[Archive] TomTom v9 Firmware and Maps

[ahkp3]

Originally Posted by ahkp3

Ok, then I can confirm that this process seemed to have totally failed on my TomTom One (V2), as the PNDNavigator.clean survived several cold boots until it was finally renamed by me, thus suddenly enabling TTS without further tampering. But I will investigate this thoroughly another time and get back to you.

Originally Posted by dsa8310

Without PNDNavigator TT does not run.

Ok, I am so sorry, that was my mistake. I must have messed up a lot of things. You are indeed right, without PNDNavigator the PNA does not run and simply stalls at the TomTom drum screen without any reaction upon booting the Navcore.

I will now switch over to DSA-12June and try to confirm or deny that there is absolutely no TTS with hardwired-sound on my device ONE (V2).

Regards,

ahkp3

[ahkp3]

Originally Posted by dsa8310

If there is such device which gets loquendo only without HS (and looses loquendo with HS) then the HS checkbox will be preserved.

After running several tests with DSA June12 whether or not Loquendo is possible with HS, I can confirm that on my device, ONE (v2), the only way to get TTS output is to clear the HS checkbox upon generating the Navcore.

With HS turned on, I ran each and every trick to get TTS somehow to run, but to no avail. With HS turned off, TTS works right out of the box.

So, may I suggest, that you keep that box, please. I will start beta field test during the next few days and I am pretty sure, that there are a couple of old devices, especially TT PNAs that didn't come with factory TTS enabled, that will need the sound emulation.

I will keep you posted with more HardwareIDs that need to have HS turned of in order for TTS to work. So far, the list is very short:
TT ONE (v2)

Additionally, I ran some tests with a swapfile of 16 MB on a 32 MB TT and it crashed 2 out of 7 times. However I never noticed additional free RAM compared to a 12 MB swapfile. I will try to alter the script to run the kickboxing mechanism with fixed file sizes of 8, 14 and 18 MB, to find out if there are indeed additional mempages that can be swapped beyond the 17 MB free RAM barrier.

Thanks again for your patience and effort with this marvelous tool,
ahkp3

[dsa8310]

Link to dsa-17June2012.zip:

dsa-17June2012.zip

New: different, hopefully better MB implementation.

Make room for 32MB of free contiguous storage for swap.
All devices, regardless of RAM size, use the same MB algorithm.


Please report issues if any on 32MB devices (I have a 64MB device).
The swap file could be much smaller if space is an issue - as low as 5MB, but with diminished results.

HS is still in place pending further testing - as I said before, if there are devices which don't get LoquendoTTS with HS selected, but only with it unselected, then HS should stay. Otherwise it is not necessary (by default HS code would be on).

[dsa8310]

Essentially, the "stress" file is only 1MB larger than the current free memory, the loop is repeated every 10 seconds, and it ends when free swap gets lower than 4MB or for 7 times in a row there is no decrease in free swap size - so use your device as much as possible in the initial phase, e.g. play the Jukebox, browse menus, etc.

As a result the MB process is much discreet and never causes system restart - hopefully

If necessary, shrink the swap file - under Windows, with your best abilities... under Linux or Mac OSX, with dd.

[dsa8310]

Aha, so HS stays!

I could see the RSS size of ttn (aka PNDNavigator) going as low as 15MB (with top), so there is hope for significant RAM gains even for 32MB devices.
On my go910 with map NA_2GB, TT reports about 32MB of free memory when the MB process is done, but admittedly I went overboard here...

[voyteke]

Originally Posted by dsa8310

different, hopefully better MB implementation.
Make room for 32MB of free contiguous storage for swap.
All devices, regardless of RAM size, use the same MB algorithm.
Please report issues if any on 32MB devices (I have a 64MB device).
The swap file could be much smaller if space is an issue - as low as 5MB, but with diminished results.

as always, much obliged for your work, and help offered

dumb question #1:

is that of benefit on TT720? how do I 'observer' the benefit(s)?
what switches to use on install ?

dumb question #2:
ahem, how can I tell from device what emulation I've installed...?

if I call up skins, will the 1st skin that comes is, does that represent what I'm running...?

thanks

going to re-dsa now

TT720 9.430; 65MB. 24.5 free; 880.3865

[voyteke]

Originally Posted by voyteke

TT720 9.430; 65MB. 24.5 free; 880.3865

is the sed error any problem ?

ooops, doesn't boot anymore...
what was the deal, I need to remove... skins directory.. ?
-----------
Windows
Processing 9.430.1-2008.go.CAB-navcore-GO.cab, please wait...
patching file etc/rc
Hunk #1 succeeded at 42 (offset 5 lines).
Hunk #2 succeeded at 813 (offset 108 lines).
Hunk #3 succeeded at 843 (offset 110 lines).

sed: cannot rename etc//sed2xsAyR: Permission denied

Patching PNDNavigator... done.

