lattjeful
Like Like
- Pronouns
- he/him
This shit should be pinned lol. This really sums up the Steam Deck/Switch 2 comparison.
-
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| (New Staff Post, Please read)
- Thread starter Dakhil
- Start date
oldpuck
Like Like
- Pronouns
- he/they
Y’all are making my point
You can’t give a simple answer that is also nuanced.The question was, will handheld Drake be more powerful than SteamDeck and the answer is absolutely not. It’s not a question.
On every single benchmark, AMD outperforms Nvidia on similarly specced hardware.
On single core performance, Steam Deck’s CPU will stomp Drake’s. It benchmarks better and it’s clocked much higher.
On multi-core performance it’s the same. SSD performance? Well ahead. Memory? Lucky if Nintendo is tied.
“How much will it matter” is a much more complicated question.
epicmartin7
Octorok
I think the big mystery box surrounding the Switch 2 is definitely that DLSS capability. Depending on how it's implemented, it could either be a small nudge, or a big difference.Y’all are making my point
The question was, will handheld Drake be more powerful than SteamDeck and the answer is absolutely not. It’s not a question.
On every single benchmark, AMD outperforms Nvidia on similarly specced hardware.
On single core performance, Steam Deck’s CPU will stomp Drake’s. It benchmarks better and it’s clocked much higher.
On multi-core performance it’s the same. SSD performance? Well ahead. Memory? Lucky if Nintendo is tied.
“How much will it matter” is a much more complicated question.
- Pronouns
- he / him
Is this the case? I didn't think there had been too many benchmarks available for A78C-or-comparable, but from what there were I thought it was the case that Switch 2 compared to PS5 would at least not be any worse off than Switch 1 vs PS4. And Steam Deck CPU seems basically like PS5 CPU cut in half.
gonanzir
Nintendo doing Nintendo.
The cpu comparison is a complicated thing, but we can run a cpu benchmark to make comparison, for example, in geekbench5, the steam deck's multi-core run score is 3724.compared to the MediaTek Dimensity 8200,the chip has 1*[email protected]+3*[email protected]+a55*[email protected], and the final geekbanch5 score is 3947. Note that the a55 doesn't offer much of an advantage.
Yes, but as far as I understand, that's mainly with regard to desktop GPUs, and only when looking into raw performance. Devices like the Series consoles and SD which use RDNA2 in their APUs lack a few things the desktop GPUs have, like Infinity Cache. So in the non-desktop realm, things may be a bit different.
Multi-core? You sure? You had done the comparison between PS5 vs Orin AGX.
Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)
having instant access to local 2 player co-op is not something they will just get rid off
famiboards.com
Found the entry for the Steam Deck, and compared it to the Orin AGX you used.
| CPU | Geekbench 6 Single | Geekbench Multi |
|---|---|---|
| 8x A78AE @ 2.19Ghz | 1001 | 5449 |
| 4c/8t Zen2 @ 2.8Ghz | 1153 | 4222 |
| Diff (%) | 86.8% | 129.1% |
While this is using a slightly higher clock than what we expect for Drake, it's also based on the A78E, which would definitely underperform in multi-core compared to the A78C.
NineTailSage
Bob-omb
Y’all are making my point
The question was, will handheld Drake be more powerful than SteamDeck and the answer is absolutely not. It’s not a question.
On every single benchmark, AMD outperforms Nvidia on similarly specced hardware.
On single core performance, Steam Deck’s CPU will stomp Drake’s. It benchmarks better and it’s clocked much higher.
On multi-core performance it’s the same. SSD performance? Well ahead. Memory? Lucky if Nintendo is tied.
“How much will it matter” is a much more complicated question.
Yes, but as far as I understand, that's mainly with regard to desktop GPUs, and only when looking into raw performance. Devices like the Series consoles and SD which use RDNA2 in their APUs lack a few things the desktop GPUs have, like Infinity Cache. So in the non-desktop realm, things may be a bit different.
Multi-core? You sure? You had done the comparison between PS5 vs Orin AGX.
Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)
having instant access to local 2 player co-op is not something they will just get rid off
Found the entry for the Steam Deck, and compared it to the Orin AGX you used.
CPU Geekbench 6 Single Geekbench Multi 8x A78AE @ 2.19Ghz 1001 5449 4c/8t Zen2 @ 2.8Ghz 1153 4222 Diff (%) 86.8% 129.1%
While this is using a slightly higher clock than what we expect for Drake, it's also based on the A78E, which would definitely underperform in multi-core compared to the A78C.
