You've Got Mail.........DM42L

[Paul Dale]

(02-24-2016 11:26 PM)BarryMead Wrote:  How would one investigate the stack usage (or increase the stack size) on a real device? I don't own an ICE debugger for the ARM cpu yet.

You can't increase the stack size. Without a debugger I don't think there is a way to investigate the stack usage. If it is stack use, I'd expect lots of other values to cause problems.

About the only useful suggestion would be to pull the series expansion and Tan code from the 31S branch and put them into the 34S trunk and see if that has a similar problem. The 31S doesn't use extended precision for its trig and those routines use less memory -- I haven't checked but they quite possibly are accurate enough for double precision.


Pauli

[BarryMead]

Paul:

The problem is more wide-spread than I had originally thought. The flash image that I downloaded from the SVN repository along with version 3884 was not in-fact recently compiled. It was back at version 3869. When I compiled the 3884 realbuild version WITHOUT Bit's patches it ALSO has the RESET BUG.

So this means that EVERYONE who uses the new version to flash a calculator will get a reset when they try to take the tangent of .9999999

Sorry for my earlier assumption errors. I thought that the realbuild directory was always kept up-to-date in the SVN repository, so I ASSUMED (incorrectly) that the
stock flash image was at revision 3884 and error free, which was not the case.

Can you try to compile and flash the current version into a calculator to see if you see the same result. If so and you have a debugger at your disposal, perhaps it would be possible to find and fix the issue!
I created a separate thread as this issue was not really very related to the original thread.

[Paul Dale]

I can't build firmware for the 34S and don't have a debugger -- Marcus has handled this in the past.


- Pauli

[walter b]

(02-23-2016 02:12 AM)rpn fan Wrote:  Now, if we were talking about a premium body for a WP-xxS, I would pay premiums for that. Because they would be moving the ball forward.

A DM42L can only be a projection of an excellent calculator of 1988 on a form factor of 1981. And even an upcoming DM42P will only reproduce a calculator of 1988. Far better than nothing, for sure, and very much appreciated.

Though it will be more important in another way: A DMxxP will offer a platform allowing to extend the life of the WP 34S of 2011. And due to the more versatile display (HP, listen!) it will already allow for an upgrade of said 34S. It will be something like a repurposable HP-17bii+ Silver HP failed offering.

d:-)

[Luigi Vampa]

(02-26-2016 09:17 PM)walter b Wrote:  A DMxxP will offer a platform allowing to extend the life of the WP 34S of 2011.

Oooops!
You have just used 'will', instead of 'would/could'. It might have been your subconscious... or it might haven't ;0)

[walter b]

Ciao Luigi,

That may have been a presumption happening to me frequently: I presume rational people and reasonable developments. May not become reality but would be ... ummh ... reasonable, wouldn't it? And at least two fellow members reported on a future DMxxP, didn't they?

Cordialmente,

d:-)

[Thomas Okken]

(02-19-2016 09:50 PM)JimP Wrote:  

(02-19-2016 09:38 PM)Den Belillo (Martinez Ca.) Wrote:  Folks who have purchased something from Swiss Micros are receiving notice about the coming DM42L.
There's a link to a Youtube vid: https://www.youtube.com/watch?v=314JMYJE...e=youtu.be
Great news.

OK I was toying with getting a DM-41L but I think this is going to be the one I will get. Is there any indication on how much memory this new beauty will have, or will it be precisely the same as the HP42S?

I'm a bit late joining this thread, but since it seems that this specific question has gone unanswered so far, I thought I'd chime in.
Swiss Micros contacted me a few weeks ago, sending me the link to that same video. The reason they contacted me is because DM42L runs Free42.
So, no, not *precisely* like the HP-42S. They didn't tell me about hardware specifications, but I assume those aren't a secret. In any case, Free42 uses RAM at about twice the rate of the 42S, because variables take up 16 bytes instead of 8, programs are stored in a less compact format, and system overhead is larger. Still, even with 32 kilobytes of available RAM, you'd already be able to run anything that anyone has ever written for the 42S, and it seems likely that DM42L will have significantly more than that.

[Dwight Sturrock]

(03-20-2016 05:23 PM)Thomas Okken Wrote:  Swiss Micros contacted me a few weeks ago, sending me the link to that same video. The reason they contacted me is because DM42L runs Free42.
So, no, not *precisely* like the HP-42S. They didn't tell me about hardware specifications, but I assume those aren't a secret. In any case, Free42 uses RAM at about twice the rate of the 42S, because variables take up 16 bytes instead of 8, programs are stored in a less compact format, and system overhead is larger. Still, even with 32 kilobytes of available RAM, you'd already be able to run anything that anyone has ever written for the 42S, and it seems likely that DM42L will have significantly more than that.

