GPD Pocket


Well GPD have just said on the Chinese forum that there will not be a uSD port. They feel the 128GB emmc storage is sufficient . I hope they change their mind.

Also, it looks like they are aiming to start an Indiegogo Campaign on Feb 4th, Indiegogo price looks like it might be $399
 
Last edited:
I also owned a 901 (1 ssd of 16Gb version) and really liked it. But the keyboard was the worst I've ever had on any laptop. It even failed and had to be replaced once.
I own a 901 too. I just hate it. It's slow, both on the SSD side and CPU side (not even taking about the GPU here). And the SSD size is just too small (I had the 4+8 GB). It was just a complete waste of money. I found the keyboard fine for the size of the device, but that's pretty much it.
All in all, the Pandora (even the CC model) is much more usable than this,
 
I own a 901 too. I just hate it. It's slow, both on the SSD side and CPU side (not even taking about the GPU here). And the SSD size is just too small (I had the 4+8 GB). It was just a complete waste of money. I found the keyboard fine for the size of the device, but that's pretty much it.
All in all, the Pandora (even the CC model) is much more usable than this,
The 4+8 had a really bad SSD from what I remember.
 
I'm still using my 901 (4+8GB model). I replaced the 8GB SSD pretty early on for a 64GB one and upgraded the RAM to the max. 2GB (I guess every 901 owner upgraded the RAM). I find the keyboard quite good for its size and the price of the device, although I miss a backlight. That's my biggest concern with the Pyra. I hope it will at least somewhat be able to keep up with the 901 in terms of keyboard usability.
Sure, the SSDs aren't blazing fast, but I get around 20MB/s in both directions which is enough in pretty much any scenario I'd run on that machine.

The bottleneck for me is usually the CPU. It has no VT-x which makes running VMs no fun and the bloated websites that make up the majority of the web these days don't make it any better.
In any case, the 901 is way more powerful than my Nokia N900 (OC'ed to 900MHz). As the N900 is comparable to the Pandora in that regard I never saw a reason to buy a Pandora.
 
I'm still using my 901 (4+8GB model). I replaced the 8GB SSD pretty early on for a 64GB one and upgraded the RAM to the max. 2GB (I guess every 901 owner upgraded the RAM). I find the keyboard quite good for its size and the price of the device, although I miss a backlight. That's my biggest concern with the Pyra. I hope it will at least somewhat be able to keep up with the 901 in terms of keyboard usability.
Sure, the SSDs aren't blazing fast, but I get around 20MB/s in both directions which is enough in pretty much any scenario I'd run on that machine.

The bottleneck for me is usually the CPU. It has no VT-x which makes running VMs no fun and the bloated websites that make up the majority of the web these days don't make it any better.
In any case, the 901 is way more powerful than my Nokia N900 (OC'ed to 900MHz). As the N900 is comparable to the Pandora in that regard I never saw a reason to buy a Pandora.
My Gigahertz is SO MUCH faster / useable than my old eeePC 901! Be it on productive software, even browsing, and gaming!
 
My Gigahertz is SO MUCH faster / useable than my old eeePC 901!
That's seems unlikely. notaz has published some nbench [1] figures for the Pandora here. [2] He ran it at 500MHz, so for a fair approximation his numbers need to be doubled.

This is the result I just got from my EEE 901 running Debian Jessie:
Code:
BYTEmark* Native Mode Benchmark ver. 2 (10/95)
Index-split by Andrew D. Balsa (11/97)
Linux/Unix* port by Uwe F. Mayer (12/96,11/97)

TEST                : Iterations/sec.  : Old Index   : New Index
                    :                  : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT        :          505.64  :      12.97  :       4.26
STRING SORT         :          75.612  :      33.79  :       5.23
BITFIELD            :      2.0761e+08  :      35.61  :       7.44
FP EMULATION        :          86.879  :      41.69  :       9.62
FOURIER             :          6529.2  :       7.43  :       4.17
ASSIGNMENT          :          13.555  :      51.58  :      13.38
IDEA                :          2029.3  :      31.04  :       9.22
HUFFMAN             :          1074.2  :      29.79  :       9.51
NEURAL NET          :          7.9561  :      12.78  :       5.38
LU DECOMPOSITION    :          326.81  :      16.93  :      12.23
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX       : 31.536
FLOATING-POINT INDEX: 11.712
Baseline (MSDOS*)   : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
CPU                 : Dual GenuineIntel Intel(R) Atom(TM) CPU N270   @ 1.60GHz 1600MHz
L2 Cache            : 512 KB
OS                  : Linux 3.16.0-4-686-pae
C compiler          : gcc version 4.9.2 (Debian 4.9.2-10)
libc                :
MEMORY INDEX        : 8.044
INTEGER INDEX       : 7.741
FLOATING-POINT INDEX: 6.496
Baseline (LINUX)    : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.
Even after doubling notaz' results it still shows that the N270 is clearly faster than the OMAP3530 in the original Pandora in every aspect except for FP emulation. And afaik the 3570 in the 1GHz model scales proportionally. notaz' 500MHz results also line up quite nicely with the N900 (OMAP 3430 stock at 600MHz) results that are in the nbench list. [3] So I assume his numbers are realistic and my experiences from the N900 are transferable to the Pandora.
nbench doesn't account for smp, which the N270 has in form of hyper threading but the OMAP3xxx lacks. So this benchmark even favours the OMAP3xxx.

