• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

KeplerL2: Valve is targeting 2028 for the Steam Deck 2

...so the SD2 and PS6 handheld are the same device...
The Wire Wow GIF
 
They'll likely use Zen 6 and RDNA 5 for Z3 Extreme, which would be a massive power boost from the OG Deck.
Z3 Extreme is Zen 6 + "RDNA 4m" (RDNA3.5+ RDNA4 ML) [Medusa Point 1, what else would it be?]
 
Last edited:
Knock off the trolling
The Steam Deck 2 feels completely unnecessary. By then, the upcoming PS handheld will likely dominate both in sales and technology. Valve is basically walking into a losing battle.

They'd honestly be better off canceling it. No one is going to need another PC handheld when there's a proper PS HH offering access to your entire PS library.
 
Last edited:
The Steam Deck 2 feels completely unnecessary. By then, the upcoming PS handheld will likely dominate both in sales and technology. Valve is basically walking into a losing battle.

They'd honestly be better off canceling it. No one is going to need another PC handheld when there's a proper PS HH offering access to your entire library.

PS doesn't include the "entire library" of Steam.
 
Yeah I might get the PS handheld, depending on their execution, but I definitely need a pc handheld. This generation, I really started building up my steam library

Same. I'd love a PS handheld, but I've got so much on Steam that I can't play on PS.
 
Sony locking cloud saves behind a PS+ membership that they keep jacking up the price on is another reason I'd rather pick up SD2.
 
The Steam Deck 2 feels completely unnecessary. By then, the upcoming PS handheld will likely dominate both in sales and technology. Valve is basically walking into a losing battle.

They'd honestly be better off canceling it. No one is going to need another PC handheld when there's a proper PS HH offering access to your entire PS library.
PSP and VITA disagree
 
The Steam Deck 2 feels completely unnecessary. By then, the upcoming PS handheld will likely dominate both in sales and technology. Valve is basically walking into a losing battle.

They'd honestly be better off canceling it. No one is going to need another PC handheld when there's a proper PS HH offering access to your entire PS library.

Satire right? Because that pretty much rings true only if you own a PS5 and not any other system. Otherwise, I doubt your PS library is the only thing that matters to you.

What about Nintendo games, their handheld is arguably more essential than this PS6 one, even disregarding the PC/Deck (which can run the Nintendo games, unlike the Sony handheld).

The whole dominating in sales & tech is the giveaway, because the two are historically almost never linked. The best tech rarely ever combines with the best sales. Unless maybe you're not counting the devices with more sales, like the Nintendo ones.
 
Last edited:
The Steam Deck 2 feels completely unnecessary. By then, the upcoming PS handheld will likely dominate both in sales and technology. Valve is basically walking into a losing battle.

They'd honestly be better off canceling it. No one is going to need another PC handheld when there's a proper PS HH offering access to your entire PS library.
Is it bait or are Sony fanboys really that limited in their believes?

Steam Deck is an open platform, the vast majority of Steam users don't give a fuck about Sony's closed ecosystem or their woke exclusives.
 
Yes that's the biggest problem right now for future handhelds. I am waiting to see handhelds with a panther lake chip with cut down cpu cores.
According to rumors "Medusa Halo Mini" will have 24 Rdna 5 CU's and presumably 12 zen6 cores. Maybe Valve can get a semi-custom solution again and cut that chip down by a third and build a handheld below 1K euros, but that's a big maybe
 
Z3 Extreme is Zen 6 + "RDNA 4m" (RDNA3.5+ RDNA4 ML) [Medusa Point 1, what else would it be?]
Isn't that a chip with 8 CU? Z2 Extreme has 16. I know AMD segmentation and product naming make about as much sense as an acid trip but so far they haven't shipped a sequentially named successor to a gaming product with considerably less performance than its predecessor.
 