I think that was going to be my question initially as well, when comparing Drake 8 core CPU to SD 4core/8thread CPU.
Which set-up benefits a pure gaming device more
I found this article on Anandtech having discussions about muti-threading on Zen3 and test multi-core vs muti-threading
Here's a summary
"We’ve known for many years that having two threads per core is not the same as having two cores – in a worst case scenario, there is some performance regression as more threads try and fight for cache space, but those use cases seem to be highly specialized for HPC and Supercomputer-like tasks. SMT in the real world fills in the gaps where gaps are available, and this occurs mostly in heavily multi-threaded applications with no cache contention. In the best case, SMT offers a sizeable performance per watt increase. But on average, there are small (+22% on MT) gains to be had, and gaming performance isn’t disturbed, so it is worth keeping enabled on Zen 3."
Investigating Performance of Multi-Threading on Zen 3 and AMD Ryzen 5000
BreakAtmo
Bob-omb
- Pronouns
- He/Him
Where are you getting the 2.8GHz from? My understanding was that the Deck CPU can technically go up to 3.5GHz, but in practice the full 15W power draw works by getting the GPU to its 1.6GHz cap, then the CPU gets whatever is left over (unless you lower the GPU cap yourself in the settings).
I remember hearing that the CPU with that leftover power would only be about 2GHz, but that could be totally wrong. Also, the Deck OLED probably lets it go higher due to 6nm.
NintenDuvo
Bob-omb
- Pronouns
- He him
them Nintendo Lawyers have made their presence known and its so frustrating because we have nothing to discuss and Nintentubers are reaching in the bottom of the barrel for content
Valve lists the CPU from 2.4Ghz to 3.5Ghz. 2.4Ghz is considered the base. 3.5Ghz is supposedly for Turbo involving a single core. 2.8Ghz is supposedly Turbo for all 4 cores.
oldpuck
Like Like
- Pronouns
- he/they
Typing on my phone in an airport made my language ambiguous. By "the same" I meant "Steam Deck and Drake perform the same" not "Steam Deck stomps, the same as in single core"
This is my reference comparison now that Orin NX is out (I think I did the old comparisons based on AGX because the NX/Nano products weren't yet available, IIRC). The three big asterisks are *Steam Deck has dynamic clocks which can go higher, *A78C should slightly outperform on A78AE on multicore, *Switch clocks unknown 2.0 GHz probably optimistic.
"Which setup will age better" is a genuinely interesting question, but it comes back to my original point - Steam Deck has a horsepower advantage, Drake has an architecture advantage, good ports can use the one to overcome the other. Love to really parse that, but I assumed the original question asker wasn't up for that.
Let me put it this way - right now, if a game is on both platforms, I get it for Steam Deck. I expect that to change.
Using Win11 instead of SteamOS results in those better scores for Steam Deck? Honestly, I thought SteamOS was the more light-weight OS of the two. Whereas a 10% drop in core clock for A78AE results in a score drop of 14%/single and 24%/multi, though I don't know how they calculate the scores.
NineTailSage
Bob-omb
It's definitely an interesting situation to discuss and judging most of those multi-core performance metrics, the muti-threading doesn't hold up as well as it should given AMD's single-core advantage. Drake also having a dedicated FDE on board frees up processing power that the SD will have to contend with CPU wise...
All in all with games being more tailor made for Switch 2 vs SD, I expect that the perceived performance value will look much larger in actuality than it does on paper.
ShadowFox08
Paratroopa
It might actually be closer to 7x A78 cores vs 3x Zen 2 cores... Cause the one core is for the OS... It will be interesting to compare the two for sure. If we see how do get 4nm TSMC, I see no reason why switch 2 wouldnt have a performance vs wattage ratio. Even if Nintendo takes 8-10 watts for a 1.6 TFLOPs GPU and 2ghz for 7 A78 cores for gaming, that would be very impressive compared to SD using 15 watts (not sure how much power draw the OLED version is).Yes, but as far as I understand, that's mainly with regard to desktop GPUs, and only when looking into raw performance. Devices like the Series consoles and SD which use RDNA2 in their APUs lack a few things the desktop GPUs have, like Infinity Cache. So in the non-desktop realm, things may be a bit different.
Multi-core? You sure? You had done the comparison between PS5 vs Orin AGX.
Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)
having instant access to local 2 player co-op is not something they will just get rid off
Found the entry for the Steam Deck, and compared it to the Orin AGX you used.
CPU Geekbench 6 Single Geekbench Multi 8x A78AE @ 2.19Ghz 1001 5449 4c/8t Zen2 @ 2.8Ghz 1153 4222 Diff (%) 86.8% 129.1%
While this is using a slightly higher clock than what we expect for Drake, it's also based on the A78E, which would definitely underperform in multi-core compared to the A78C.
That's part of it, but the real mystery box imo is NVN2 (where DLSS is one of many features), as NVN was with the original Switch. SD runs PC code, while Switch 2 games will be made using custom API/ Tools.
Danny
Like Like
- Pronouns
- He/Him
The only thing steam deck being stronger than Switch 2 will cause is a lot of discussion about how weak the Switch 2 is that will happen when the specs are revealed. And of course everyone will be talking about Nintendo once again cheaping out by releasing hardware weaker than several years old previous hardware.
Last edited:
ERezaSwitch
Piranha Plant
2015 pic of the Wii U
Darknut85
#TeamRemake
I didn’t know they had Switch 2 prototypes that far back. So they‘ll call it SWii Utch?/s
Last edited:
Steve
Paratroopa
- Pronouns
- he/him
Nuh uh.
Steve
Paratroopa
- Pronouns
- he/him
Yea it’s stated that developers start working on the next system when their main consoles launches.
As usual Nintendo are ahead of the curve
/s
7hatDeadCat
Tektite
- Pronouns
- he/him
Darknut85
#TeamRemake
For us Nintendo fans the whole year is January with every day being the first April.late april fools
SoldierDelta
Designated Xenoblade Loremaster
- Pronouns
- he/him
I think we've hit the late-stage of Nintendo insanity... Next week we're going to see a proto-PSPortal and go insane over it.
Danny
Like Like
- Pronouns
- He/Him
If AMD is better, why is Nintendo partnering with Nvidia instead? Is Nvidia cheaper?Y’all are making my point
The question was, will handheld Drake be more powerful than SteamDeck and the answer is absolutely not. It’s not a question.
On every single benchmark, AMD outperforms Nvidia on similarly specced hardware.
On single core performance, Steam Deck’s CPU will stomp Drake’s. It benchmarks better and it’s clocked much higher.
On multi-core performance it’s the same. SSD performance? Well ahead. Memory? Lucky if Nintendo is tied.
“How much will it matter” is a much more complicated question.
ReddDreadtheLead
#TeamLate2025WithAPotentialForEarly2026
- Pronouns
- He/Him
I wouldn’t even really necessarily look at it like that, as the flops is a complex answer that doesn’t really portray the capabilities of the system.
If we take into account that Drake would have 12SMs, the Deck has 4WGPs, Drake has a GPU that is 3 times larger than the deck but will be clocked slower. The deck has the clockspeed advantage but Drake has way more hardware resources in the GPU alone that handling complex graphics should not be an issue at all for it.
Larger amounts of cache, more bandwidth, more formats supported by the GPU, etc.
RDNA2 is a good uArch and Ampere is also a good uArch, but the GPU in the deck is way smaller so it won’t be able to handle things well even if it achieved the same FLOPs. Drake will handily outperform it simply due to that fact.
With RDNAx, they manage to share resources in a WGP (2 CUs) that act more like an SM would in how it shares its resources. Which is why I’m comparing it in such ways.
To go further, the Series S has 10 WGPs but it has significantly higher bandwidth and is clocked higher, so in a scenario such as that Drake will take longer to do a similar task that the series S does. But I will get back to this with an Asterisk.
PS5 has 18WGPs so on top of having 50% more WGPs, it would also be clocked >2x faster than Drake ever would in TV mode, so it would be 3x the performance on hardware resource value (which matters a lot).
Series X would have >2x the hardware resources and is likely clocked to 80 faster than Drake would be, so about 4x the performance Drake offers. Again, asterisk here.
Note that I’m using TV numbers, in portable you double everything here to get an idea of how much of a gap there will be.
And I’m saying this to reign some expectations because some people will get carried away and look at flops and think it’s the end all be all, but different architectures work differently with different flops. What Ampere does and what RDNA2 do is very different and they shouldn’t be seen as a 1:1, ever. Only when they are within the same architecture does it make sense to look at flops and even then, we need to account for other stuff such as memory and memory bandwidth. It’s all a piece of a very large and complex puzzle that quite frankly most people here will not get (neither will I as I’m not a game dev but know people!). It’s a lot of guesswork to get an idea.
