GPD Win (x86 Computer / Palmtop)


I think as I said a long while back, there are way to many players in the market.

In the handheld SoC market? There are effectively two: ARM and Intel.

OK, so there's technically MIPS, AMD, and VIA, but they're so far out of the running that they don't really count as players. All the dozens of ARM chipmakers just license the cores from ARM (plus GPU IP owners), and they can come and go as they please without much overall effect on the market or on technological advancement. With intel effectively abandoning the handheld space, there is really only ARM left. We have to hope that either intel changes their mind, or that AMD (or VIA, ha!) actually defies the odds and can get a 2W part to market sometime soon.
 
Last edited:
ARM as a company represents tons of competition within that space as they only design and license, they're basically all R&D as a product, where Intel both designs and manufactures.

I'm actually happy this is the situation for ARM, and sort of wish the x86 market was as diverse. A manufacturer can scale its production on their target audience. Intel has to gauge the entire market rather than just segments like ARM manufacturers can do.

Competition breeds innovation. Intel in some regards are just off in left field looking at themselves and trying to figure out why not as many people like them even though they bring a lot to the table.
 
All the dozens of ARM chipmakers just license the cores from ARM (plus GPU IP owners)

Not true for SoCs from Samsung, Qualcomm, Apple nVidia, and some others that use custom ARM cores. And I think you underplay the effort needed to make an SoC that uses licensed a CPU and GPU, there is still much else to be added and much work in the physical implementation is needed to make it not suck. For example of a big company that failed to make the ARM cores not suck, see Snapdragon 810.

The effort was big enough that a lot of players like TI and ST-Ericsson had to drop out of doing mobile oriented SoCs because they couldn't keep up.
 
The effort was big enough that a lot of players like TI and ST-Ericsson had to drop out of doing mobile oriented SoCs because they couldn't keep up.

Maybe because it was a lot of effort for very little improvement over simply throwing cores at a wafer and seeing what sticks. TI and Ericsson spent a ton of money developing the equivalent of a Merc S-class and charged S-class money for them. But most people are more than happy with a Toyota Corolla. Chinese chipmongers like allwinner and rockchip license cores, assemble an SoC with minimal fuss, and sell them for a couple bucks each. Sure, they're not as fast or efficient as a heavily engineered chip, but they're good enough for most applications.
 
Was there even plans for an x86 Pyra module? My guess Intel dumping the Atom line would be a bigger influence.
Not a concrete plan but it was mentioned multiple times that this possibility is one of the upsides of the modular designs.
Being able to play all my DRM-free Linux desktop ports on a secondary (x86) Pyra in another color some years down the road is a nice thought although not really neccessary.
Having more native ARM ports would be the best scenario of course but is only realistic with open sourced games, I guess.
 
Maybe because it was a lot of effort for very little improvement over simply throwing cores at a wafer and seeing what sticks. TI and Ericsson spent a ton of money developing the equivalent of a Merc S-class and charged S-class money for them. But most people are more than happy with a Toyota Corolla. Chinese chipmongers like allwinner and rockchip license cores, assemble an SoC with minimal fuss, and sell them for a couple bucks each. Sure, they're not as fast or efficient as a heavily engineered chip, but they're good enough for most applications.

But Allwinner and Rockchip are not dominating the phone market. They grabbed a big chunk of the tablet market a couple years ago, but that market has been in freefall since.

TI, ST-Ericsson and others (to a large extent nVidia) folded out of mobile largely because they couldn't keep up with Samsung, Qualcomm and MediaTek in the phone space. Not so much because of Allwinner and Rockchip. Their lack of modem integration also keeps them from really competing.
 
(If ST-Ericsson fell out on lack of modem integration, then that's hilarious, since wireless comm is pretty much all that's keeping Ericsson alive nowadays)
 
(If ST-Ericsson fell out on lack of modem integration, then that's hilarious, since wireless comm is pretty much all that's keeping Ericsson alive nowadays)

Well they couldn't keep up there either, they folded modem development in 2014:

https://www.ericsson.com/news/1856711

But my comment about lack of modem keeping them from competing was actually about Allwinner and Rockchip, and why they have virtually or exactly 0% of the phone market. ST-Ericsson had modem integration, but they simply could not keep up in iterating products. If you look back several years you'll find TI or ST-Ericsson or others that have since folded on mobile would announce a product that would show up in devices like a year and a half later at best. Their release cycle needed to be at least 3-4x faster than it was to compete with the others and they couldn't manage that.
 
