Kobo SOC upgrade ever?

masterz87 · 12-23-2018, 09:28 PM

they've literally been using the same SOC or variants of it for literally 6+ yrs now. I was just wondering if/when they'll ever go to a dual core CPU anytime within the next decade.

PeterT · 12-23-2018, 10:21 PM

Maybe I'm missing something. What benefit would there be in Multi core processors for an e-reader?

masterz87 · 12-24-2018, 12:58 AM

Quote:

Originally Posted by PeterT

Maybe I'm missing something. What benefit would there be in Multi core processors for an e-reader?

A multicore CPU or a higher clocked CPU so that page turns can be faster with a better overall UI experience especially when reading. Since currently page turning is far too slow on my aura h20 and I know e-ink can be refreshed faster than that.

That along with better loading of the assets, the ability to use zRAM for caching data in ram since kobo is using bottom-of-the-bucket grade sd cards that are only class 4(from the teardowns I've seen) being able to get the whole thing to just work better.

Finally something with big.LITTLE would keep or likely reduce power usage from the SOC as the system would be able to do more things with itself.

TL;DR
Maybe I'm just ranting but having used basically the same-exact arm cortex tier from my OG nook I want more speed.

The not ereader is something that I want from a premium ereader(considering the costs that I'm paying for the system). Basically that's it, faster soc to improve the screen refreshes book loading etc.

GeoffR · 12-24-2018, 07:47 AM

Quote:

Originally Posted by PeterT

Maybe I'm missing something. What benefit would there be in Multi core processors for an e-reader?

This guy.might buy one.

Quoth · 12-24-2018, 10:55 AM

Quote:

Originally Posted by PeterT

Maybe I'm missing something. What benefit would there be in Multi core processors for an e-reader?

Almost none.
Indexing isn't very often compared to reading. I read a lot. I don't find ANY of my eReaders too slow for actual reading. The speed of the Flash memory may be more of a limit than CPU for indexing and USB transfer anyway.

Tuneraider · 12-24-2018, 11:06 AM

Quote:

Originally Posted by masterz87

they've literally been using the same SOC or variants of it for literally 6+ yrs now. I was just wondering if/when they'll ever go to a dual core CPU anytime within the next decade.

https://goodereader.com/blog/e-book-...months-of-2018

A couple of paragraphs from that article:

eBook sales generated $358 million dollars and declined by 3.8% and during the same period last year the format brought in $373 million dollars. Digital audiobook sales were $137 million dollars from January to April 2018 and this was an increase of 36%. This was roughly an increase of $37 million from the same time last year.

Declining ebook sales are not just an anomaly in 2018, but have been occurring for quite some time. Traditional publishers sold 10% fewer ebook units in 2017 compared with the previous year, according to data released by PubTrack Digital. Total sales were 162 million in 2017 rather than the 180 million units sold the year before. Nielsen’s reports put 2016 ebook unit sales from the top 30 traditional publishers down a full 16% from their 2015 numbers.

----

I'm guessing no. Declining sales = less R&D. In a decade instead of reading and listening to a ebook it will probably evolve to something like a virtual reality/3D format where your are Jack Reacher that's kicking azz and taking names.

Quoth · 12-24-2018, 12:57 PM

The eBook figures are based only on Traditional publisher sales and some of those include audio books!
Amazon does not publish their eBook sales. Smashwords do.
The sales of eBooks may have actually risen. However the majority are read on phones, then tablets and eink screen ereaders are third. Amazon might have 90% of eBook market and ONLY supports their app and Kindle.

An ereader is really very simple. The actual software (firmware) could have been much better than today over 10 years ago. Probably 15 years ago. The expensive and tricky part is the eink. Amazon bought and buried the Philips' eInk spin off that had produced a folding mobile phone. Almost all eink readers ever have used https://en.wikipedia.org/wiki/E_Ink_Corporation eink. Sony had first generally available eInk reader in 2004 and it was a little grey but fine in good lighting. I tried it in shop but couldn't afford it. Amazon came late with SW based on Mobi (terrible language support) in 2007.
There is almost no HW (hardware) R&D needed, as the design is very simple and based on panel makers reference design. Most changes have been simply newer panels and cosmetic.