Finally the asterisk, due to MS screwing up they (at the start for the few years) have, um, how do I put this where it doesn’t sound crazy? Ok so, because their system wasn’t quite ready, tools and such and all, they had performance issues with games. Not because the series X isn’t stronger than the PS5, because it is, but because there system was behind. And the DX12U API that the consoles run, even customized for it, will never run as close to the metal as the GNM successor API for the PS5 or NVN2. So two stones killed a bird here, but as time progresses and companies move to newer environments for development (one hopes anyway!), one element gets solved. But the API thing is still pretty high level compared to the other systems and that will never get to the same level imo due to the structure of Microsoft and Xbox.
That’s all I have today folks! I hope people understand the idea that the deck doing 1.6TFLOPs is irrelevant to Drake doing 1.6TFLOPs.
They're not better, they just have different tradeoffs.
AMD has more raw razter performance per teraflop.
Nvidia has better AI/ RT acceleration.
KarlRemarx
Cappy
They're not "better" in any way that really matters due to tradeoffs (definitely not enough to throw away an existing partnership) and are often only "better" in scenarios where Steam Deck will suck battery beyond what Nintendo would tolerate, and nvidia's advantages in other areas (DLSS) are uniquely beneficial to a hybrid that will operate at two different resolution targets. Which is why oldpuck keeps saying it's complicated.
They also aren't going to throw away a partnership that gives them hardware continuity across generations and with a company that is generally technologically ahead of AMD. Nintendo have the right partner in this scenario.
All of this is also completely meaningless, because the primary graphical metric that will matter when it comes to selling Switch 2 will be whether it delivers a sufficient uplift from Switch 1 that is noticeable enough to the mass market, not performance relative to Steam Deck.
There again proving Oldpucks point lol.
"The problem is what happens when someone asks a reasonable and genuine question - "which is more powerful" - you kinda have to admit that all these numbers are liars. Any sort of nuance, and it's a three page reply. Any "short for non-techies" answer reply is gonna miss the nuance"
Because AMD didn't have anything when the Switch was being designed
Stinky Horse
Bob-omb
TSR3
Pilotwings
This ...
... is what it's all been about, at least for me anyway. Whether a modern Nintendo console can look good on modern displays, whether 4K TVs or HD tablet screens. Nintendo's blue ocean strategy means the other handheld gaming devices have never been direct competitors competitors to them, at least for the mass market.
Aufhebung RPG
History is tragedy becoming comedy
Haha, I can tell if it's fake news even without looking at the machine because Shigeru Miyamoto's hair is basically all gray and his face is sagging nowSneak peek of next week's news
Aufhebung RPG
History is tragedy becoming comedy
No, I need to correct myself that Nintendo strategically executed blue ocean strategy only during the wii period, from the wiiu onward Nintendo's strategy was both core gamers and blue oceans
Darknut85
#TeamRemake
Japanese people age differently./s
man the Xperia Go was such a weird device. it was more of a sneak peek at the Vita 3GSneak peek of next week's news
KarlRemarx
Cappy
They're "competing" in the same way every videogame company is competing for attention/entertainment hours from customers. At the end of the day someone who bought a Steam Deck is absolutely a potential Switch 2 owner to be won over - it's just that a Steam Deck owner is probably the easiest possible target Nintendo could go for regardless of any power differential, because the Venn diagram of "someone with the cash and entertainment interests to buy a Steam Deck" and "someone who would pay money for a 3D Mario in year one" is almost a circle. If you look at Steam Deck's most played games they are primarily indie games that could (and do) run on Switch 1, and AAA games that simply cannot run at 1080p (or even native 800p) at 60fps on Steam Deck (thus proving that hardware power is actually meaningless at a certain point, if sub-HD and sub-60fps are acceptable to even hardcore audiences). Steam Deck actually proves Nintendo's approach is correct - it just benefits from having been manufactured years after Switch 1.
What matters for Nintendo hardware is being:
1 ) good enough to keep pace with (diminishing) advances in customer expectations for graphics, to keep AAA game engines viable so 3rd parties remain aboard (although I would argue that indies are actually more important than AAAs when it comes to supporting a hybrid)
2) available at a mass market price
From all available information we have, Nintendo/nvidia have largely solved this. What will be most interesting is the software and interface Nintendo deliver around this spec, which will be what actually decides how they perform in the market.
