News Valve confirms the Steam Deck won't have annual releases — Steam Deck 2 on hold until a generational leap in compute performance takes place

[Admin]

A Steam Deck designer has confirmed that Valve will not follow an yearly release cycle for the Steam Deck.

Valve confirms the Steam Deck won't have annual releases — Steam Deck 2 on hold until a generational leap in compute performance takes place : Read more

 

[hotaru251]

ngl I wish everyone had this mentality.

there doesnt need to be a sequel until its a noticeable improvement.

 

[Pierce2623]

I just want the ARM version of Proton. X86 handhelds still suck because of how bulky they are. I mean, don’t get me wrong, I’ve got a Steam Deck. It just sucks.

 

[doughillman]

I just want the ARM version of Proton. X86 handhelds still suck because of how bulky they are. I mean, don’t get me wrong, I’ve got a Steam Deck. It just sucks.

🤡

 

[Notton]

Steamdeck (LCD) launched on Feb 22', so 2yrs have passed.
The APU used was Zen2 + RDNA2, so there's been substantial improvements there too.

But I can totally understand waiting for a cheaper variant of Zen5 + RDNA3.5, or the Z2 chip, to ship.

 

[Pierce2623]

🤡

The clown emoji is only suitable for those that have convinced themselves x86 handhelds are already good. I’ve got a Deck and a Ally both just collecting dust.

 

[Giroro]

FINALLY! Valve finally had the courage to confirm the Steam deck didn't get an annual update when it was a year old ... 2 years ago.

 

[thestryker]

Nothing off the shelf can compete, but Van Gogh wasn't off the shelf either. I have a hard time believing something based on Zen 4c/5c with RDNA 3.5 wouldn't be a lot higher performance especially if they added on some cache for the GPU. Power consumption for quad channel might be too high, but that would be another potential lever. Whether there's a new Steam Deck or not next year ought to be interesting for handhelds with Lunar Lake and Zen 5 based devices.

 

[Giroro]

Really though, I don't care too much if Valve makes the Steam Deck more powerful, but I very much would like it to be smaller.

 

[palladin9479]

Hmm we had this debate already, ISA only really matters at super low power. Well more like sub 500mhz or so. When we approach 1ghz (one billion operations per second) every scalar processor hits the same I/O issues and requires the same solutions. Those solutions are complex and erase all ISA distinctions.

 

[cryoburner]

ngl I wish everyone had this mentality.

there doesnt need to be a sequel until its a noticeable improvement.

While they might not 'need' to frequently release significantly improved versions of the hardware, unlike with locked-down console ecosystems, any company can release their own competing PC handheld. So after a few years, the hardware is inevitably going to start looking a bit dated compared to newer products from competitors. Unlike with the Nintendo Switch, they can't go anywhere close to 8 years before a successor to the Steam Deck is released. Though, technically, I suppose Valve could just discontinue the product line and never release a successor like they have had a tendency of doing with all their other hardware. : P

There are some rumors about a potential Index 2 in the works, though unless they make the cost far more palatable, it's unlikely to do much to revitalize the stagnating PC VR ecosystem.

I just want the ARM version of Proton. X86 handhelds still suck because of how bulky they are. I mean, don’t get me wrong, I’ve got a Steam Deck. It just sucks.

The problem with that idea is that the PC games you would be running on it are designed for x86, and would in many cases see a notable hit to performance with the code being translated to run on ARM architecture. On a higher-powered desktop or laptop you might have performance to spare to allow for that sort of emulation, but on a handheld you are already being limited by the hardware even for native code, so the hardware would need to be overbuilt to counteract the performance loss, and would see a reduction in efficiency as well, probably countering any potential gains that the alternate architecture might have to offer.

 

[Makaveli]

Probably see a release based on Zen 6 and RDNA 5 IP

 

[palladin9479]

So the thing with these sorts of custom bespoke solutions is they require heavy investment in design and testing. This isn't like PC building where you can put arbitrary parts together and expect it to work, they need to design new boards, firmware, drivers then certify it all as not gonna break twelve months after purchase.

