Page turning latency benchmarks?

[ottischwenk]

Originally Posted by shalym

We get it.

OK.

Your e-ink android tablets are better than any dedicated reader..

Of course - dedicated readers are technically extremely out of date.
Only the screen has technically changed for 10 years: and because the SD cards have become cheaper and small cards are no longer available, the storage has become larger.
This goes so far that Kobo, because there are no more 8 GB cards, 16 GB built in but only partitions 8 GB

.in your opinion.

No - its a fact.

Thank you for enlightening us.

It was my pleasure.

Dedicated readers are technically cheapest garbage.

[PeterT]

I wonder if there is a difference in page turn speed with swipe as opposed to touch?

Sent from my Pixel 4a using Tapatalk

[John F]

Originally Posted by meeera

I fully admit that I'm not using Science! to measure the speed, but I did wait before flipping, to allow for the page-reading. A full refresh is a touch slower, but on a non-refresh turn, if I'm counting "one-and" for seconds, I don't quite get to the "and". And my device is pretty old now.

I don't think I see any difference in page turn speed on a full refresh. Just a longer time if I don't turn the page quickly.

[John F]

Originally Posted by meeera

...

Faster CPU means a battery life trade off, ...

Aren't the CPUs used in a lot of the ereaders quite old? You could probably use an updated CPU with better speed and better baterry life?

I'm just guessing.

But in general, I believe that for these portable devices, the CPUs get faster with each generation and the power consumption varies little. I'm thinking of my sony 505 vs my glo hd, or my iphone se vs iphone 11.

[salamanderjuice]

Originally Posted by John F

Aren't the CPUs used in a lot of the ereaders quite old? You could probably use an updated CPU with better speed and better baterry life?

I'm just guessing.

But in general, I believe that for these portable devices, the CPUs get faster with each generation and the power consumption varies little. I'm thinking of my sony 505 vs my glo hd, or my iphone se vs iphone 11.

Yes they are quite old. The Kindle Paperwhite 4 uses a 1GHz iMX.6 sololite single core CPU built on a 40nm process. Amazon put the same CPU in the 2nd gen Paperwhite back in 2013. Modern iPhones use a 5nm process which allows for a much faster CPU at much lower power usage than would be available with a 40nm process.

The Boox devices use newer low-to-mid range Qualcomm Snapdragons built on 14nm processes. To me, my Nova 2 feels considerably faster than my Kobo Aura One was (which has the same CPU as a Kindle Paperwhite 4!).

[Ereadlog]

Originally Posted by Hyphen-ated

In the spirit of being the change I want to see in the world, I have measured the latency of my Kobo Aura HD, running software version 4.25.15875

Results:
epub:
avg latency: 1.658s
min latency: 1.574s
max latency: 1.975s

That's interesting. I made quick test just now; it isn't as accurate as yours but still I don't think inaccuracy rate is greater than 10% (+/- more likely real result is lower than measured since imperfect reflex time raises it)

device: Boox Note Air, font 6, setting "V2 engine that supports original layout" or something like that, normal refresh mode

test:
open epub default reading app with some epub (383 pages) and let it index (takes few secs) the book
clear cache / close all apps
open this book again and go to middle of the book
for each page: tap page and stopper and then again stopper when page is turned

results:
1 0,92
2 1,07
3 1,07
4 1,17
5 1,14
6 1,08
7 1,03
8 1,05
9 1,15
10 1,01
11 0,94
12 1,08
13 1,13
14 1,13
15 1,09
16 1,17
avg 1,076875


I made similar test with 818 page pdf and results were the same

Then I also made 2 tests with same books but with "speed mode" and avgs were around 0.6-0.8 . In this mode there is almost no ghosting, just sharpness a bit lower but still v.good for reading.

I owned P10 in the past and compared side by side both devices and actually P10 had even faster page turns (normal modes); FYI P10 is not stable device (hang/freeze) so I returned it

[Ereadlog]

Originally Posted by ottischwenk

Hardly anything has happened in the past 10 yearsAlmost all of them (Kindle, Kobo, aso) - only Pocketbook devices could be more modern (dual core CPU); but I do not know for sure.Fast enough for me - so I have no need to measure it.
My EInk Android tablets are equipped with quad / octa cores, from 1.5 GHz and usually from 2 GB Ram upwards and that is noticeable.

in term of price/performance ereaders are atrocious; the only reason to buy them is eink screen;

if we do not account for the screens: there is new xiaomi 11' tablet - it got 6GB ram, snapdragon 860, 128GB rom
compare this to note air: 3GB, snapdragon 636, 32GB and 2x more expensive than xiaomi

so given pure performance (eg for web browsing) is 2-3x better with xiaomi the price/performance of note air is 4-6x worse; this is really huge and it all comes down to eink monopoly + smaller market for ereader tablets

[Hyphen-ated]

Originally Posted by Ereadlog

device: Boox Note Air
...
avg 1,076875

Interesting. So this android device also has awful performance.
Although your timing method was pretty bad...