The SW (FW, Firmware, software) is usually based on GNU Linux or Android (Linux Kernel), a basic pair of readers (sometimes the Adobe one), one for eBooks and one for PDFs and a simple GUI. A very small R&D can build that quite quickly (and it shows).
Most of the development budget seems to be on tooling for the case, then PCB.

With a team of THREE, I built a 4G proof of concept VOIP phone with Firefox, Debian Linux, email, choice of Tile or Desktop GUI (both Trolltech Qtopia), SD card slot, USB host and slave sockets, touch LCD (eink was available, but no good for colour or video), PDF reader, HTML based eBooks, media player etc. Six months!

Forget Mirasol, it's not as nice as eInk for text and crippled by Qualcomm's royalties model even though it's colour and sort of does video

The fact is that there little incentive for dedicated ereader R&D. It's a niche. Tablets and phones are cheaper to make and commodity items. That's why the Amazon Fire is a crippled Android Tablet. Most eBooks are read on the kindle App on Phones & Tablets.

All the eBook growth is small publishers, independents and self publishing. The traditional publishers are focused on Retail who have no ability or interest in ebooks at all all (apart from maybe Barnes& Noble who are clueless).

Summary:
The Traditional Publisher figures include audio books and are not representative. They charge almost as much as for paper and almost always have DRM.
Amazon has maybe 90% of the eBook market and doesn't say how many they sell.
Indications are that Independents, small publishers and self-publishing is seeing massive growth in eBooks because they are cheaper, often DRM free and on more platforms, but not in bricks & mortar Retail.
Dedicated eReader sales may have declined because almost everyone has a smartphone and/or tablet (sales of those, esp Apple, have declined because no need to upgrade and market saturated). Over half eBooks are read on phones.
Amazon is a danger to publishers and consumers due to their market power and strategies and lack of USA consumer vs corporate protection.
https://en.wikipedia.org/wiki/List_o...ions_by_Amazon

Many makers have abandoned production of eReaders (Sony being most notable). Sony has been crippled ever since they became a media company as well as electronics (stupid DRM on Mini disc, stupid proprietary DRM memory sticks on cameras, CD root kits, DVD, Bluray DRM, reliance on ebook shop for ereader).
Current Amazon dominance and only allowing Amazon app on general purpose gadgets pretty much means anything other than Kindle is doomed. The Basic Kindle Touch has to be close to cost. I don't think anyone can retail an eInk reader at that price and make a profit. Amazon has some premium models partly for the look of it and because it's no effort to do so. Almost everyone is using the same panels.
see https://en.wikipedia.org/wiki/Comparison_of_e-readers

Quoth · 12-24-2018, 01:01 PM

Deleted

DNSB · 12-24-2018, 02:39 PM

Quote:

Originally Posted by masterz87

That along with better loading of the assets, the ability to use zRAM for caching data in ram since kobo is using bottom-of-the-bucket grade sd cards that are only class 4(from the teardowns I've seen) being able to get the whole thing to just work better.

From my testing, the class of the card is of little relevance. What is critical is the small block read/write speeds. As an example, years back, I tried a Patriot 32GB class 10 with great large block specs and Crystal DiskMark tested abysmal small block read/write speeds. When reading, it caused noticeably slower operation. We are not writing video at 10MBPS, we are running a Linux OS which uses a lot more small block operations (4K or smaller reads/writes).

You might also want to compare the specs for the SanDisk eMMC in the KA1 to the specs for a class 10 uSD card before getting too enthused about the value of any increase in storage speed. From what I see of the part number (SDIN8) and searching on SanDisk's site, as a minimum, it would beat a class 10 uSD card with a minimum 17MB/s read, 90MB/s write (SDIN8DE1) and quite possibly beat a UHS 3 uSD card with a 30MB/s read, 115MB/s write (SDIN8DE2).

masterz87 · 12-24-2018, 04:27 PM

Quote:

Originally Posted by DNSB

From my testing, the class of the card is of little relevance. What is critical is the small block read/write speeds. As an example, years back, I tried a Patriot 32GB class 10 with great large block specs and Crystal DiskMark tested abysmal small block read/write speeds. When reading, it caused noticeably slower operation. We are not writing video at 10MBPS, we are running a Linux OS which uses a lot more small block operations (4K or smaller reads/writes).

