer sessions. I will say that the ingame- load times of Lego City were pretty horrid though.
Hmm... Maybe the HBM stack is for Xavier? I was reading Drive PX2 specs and had it 8GB of 128-bit LPDDR4 RAM for system memory and 4GB of GDDR5 RAM for Graphics memory.
HBM is probably overkill for Switch, so the other option is higher LPDDR4 RAM Memory Bandwidth (so 128-bit, 2 chips) and more cache.
____
16nm for the GPU's of the Slim consoles makes sense, and likely for PS4 Pro.
Are you sure on X1's CPU cores not being able to all be active simultaneously? I remember Anandtech saying that all 8 cores are exposed by the OS and so usable at the same time in Shield TV.
- Status