[Arm-netbook] linus torvalds says "2016 is the year of the arm laptop"

[Luke Kenneth Casson Leighton]

On Wed, Oct 14, 2015 at 3:58 PM, Lauri Kasanen <cand at gmx.com> wrote:
> On Wed, 14 Oct 2015 14:07:53 +0000
> joem <joem at martindale-electric.co.uk> wrote:
>> On Wed, 2015-10-14 at 10:48 +0100, Luke Kenneth Casson Leighton wrote:
>> > >> > I have 8" and 10" ARM phablets with that resolution display
>> > >> > and I'm sure there must be some youtube out there of similar
>> > >> > devices.
>> > >>
>> > >>  what's the model / make and  what's the internal processor?
>> > >
>> > > Its at home now - It was one of these or a similar model
>> > > http://www.geekbuying.com/item/Onda-V919-4G-Air-9-7-Inch-Android-4-4-Phablet-MTK8752-Octa-Core-2-4GHz-2048-1536-2G-3G-4G-2GB-32GB-Tablet-PC-WiFi-GPS-BT---Silver-344501.html
>> >
>> >  ahh mediatek gpl-violating with 100% certainty *sigh*.
>> >
>> >  ok so it's technically doable.  that's a quad-core 64-bit with an
>> > A53... yowser the peak power consumption is going to be enormous
>> > though.   it'll be a MIPI or eDP display.
>>
>> Most laptops last only 2 / 3 hours with gaming and get very hot.
>> These phablets last more than that watching videos.
>> I would plug in the adaptor while playing games or doing any
>> serious work, and I can imagine a small fan inside to remove heat
>> while playing games.
>>
>> Still difficult to see what the problem is bringing out ARM laptops with
>> high definition displays.
>
> Is there really much point to them?

 _yes_.  for programming, yes and double-yes.

 the reason is simple: the larger the amount of information you can
get on-screen makes you a more effective programmer.   we're not like
"normal" people - we need to be able to develop "normal" things for
those "normal" people...but also to have the tools *to* develop those
"normal" things on-screen *simultaneously*.

 so the additional resolution allows for more and more side-by-side
programs *without* needing to hit alt-tab... or touch the mouse... or
click on a task bar.... or play "hunt the app".  everything's just
*there*.  no f*****g about.

and that means "less time wasted".   and less mental exertion spent.
and that means "you are a more effective programmer".


here's a few screen-shots to give you some examples.  note the
effectiveness of the massive resolution...

 ....oh wait!  lovely irony: you don't have a 2560x1600 screen, so you can't! :)

 i'll upload them anyway and describe them to you.

 http://hands.com/~lkcl/scrn.png

 a side-by-side comparison of two near-full-screen A4 datasheets,
sized 1200 pixels by 1500 pixels, clearly visible with full detail.
the only reason i shrunk these to 1600 wide is out of concern and
respect for the recipient, who may not have had such a high-resolution
LCD.

 in the bottom right corner you can see the fvwm2 6x4 grid.  i have
*twenty four* virtual screens, each with a different setup.  ps ax |
grep xterm | wc shows *thirty six* 80x60 terminals open.  ps ax | grep
vi | wc shows *sixty two* vi sessions open.

 http://hands.com/~lkcl/scrn2.png

 here i have 4 xterms - all of them are 80x60 - clearly visible, next
to a whopping 1200 x 1400 sized web browser, which is _more_ than
enough to see 99.9% of web sites full functionality without needing a
horizontal scrollbar.  there are 2 more underneath (which i consider
to be inconvenient but it is unavoidable.  i would _like_ a much
higher-resolution screen to be able to deal with this).

 in the 6 xterms i am using:

 * one for general-purpose things (currently editing the rhombus-tech
wiki pages)
 * one for helping with a documentation issue in libopencm3, but
normally this one has been used for editing the firmware
 * below that in the bottom right is the window for uploading firmware
to the stm32f072 boards.
 * to the right is a 32-bit chroot where i can do "make bin" to
compile the firmware.  as that's in the command-history i can do
"up-arrow, return".
 * underneath are two xterms which have library files open in
different subdirectories.  each has about 10 vi sessions open and
ctrl-z'd

 http://hands.com/~lkcl/scrn3.png

 this is where i've been doing openscad development.  note how i can
get an entire 80x60 xterm in at the left, with another 80x60 xterm
*underneath*.... with *NO OVERLAPS*... then a huge openscad window at
something like 2000 x 1600.  the only reason for the 3rd xterm
overlapping is because i have temporarily paused the 3d development of
the laptop casework source code.