You might also want to compare the specs for the SanDisk eMMC in the KA1 to the specs for a class 10 uSD card before getting too enthused about the value of any increase in storage speed. From what I see of the part number (SDIN8) and searching on SanDisk's site, as a minimum, it would beat a class 10 uSD card with a minimum 17MB/s read, 90MB/s write (SDIN8DE1) and quite possibly beat a UHS 3 uSD card with a 30MB/s read, 115MB/s write (SDIN8DE2).

I know that eMMC is going to be leaps and bounds beyond any uSD card as they're literally the bottom of the bin in terms of flash performance.

Finally to the other people talking about why it's pointless to have a better SOC, then explain to me how in the hell the chinese folks at dasung are doing the NOOK fast-refresh 1bit mode in what is at least 4 colors but from the video looks like 8 colors. That's what I want, I want page turns to be faster the UI to be snappier as it stands the whole thing is sluggish.

NiLuJe · 12-24-2018, 04:57 PM

Making videos look cool on an eInk device has been relatively "easy" to do, even on crappy old Pearl panels & the original Epson EPD controller (well, maybe not the original one, but, still, early Pearl days stuff ;p).

You'll find that doing the same with real-world, unprocessed content, and in real time, while not looking like utter crap in person, is another thing entirely.

FWIW, recent devices are heaps and bounds ahead of the original H2O on every front (in Kobo-land, that's Mk.7 devices (most prominently the Clara & Forma). In Kindle-land, the Oasis 2 and the PW4).

bgalbrecht · 12-25-2018, 12:46 AM

I know I'm an outlier, but now that Amazon, Kobo, and B&N no longer include external micro SD card slots, I'm glad that Kobo is still using an internal SD card slot instead of a SOC with 4 GB or 8 GB of memory.

masterz87 · 12-31-2018, 06:35 PM

Quote:

Originally Posted by NiLuJe

Making videos look cool on an eInk device has been relatively "easy" to do, even on crappy old Pearl panels & the original Epson EPD controller (well, maybe not the original one, but, still, early Pearl days stuff ;p).

You'll find that doing the same with real-world, unprocessed content, and in real time, while not looking like utter crap in person, is another thing entirely.

FWIW, recent devices are heaps and bounds ahead of the original H2O on every front (in Kobo-land, that's Mk.7 devices (most prominently the Clara & Forma). In Kindle-land, the Oasis 2 and the PW4).

I can't imagine that there's much _real_ difference in terms of fluidity because from what I can see it's still the same SOC under the wraps and only the lighting/panel have changed.

NiLuJe · 01-01-2019, 01:54 PM

@masterz87: It's *really* not.

H2O (Mk.5):

Code:

ntx_hwconfig -v -s /dev/mmcblk0 CPU
[27] CPU='mx50'

Code:

cat /proc/cpuinfo
Processor       : ARMv7 Processor rev 5 (v7l)
BogoMIPS        : 166.57
Features        : swp half thumb fastmult vfp edsp neon vfpv3 
CPU implementer : 0x41
CPU architecture: 7
CPU variant     : 0x2
CPU part        : 0xc08
CPU revision    : 5

Hardware        : Freescale MX50 Reference Design Platform
Revision        : 50011
Serial          : 0000000000000000

Forma (Mk.7):

Code:

ntx_hwconfig -v -s /dev/mmcblk0 CPU
[27] CPU='mx6sll'

