News Valve: No Performance Upgrades for the Next-Gen Steam Deck

[Admin]

Valve seeks to improve display and battery life of Steam Deck, but not performance.

Valve: No Performance Upgrades for the Next-Gen Steam Deck : Read more

 

[Elusive Ruse]

I think they have their priorities straight, a handheld gaming device's battery needs to be top notch. What's the point of a portable gamer if it needs to be plugged in most of the time?

 

[gman68]

I think they have their priorities straight, a handheld gaming device's battery needs to be top notch. What's the point of a portable gamer if it needs to be plugged in most of the time?

I agree. Also, if you are gaming at home and you really want to use the steam deck, you can stream from a more powerful desktop or gaming laptop.

 

[thestryker]

I'm confused by them saying this because AMD's newer architectures (Zen 3/4) are much higher performing and much more efficient. I certainly understand them not changing the GPU/RAM situation, but if they were to leave the CPU side alone it would be for pure profit reasons and nothing else.

 

[prtskg]

I'm confused by them saying this because AMD's newer architectures (Zen 3/4) are much higher performing and much more efficient. I certainly understand them not changing the GPU/RAM situation, but if they were to leave the CPU side alone it would be for pure profit reasons and nothing else.

GPU has changed by one generation only, rdna3 from rdna2 and CPU change is not that important for gaming, so it's understandable that they're not changing performance.

 

[The Historical Fidelity]

I'm confused by them saying this because AMD's newer architectures (Zen 3/4) are much higher performing and much more efficient. I certainly understand them not changing the GPU/RAM situation, but if they were to leave the CPU side alone it would be for pure profit reasons and nothing else.

 

[TerryLaze]

but if they were to leave the CPU side alone it would be for pure profit reasons and nothing else.

They had a pre order cost of $5 and the decks are so cheap that they probably don't even make any money from them directly.
Keeping the cost down will be great for users that don't want to be paying nvidia money for a deck.

 

[thestryker]

GPU has changed by one generation only, rdna3 from rdna2 and CPU change is not that important for gaming, so it's understandable that they're not changing performance.

In the case of the Steam Deck the CPU side is very important given the limited power profile of the device. The current CPU cores in it are the weakest point of the device and moving to a newer architecture would both increase performance and allow it to be more efficient (they don't need to increase the number of cores just the type). If you need an example of the type of performance difference possible just look at the 4700G/4750G vs 5600G (the latter has fewer CPU/GPU cores, and even the 5300G can periodically beat the 4700G/4750G).

They had a pre order cost of $5 and the decks are so cheap that they probably don't even make any money from them directly.
Keeping the cost down will be great for users that don't want to be paying nvidia money for a deck.

Valve almost certainly isn't losing money on the Steam Deck, but they may not be making a lot of profit (I'd assume the profits are mostly on the higher models). Keeping the existing APU as it is or with a node shrink would be a way to save money on development so they don't have to make it back with a new Steam Deck. I'm not suggesting they should use a 6xxx series APU in these as that'd be a terrible choice price/perf/efficiency wise.

If they were really worried about battery life though they'd be looking to remake the APU with Zen 3/3+ based cores as that could have as much of an impact as a larger battery/node shrink. This shouldn't cost them anywhere near as much as the original APU did development wise (if they leave GPU/Memory configuration), but the result would be higher performance with a lower power envelope.

 

[ezst036]

Valve has the correct priority, I think. Mobile is mobile, so it needs to be mobile. That would make battery life the obvious target. Keeping it constantly plugged in is counterintuitive.

Since Steam Deck uses the Linux-based SteamOS, Valve doesnt have to give up on the home gamer. They just have to release an ISO and leverage the flexibility that comes standard with the Linux platform. Now all PCs are potential Steam boxes.

An Aerith with a die shrink and gobs of X3D gaming cache would make a lot of sense.

 

[kyzarvs]

In the case of the Steam Deck the CPU side is very important given the limited power profile of the device. The current CPU cores in it are the weakest point of the device and moving to a newer architecture would both increase performance and allow it to be more efficient (they don't need to increase the number of cores just the type). If you need an example of the type of performance difference possible just look at the 4700G/4750G vs 5600G (the latter has fewer CPU/GPU cores, and even the 5300G can periodically beat the 4700G/4750G).