So where is your "SO MUCH faster" coming from?


[1] http://web.archive.org/web/20160308152014/http://www.tux.org/~mayer/linux/bmark.html
[2] https://pyra-handheld.com/boards/th...-performance-numbers.42783/page-3#post-660242
[3] http://web.archive.org/web/20160307052709/http://www.tux.org/~mayer/linux/results2.html
 
Gigahertz model has a larger cache...

But I'm not speaking about synthetic benchmark. I speak about real world experience, where you actualy launch and (try to) use apps or games.
Launch a Browser, launch Jedi Knight, ReVolt, on both models, and you'll see.

Anyway, this is a thread about GPD Pocket, not about comparisons of out-of-productions computers like the eeePC 901 or the Gigahertz Pandora.
 
Gigahertz model has a larger cache...
32kB L1 + 256kB L2. The N270 has the same amount of L1 cache and double the L2 cache. So that hardly accounts for your claim of the Pandora being faster than an EEE 901.

But I'm not speaking about synthetic benchmark. I speak about real world experience, where you actualy launch and (try to) use apps or games.
You have to start somewhere. And I think synthetic benchmarks are a reasonable first step if you want to compare different systems.

Launch a Browser, launch Jedi Knight, ReVolt, on both models, and you'll see.
If you want to compare different systems you have to use the same software on both of them. Otherwise you're comparing apples and oranges.
You just released netsurf for the Pandora, so let's use that for conmparison! This is my result for starting netsurf-gtk as it's delivered by Debian on the 901 (I know the versions don't match, but it's just a first glance at "real world experience"):
Code:
$ /usr/bin/time -f '%S %U' netsurf-gtk
0.09 0.98

Anyway, this is a thread about GPD Pocket, not about comparisons of out-of-productions computers like the eeePC 901 or the Gigahertz Pandora.
You're the one who claimed the Pandora is faster than an EEE 901. Don't weasel out by calling this off-topic (which is very well is)! Show some proof that supports your claim!
 
32kB L1 + 256kB L2. The N270 has the same amount of L1 cache and double the L2 cache. So that hardly accounts for your claim of the Pandora being faster than an EEE 901.

You have to start somewhere. And I think synthetic benchmarks are a reasonable first step if you want to compare different systems.

If you want to compare different systems you have to use the same software on both of them. Otherwise you're comparing apples and oranges.
You just released netsurf for the Pandora, so let's use that for conmparison! This is my result for starting netsurf-gtk as it's delivered by Debian on the 901 (I know the versions don't match, but it's just a first glance at "real world experience"):
Code:
$ /usr/bin/time -f '%S %U' netsurf-gtk
0.09 0.98

You're the one who claimed the Pandora is faster than an EEE 901. Don't weasel out by calling this off-topic (which is very well is)! Show some proof that supports your claim!
I was comparing the cache of the CC/Rebirth with the Gigahertz model (I think the Pandora bench wasn't from a gigahertz model).

You obviously are happy with your eeePC, good for you. I wasn't with mine. That's it.
 
I was comparing the cache of the CC/Rebirth with the Gigahertz model
I still doubt that this cache increase leads to an increase in speed that justifies calling a Pandora "SO MUCH faster" than a 901.
 
I've only really played VVVVVV and emulated some GBA and PCE games in the years I've owned it. The Pandora certainly has the library of easy to find games versus the Arch install I'm currently rocking. But I found VVVVVV to run smoother on my 901 than my 1GHz, and since I switched to Arch this machine can now play 1080p youtube videos fine (though I usually watch them scaled, because of the 600 line screen) - I don't actually know how the Pandora scores on that front.

I first ran Ubuntu (9.04 I think) which was a crashy mess at times. I then switched to Debian Squeeze (6) but that also took a bit of work to find a stable config. Since I've been running arch I've had the one system breaking update, but other than that it's all working better than ever.
 
Is Ubuntu going to be officially supported to the same level as Windows, or are they counting on it being community supported?

Can someone translate the difference between the $399 and the $599 versions? I'm betting that the SoC stays the same, but that there could be two memory/storage price points.

There is a lot of information in those pictures and comparative...

It appears to be a three board design. Keyboard, display back light, motherboard (SoC, memory, storage, ports all on one).
Screwed together aluminum case design. Should be strong, but I question how they're planning to get decent WiFi signal. But, this should at least make the battery changeable.
1920x1080 7" screen should be a nice desktop - though not overly 'roomy'.
23mm thickness is still 'pretty thick', but not 'bad'. I still question the pocketability of this.