Code:

cat /proc/cpuinfo
processor       : 0
model name      : ARMv7 Processor rev 1 (v7l)
BogoMIPS        : 1.50
Features        : half thumb fastmult vfp edsp neon vfpv3 tls vfpd32 
CPU implementer : 0x41
CPU architecture: 7
CPU variant     : 0x4
CPU part        : 0xc09
CPU revision    : 1

Hardware        : Freescale i.MX6 SoloLite (Device Tree)
Revision        : 0000
Serial          : 0000000000000000

Relevant to our interests, the former is a Cortex A8 paired with one or the earliest iteration of the EPDC to support Carta screens, while the latter is a Cortex A9 paired with the very latest EPDC.
Even if you forget the CPU (which you may, as the main drawing point of the A9 is better power efficiency, and we've established that, at least given Kobo's current SW platform and constraints, parallelism wouldn't help much, so an i.MX7D like on the Oasis 2 wouldn't bring much), there has been *massive* improvements between these two revisions of the EPDC.

It really helped in closing the gap between Amazon and Kobo in terms of how well the screen is exploited (Amazon is still miles ahead, because $$$, but, still).

Each new Kobo 'Mark' introduced a new SoC. Mk.6 was basically just Kobo playing catch up by following Amazon in moving towards i.MX6 boards (A9, slightly more recent EPDC).

sherman · 01-01-2019, 03:15 PM

Quote:

Originally Posted by NiLuJe

@masterz87: It's *really* not.

H2O (Mk.5):

Code:

ntx_hwconfig -v -s /dev/mmcblk0 CPU
[27] CPU='mx50'

Code:

cat /proc/cpuinfo
Processor       : ARMv7 Processor rev 5 (v7l)
BogoMIPS        : 166.57
Features        : swp half thumb fastmult vfp edsp neon vfpv3 
CPU implementer : 0x41
CPU architecture: 7
CPU variant     : 0x2
CPU part        : 0xc08
CPU revision    : 5

Hardware        : Freescale MX50 Reference Design Platform
Revision        : 50011
Serial          : 0000000000000000

Forma (Mk.7):

Code:

ntx_hwconfig -v -s /dev/mmcblk0 CPU
[27] CPU='mx6sll'

Code:

cat /proc/cpuinfo
processor       : 0
model name      : ARMv7 Processor rev 1 (v7l)
BogoMIPS        : 1.50
Features        : half thumb fastmult vfp edsp neon vfpv3 tls vfpd32 
CPU implementer : 0x41
CPU architecture: 7
CPU variant     : 0x4
CPU part        : 0xc09
CPU revision    : 1

Hardware        : Freescale i.MX6 SoloLite (Device Tree)
Revision        : 0000
Serial          : 0000000000000000

Relevant to our interests, the former is a Cortex A8 paired with one or the earliest iteration of the EPDC to support Carta screens, while the latter is a Cortex A9 paired with the very latest EPDC.
Even if you forget the CPU (which you may, as the main drawing point of the A9 is better power efficiency, and we've established that, at least given Kobo's current SW platform and constraints, parallelism wouldn't help much, so an i.MX7D like on the Oasis 2 wouldn't bring much), there has been *massive* improvements between these two revisions of the EPDC.

It really helped in closing the gap between Amazon and Kobo in terms of how well the screen is exploited (Amazon is still miles ahead, because $$$, but, still).

Each new Kobo 'Mark' introduced a new SoC. Mk.6 was basically just Kobo playing catch up by following Amazon in moving towards i.MX6 boards (A9, slightly more recent EPDC).

Let's be brutally honest though. The Cortex A9 is an ancient CPU design. It was announced in 2007, over ten years ago!

12-23-2018, 09:28 PM	#1
masterz87 Connoisseur Posts: 56 Karma: 2072578 Join Date: Jan 2017 Device: Kobo Aura H20	Kobo SOC upgrade ever? they've literally been using the same SOC or variants of it for literally 6+ yrs now. I was just wondering if/when they'll ever go to a dual core CPU anytime within the next decade.