According to rumors "Medusa Halo Mini" will have 24 Rdna 5 CU's and presumably 12 zen6 cores. Maybe Valve can get a semi-custom solution again and cut that chip down by a third and build a handheld below 1K euros, but that's a big maybe
It's not called that it's Medusa Premium (2028).
But those parts are not really Ryzen but more Radeon with CCD glued. Very good chance they get outright cancelled.

But really remember it's Medusa Premium the pricing doesn't lend itself to a steam deck 2.

I don't know why it's hard to read between the lines of what Kepler is saying. So I will say it.

Steam Deck 2 is 99.9% a medusa point derivative with RDNA3.5 (+RDNA4 ML).
 
Last edited:
It's not called that it's Medusa Premium (2028).
But those parts are not really Ryzen but more Radeon with CCD glued. Very good chance they get outright cancelled.

But really remember it's Medusa Premium the pricing doesn't lend itself to a steam deck 2.

I don't know why it's hard to read between the lines of what Kepler is saying. So I will say it.

Steam Deck 2 is 99.9% a medusa point derivative with RDNA3.5 (+RDNA4 ML).
I guess I didn't want to believe they would make us wait 6+ years for a machine with rdna 3.5 but I needed to come back to earth and see the writing on the wall i.e the Steam Machine anemic specs
 
I guess I didn't want to believe they would make us wait 6+ years for a machine with rdna 3.5 but I needed to come back to earth and see the writing on the wall i.e the Steam Machine anemic specs
They didn't want to. But that's all AMD has. Valve can't guarantee big enough volume to make a chip from scratch.

Their major issue is their blanket unwillingness to work with Nvidia lmao. And it's also unlikely that a company that has been so hostile to GeForce would get the Nintendo treatment. They are stuck with AMD whose focus is on datacenter and whose consumer product line moto is basically "minimal viable product".
 
They didn't want to. But that's all AMD has. Valve can't guarantee big enough volume to make a chip from scratch.

Their major issue is their blanket unwillingness to work with Nvidia lmao. And it's also unlikely that a company that has been so hostile to GeForce would get the Nintendo treatment. They are stuck with AMD whose focus is on datacenter and whose consumer product line moto is basically "minimal viable product".

Nvidia would provide them a chip? When are the Intel + Nvidia APUs coming out. Those could be interesting products for handhelds.
 
Nvidia would provide them a chip? When are the Intel + Nvidia APUs coming out
Late 2028 lol. It's a titan lake Variant.

Nvidia has N1 | N1X out soon (18 months late, were supposed to launch with Blackwell)

Also N2 | N2X (supposed to launch Late 27 with Rubin Gaming).
 
Late 2028 lol. It's a titan lake Variant.

Nvidia has N1 | N1X out soon (18 months late, were supposed to launch with Blackwell)

Also N2 | N2X (supposed to launch Late 27 with Rubin Gaming).

Are those x86 parts? Not interested in ARM (compatibility).
 
Are those x86 parts? Not interested in ARM (compatibility).
N1 | N1X | N2 | N2X are ARM.

Nvidia makes GPU Tile and MediaTek does the CPU tile with some borrowed Nvidia IP (this arrangement turned out to be a disaster for N1X).
 
Last edited:
Specifically for I was coping for Nvidia N1 to be chosen for Steam Deck 2.

Basically Small ARM CPU + RTX 5030 iGPU 🔥 on TSMC N3E.

PlayStation handheld would have been embarrassed. Alas.

I think a Nvidia iGPU + Intel x86 CPU would be highly desired in the handheld space. Asus and Lenovo could incorporate them in their Windows handhelds. Xbox Ally X2?
 
They didn't want to. But that's all AMD has. Valve can't guarantee big enough volume to make a chip from scratch.