Not a concrete plan but it was mentioned multiple times that this possibility is one of the upsides of the modular designs.
Being able to play all my DRM-free Linux desktop ports on a secondary (x86) Pyra in another color some years down the road is a nice thought although not really neccessary.
Having more native ARM ports would be the best scenario of course but is only realistic with open sourced games, I guess.

I remember some of the talks/debates on the forums. I agree with ED and the community. It was down to a few points:

(1) Backwards/forwards compatibility - say goodbye to what we've built with the PND's and software and support

(2) x86 architecture still draws way too much power. This was a bit before the IntelTVsticks and other things hit the marketplace, but I still don't see a battery pack for those, and they do have tiny fans in there. It's not passive cooling.

(3) Will you be able to have a truly OPEN device? x86 chipsets have backdoors and though I need them for work sometimes, I don't trust my Wintel Overlords.

There were other more salient points...I'm just going by my memory, which is human and not perfect.
 
Rather than post the lot, here is a list of GPD WIN test videos a Chinese guy has . They are for tests done with the Z8500 GPD WIN prototype, so unfortunately they are not representative of the final product which will have the Z8700/Z8750. You can see tests for CS6 Photoshop opening up, and him doing some scaling. A test of PS 2 emu game, Halo etc

http://i.youku.com/i/UMzI0NzMzOTMzMg==
 
Last edited:
Well, right now, people have problems getting those Atom SoCs to properly work with Linux.
So unless that's fixed...

It's fixed. The only real sticking point had been audio over HDMI (which is a dealbreaker for stick/STB devices), but that's been resolved. All of the built-in functionality in the cherry trail chips is now supported. Extra stuff (touchscreen, gaming controls, wireless radios) is still up to what hardware ends up in the production units. Unfortunately, with linux not being a priority for GPD, it's even odds whether the parts they choose (whatever's cheapest that week) will play nice or not.
 
They are for tests done with the Z8500 GPD WIN prototype, so unfortunately they are not representative of the final product which will have the Z8700/Z8750.
There might be a 10%-20% performance boost but not much more. It might in some cases make the difference between playable and smooth, but I think the prototype gives a good idea of what to expect.
 
Will this also Play Tetris ??
The Batterie Live, on Dragonbox isnt that much.., up to 5 Hours, even my CC Pandora whit its original Akku and the Screenbacklight ISSUE, where this is permanently hight runs much longer..
An Usefull Windows Mini PC, but a bit diverent too the Pyra..
 
(2) x86 architecture still draws way too much power. This was a bit before the IntelTVsticks and other things hit the marketplace, but I still don't see a battery pack for those, and they do have tiny fans in there. It's not passive cooling.
Even if it is repeated over and over again, it doesn't become true all of a sudden. Even the previous generation of Atoms had a better performance per watt than the most efficient ARM SoCs and idle power consumption was already really good on even older Atoms.

(3) Will you be able to have a truly OPEN device? x86 chipsets have backdoors and though I need them for work sometimes, I don't trust my Wintel Overlords.
That's a valid point. The whole management engine issue is worrying. Unfortunately ARM SoCs have their own share of issues, like incomplete or completely missing graphics drivers. So unless you are fine without any 3D acceleration, you have to use binary only code on most ARM SoCs as well, and certainly on the PowerVR GPU of the Pyra.

I'd probably preorder the GPD Win from ED if I could be certain that it would run Linux distros without any major issues. I have no intention on using Windows 10 whatsoever, so unless someone tries Linux on the thing in a review, I'll probably wait and see how it turns out. The Atom chip itself should be fairly well supported on Linux by now, but I'm not sure about the input devices.
 
Will it also Play Rigs of Rods??
And 18 Wheels of Steel??
And what about Spinntires, the Russian Offroad Truck Simulator..

One of the reason i like the Pandora/Pyra is, that it should work, when it was portet for the device
The same like on the consoles, a PS2 Game per exemple works on every PS2..

Also the whole history around pandora/pyra dos it make lot more sympathic..

Also the Win is just a tiny Windows Netbook, nothing spezial, while the Pandora is something spezial..
 
Back
Top