12-24-2018, 04:57 PM	#11
NiLuJe BLAM! Posts: 13,506 Karma: 26047202 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	Making videos look cool on an eInk device has been relatively "easy" to do, even on crappy old Pearl panels & the original Epson EPD controller (well, maybe not the original one, but, still, early Pearl days stuff ;p). You'll find that doing the same with real-world, unprocessed content, and in real time, while not looking like utter crap in person, is another thing entirely. FWIW, recent devices are heaps and bounds ahead of the original H2O on every front (in Kobo-land, that's Mk.7 devices (most prominently the Clara & Forma). In Kindle-land, the Oasis 2 and the PW4). Last edited by NiLuJe; 12-25-2018 at 02:48 AM.

01-01-2019, 01:54 PM	#14
NiLuJe BLAM! Posts: 13,506 Karma: 26047202 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	@masterz87: It's really not. H2O (Mk.5): Code: ntx_hwconfig -v -s /dev/mmcblk0 CPU [27] CPU='mx50' Code: cat /proc/cpuinfo Processor : ARMv7 Processor rev 5 (v7l) BogoMIPS : 166.57 Features : swp half thumb fastmult vfp edsp neon vfpv3 CPU implementer : 0x41 CPU architecture: 7 CPU variant : 0x2 CPU part : 0xc08 CPU revision : 5 Hardware : Freescale MX50 Reference Design Platform Revision : 50011 Serial : 0000000000000000 Forma (Mk.7): Code: ntx_hwconfig -v -s /dev/mmcblk0 CPU [27] CPU='mx6sll' Code: cat /proc/cpuinfo processor : 0 model name : ARMv7 Processor rev 1 (v7l) BogoMIPS : 1.50 Features : half thumb fastmult vfp edsp neon vfpv3 tls vfpd32 CPU implementer : 0x41 CPU architecture: 7 CPU variant : 0x4 CPU part : 0xc09 CPU revision : 1 Hardware : Freescale i.MX6 SoloLite (Device Tree) Revision : 0000 Serial : 0000000000000000 Relevant to our interests, the former is a Cortex A8 paired with one or the earliest iteration of the EPDC to support Carta screens, while the latter is a Cortex A9 paired with the very latest EPDC. Even if you forget the CPU (which you may, as the main drawing point of the A9 is better power efficiency, and we've established that, at least given Kobo's current SW platform and constraints, parallelism wouldn't help much, so an i.MX7D like on the Oasis 2 wouldn't bring much), there has been massive improvements between these two revisions of the EPDC. It really helped in closing the gap between Amazon and Kobo in terms of how well the screen is exploited (Amazon is still miles ahead, because $$$, but, still). Each new Kobo 'Mark' introduced a new SoC. Mk.6 was basically just Kobo playing catch up by following Amazon in moving towards i.MX6 boards (A9, slightly more recent EPDC). Last edited by NiLuJe; 01-01-2019 at 02:11 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
kobo aura or kobo glo hd? worth an upgrade	wakkaday	Which one should I buy?	52	07-10-2015 07:53 AM
Why can’t I complete the Kobo Touch firmware upgrade or Desktop upgrade?	DarrellAtKobo	Kobo Reader	31	08-31-2012 10:45 PM
Trade Anyone looking to upgrade their Kobo?	Dinodoc	Flea Market	1	11-15-2010 02:48 PM

12-23-2018, 10:21 PM	#2
PeterT Grand Sorcerer Posts: 13,510 Karma: 78910112 Join Date: Nov 2007 Location: Toronto Device: Libra H2O, Libra Colour	Maybe I'm missing something. What benefit would there be in Multi core processors for an e-reader?