Their major issue is their blanket unwillingness to work with Nvidia lmao. And it's also unlikely that a company that has been so hostile to GeForce would get the Nintendo treatment. They are stuck with AMD whose focus is on datacenter and whose consumer product line moto is basically "minimal viable product".
Z2 Extreme is around 90% faster than van gogh at 35 W using 16 rdna 3.5 CU. It sounds like the Steam Deck 2 will be quite disappointing performance wise. If Asus, Lenovo or whoever slaps Medusa Premium into a chassi they'll embarrass Valve even if the cost is 2-3 times higher. One could hope for an official SteamOS version with that chipset
 
I think a Nvidia iGPU + Intel x86 CPU would be highly desired in the handheld space. Asus and Lenovo could incorporate them in their Windows handhelds. Xbox Ally X2?
A slimmer of hope is that there are some vague rumors of Xbox preparing a handheld for their next official hardware lineup, with a custom chip this time
 
Z2 Extreme is around 90% faster than van gogh at 35 W using 16 rdna 3.5 CU. It sounds like the Steam Deck 2 will be quite disappointing performance wise. If Asus, Lenovo or whoever slaps Medusa Premium into a chassi they'll embarrass Valve even if the cost is 2-3 times higher. One could hope for an official SteamOS version with that chipset

Yeah, but Valve really hasn't shown any interest in winning the battle in power. They have hit the sweet spot in price and battery life with the first SD, which is why it has sold so well. I expect they will do the same with SD 2. Any handheld can run Steam OS so doesn't really make much difference.
 
Last edited:
Yeah, but Valve really hasn't shown any interest in winning the battle in power. They have hit the sweet spot in price and battery life with the first SD, which is why it has sold so well. I expect they will do the same with SD 2. Any handheld can run Steam OS so doesn't really make much difference.

Steam Deck got popular because it smacked competing devices in terms of performance/watt/price balance, including Nintendo's Switch 1.
 
Steam Deck got popular because it smacked competing devices in terms of performance/watt/price balance, including Nintendo's Switch 1.

There really weren't any mainstream devices when Steam Deck launched. Just the direct stuff out of China like Onexplayer and GPD. But yeah, I agree the overall balance of the device is what made it obtain wider appeal. Asus and Lenovo have provided solid options though. Competition has been good.
 
There really weren't any mainstream devices when Steam Deck launched. Just the direct stuff out of China like Onexplayer and GPD. But yeah, I agree the overall balance of the device is what made it obtain wider appeal. Asus and Lenovo have provided solid options though. Competition has been good.
I was thinking about this the other day and was like, what if Valve ordered a chip from AMD that kept the same core count for the cpu and gpu, but updated it to Zen 5 and RDNA 3.5. Would a "custom chip" like that really be that expensive to design?

A big part of the steam deck APU's efficiency over Z1 and Z2 extreme is because the CPU core count is low compared to the Z1 and Z2 extreme.
 
Yeah, but Valve really hasn't shown any interest in winning the battle in power. They have hit the sweet spot in price and battery life with the first SD, which is why it has sold so well. I expect they will do the same with SD 2. Any handheld can run Steam OS so doesn't really make much difference.
They've been talking about a "generational leap". If that turns out to be a 60% performance bump I think people will be disappointed. Many current games will simply not be playable in any decent state at that performance level, never mind next gen games.
 
They've been talking about a "generational leap". If that turns out to be a 60% performance bump I think people will be disappointed. Many current games will simply not be playable in any decent state at that performance level, never mind next gen games.

A lot is going to depend on how things shake out in the market. Those powerful handhelds are now over $1k. I'd be more disappointed in a high priced Steam Deck 2 than a handheld that doesn't stand toe to toe with the most powerful on the market. But we will see.
 
I'm sure Nividia were ecstatic to offload some ancient T239s on 8nm.
There is no offloading. T239 was made for Nintendo. It could have been fabricated on a TSMC EUV Node, Samsung EUV node, etc.

They chose SEC 7LPH | 8NP because it was the cheapest node they can use. It's Nintendo.
 
Last edited:
Valve seems to be ok with not going bleeding edge tech wise judging from what steam machine specs were to be, also they probably don't want to go the route of having a handheld in the 1k range. So it wouldn't be surprising if steamdeck 2 is not the most powerful handheld device.
 
