-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Switch 2 battery to last 4 hours and why AMD lost out
- Thread starter Dorfdad
- Start date
64bitmodels
Reverse groomer.
compare it to other ARM devices then
Rambone
Member
As much as I would hate for this to be true, it's a total plausible Nintendo move to make.
mckmas8808
Mckmaster uses MasterCard to buy Slave drives
Which is really surprising. I could see if the got 6 hours of battery life out of it. But 5 watts, just for it to last 4 hours? Not great.
mckmas8808
Mckmaster uses MasterCard to buy Slave drives
How Long Does the Battery Charge Last?
Important: As with most lithium-ion batteries, the console battery life will gradually decrease over time. After about 800 charge cycles, the console battery life will decrease to about 80% compared to the battery on a new console.www.nintendo.com
So the 2017 console with 4-6.5 hours (or 4-9 hours after the first SoC revision) of battery life with a successor likely launching 8 years later for a likely higher MSRP touting up to 4 hours battery life as something to shout about? Aaaaalright…
Would $400 make this a reasonable device?
jm89
Member
They are deffo gonna skimp somewhere, and probably not just the LCD screen. Assuming they are aiming for $400, they need to keep the hardware costs within that but also leave space for profit.
Last edited:
Minsc
Gold Member
It's super crazy, as at wattages that low you can get 12 hours out of the OLED Deck.
gokurho
Member
Seems AMD was in the running for the switch 2 but lost the bid do to efficiency of their chips and wanting to force Nintendo into 15watt performance, but Nintendo wanted 5watts. They also state that the battery in the switch 2 is slightly better than switch 1 because the new chip is more efficient. So we could be looking at 4 hours of time now.
Might explain why OLED is missing also. 5 watts sounds very low for switch 2 unless that chip is magical!
Battery life is likely to be one of the highlights of the Nintendo Switch 2. Judging from last week's leaks, the battery should provide around 20 wh, a slightly higher capacity than the original Switch battery. However, the system will consume way less power in handheld mode than its predecessor, granting much better battery life than the current generation console.
Nintendo Switch 2 AMD Bid Didn't Go Through Due to Handheld Mode Performance and Efficiency; Leaks Suggest Excellent Battery Life
According to rumors, AMD also bid for the Nintendo Switch 2, but eventually lost due to performance and efficiency in handheld modewccftech.com
UPDATED POST SEE #141 for new USBC ADD-ON ATTACHMENT
Switch 2 battery to last 4 hours and why AMD lost out
See, that's what I thought(although for more intensive games like TOTK it was reported that the switch in docked mode would peak at around 16w). So the question becomes; Why would Nintendo limit the switch 2 at 5w in handheld mode, when it was much higher on the OG switch? Unless of course...www.neogaf.com
jm89
Member
marquimvfs
Member
5W mode could be the power they were pursuing for retrocompatibility, I guess. Given that the AMD couldn't achieve that power because of the emulation layers, they were discarded. Also, there's nothing impressive at those marks alone, it could be just Nintendo wanting to go cheap on battery and cooling. It only will be a remarkable thing if that power level translates to a high performance. On the other hand, if the 5W is only about the retrocompatibility, if they manage to emulate switch with that power level, while improving the assets quality, that will be sort of amazing.
Yes, so if peak efficiency is at 540Mhz, it will be reduced at 300 MHz! Therefore Nintendo could have designed a smaller and cheaper chip for the same performance! That's why 8nm doesn't make sense for a chip of T239's size. See the below analysis by Thraktor.
Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)
Oldpuck's leaving #teamleapday? Guess I'll have to continue flying the flag solo. On the time of 18 to 24 months from sampling to release, I assume you're talking about the TX1 and the original Switch? If so, I don't think it's really comparable, as TX1 was already entering production...
famiboards.com
Also, TFs on Ampere are not "fake". You can achieve them in compute applications, but since the extra FP32 unit can only be used at the expense of an INT32 one, the real world performance gains in games are less. You should multiply the Ampere TF figures by ~0.7 to get the equivalent RDNA 2 part or ~0.875 to get the equivalent GCN part. For example the 5 TF - 6 TF 3050 Ti is about as fast as the 5 TF (GCN-based) 470 in the TPU benchmarks.
Last edited:
kid friendly or not, you’re delusional if you think they don’t look at the marketplace and potential competition when they make their hardware.
And one key thing they’d have focused on for the Switch 2 would be giving it capable enough specs to guarantee many more third party ports in the years to come. There’s money being left on the table if that’s not a focus. Significant amount of money.
‘Cheap’ is debatable too. I strongly doubt a Switch 2 would come in at a penny less than the $349 launch price of the Switch OLED. Especially now they’ll have to ship it with decent sized, relatively high speed storage.
Fully agreed.
CrustyBritches
Gold Member
MLID is the perfect example of 'Fake it till you make it". He was all bullshit and repackaging existing info for years until the PS5 Pro info he accurately leaked.
As far as the Switch 2 goes, the best sources have been the Nvidia leak and Kopite7Kimi. In this case, the power limits could just be assumed based on the Switch 1.
rnlval
Member
FYI, AMD has Samsung Galaxy S24's Exynos 2400 SoC which has RDNA 3.