Comparing pocketability using Pandora as a baseline:
Pandora: 140x83x29
GPD Pocket: 180x106x23.3
GPD Win: 155x96x24
Pyra: 139x87x32
Samsung Galaxy Note 3 (A big handy to me phone): 151.2x79.2x8.3

Device pocket length (X dimension, left-right over landscape perspective): When inserted into a front pants pocket, the device is turned long-way-up. So, the additional 40mm in width (height in pocket) may become an issue for the 180mm GPD Pocket. In pants with shallow pockets I have found the Note 3's 151mm in this dimension to hit the limit for being able to comfortably sit down without getting impaled in the gut.

Device pocket width (Y dimension, front to back over landscape perspective): From some quick tests of phone in pocket and figuring out how much space is left, I could see a device up to 15mm width working, but not much more than that - on a 'thin' device'.

Device pocket thickness (Z dimension, lid to base): This dimension winds up being pretty critical in 'printing'. I.e. an 8.3mm thick phone can 'disappear' into a pocket, but anything in the 20mm or more thickness range starts to look like you're smuggling a religious text. The Pyra and Pandora suffer a bit here as they reach into the limits of pocketable thickness.

When considering pocketability, though, the Y and Z dimensions should be combined into a cross sectional area. Essentially a pocket is a fabric tube with a finite cross section. On this basis, the GPD Pocket is a bit less pocketable than a Pandora, but more pocketable than a Pyra. All of the keyboard & controls based devices are, of course, considerably less pocketable than a 'big phone' to where a person could carry 2-3 'big phones' in the same volume, not that anyone would want to.

Pandora: 83x29 = 2407
GPD Pocket: 106x23.3 = 2469.8
GPD Win: 96x24 = 2304
Pyra: 87x32 = 2784
Note 3: 79.2x8.3 = 657.36

So, the GPD Pocket is not significantly 'less pocketable' than any of the other 3 clam shell devices IF you're wearing loose fitting pants with deep pockets. If you're wearing standard pants or jeans, it might 'fit', but don't count on being able to sit down or board a car with it in there. 'Skinny jeans' isn't going to work with any of them.
 
Is Ubuntu going to be officially supported to the same level as Windows, or are they counting on it being community supported?

Can someone translate the difference between the $399 and the $599 versions? I'm betting that the SoC stays the same, but that there could be two memory/storage price points.


Yes , GPD will develop the drivers and fully support Ubuntu on the GPD Pocket. Actually they seem to be pushing it harder than Windows 10. I think they realize it has more uniqueness as a Ubuntu product than a Windows product

Someone did an image translate on the chart , and it appears that the $399 will be the Indiegogo price - scheduled I think for Feb(may move). $599 is what they say will be the RRP which doesn't mean much TBH. GPD always throw out high RRPs with the actual retail price being much lower. Take the GPD WIN for example. The Indiegogo price was $299 - the RRP they quoted for it was $499. It has sold at retail from day one between $300 - $360. They did the same sort of thing with the GPD XD. I'd say that if the Indiegogo price turns out to be $399, then the shops will sell it for $450- $480 would be my guess

At first I was skeptical about it's pocketability, but now that they have given the dimensions, I can see it would even fit in the pockets of the shorts I'm wearing. Not going to fit every pocket, but yeah, good to see.
 
Last edited:
If this is all true then I am seeing many positives. We will have to wait until after it is released to see how much support they give it. Ubuntu isn't my cup of raktajino, but it beats the alternative a bit, I guess. If the build quality, user serviceability, and keyboard (all the hardware, really) are decent, I might be more interested in it, especially if it is relatively widely supported (even if only by fans and hobbyists). Before I would really consider buying it I would need to see it as an improvement over whatever else I have (hopefully the Pyra, and if I have mobile phone functionality it will take a lot to replace it). I will definitely keep my eye on the Pocket. At least GPD has been creative with their recent device names...
 
So, the GPD Pocket is not significantly 'less pocketable' than any of the other 3 clam shell devices IF you're wearing loose fitting pants with deep pockets. If you're wearing standard pants or jeans, it might 'fit', but don't count on being able to sit down or board a car with it in there. 'Skinny jeans' isn't going to work with any of them.
Depending on the pocket different quantities are important. For standard jean pocket the curvature of the thigh is also a concern, so you don't want something too wide there. For coat/jacket inside pocket, weight is the primary driver IMO, if it will even fit. For cargo leg pocket or coat outside pocket, I reckon you've got it though.
 
What did MS learn from the surface 3. Nobody wants junk overeating hardware with win10 on it.
I feel there is a lesson in there somewhere.

GPD does not have the kind of money to subsidize win10 into the market. This is lesson 3 of the surface.

Design a good keyboard, make the cooling passive. Go linux only. This is low hanging fruit in a world where they are failing so extensively.

Bonus points for making the PCB modular in any fashion deemed suitable.
People would put their rpi3's in that for good money left and right. Or dodge the intel.
More bonus points for real keys and a matte screen and user replaceable ram. Throw in a good software logo for good measure.
 
Last edited:
Matches the bloated software.
If your product is being matched up to one of the costliest mistakes of computer history and it doesn't stack up really well, there might be a problem. One of which is it fails for the same reasons.
 
Tbh, this running Win10 is probably my dream handheld. Me buying a pyra is looking less and less likely.
 
Back
Top