Specs for a hypothetical Xbox portable

I reckon they’ll add the s/x ssd card expansion slots to the dock, along with power and sync button.

399 for handheld which is a refined series S with most of the bottlenecks removed. 199 for an external dock which has a beefier GPU and cooling support for the main unit.

And sell a system for 499 which is just a traditional console using the chipset of Handheld and Dock in one box unit.

1.5 Chips - 2 SKUs - 3 Form factor

Another approach could be to design a beast of a handheld by putting all the power in the handheld. But downclock and limit other aspects in handheld mode to conserve power. Then let the beast unleash in docked mode by overclocking the hell out of it and removing all the power and thermal limitations using an innovative dock.

1 chip - 2 SKUs - 3 Form factor

In both cases, starting from 399 and going upto 599 with all the components.

The external dock needs PCIE 6.0 x 16 connection to the handheld, 8GB of GDDR7, along with a beefy GPU. That’ll mostly likely cost over $299 if not $399.

The beast of a handheld might not work due to ram bw limitations, but they can add 128mb of infinity cache to address that. I wonder if they can just shut off half the handheld’s GPU in portable mode.

Dock mode these days is possible with thunderbolt 4.0 support via USB 4. Don’t know why would they need to go to Pcie 6.0 which has not been used yet in any console or gaming PC.

The next gen might move to pcie 5 and external GPU connection won’t be as expensive as you suggest.

Edit: rtx 4060 which is better then series X only use half of the pcie 4.0 bandwidth for it’s operations for example. People can literally add ssds on there rtx 4060 by utilising the other half of the pcie 4.0 bandwidth.

These low to mid range systems does not need much gpu bandwidth.

PCIE 5.0 is available today and PCIE 6.0 hardware is available this year.

I want the connection to be as fast as possible to make the docked experience as close to the standalone console as possible for the developer.

Also the docked GPU is a high end system. Should be at least 2x the handheld.


What about this.

The portable uses HBM3 which solves the bandwidth issue completely. HBM3 is also lower power and has low physical footprint. During portable mode all the cpu, gpu, memory clocks are aggressively lowered, and majority/half of the shader-engines+ROPs get switched off.

When docked, the GPU gets 3-8x boost. The CPU gets a 50% boost. The ram bandwidth gets a 2-3x boost.

Not sure about HBM3. It is the perfect memory system for performance and efficiency. But it’s cost and yield makes it too expensive. It is high tech product. Don’t think anyone will invest in it to make a loss generating device. But you never know.

The downclock and boost up strategy is the way to go. Some handhelds even function in this way already.

I have already proposed a base series S Tflops machine but refined one with newer IP blocks and better memory support as a handheld. And then overclock the hell out off it while in docked mode where dock has an innovative cooling solution.

RDNA 4 is expected to have game clocks of upto 3.2 Ghz. Simply cut the clocks by half and cut the CUs by one-third and you may have a power efficient machine ready in your hand. In docked mode it will have 60CUs @3. 1+Ghz with RDNA5 architecture. In handheld it will have 20CUs @1.55+Ghz with RDNA5 architecture.

Since it is a hybrid, physical memory will be the same. Depending upon the requirement, switch off and on memory chips to have desired bandwidth.

I don’t care about battery life much. 1 hour is sufficient for me. If I am playing more then that then it means I am not moving and can plug it in.

1 Like

I suppose by 2026, on 3nm or 2nm they can fit a Octocore Zen or a 16 core ARM + 24 WGP or equivalent + 128mb of infinity cache in under 200mm.

128 bit bus LPDDR6 9600mhz for ~156 GB/s bandwidth.

2WGP are disabled for yield.

Peak compute when docked would be around ~34 TFs FP32 at ~3000mhz. Portable compute might reach ~11TFs PF32 at ~2000mhz. For reference, ROG Ally does something like ~6TF FP32 sustained at 30 watt APU TDP. Keep in mind these are doubled FP32 throughput introduced in RDNA3.

Portable infinity cache would be clocked at 2000mhz and maybe half it gets disabled too.

Real world perf is 50% over a S and X depending on not docked vs docked.

Infinity cache allows for scaling BW without investing in HBM or a 512bit bus.

Last week on the podcast they were talking about the handheld haveing the same specs as the console and just being more expensive. I wonder how powerful that can be in a few years? I think if they can make a handheld stronger than the one ps makes they can get a good amount of marketshare back from switch and ps

I don’t think that rumor is accurate.

I think Xbox will always struggle against the juggernauts that are Nintendo and PlayStation. The brand recognition is simply not the same and it will take a long time to get it there.

But that’s why putting Xbox on every screen is their goal and makes sense. That’s how they can grow. Trying to win over old-man console warriors is futile. They need to appeal to the kids who are just getting into gaming.

A “Game Pass handheld” would certainly help, if they market it effectively. So would growth on mobile and tablet devices, which is where most kids play these days. Xbox will keep casting their net as wide as possible while Nintendo and PS are still focusing on their consoles for another generation or two.

My opinion is that the Xbox handheld and the Switch handheld will not close competitors due to library, OS, specs, ergonomics, and expectations.

The only threat to a Xbox handheld is really a PS one, and Sony doesn’t have good options unless they do something that’s at best in between PS4 and PS5. That’ll have PS4 BC and zero compatibility for PS5 games. Studios will have to downgrade their games to run on the PS portable and such a task will be inherently more difficult than a upgrade. I foresee such upgrades only happening for recent or new AAA games. This will also introduce a third developer target after the PS5 and the PS5 pro.

It’s likely still something worth doing for them.

Specs wise the bandwidth and the number of compute units for the PS4 is really close to the series s. A PS portable would not be arriving earlier than a series s level portable based on the available tech.

If the Xbox portable uses ARM CPU cores, the Xbox handheld should be able to have a better battery life.

Whether or not the Xbox handheld being a better handheld by key metrics (compatibility and battery life) translates into better sales. Who knows.

1 Like

I feel like MS has the upper hand here. I somehow highly doubt that by even 2028 there will be a handheld with enough power to reproduce either PS5 or Xbox Series X in portable form. That means that even if Sony wanted to release a Portable PS6, I don’t know how they’d be able to have specs that could run PS5 games natively and it would be next to impossible to have all games ported over, which means they’d lose BC. MS already has the Series S, which means that as long as they hit that performance target, all next gen games and everything so ce the first Xbox will work on their handheld.

The more I think about it, the more I think the handheld has potential, especially with ABK added into the mix and Gamepass. If marketed properly and at a decent price, this thing could be huge for Xbox!


Exactly Xbox would have the high performance portable console market to itself.

1 Like