Stinky Horse
Bob-omb
The one in' the pic is the PSP Go! Which would have been actually been a sneak peak at the Vita design if they'd stuck to the Vita protype.
kvetcha
hoopy frood
- Pronouns
- he/him
I have a friend who just picked up a PSP Go and has been playing Grandia on it. I always thought it was a cool little thing.
eye0h
"aggressively persue a multiplatform strategy"
- Pronouns
- He/Him
I wanted an xperia play so bad, remembered spamming the crash bandicoot demo in store while begging my dadeasy mistake to make
- Pronouns
- He/Him
Yeah it really helped Switch to have this kind of title right at the start. It felt like Nintendo saying, "Yes we knew that the last few years we were holding back and reorganising ourselves, but now we are back with full steam ahead."I'll always remember the day i got my switch and played BOTW.
Playing a huge open world game on a handheld, that's visually beautiful will always hold dear to me.
Hopefully the Switch 2 launch title will to something similar in which i react
,,how did they fit this in a portable system''
I think one of the main reason we got an internal delay, is mostly because Nintendo wants to have another BOTW moment, in which that sole game will carry the system legacy and be the system seller.
Hopefully this time around it'll be the next 3D Mario
Remember, after WiiU many people thought that it was impossible for Nintendo to have a runaway success like DS and Wii again, but then it surpassed all expectations. Just imagine what the narrative about consoles would have been, without a successful Nintendo console and PS4/XBOX One winding down. I don‘t wanna say that Nintendo saved the console market, but it gave it new life.
Switch really was the perfect storm. From BOTW to at least Mario Wonder the Switch train had zero slow downs, at least for me.
That‘s why I think Nintendo is taking their time now. I believe that it is impossible to come up with a successor that will do the same as the Switch and they for sure know that. They gained a lot of trust with it but also st a certain level of expectations. I am very certain that there will be some big title right away with an interesting spin and a new hardware that isn‘t just "more power".
Quangcute03
Rattata
- Pronouns
- He
About the SD vs Switch 2 handheld mode performance, I think it's not as simple as "SD is miles ahead of Switch 2 handheld mode" or "SD is better". It's complicated, and I think there are reasons that make the comparison much more muddy.
On the GPU side, assuming Sw2 clocked the GPU around 550MHz and higher, Sw2 GPU is gonna have a compute performance advantage against SD Van Gogh RDNA2 GPU. Not by much if it's around 550MHz, but the gap is gonna wider if Sw2 GPU clocked higher. However, SD narrow-but-fast design means that other part of SD GPU will be much more performant than Sw2 GPU equivalent part. Stuff like ROPs will see SD blows away Sw2 in pixel fillrate perf, as SD have 16 ROPs and Sw2 GPU also seems to have 16 ROPs. That extend to other fixed-function part of the GPU, like TMUs, as well as interconnects, etc... Moreover, Ampere INT32/FP32 + FP32 math pipe design means occupancy or in another words, shader ultilization will not reach the maximum throughput. All in all I do believe that SD will outperform Sw2 in terms of raw rasterization performance. BUT, that's not the end of it, as Sw2 GPU will have much better RT perf (though I don't know if RT perf is enough to be viable at all in handheld mode), better upscaling technology (DLSS3.x vs FSR3.x), and potentially more extra feature set that I'm not aware of. On top of that, we have to keep in mind that Sw2 will have a much more optimized software stack as well as close-to-the-metal API compared to SD which have to run it's game through a translation layer (or DX11/12 if we run Windows), which both have a sizable overhead penalty.
As for the CPU, I'm not sure. Maybe it's the same in terms of multicore performance. Its gonna boil down how the CPU cluster is being configured (for example if Nintendo decide to cheap out by cutting L2/L3 cache size; we simply don't know for sure at the moment), clock speed that the CPU core is running,...
Also if anyone else think T239 might have a SLC (System Level Cache) of some sort? Or it'll likely that Nintendo will cheap out on this?