Exynos 2400 SoC beats the old NVIDIA Tegra X1.
Wasn’t Wii an AMD console?
kevboard
Member
the original Switch had about 3h on average. lower or higher depending on how much GPU power a game needed.
only with the Mariko revision did the battery life go above 4h on the Switch 1.
Breath of the Wild, the big launch title, was 3h or less depending on your screen brightness. it being around 2 hours on max brightness.
I personally always got around 2h 30min with my screen settings (around 40%) on my launch system.
also, 4h of portable use is totally fine. I can't think of a situation where I would be 4h away from a power source while using a handheld system.
Last edited:
rodrigolfp
Haptic Gamepads 4 Life
Power consumption maximum of the entire system.
chromhound
Member
So Switch+ ?
CrustyBritches
Gold Member
The actual limit I wouldn't know. Typical total system power consumption in a 'AAA' game is around 20-25W. GOW and HZD can hit 27-28W. TW3 can hit 25W in certain areas.
rodrigolfp
Haptic Gamepads 4 Life
When I was looking for a power bank for mine I read on sites that can go up to 45W. And that makes sense as the og charger is 45W and battery sometimes doesn't charge or even discharge when I am plugged and playing something very power intensive...
Yes. It was based on the Radeon 4,000 series.
Haint
Member
Yes, so if peak efficiency is at 540Mhz, it will be reduced at 300 MHz! Therefore Nintendo could have designed a smaller and cheaper chip for the same performance! That's why 8nm doesn't make sense for a chip of T239's size. See the below analysis by Thraktor.
Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)
Oldpuck's leaving #teamleapday? Guess I'll have to continue flying the flag solo. On the time of 18 to 24 months from sampling to release, I assume you're talking about the TX1 and the original Switch? If so, I don't think it's really comparable, as TX1 was already entering production...
Also, TFs on Ampere are not "fake". You can achieve them in compute applications, but since the extra FP32 unit can only be used at the expense of an INT32 one, the real world performance gains in games are less. You should multiply the Ampere TF figures by ~0.7 to get the equivalent RDNA 2 part or ~0.875 to get the equivalent GCN part. For example the 5 TF - 6 TF 3050 Ti is about as fast as the 5 TF (GCN-based) 470 in the TPU benchmarks.
Yes that's why I said actual performance is around 70% less than the "fake" TF figure would imply. Only thing Nintendo cares about is price, and Samsung 8nm wafer is heavily rumored to be at least 1/3rd the price of a 5nm TSMC wafer, $5K - 6k range Vs. $15K - $18K. I don't even think a 5nm wafer will produce 2x more chips per wafer, never mind 3x or better. Nintendo's miser ass is most definitely NOT going to pay a 125%+ premium per chip for better clocks or power efficiency, not a chance in hell.
Last edited:
Kataploom
Gold Member
The Nvidia long term partnership started with the Switch and Switch 2 was always gonna have BC for which Nvidia was necessary. The sole reason Nintendo contracted DeNA for the current account system was for them to avoid having to start over again next gen with the installed base, as per Iwata words iirc so the BC and long term deal with Nvidia was basically a given since the Switch mere conception.
Fake
Member
Game Cube as well, but was before AMD got ATI.
Rambone
Member
Yea, I suppose I won't try and dwell on whether or not we will get yesteryears sloppy seconds and just be happy to be getting (much needed) new hardware.
kevboard
Member
ATi technically. that was before AMD bought Radeon.
but the CPU was an IBM PowerPC design.
making the 7th gen the IBM PowerPC generation lol, as all 3 consoles had IBM CPU cores, with the PS3 and 360 even using the exact identical core design, with the only difference being that the PS3 only had 1 main CPU core while the 360 had 3 of these cores. the PS3 of course relying on the Cell's SPEs to help both the main core and the GPU
Last edited:
existensmaximum
Member
Guess not. But are you really suggesting the Nintendo Switch or it's successor is in direct competition with handheld gaming PCs..?
SweetTooth
Gold Member
Yes, with handheld PCs, with the rumored handhelds from Microsot and Sony.
itsjustJEFF
Member
that's 4 hours of Nintendo... that's different.4 hours is impressive?
LordOfChaos
Member
Yes, 1536 Ampere Cuda Cores, 48Tensor cores and 12 ray-tracing cores
SweetTooth
Gold Member
Nintendo gets a free pass even on battery life4 hours is impressive?
twilo99
Member
As big of a blasphemy as low end Zen2 CPU in a high end gaming console in 2025?
I think not.
CrustyBritches
Gold Member
For sure. I think that would be considered power consumption "from the wall" while charging, as opposed to total system power consumption. The Deck will show you it in a few of the overlay modes next to the battery percentage and estimated battery time remaining.
I have a kill-a-watt/power meter and when you are playing unplugged the Deck says 25W for TW3, then when you plug in the charger the power meter shows ~45W. If you turn off the Deck and charge it the power meter says ~25W.
Last edited:
lachesis
Member
This whole wattage count just reminded me of this.. 