Valve almost certainly isn't losing money on the Steam Deck, but they may not be making a lot of profit (I'd assume the profits are mostly on the higher models). Keeping the existing APU as it is or with a node shrink would be a way to save money on development so they don't have to make it back with a new Steam Deck. I'm not suggesting they should use a 6xxx series APU in these as that'd be a terrible choice price/perf/efficiency wise.

If they were really worried about battery life though they'd be looking to remake the APU with Zen 3/3+ based cores as that could have as much of an impact as a larger battery/node shrink. This shouldn't cost them anywhere near as much as the original APU did development wise (if they leave GPU/Memory configuration), but the result would be higher performance with a lower power envelope.

But as the article states, this would mean 2x compatibility programs, 2x architectures for developers to support etc etc. If keeping architectures works for the market leader (Nintendo), I think from a sales point of view it may well be the way to go.

 

[TerryLaze]

But as the article states, this would mean 2x compatibility programs, 2x architectures for developers to support etc etc. If keeping architectures works for the market leader (Nintendo), I think from a sales point of view it may well be the way to go.

Huh?! The what now?!
We are talking about Steam and PC games. There are dozens of "compatibility" and "architectures" they have to design for anyway, not that they ever do any optimization but at least in theory.
AMD makes linux drivers anyway and even if they wouldn't steam would only have to make drivers.
The only additional thing for the deck would be for everything to be controllable by the controller and for the writing to be big enough to be visible, both of those don't change no matter of the CPU changes or not.

 

[-Fran-]

Well, if they swap the APU for an equivalent one at the same TDP, it shouldn't be a problem at a high level, but the Deck is so well integrated that it may need additional consideration. Not just a cost (of the new APU) perspective, but also software support (Kernel) for it. It would be a good idea to make the APU more efficient though. Maybe keep the same overall performance envelope, but reduce the power even further; this is what consoles usually do. So, in that sense, there is a case to be made for a new APU in the Deck, but not necessarily more performance. Also, on this point, the performance of the Deck is fine for now. More is always nice, but more battery life and better screen are definitely the good priorities.

EDIT: Another thought... I wonder if they'll make the new screen compatible with the current Deck. Same with the battery.

Regards.

 

[thestryker]

Well, if they swap the APU for an equivalent one at the same TDP, it shouldn't be a problem at a high level, but the Deck is so well integrated that it may need additional consideration. Not just a cost (of the new APU) perspective, but also software support (Kernel) for it. It would be a good idea to make the APU more efficient though. Maybe keep the same overall performance envelope, but reduce the power even further; this is what consoles usually do. So, in that sense, there is a case to be made for a new APU in the Deck, but not necessarily more performance. Also, on this point, the performance of the Deck is fine for now. More is always nice, but more battery life and better screen are definitely the good priorities.

Zen 2 to Zen 3 is around 20% minimum increase IPC wise so I can't imagine not getting higher performance along with better efficiency. I suppose they could drop the clockspeeds compared to where they currently are, but I'm not sure how much would really be gained there.

EDIT: Another thought... I wonder if they'll make the new screen compatible with the current Deck. Same with the battery.

Regards.

I could see the screen potentially being compatible, but I think the battery might be harder for them to swing if they're increasing capacity.

 

[SethNW]

I am with them on that one. Like Steam Deck really does need better battery life abd some other tweaks to it way more than better performance at cost of lower battery life. Sure eventually they will have to go to new SoC, but I think for performance target, they are doing fine. Plus tweaks here could do wonders for future, when they actually do get new SoC. Plus as mentioned, it is more user friendly if they stick to same performance.

As for Switch, I don't think it compares as well, since it is closed ecosystem and Nintendo could mandate compatibility with original, bit like with PS4 Pro didn't make PS4 obsolete. Only game that actually requires it is CP2077, die to poor performance. So Nintendo could have easily pulled it off. Unlike Valve, who doesn't have as much control over released games. Sure they could mandate it on Steam, but PC is open platform, there are other stores you could release on.

Still, I think Valve has right idea here for their handheld, it is first generation and naturally that they couldn't hit everything perfectly, so improving on flaws with major revision update is right way to go here. Since performance wise, handhelds will always need compromises anyway, so you might as well focus on what it makes portable instead of what would further tie you to power socket