Installing to /cygdrive/G/, please wait.
Extracting 9.430.1-2008.go.CAB-navcore-GO.cab
rm: failed to return to initial working directory: No such file or directory
rm: failed to return to initial working directory: No such file or directory
rm: failed to return to initial working directory: No such file or directory
Creating swap...

32+0 records in
Unzipping 9-430-890926-1.win.2.2.xpi
32+0 records out
33554432 bytes (34 MB) copied, 53.455 s, 628 kB/s

Done.

[dsa8310]

Maps are at v890 now.

MB might not work with all modes and devices with a v9 NC, because of the suboptimal kernel.

Nowadays, you may select skins of interest (with Ctrl/click or Alt/click - Mac), e.g. the GO skins and Self, but furthermore, the initial mode could be determined with keys TAB, back-TAB (and mouse) - which set the anchor (current) cell in the mode table. Subsequently, skins are changed with the Skin Chooser as usually.
(no skin selection still means all skins with self as the initial mode)

[dsa8310]

Originally Posted by voyteke

is the sed error any problem ?

ooops, doesn't boot anymore...
...
sed: cannot rename etc//sed2xsAyR: Permission denied
...

This was dubious, the others are OK - please reinstall dsa as recommended.

Note that MB with v9 on go720 does not work - because of kernel; try the initial 25 mode.

On my go910, nc v9.450 (probably v9.430 also) works with MB at least in the 25 (x30) mode.

...
I will switch under Windows to try to reproduce your error.

[voyteke]

Originally Posted by dsa8310

I will switch under Windows to try to reproduce your error.

I deleted skins, re-patched, all's well:

//little later//

deleted skins, repatched:

Windows

Processing 9.430.1-2008.go.CAB-navcore-GO.cab, please wait...

patching file etc/rc
Hunk #1 succeeded at 42 (offset 5 lines).
Hunk #2 succeeded at 813 (offset 108 lines).
Hunk #3 succeeded at 843 (offset 110 lines).

Patching PNDNavigator... done.

Installing to /cygdrive/G/, please wait.
Extracting 9.430.1-2008.go.CAB-navcore-GO.cab
rm: failed to return to initial working directory: No such file or directory
rm: failed to return to initial working directory: No such file or directory
rm: failed to return to initial working directory: No such file or directory
Creating swap...

32+0 records in
Unzipping 9-430-890926-1.win.2.2.xpi
32+0 records out
33554432 bytes (34 MB) copied, 45.284 s, 741 kB/s

Done.
=============================================
24 x40
after boot: free 25MB

[dsa8310]

The "rm" errors happen under Windows, they are not fatal (Cygwin something).

[omolino]

Today i test the old DSA 6 jun with 9430Nav for One (SKIN 34) with MB and HS

It works very well, fast and loquendo v7 speek perfect without stops
in One V6 i have 6Mb of free ram and NO problems with space in HD

tHANs very much for your work, the new N9 ARE so fast than the old V8

now i realice that there are a new versión of DSA 17 june, and i'm sure that it will be better, now i going to do a new nav Ultra light for ONE

THANKS very much for your work !!!

[omolino]

I have not tested yet, but i did a DSA with HS and MB (skin 34)

I delete the extras and some files that are not necessary to have a small nav simple but fast for devices that have 32Mb and no much resources !!!!

Thanks to dSA !!! and thanks again and again....

if you want to test it in One, start... etc small devices :

DSA ULTRA LIGHT 3 (By Dsa 17june)
===========================
Zippyshare.com - 9430 UL3.rar

Thanks again !!!

[ahkp3]

Originally Posted by dsa8310

Essentially, the "stress" file is only 1MB larger than the current free memory, the loop is repeated every 10 seconds, and it ends when free swap gets lower than 4MB or for 7 times in a row there is no decrease in free swap size - so use your device as much as possible in the initial phase, e.g. play the Jukebox, browse menus, etc.