Not just CPU - but whole system engineers must be going thru something like this to keep the wattage usage under given amount of set allowance... much respect to them.
I'm honestly not expecting ground breaking - heck, PS4 Pro level at docked mode = I think it's plenty good enough for me as I don't really care much for RT or whatnot. Hope it would just keep consistent frame rate, hopefully 60fps for most games - at least 1080p. That's all I'm asking for.
Two sku rumors actually made me thinking.. that what if they are gunning for "lite" version that's similar to current "lite" ver? in size... and "XL" version as 2 skus? Lite being cheaper one, and XL being the expensive model...
Highly doubt if they are gunning at simultaneous 2 sku launch other than minor differences in included storage like Wii U model though... but one can imagine, I guess.
Not just CPU - but whole system engineers must be going thru something like this to keep the wattage usage under given amount of set allowance... much respect to them.
I'm honestly not expecting ground breaking - heck, PS4 Pro level at docked mode = I think it's plenty good enough for me as I don't really care much for RT or whatnot. Hope it would just keep consistent frame rate, hopefully 60fps for most games - at least 1080p. That's all I'm asking for.
Two sku rumors actually made me thinking.. that what if they are gunning for "lite" version that's similar to current "lite" ver? in size... and "XL" version as 2 skus? Lite being cheaper one, and XL being the expensive model...
Highly doubt if they are gunning at simultaneous 2 sku launch other than minor differences in included storage like Wii U model though... but one can imagine, I guess.
kevboard
Member
tensor cores and RT cores yeah
SweetTooth
Gold Member
Tell me one instance of a console manufacturer changing the architecture of the CPU mid-gen. Increasing clocks of the same architecture doesn't count.
Once you answered, you will realize how clueless your comment sound.
You know making drastic changes to console CPUs entails changing the whole development suite, libraries..etc making developers life hell on earth like developing for a while new console!
Your second mistake is treating whatever CPU in consoles like the one used in PCs. consoles games are coded on a lower level program, there are also more specialized HW not found on PC (decompression blocks equals to 11 Zen cores on PS5) plus the eight Zen 2 cores. Of course you haven't seen DF article with GoW Ragnarok porting devs!
Believe me hardware designers at these companies are way smarter than you and me, if it was so easy and hassle free to "upgrade" the CPU, Microsoft would have done it for One X and Sony would have done it for PS5 Pro.
Now back yo my question, which midgen console refresh changed the CPU?
Last edited:
James Sawyer Ford
Gold Member
Makes sense for Nintendo to go with NVIDIA since DLSS is going to be super important for them and as far as we know AMD doesn't have a good competitor for that yet.
But yeah, I don't think AMD was ever in the equation to begin with.....due to various reasons like BC. Could just be another "we had to get a competitive bid" to lower the asking price a tad for NVIDIA, though reality is NVIDIA doesn't need to be competitive with anyone in this market.
But yeah, I don't think AMD was ever in the equation to begin with.....due to various reasons like BC. Could just be another "we had to get a competitive bid" to lower the asking price a tad for NVIDIA, though reality is NVIDIA doesn't need to be competitive with anyone in this market.
existensmaximum
Member
Mmm yeah, no. Not unless these handhelds cost max $400 and include new Zelda, Mario and Mario Kart games. There's a reason why the Switch has sold 143 million units and the most successful handheld PC has sold around 3 million.
SweetTooth
Gold Member
Bolded didn't save Nintendo when the big boys arrived to their space.
Panajev2001a
GAF's Pleasant Genius
Price is one factor, granted there is inflation, but still a lot of technology evolution too… not sure…
rnlval
Member
Per SM basis, Ampere's 2X FLOPS increase is "fake" when NVIDIA didn't scale TMU by 2X. Modern graphics gaming workloads mostly operate on floating point data formats.Yes, so if peak efficiency is at 540Mhz, it will be reduced at 300 MHz! Therefore Nintendo could have designed a smaller and cheaper chip for the same performance! That's why 8nm doesn't make sense for a chip of T239's size. See the below analysis by Thraktor.
Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)
Oldpuck's leaving #teamleapday? Guess I'll have to continue flying the flag solo. On the time of 18 to 24 months from sampling to release, I assume you're talking about the TX1 and the original Switch? If so, I don't think it's really comparable, as TX1 was already entering production...
Also, TFs on Ampere are not "fake". You can achieve them in compute applications, but since the extra FP32 unit can only be used at the expense of an INT32 one, the real world performance gains in games are less. You should multiply the Ampere TF figures by ~0.7 to get the equivalent RDNA 2 part or ~0.875 to get the equivalent GCN part. For example the 5 TF - 6 TF 3050 Ti is about as fast as the 5 TF (GCN-based) 470 in the TPU benchmarks.
TFLOPS are useless by themselves without the read and write units such as TMUs and ROPS.
Last edited:
existensmaximum
Member
Their space is the handheld market.
rnlval
Member
Wii U's GPU was based on DirectX9 class Radeon HD 4xx0 with 160 stream processor scale. The original Wii GPU was a DirectX 7 class GPU.