Looks like a marriage made in heaven, Free42 plus modern hardware. Portrait preferable, but landscape would be ok too.

[Paul Dale]

(03-20-2016 05:23 PM)Thomas Okken Wrote:  In any case, Free42 uses RAM at about twice the rate of the 42S, because variables take up 16 bytes instead of 8, programs are stored in a less compact format, and system overhead is larger.

Thomas, any idea as to the effort require to make Free42 memory use more compatible with an embedded system? Not so much the extra RAM uses mentioned above, which aren't a major issue. More the freer allocation/freeing of RAM and the possible fragmentation risk.


- Pauli

[Vtile]

So where I can send the order. :o

[Thomas Okken]

(03-20-2016 10:48 PM)Paul Dale Wrote:  

(03-20-2016 05:23 PM)Thomas Okken Wrote:  In any case, Free42 uses RAM at about twice the rate of the 42S, because variables take up 16 bytes instead of 8, programs are stored in a less compact format, and system overhead is larger.

Thomas, any idea as to the effort require to make Free42 memory use more compatible with an embedded system? Not so much the extra RAM uses mentioned above, which aren't a major issue. More the freer allocation/freeing of RAM and the possible fragmentation risk.

- Pauli

Hi Pauli,

I'm not sure what you mean by "freer allocation/freeing of RAM" (garbage collection maybe?), but as far as fragmentation is concerned, that wouldn't be difficult to deal with in an embedded environment.

The only functions called by the Free42 core that aren't in the Free42 shell are malloc(3), realloc(3), and free(3). With a bit of C++ magic, those could all be overloaded to use movable memory blocks, and because the Free42 core is single-threaded, that wouldn't require any code changes to lock and unlock those memory blocks.

This wouldn't be something you could do in an afternoon or a weekend, but it shouldn't be a major rewrite either.

- Thomas

[Thomas Okken]

(03-21-2016 02:41 AM)Vtile Wrote:  So where I can send the order. :o

https://www.swissmicros.com/

DM42L is not an actual product yet, though, so I don't think you can place an order just yet... But you could let them know you're interested.

[Paul Dale]

Thomas, thanks for the quick response. C++ ought to be able to hide the implementation details of reallocatable memory blocks with a (small) performance loss.


- Pauli

[JeffJ]

On 9 May 2016, I asked Michael Steinmann about the status of the DM42L project and he said it was taking a little longer than expected, but a release date in September should be possible. So we're getting closer. I already have my money set aside to purchase one.

[rprosperi]

(05-12-2016 06:45 PM)JeffJ Wrote:  On 9 May 2016, I asked Michael Steinmann about the status of the DM42L project and he said it was taking a little longer than expected, but a release date in September should be possible. So we're getting closer. I already have my money set aside to purchase one.

Thanks for sharing the update Jeff.

Hmmm.... HHC just happens to be in September....

[walter b]

See, sometimes developments are simply reasonable.

d:-)

[Alejandro Paz(Germany)]

Quote:So we're getting closer. I already have my money set aside to purchase one.

How much do you think/guess it will cost ?

[JeffJ]

Check the SwissMicros web site for similar models already for sale. The DM-41L is approximately $130 US, so I would expect the DM-42L to be in the same price range.

[vido]

(08-25-2016 05:54 PM)JeffJ Wrote:  Check the SwissMicros web site for similar models already for sale. The DM-41L is approximately $130 US, so I would expect the DM-42L to be in the same price range.

The 42 will have a multiline screen and its keyboard design is also different. Also, it is apparently requiring some lengthy development time, so I would expect it to be more pricey than the 41L, which was leveraging the hardware design of the 1xL series.
Anyway I'm waiting for it to be available to order from Swissmicros, of course :-)

[Massimo Gnerucci]

(08-26-2016 06:44 PM)vido Wrote:  

(08-25-2016 05:54 PM)JeffJ Wrote:  Check the SwissMicros web site for similar models already for sale. The DM-41L is approximately $130 US, so I would expect the DM-42L to be in the same price range.

The 42 will have a multiline screen and its keyboard design is also different. Also, it is apparently requiring some lengthy development time, so I would expect it to be more pricey than the 41L, which was leveraging the hardware design of the 1xL series.
Anyway I'm waiting for it to be available to order from Swissmicros, of course :-)

You can check what Michael said on the other thread...