Last edited:
There is no offloading. T239 was made for Nintendo. It could have been fabricated on a TSMC EUV Node, Samsung EUV node, etc.

They chose SEC 7LPH | 8NP because it was the cheapest node they can use. It's Nintendo.
Its a cut-down derivative of the T234 automotive chip and a relatively ancient fossil from back in 2021....

 
There is no offloading. T239 was made for Nintendo. It could have been fabricated on a TSMC EUV Node, Samsung EUV node, etc.

They chose SEC 7LPH | 8NP because it was the cheapest node they can use. It's Nintendo.
Why would nvidia bother making a custom chip specifically for Nintendo and then selling it at cost? What do they stand to gain from this?
 
How powerful is the Medusa chip compared to AI Max +395?

I see 5070 vs 4060 thrown around. Who knows, maybe 50% best case? It's supposed to be targeting the same TDP but with better specs (also keep in mind these TDPs are higher than what normal handhelds run - they target 60-80W and can scale up to 160W - handhelds normally run around 20-40W and in Deck 2's case probably closer to the 20W). Which would be pretty amazing, but the bigger question is at what cost. I don't know if a $1500 base model is really going to convince that many people to jump on board, with the way prices are going.

Really cost and actual availability are the biggest issues. We're talking about 2 years out still, and who knows if it'll be delayed or not. Same with Deck 2. I see the thread is putting 2028, but could easily be 2029 or 2030 if the prices aren't aligned with the number Valve wants to release at.

If the Deck 2 BOM in 2028 is $1200 and Valve wants to release at $600, then they wait. Maybe 2029 BOM is $1500 and 2030 BOM is $1900, and we don't see Deck 2 for another 10 years when prices actually start going down again. Valve has more patience than you or I do, guaranteed. They win the waiting game. Only way to lose is actually believe they will release it in 2028 until they themselves say they will.
 
Last edited:
Why would nvidia bother making a custom chip specifically for Nintendo and then selling it at cost? What do they stand to gain from this?
Getting their technologies and hardware widely used and to counterbalance AMD running to console after getting crushed on PC. The first partnership was more a coincidence (tegra X1 flopped), but Switch1 sold like wildfire. And Nvidia has deemed it a worthwhile partnership to continue.

For the record, AMD also defacto sells into consoles at cost. Had Nvidia demanded their usual juicy margins, Nintendo is under no obligation to continue using Tegra.

It's also similar to T234 that they already had planned to make for Tegra. It's a very low cost effort.
 
What Medusa Chip?

Steam Deck 2? 25% of the GPU Performance at most.
Medusa Point 1? 35%
Etc

Oh Medusa Point - then yes, I was thinking Medusa Halo vs Strix Halo, so no increases at all for Medusa Point I'd say.

It looks like Medusa Point will be worse than Strix Halo, largely because of the TDP drop, and also because it has less cores and 1/4th the GPU compute units (8-12 vs 40 in Strix Halo).

I don't see any world where a Ryzen AI Max+ 395 running at 100W+ is beaten by a Medusa Point running at 30W, with a fraction of the resources the Strix Halo has - I totally missed the comparison was putting the high-end Strix APU meant to compete with dGPUs vs a mainstream APU meant for handhelds and ultrabooks. Versus a Medusa Halo, then yes, there's a big leap, but will any handhelds see the Medusa Halo? Certainly not the Deck 2 I'd say.
 
Last edited:
They were considering Qualcomm which is even more 🤮

Bruh...

How the hell do you run your existing library, with an architecture jump, especially modern games? Brute force x86 emulation on arm? I know Windows emulation has come so far these days on arm devices, but those phones are flagships and we can bet our bottom dollar that Valve won't go for high end Snapdragon soc's.

Is their recent work on Proton, adding the FEX compatibility layer for arm64, related to this?
 
Last edited:
Top Bottom