I am willing to do the frame counting for any test videos anybody wants to post. Most phones should be able to do 60fps nowadays, but 30fps would still be totally usable

Here's an example of what it should look like https://www.youtube.com/watch?v=M3rbWfmHpQk

[meeera]

Originally Posted by Ereadlog

in term of price/performance ereaders are atrocious; the only reason to buy them is eink screen;

if we do not account for the screens: there is new xiaomi 11' tablet - it got 6GB ram, snapdragon 860, 128GB rom
compare this to note air: 3GB, snapdragon 636, 32GB and 2x more expensive than xiaomi

so given pure performance (eg for web browsing) is 2-3x better with xiaomi the price/performance of note air is 4-6x worse; this is really huge and it all comes down to eink monopoly + smaller market for ereader tablets

It comes down to having choices, and using the right tool for the job. Not "accounting for the screens" is basically missing the entire point. People don't buy eink readers for web browsing; they're for reading static pages. I've got an iPad mini that I know has a faster CPU than than my Kobo, and will turn ebook pages faster; but I prefer the Kobo for reading, and the page turns are fast enough. Pretty simple, really. I also know it won't die if I spill coffee all over it, and it's light enough to hold in my hands for long periods of time. Those are pretty killer reader features for me.


If someone prefers their different device, more power to them, and I hope they're enjoying their reading. I'm really sick of the black & white personalities who insist that X "is better" for everyone and in all circumstances, instead of realising that different people want different feature sets from their tools.

[ottischwenk]

Originally Posted by meeera

I'm really sick of the black & white personalities who insist that X "is better" for everyone and in all circumstances, instead of realising that different people want different feature sets from their tools.

I have to adapt myself more or less to every "dedicateted reader" - no device really fits. If people are content with that, then it's nice for them.

I adapt an EInk Android tablet to my needs - and that makes the difference in comfort.

[meeera]

Originally Posted by ottischwenk

I have to adapt myself more or less to every "dedicateted reader" - no device really fits. If people are content with that, then it's nice for them.

I adapt an EInk Android tablet to my needs - and that makes the difference in comfort.

Happy reading.

[ottischwenk]

Originally Posted by meeera

Happy reading.

Thank you - that's the point.

[Hyphen-ated]

Originally Posted by Ereadlog

device: Boox Note Air, font 6, setting "V2 engine that supports original layout" or something like that, normal refresh mode
...
avg 1,076875

Oh interesting, so that's an android device that also has horrendous latency.

Although your testing methodology wasn't fantastic...

You know what, I'm willing to do the frame counting work to analyze any test footage that anybody posts. Most phones can take 60 fps video nowadays, but 30 fps is not even that bad when we're talking about latencies THIS high.

You can see an example of what footage should look like at the youtube video with id M3rbWfmHpQk
(This is the second time I've written this post, I linked to youtube before but I think that made a spam filter eat the post)

If I get enough responses I'll make a little website with a nice overview of the data

[Hyphen-ated]

It would probably be a good idea to use the same public domain ebook in tests. Since display lag is my white whale, how about Moby Dick. Specifically the "compatible epub" version from standardebooks dot org. (I would just link to it, but apparently I'm not allowed to link to things)

[JSWolf]

Originally Posted by Hyphen-ated

In the spirit of being the change I want to see in the world, I have measured the latency of my Kobo Aura HD, running software version 4.25.15875

Procedure:
* Get an epub of "Dracula", and go to the first page of chapter 1.
* Record with a 60 fps camera.
* Smack the screen with a finger nice and quickly to get an easily measurable moment of contact.
* Wait for the page to turn.
* Repeat a total of 10 times.
* Play back the video frame by frame and measure from when the finger touches to when the page has turned and is comfortably readable.
* Convert the book to kepub and repeat the whole thing

Results:
epub:
avg latency: 1.658s
min latency: 1.574s
max latency: 1.975s

kepub:
avg latency: 1.914s
min latency: 1.832s
max latency: 1.964s

I'm surprised to see that kepub did worse. I have a very distinct memory of a noticeable improvement after converting to kepub. Maybe some of my books were pathologically slow for some technical reason, and those books were handled better by kepub.

I consider these numbers to be atrocious. Under 0.5s would be more bearable, but still objectively bad.

It's easy to prove that a majority of this slowness is because of poorly written software, not because of inherent limitations of the display. If I immediately tap to turn the page right after a previous page turn finishes, it's way way faster.

Are you refreshing the screen on every page turn? I don't and I know I don't get page turn speed that slow.