12-24-2018, 12:57 PM	#7
Quoth Still reading Posts: 14,016 Karma: 105092227 Join Date: Jun 2017 Location: Ireland Device: All 4 Kinds: epub eink, Kindle, android eink, NxtPaper	The eBook figures are based only on Traditional publisher sales and some of those include audio books! Amazon does not publish their eBook sales. Smashwords do. The sales of eBooks may have actually risen. However the majority are read on phones, then tablets and eink screen ereaders are third. Amazon might have 90% of eBook market and ONLY supports their app and Kindle. An ereader is really very simple. The actual software (firmware) could have been much better than today over 10 years ago. Probably 15 years ago. The expensive and tricky part is the eink. Amazon bought and buried the Philips' eInk spin off that had produced a folding mobile phone. Almost all eink readers ever have used https://en.wikipedia.org/wiki/E_Ink_Corporation eink. Sony had first generally available eInk reader in 2004 and it was a little grey but fine in good lighting. I tried it in shop but couldn't afford it. Amazon came late with SW based on Mobi (terrible language support) in 2007. There is almost no HW (hardware) R&D needed, as the design is very simple and based on panel makers reference design. Most changes have been simply newer panels and cosmetic. The SW (FW, Firmware, software) is usually based on GNU Linux or Android (Linux Kernel), a basic pair of readers (sometimes the Adobe one), one for eBooks and one for PDFs and a simple GUI. A very small R&D can build that quite quickly (and it shows). Most of the development budget seems to be on tooling for the case, then PCB. With a team of THREE, I built a 4G proof of concept VOIP phone with Firefox, Debian Linux, email, choice of Tile or Desktop GUI (both Trolltech Qtopia), SD card slot, USB host and slave sockets, touch LCD (eink was available, but no good for colour or video), PDF reader, HTML based eBooks, media player etc. Six months! Forget Mirasol, it's not as nice as eInk for text and crippled by Qualcomm's royalties model even though it's colour and sort of does video The fact is that there little incentive for dedicated ereader R&D. It's a niche. Tablets and phones are cheaper to make and commodity items. That's why the Amazon Fire is a crippled Android Tablet. Most eBooks are read on the kindle App on Phones & Tablets. All the eBook growth is small publishers, independents and self publishing. The traditional publishers are focused on Retail who have no ability or interest in ebooks at all all (apart from maybe Barnes& Noble who are clueless). Summary: The Traditional Publisher figures include audio books and are not representative. They charge almost as much as for paper and almost always have DRM. Amazon has maybe 90% of the eBook market and doesn't say how many they sell. Indications are that Independents, small publishers and self-publishing is seeing massive growth in eBooks because they are cheaper, often DRM free and on more platforms, but not in bricks & mortar Retail. Dedicated eReader sales may have declined because almost everyone has a smartphone and/or tablet (sales of those, esp Apple, have declined because no need to upgrade and market saturated). Over half eBooks are read on phones. Amazon is a danger to publishers and consumers due to their market power and strategies and lack of USA consumer vs corporate protection. https://en.wikipedia.org/wiki/List_o...ions_by_Amazon Many makers have abandoned production of eReaders (Sony being most notable). Sony has been crippled ever since they became a media company as well as electronics (stupid DRM on Mini disc, stupid proprietary DRM memory sticks on cameras, CD root kits, DVD, Bluray DRM, reliance on ebook shop for ereader). Current Amazon dominance and only allowing Amazon app on general purpose gadgets pretty much means anything other than Kindle is doomed. The Basic Kindle Touch has to be close to cost. I don't think anyone can retail an eInk reader at that price and make a profit. Amazon has some premium models partly for the look of it and because it's no effort to do so. Almost everyone is using the same panels. see https://en.wikipedia.org/wiki/Comparison_of_e-readers

12-24-2018, 01:01 PM	#8
Quoth Still reading Posts: 14,016 Karma: 105092227 Join Date: Jun 2017 Location: Ireland Device: All 4 Kinds: epub eink, Kindle, android eink, NxtPaper	Deleted

12-25-2018, 12:46 AM	#12
bgalbrecht Wizard Posts: 1,806 Karma: 13399999 Join Date: Aug 2007 Location: US Device: Nook Simple Touch, Kobo Glo HD, Kobo Clara HD, Kindle 4	I know I'm an outlier, but now that Amazon, Kobo, and B&N no longer include external micro SD card slots, I'm glad that Kobo is still using an internal SD card slot instead of a SOC with 4 GB or 8 GB of memory.

Advert

Advert