 those two xterms i can make edits in the top one and check any
additional files (util libraries) *WITHOUT* having to flick
back-and-forth.  display of both the calling and called function can
be on-screen at the SAME TIME.

 http://hands.com/~lkcl/scrn4.png

 this is where i've been managing the 3D printing.  note how
pronterface, with a much better 3D display than repsnapper, is set up
to cover the majority of the screen (around 1800 x 1500) whereas
repsnapper is set to a minimalist functional size (around 640x480)
that *still* doesn't interfere with repsnapper.

 http://hands.com/~lkcl/scrn5.png

 this is where i manage WIFI and sound connections.  it doesn't
interfere with anything else.  no windows "overlap".

 that's just five out of the 24 screens i could potentially use, and
you can see i am *genuinely* using 14 of those.... on a regular basis.
that's not 14 screens where things are abandoned, that's 14 screens
where i go to those 14 screens *every single day* because i am
handling so many simultaneous different tasks.

 but... unfortunately, developers like this are "not the norm".  it's
only slowly coming into the average end-user's consciousness that
workflow is more productive when things are side-by-side with more
information on-screen.

 most developers substitute multiple-screen setups for
higher-resolution setups.  i've had a 4-screen setup in the past,
which included a laptop's screen, an HDMI output, a DVI output and a
USB Displaylink UD-160A.  the laptop screen, because it was running
mac osx, was actually the most ineffective, and was relegated to
running low-use applications.  all the others were huge.  results were
shown on the right screen (full-screen web browser), editing in 4 long
80x65 xterms took place on the 1920x1080 HDMI monitor.

 it wasn't good, because the end xterm had to overlap.  i really did
need a much higher-res screen in order to get 4 terminals
side-by-side, and i really did need way more than 4 terminals because
the source code really was that complex.


 so does that give you some illustration as to why developers need as
much screen real-estate as they can possibly get?


> To me the quest for DPI passed the
> "mine's bigger than yours" point long ago. Of course I have bad
> eyesight, and can barely see the pixels on 1920x1080 at 24",

 so did i think that i had "bad eyesight"... and i had a 1920 x 1200
24in mac... then i got a 13in macbook air, replacing the OS with
debian, it runs a 2560x1600 LCD... after only 4 hours going through
the install and configuration process i went back to the 24in mac and
was freaked out by the fact that i could see staggered lines on the
24in mac's screen!

 about 2/3 of the way down, the screen's pixels shifted over by about
1/5 of a pixel and i had *never noticed*.  additionally my eyes were
able to perceive individual red, green and blue pixels.


> but at the
> resolutions the current phones are using you'd need to hold it glued to
> your eyeball and still couldn't see it all.

 ... and then you go back to a coarser-res screen and you can, if you
care to look, make out individual red, green and blue.  but only if
you've trained yourself, first, on the higher-resolution screens, and
haven't "bought the hype".

 the whole "it's a retina screen therefore you can't discern
individual pixels isn't that amazing buy it now it's better than
anything else no really" marketing hype is, frankly, total bollocks.

 during the 2nd world war (before radar), lookouts used to be able to
tell what type of plane was approaching, when it was literally a mere
dot.  analysis of people who had developed this ability showed that
they were able to correctly identify a plane when there were quite
literally only around 100 visual cortex neurons firing.

 humans *do* have the ability to discern more information than we are
*told* we can discern.  it just takes training.


> I couldn't find any sales numbers for Chromebook Pixel. Mac sales have
> gone down despite the Retina screens. The data does not support the
> case that a higher-res screen is the missing point.

 that's because the majority of people are (a) not programmers and (b)
honestly... not very observant and (c) tend not to go "back" to what
they had before, to make a comparison.

 the majority end-users simply don't have the need for higher-res.
1600 wide is "good enough".  hell, 1366x768 is still "good enough" to
do email by internet browsing, watch a few films, and do document
editing.  why would the average end-user _ever_ want to put two
full-screen apps side-by-side when their minds can't cope (because
they've never trained to do it) with more than one thing at a time?

 by complete contrast, my next development laptop - which will be in
several years time - i will be getting the lightest 15in that i can
find... but i will be prioritising screen resolution.

l.