As a result the MB process is much discreet and never causes system restart - hopefully

I just tested the new algorithm, but I am sorry to inform you, that it crashes after 8-16 minutes. This is reproducable every single time if you just let the TomTom sit idle. Highest free RAM I saw was 19.1 MB, but only for a split second, which is around 2.5 MB up from the previous memboostcode.

Looking at the code, I think the ending conditions of the script are very hard to achieve on a 32 MB TomTom:

- condition 1 is met when free swap file space is below 4 MB. With a 32 MB swapfile and around 19-20MB of data to swap at the very maximum (deduced from observed max free RAM above), there will always be at least 12 MB unused within the swap file. Therefore this condition is very unlikely to be met at all and therefore trigger the end of the script.

- condition 2 kicks in if for 7 occasions in a row, the free swap file size has not been reduced. This condition, on my opinion, is also very hard too meet:

What happens when free RAM has practically maxxed out at around 19 MB, is that the next kickboxingfile will force the kernel to swap mempages, that are likely to be used again during the next seconds. So these will make the free swap size smaller for a fraction of a second, thus resetting the counter if it even has accumulated anything so far. However during the next computing cycles these mempages will be pushed back to memory thus increasing free swap file size by around 2-3 MB.

I did observe that after maxxing out the free RAM at around 18-19 MB it suddenly drops to around 15-16.5 MB, as soon as you just switch back to the navigation view (If you don't do that, the TomTom crashes while being idle). The memboost script kicks in again regaining these 2-3 MB and resetting the counter back to zero each time.

Therefore, from my opinion, it is very unlikely, that a counter of 7, which is the exit value, will ever be achieved, if the memboost script operates right on the margin of maximized free RAM. In this area, the script simply swaps mempages, that are not really swappable because they will be used again shortly, back and forth from RAM to swap, simply to fulfill the demands of the script.


My suggestions:

- swapfilesize: While it may sound great to have a 32 MB swapfilesize for a 32 MB TomTom, I cannot be too fond of that. As you pointed out, the RSS size of ttn (aka PNDNavigator) went as low as 15MB. Given a 32 MB TomTom maximized free RAM should be theoretically around 17-18 MB then. In fact with a 12 MB swapsize I already happened to have around 16 MB free RAM with the old memboost code and 9.430. So, from my opinion, anything larger than a 16 MB swapfilesize is very questionable regarding its efficiency. But given that the memboost code should also work for 64 MB TomToms as well, I am okay with the size of 32 MB.

- exit code:
I was thinking about altering the exit code in a way that not "no decreases" in free swap file size are counted, but actual increases. With the memboost script working, the sudden increase of free swap file size indicates, that non-swappable files were indeed swapped during a kickboxing cycle and are now being pulled back into RAM. This will usually occur only when free RAM is close to its maximum. If this happens one more time, then the script should exit immediately to avoid crashing.

But I am not sure if this also works, if someone is using the Tomtom right away instead of letting it sit for the first three or four minutes. Maybe there should be another condition to be met, like "free RAM must be at least 12 MB, too".

Instead of crossing the line to approach maxximum free RAM, I could think of an approach of diminishing differentials in free swap file space. When the decrease in free swap file space is anything smaller than let's say 512 kB for let's say the third time (not necessarily in a row, just total!), the script should stop. This would also cover the fact that free swapfile space increased.

From my point of view, it makes absolutely no sense to run the script for a forth or fifth time again if the outcome is 512 kB or less each time.

The routine would then look like this:
- kickboxing mechanism
- sleep 15 seconds to allow kernel for swapping
- if (nfs - fs) is smaller than 512 kB (that should include negative values), then add one to counter
- if counter is three, then exit script; else repeat kickboxing

I think, there should definitely be a sleep pause between kickboxing and measuring free swap file space, to allow the kernel to finish swapping

But maybe there is a much smarter way... What do you think?

Regards,

ahkp3

[ahkp3]

Originally Posted by ahkp3

- if (nfs - fs) is smaller than 512 kB (that should include negative values), then add one to counter

Sorry, I was still asleep when I wrote that.

Should be:
- if (fs - nfs) is smaller than 512 or negative