What they are saying is that the performance benefit of going to a newer design does not out weight the upfront design costs. They are waiting for a bigger generational leap in performance-per-watt before paying those costs again.

 

[John Kiser]

I just want the ARM version of Proton. X86 handhelds still suck because of how bulky they are. I mean, don’t get me wrong, I’ve got a Steam Deck. It just sucks.

All well and good until you realize that they'd probably still need to draw quite a bit of power to compete at nearly the same level.

The problem saying you want arm is that the power envelope of arm isn't that much smaller than these handhelds actually are. The snapdragon 8 gen 4 can draw over 20 watts of power on its own and you need to consider thst the current gen in these things is about a 30 watt powder draw on some of rhwaw with a peak of 54 watts. They don't draw as much power as you seem to think.

The snapdragon in question draws over 20w and gets exceedingly hot if you let it run which means building the same bulky solutions which exist for cooling and putting controls into the devoces...

 

[usertests]

Nothing off the shelf can compete, but Van Gogh wasn't off the shelf either. I have a hard time believing something based on Zen 4c/5c with RDNA 3.5 wouldn't be a lot higher performance especially if they added on some cache for the GPU. Power consumption for quad channel might be too high, but that would be another potential lever. Whether there's a new Steam Deck or not next year ought to be interesting for handhelds with Lunar Lake and Zen 5 based devices.

Exactly this. Nothing off-the-shelf for the Zen 5 generation seems likely for Steam Deck 2. Kraken Point has an interesting 4+4 CPU but the 8 CUs RDNA3.5 isn't enough over 8 CUs RDNA2. Maybe top Strix Point / Z2 Extreme with 16 CUs isn't enough either if Valve wants to triple/quadruple the performance at the same power to try to turn 720p30 (minimum) gaming into 1080p60. 8-core Strix Halo gets the 256-bit memory controller and great iGPU, but it could be too expensive and power-hungry.

I think Valve would want only Zen 5c/6c cores as the low-power follow up to quad-core Zen 2. I'm not sure about the iGPU because there are many levers they could pull. There may need to be an NPU for FSR4, if that ends up being one way FSR4 can be run.

 

[thestryker]

Exactly this. Nothing off-the-shelf for the Zen 5 generation seems likely for Steam Deck 2. Kraken Point has an interesting 4+4 CPU but the 8 CUs RDNA3.5 isn't enough over 8 CUs RDNA2. Maybe top Strix Point / Z2 Extreme with 16 CUs isn't enough either if Valve wants to triple/quadruple the performance at the same power to try to turn 720p30 (minimum) gaming into 1080p60. 8-core Strix Halo gets the 256-bit memory controller and great iGPU, but it could be too expensive and power-hungry.

I think Valve would want only Zen 5c/6c cores as the low-power follow up to quad-core Zen 2. I'm not sure about the iGPU because there are many levers they could pull. There may need to be an NPU for FSR4, if that ends up being one way FSR4 can be run.

Yeah I think Valve would have to be looking at a fully custom SoC with what's on the market now to make a viable successor. If the cache doesn't make a big enough difference I figured that the Zen 5c cores would be preferable due to lowering SoC size and being more efficient in their clock range. All of AMD's off the shelf SoCs with good graphics configurations have way too much compute for a power limited handheld device.

 

[williamcll]

Yeah I think Valve would have to be looking at a fully custom SoC with what's on the market now to make a viable successor. If the cache doesn't make a big enough difference I figured that the Zen 5c cores would be preferable due to lowering SoC size and being more efficient in their clock range. All of AMD's off the shelf SoCs with good graphics configurations have way too much compute for a power limited handheld device.

That seems like a massive compatibility issue.

 

[thestryker]

That seems like a massive compatibility issue.

What are you referring to here?

AMD's Zen 4c/5c cores are identical in design to their larger counterparts, but have half the L3/core. If you're talking about a custom SoC that's exactly what Van Gogh (the Steam Deck SoC) already is so it's very much not a problem.