On the GPU side, assuming Sw2 clocked the GPU around 550MHz and higher, Sw2 GPU is gonna have a compute performance advantage against SD Van Gogh RDNA2 GPU. Not by much if it's around 550MHz, but the gap is gonna wider if Sw2 GPU clocked higher. However, SD narrow-but-fast design means that other part of SD GPU will be much more performant than Sw2 GPU equivalent part. Stuff like ROPs will see SD blows away Sw2 in pixel fillrate perf, as SD have 16 ROPs and Sw2 GPU also seems to have 16 ROPs. That extend to other fixed-function part of the GPU, like TMUs, as well as interconnects, etc... Moreover, Ampere INT32/FP32 + FP32 math pipe design means occupancy or in another words, shader ultilization will not reach the maximum throughput. All in all I do believe that SD will outperform Sw2 in terms of raw rasterization performance. BUT, that's not the end of it, as Sw2 GPU will have much better RT perf (though I don't know if RT perf is enough to be viable at all in handheld mode), better upscaling technology (DLSS3.x vs FSR3.x), and potentially more extra feature set that I'm not aware of. On top of that, we have to keep in mind that Sw2 will have a much more optimized software stack as well as close-to-the-metal API compared to SD which have to run it's game through a translation layer (or DX11/12 if we run Windows), which both have a sizable overhead penalty.
As for the CPU, I'm not sure. Maybe it's the same in terms of multicore performance. Its gonna boil down how the CPU cluster is being configured (for example if Nintendo decide to cheap out by cutting L2/L3 cache size; we simply don't know for sure at the moment), clock speed that the CPU core is running,...
Also if anyone else think T239 might have a SLC (System Level Cache) of some sort? Or it'll likely that Nintendo will cheap out on this?
Last edited:
NintenDuvo
Bob-omb
- Pronouns
- He him
All anyone had to do was to google when was the 30th anniversary of super mario bros to see that happened when the Wii U was pretty much still "alive" also there was no way Nintendo let anyone with a camera near 100 feet of the prototype.
Me personally my patience for jokes is at an all time low because i need for Nintendo to acknowledge this device we been speculating for years already
Me personally my patience for jokes is at an all time low because i need for Nintendo to acknowledge this device we been speculating for years already
Had already replied to this once, but my thoughts dwelt on it some more. So far with the benchmarks, it's been purely on the CPU, and with regards to "which is stronger", Deck would beat out Drake at least in the single-core scenario, heavily because of that boost to 3.5Ghz. But this is a comparison based on benchmarking a single component, when the question was between Drake (portable) and Deck. So like what BreakAtmo was talking about, the Deck is unlikely to hit the peak of both CPU and GPU at the same time because of power draw limitations, whereas the clocks being thrown around for Drake portable are based closer to the general power draw the Switch had with the TX1 when portable. Technically, the hardware of Drake would surpass Deck if they were placed in the same situation, but even when they aren't, Deck has to juggle the performance of the components whereas for Drake, it likely to remain consistent.
Speaking of the Switch, its own portable mode had 3 GPU modes. 307.2Mhz, 384Mhz, and 460Mhz, none of which changed the CPU clock. The 550Mhz for Drake as far as I understand is more or less chosen because of it being relatively close to the theoretical peak efficiency clock, not its actual limit. Drake could have a similar arrangement for portable mode like Switch did, having a mode below and above this chosen clock (besides the clock for docked mode).
oldpuck
Like Like
- Pronouns
- he/they
Because we're looking at these chips backwards.
AMD chips can put more pixels on the screen for every core inside their GPU. Nvidia can put more cores into the same sized GPU. When you already know the core numbers, like we do for Steam Deck and Switch 2, that can make AMD seem like the better choice.
But that's backwards. Had Nintendo gone with AMD, they wouldn't necessarily have been able to go with that many cores - either they're too big, or too battery draining, or too expensive.
We can only guess at a possible AMD/Nintendo chip, because AMD doesn't have any ARM based chips on the market, and we don't know the internals of any deals that the two companies might have offered Nintendo. But AMD couldn't offer DLSS, advanced RT, or high quality backwards compat.
It's almost definitely true that those features highly outweigh any small raster advantage AMD could offer.
oldpuck
Like Like
- Pronouns
- he/they
Handy number from benchmarks, you're right that it's not the on-label spec. I went to benches first rather than double checking the spec sheet.
Semi Lazy Gamer
Like Like
Microsoft has been asking for more NPU for AI based tech from AMD, which is bottlenecking gaming based improvements.
forums.anandtech.com
Page 350 - Discussion - Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)
Page 350 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
forums.anandtech.com
EPD: Hold my beer.
The bolded actually makes me excited, because if any devs is going to make a game that's ridiculously optimised for Ampere, its some of Nintendos devs.
Last edited by a moderator:
