General technology communities have E3 and CES, but we at Cemetech have Teachers Teaching with Technology or T^3, TI Education's annual conference. This year, we're bringing you the minute-by-minute updates from T^3 2015 in Fort Worth, Texas, right in TI Education's backyard. As in the previous two years that I attended, I'm looking forward to attending a lot of fascinating sessions with educators using TI calculators and connecting with teachers and with TI. I'll also be seeing how our mission to spread graphing calculators as a programming tool can fit into what else teachers do in the classroom. As always, keep your eyes on Cemetech for the latest that I and my colleague DrDnar discover here, experiences with the new TI-84 Plus CE, and everything else education and TI-related that we explore.

T^3 2015 kicked off with a drizzly Texas morning walk to the Fort Worth Convention Center, a short distance from our hotel. As is traditional for T^3 conferences, we were first welcomed by Gayle Mujica, TI's Marketing Director, then Peter Balyta, the President of TI Education. Both stressed the utility of the T^3 conference as a way both to learn from and teach your fellow educators and a forum to understand TI's technology and provide feedback back to TI. Since the conference is being held in Fort Worth, there will be a lot more TIers than usual attending. We also heard about tomorrow's Pi Day celebrations, including a 3.14-mile run and plenty of pie. Dr. Balyta introduced the two latest updates for TI's two hero lines: the new TI-84 Plus CE, and TI-Nspire OS 4.0's new math features.

The keynote speaker this morning was Dr. Jo Baoler, a mathematics education professor from Stanford. She spoke about her work in understanding and improving math education, especially with students who think they're just not good at math. Dr. Baoler led with neurophysiological research that shows there truly are not students who are innately better or worse at math, as demonstrated by the neuroplasticity of the brain. She led the audience through a fascinating example looking at how different people look at and interpret a diagram of a set of blocks that grows quadratically. We looked at a group of at-risk teens exploring this problem, and tried to understand what made them so enthusiastic to understand and solve this particular problem. Dr. Baoler left us with a few thoughts about what is good and bad about the state of math education, especially from a neurological standpoint: mistakes are good, and every time you make a mistake and learn from it, you grow a synapse.

If you're here at T^3, I hope you'll stop me, say hello, and share some of your own experiences with programming or with T^3, and keep your eyes on Cemetech for more from the conference.

Left to right: Introductions from Gayle Mujica, Peter Balyta, and Jo Baoler
I'll be looking forward to hearing about your and DrDnar's experiences/opinions on the CE.
In the interest of not saying anything that our friends at TI would find inaccurate, I'll clarify that everything below is impressions on the pre-release TI-84+CE Classic, with the so-called "version 89" OS (OS 5.0.0.89), based on a brief amount of time spent with the device. DrDnar and I didn't load anything untoward onto the calculator other than some benchmarks to see how fast matrix multiplication is, and two small nondestructive assembly tests.

Benchmarks
- The display has gotten much faster. Running a TI-BASIC program that counts from 1 to 100 on the homescreen and displays each number yields a 2.6x speedup:
-- TI-84 Plus CSE: 26.36 seconds
-- TI-84 Plus CE: 10.15 seconds
- The math core has not gotten much faster. Counting from 1 to 10000, adding 1 to the loop variable in each iteration, with no displaying yields a 15% speedup:
-- TI-84 Plus CSE: 25.35 seconds
-- TI-84 Plus CE: 22.13 seconds
- Although matrices can now be very large with all that memory, matrix multiplication operations are limited to 400 elements to prevent long churning. 20x20 matrices can be multiplied; 21x20 cannot.

Bugs
I tested some of the more annoying bugs from the Big Bad Bundle of TI-84+CSE OS 4.0 and 4.2 Bugs.
:: Fixed: Pt-On( and related functions display their color without refreshing the graphscreen.
:: Partially fixed: Menu( with 9 items now displays 9 items, but the 8th and 9th show up as ? and yield an ERR:LBL if you click them, even if you have the correct label.
:: Fixed: Even without Doors CSE's alpha-scrolling tweak, scrolling programs is now much, much faster.
:: Still broken: Setting deltaY to 1 and Ymin to 0 still incorrectly sets Ymax to 165.
:: Still broken: The numbers at the top of the Memory menu are still not right-aligned properly.

Other notes:
- No new programming commands or tokens that I could find, other than AsmPrgm84CE
- You can drag-n-drop JPG/PNG/BMP/etc images directly into TI-Connect CE to send them as images to your calculator without explicitly converting them first. You can drag-and-drop multiple files out of TI-Connect CE and into TI-Connect CE at once, including image files, as discussed in our TI Connect CE review. In addition, I've learned that TI's SmartView emulator also lets you drag-and-drop images directly into the emulator.
- Detect Asymptotes can no longer be disabled. I made an error here. Detect Asymptotes is not (and has never been) an option in Parametric mode. In Function (Y=) mode, it is still available as before.
KermMartian wrote:
- Detect Asymptotes can no longer be disabled.

The only feature I feel that TI messed up. I played around with a few friends' CSEs and their biggest complaint was the speed slowdown which I managed to fix usually by disabling asymtotes.
KermMartian wrote:
two small nondestructive assembly tests

Oh? I didn't see you do that. And yes, I never I got around to writing any assembly test programs of my own; I've been too busy these past few weeks

I spent a lot of time trying to get a 25x25 matrix multiply benchmark to run, and at the end, an engineer tells me they've limited matrix operations to 400 elements. (It gives ERR:MEM, not ERR:DIM, so I started thinking there was some weird 16-bit code left over.) But I did make a pair of random 20x20 matrices to test and I sent them to the other calculator for comparison, so here are my results (in seconds, using startTimer and chkTimer):

Code:
                TI-84+CSE       TI-84+CE
Multiply        13              10
Inverse         15              12
On a related note, since Disabling Asymptotes has been removed, does this show a noticeable increase in graphing speed? I imagine that changing the step amount increases it considerably, but just wondering if that is any faster. Smile It is a graphing calculator, after all.
Various discussions and discoveries were had:
  • It looks like Zilog's C library is in the boot code, allowing programmers to use C.
  • Perhaps of particular note is the inclusion of Zilog's 32-bit float library, which is faster than TI's BCD routines. It's perfect for graphical routines.
  • C presumably allows TI to develop applications much faster than they could before. No doubt we can look forward to additional apps, unlike the dearth of apps we've seen for the TI-84+CSE model.
Unfortunately, I doubt it's up to the engineers whether we get TI's SDK; we shall see. Talking with somebody at the conference, I did get the impression that there are things the community can do---or more precisely, not do---to make it more likely that TI will release an SDK:
  • Avoid CAS-like functionality. The French market is much more open to it, which is why TI produced a different product just for them that includes an exact math engine.
  • Do not threaten TestGuard and the testing LED. I'd like to follow-up with some questions about how much concerns about the integrity of their counter-cheating technology affects their willingness to be open with us.
DrDnar wrote:
Various discussions and discoveries were had:
  • It looks like Zilog's C library is in the boot code, allowing programmers to use C.
  • Perhaps of particular note is the inclusion of Zilog's 32-bit float library, which is faster than TI's BCD routines. It's perfect for graphical routines.
  • C presumably allows TI to develop applications much faster than they could before. No doubt we can look forward to additional apps, unlike the dearth of apps we've seen for the TI-84+CSE model.
Unfortunately, I doubt it's up to the engineers whether we get TI's SDK; we shall see. Talking with somebody at the conference, I did get the impression that there are things the community can do---or more precisely, not do---to make it more likely that TI will release an SDK:
  • Avoid CAS-like functionality. The French market is much more open to it, which is why TI produced a different product just for them that includes an exact math engine.
  • Do not threaten TestGuard and the testing LED. I'd like to follow-up with some questions about how much concerns about the integrity of their counter-cheating technology affects their willingness to be open with us.

Wow, TestGuard is that important to TI? For the big, scary, standardized SOL tests we have to take, they simply went around and cleared all memory. (The school has TI-83 Pluses, though, but I had my own TI-84 Plus.)
Hitechcomputergeek wrote:
Wow, TestGuard is that important to TI? For the big, scary, standardized SOL tests we have to take, they simply went around and cleared all memory. (The school has TI-83 Pluses, though, but I had my own TI-84 Plus.)
I dunno, I didn't ask about TestGuard specifically; that's speculation, but I plan to follow-up about it. It is, however, clear that convincing them that the community is not a threat to their sales will take time.

And here's Kerm with the TI-84+CE:
Nice photo, taken with a nice camera! Yes, they clearly took pains to make sure that the TestGuard mode is intact, and I think it behooves us as a group to try to respect that in the interest of quid pro quo with their willingness to release an SDK and give us the ability to write Apps for the device. Trying out the device made me hope that that much faster memory-mapped LCD will be something Apps and ASM programs can both use, and I think swaying any sentiment towards us having access to development tools will be easier if we collectively act in a certain way towards the new device in the projects that we publicly release.

Edit: A quick graphing speed comparison:
Wow, why is the CSE so slow? Even my 83+ is faster than that.
The chances of TI EdTech releasing a SDK for the 83PCE / 84+CE are bleak, at best:
* TI didn't release a SDK for the 84+CSE, which became available two years ago, despite the community not attempting to threaten TI's PTT and therefore sales, AFAICT;
* the fact that the community did not, for about five years, attempt to seriously threaten TI's PTT, did not make a Nspire SDK available either. The community started attacking the PTT after TI announced, at the T^3 ( http://tiplanet.org/forum/viewtopic.php?f=49&t=8856 ), that they'd fix the hole of OS 3.1 in OS 3.2.x, despite multiple explanations of the power and usefulness of a more open platform (with native code, even) to top-level executives.
Among others, I spent the equivalent of over a full day of work on writing explanations; that amount of time would clearly have been better spent on, say, libti*. But at least, I tried. The writing, in English, is attached to the first post of http://tiplanet.org/forum/viewtopic.php?f=43&t=8857 .

I'd say, let's firmly expect that TI will release no SDK for the TI-eZ80 series. Then, if they do, we'll be happy.
The community has enough skilled Z80 programmers that the TI-eZ80 boot code + OS's contents will be analyzed over time, like the 84+CSE's code was. It will just take more time and energy than it would with an official SDK.
A relatively widely available platform containing an eZ80 processor will provide further incentive to add eZ80 support in one of the major FLOSS toolchains, too.
Wait, C? That's so cool; it's about time my first programming language became available to use. Very Happy Of course, I don't mind writing in ez80, but always nice to have options! Can't wait to give this new calculator a shot! Smile

I personally think that TI will eventually release an SDK, they seem to be pulling out all the stops on this new calculator. Razz
Lionel Debroux wrote:
The chances of TI EdTech releasing a SDK for the 83PCE / 84+CE are bleak, at best:
* TI didn't release a SDK for the 84+CSE, which became available two years ago, despite the community not attempting to threaten TI's PTT and therefore sales, AFAICT;
That's because, although no one from TI has confirmed this, I believe that they were focusing efforts on the TI-84 Plus CE since mid to early 2013, shortly after it became clear that even teachers found the TI-84 Plus CSE to be a slow calculator.
Lionel Debroux wrote:
* the fact that the community did not, for about five years, attempt to seriously threaten TI's PTT, did not make a Nspire SDK available either. The community started attacking the PTT after TI announced, at the T^3 ( http://tiplanet.org/forum/viewtopic.php?f=49&t=8856 ), that they'd fix the hole of OS 3.1 in OS 3.2.x, despite multiple explanations of the power and usefulness of a more open platform (with native code, even) to top-level executives.
I think there's an argument that had the community waited for the leadership to shift a bit, that patience could have led to more successful communication with TI. Obviously that's purely hypothetical, but I am not convinced that Ndless, NLaunchy, and other such tools from TI-Planet were created as a direct result of TI making the Nspire a closed platform and would not have been created otherwise.
Quote:
Among others, I spent the equivalent of over a full day of work on writing explanations; that amount of time would clearly have been better spent on, say, libti*. But at least, I tried. The writing, in English, is attached to the first post of http://tiplanet.org/forum/viewtopic.php?f=43&t=8857 .
Thanks for sharing that again; it's an eloquent and effective argument for open platforms, and I hope it comes in handy in this discussion as well.

Quote:
I'd say, let's firmly expect that TI will release no SDK for the TI-eZ80 series. Then, if they do, we'll be happy.
The community has enough skilled Z80 programmers that the TI-eZ80 boot code + OS's contents will be analyzed over time, like the 84+CSE's code was. It will just take more time and energy than it would with an official SDK.
A relatively widely available platform containing an eZ80 processor will provide further incentive to add eZ80 support in one of the major FLOSS toolchains, too.
Analyzing the OS will allow us to write assembly programs (for education, and of course games), but without TI's cooperation in giving us an App signing key, we'll be permanently locked out of signing Apps with the 2048-bit RSA key.
Quote:
Quote:
The chances of TI EdTech releasing a SDK for the 83PCE / 84+CE are bleak, at best:
* TI didn't release a SDK for the 84+CSE, which became available two years ago, despite the community not attempting to threaten TI's PTT and therefore sales, AFAICT;

That's because, although no one from TI has confirmed this, I believe that they were focusing efforts on the TI-84 Plus CE since mid to early 2013, shortly after it became clear that even teachers found the TI-84 Plus CSE to be a slow calculator.

Understood, they could nevertheless have provided something, even an undocumented list of BCALL names, which would have reduced reverse-engineering and couldn't have hurt the making of 84+CSE content Smile
The earliest form of TI-68k/AMS documentation, before the advent of TIFS (to the general public, that is), was precisely such a raw names list, AFAIK.

Quote:
I think there's an argument that had the community waited for the leadership to shift a bit, that patience could have led to more successful communication with TI.

The fact is that community did wait for the leadership to shift. 5 years after the initial availability of the Nspire platform on the marketplace (a huge amount of patience, isn't it ?), and less than a year after contacts and explanations to top-level TI executives. Those contacts with top-level TI executives were jump-started in the spring of 2011, as a result of my own OSLauncher experiment, which scared incompetent standardized exam testing regulators and therefore as a result TI, despite it being both unreliable and harmless precisely because of the PTT.
The announcement of closing down OS 3.2 again was an extremely negative signal (~"you'll really get nothing in return for being nice", "we spit in your face"), and predictably, the Nspire castle started falling soon thereafter: multiple independent PTT killer programs, etc. Heck, the events unfolded precisely the way I (and others) stated they would, because I (among many others) have a clue, based on past history of other platforms...

Quote:
Obviously that's purely hypothetical, but I am not convinced that Ndless, NLaunchy, and other such tools from TI-Planet

It's not the first time you're writing "tools from TI-Planet". We've already explained you why the wording carries, at least to some readers, a meaning which isn't true. Let's do it once again, and hopefully for the last time Wink

* authorship shows that Ndless is not made by TI-Planet staff. It's a fact that a number of Ndless-based programs are made by TI-Planet staff, and that updated versions of some of those programs are released at the same time as a new Ndless version - but not really Ndless itself;
* as for nLaunch and nLaunch CX, more than two years after the initial availability of the Clickpad/Touchpad version, I'm not aware that we know who made them. nLaunchy is their publicly-maintained derivative, and both Excale and myself (as a minor contributor) are TI-Planet staff, but the heavy lifting was performed by whoever made nLaunch / nLaunch CX. For all I know, it could be many known persons of the community, or unknown ones coming out of the blue anonymously for a one-off feat. The nLaunchy README shows good English writing skills, but many community members have them.

It's a fact that Ndless and nLaunchy can be downloaded from the TI-Planet file archives - but that's because it's one of the two main community sites with significant interest in the Nspire platform. The other one was active two years ago, much less so nowadays, due to its staff.

Quote:
were created as a direct result of TI making the Nspire a closed platform and would not have been created otherwise.

The 0-day exploiting and prior reverse-engineering and exploration part of Ndless and nLaunchy wouldn't exist on an open platform like the TI-Z80 and TI-68k series.
A subset of the functionality provided by Ndless could definitely exist on a hypothetical open Nspire platform, just like it does on the TI-Z80 and TI-68k series, for building upon and going further than the native code support functionality provided by TI. For instance; loading our own, more advanced ASM program formats (Zehn on modern Ndless versions), and also providing a standard library of functions: most TI-Z80 "shells", starting with yours, and TI-68k/AMS "kernels" with dynamic library support provide such functionality.
  
Register to Join the Conversation
Have your own thoughts to add to this or any other topic? Want to ask a question, offer a suggestion, share your own programs and projects, upload a file to the file archives, get help with calculator and computer programming, or simply chat with like-minded coders and tech and calculator enthusiasts via the site-wide AJAX SAX widget? Registration for a free Cemetech account only takes a minute.

» Go to Registration page
Page 1 of 1
» All times are UTC - 5 Hours
 
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum

 

Advertisement