Early this week, I had the pleasure of speaking at length with GT Springer, the Education Solutions Architect for HP's calculator division. We discussed the upcoming HP Prime graphing calculator, and I was interested to learn about quite a few unique features of the new calculator. We touched on the programmability of the device, the pedagogy of the pre-loaded applications, the intriguing graphing tools it includes, and much more. I will soon have a full set of technical specifications and some screenshots of the graphing application in action, but for now, read on for the full interview summary.

Cemetech: Please give us an initial overview of the HP Prime calculator.
GT: The HP Prime launched at this year's annual National Council of Teachers of Mathematics (NCTM) conference. It won't be available until Fall, but the software was far enough along for a demo and a few teacher presentations at the conference. The HP Prime is a multitouch, full-color graphing calculator. The architecture is based on the HP-39gII; it does have HP Apps, and it incorporates a CAS in a very different way from competitors. For HP's competitors [TI and Casio, specifically], a CAS machine always has a CAS.

For us, the CAS is more like an app, so you can chose to use the CAS, or you can just use Home View (the numerical solver) for simple math and calculations. There are also a number of specialized editors for lists, matrices, programs, etc. You can easily use data between the CAS and Home View modes. The HP Apps are the basis for the useability and learnability of the device. We enforce a fairly strict structure in Apps. All have Views, and each of those Views map directly to the NCTM standards: numerical, symbolic, etc. Views, each with dedicated buttons on the calculator. Each View has a setup screen accessed via [shift][that button]. The re-use of Views across Apps gives the device a high degree of learnability. Each of the Apps encapsulates specific functionality and uses that to store results in variables. Apps can share those variables (and users can access Apps' functionality from HomeView); programs can access those as well.

Cemetech: What Apps particularly stand out for you?
GT: There are three Apps that are unique to the Prime: (1) Dynamic Geometry, (2) Spreadsheet, and (3) Advanced Graphing. The Advanced Graphing App can handle virtually anything on an XY plane, including things like 1<0, 1>0, and any conic section. It doesn't sample x and paint y pixels [like the TI-83+/84+ and the TI-Nspire]; it uses an interval arithmetic approach. It can also handle implicit relations and inequalities. If you enter x^2+y^2<16, it will graph the disk. We have been finding a lot of interest from the teacher community about this feature. I think there's an opportunity for teachers to get excited about new areas. For example, when we showed y mod x = 3, most of the teachers hadn't thought about what a function like that would look like.

The Spreadsheet App allows us to choose whether to use simple numerical features or the CAS engine on a single-cell, single-row/column, or whole-spreadsheet granularity. You can even do things like x^row+1, and fill whole spreadsheet with that. The new Dynamic Geometry App has some parallels to Geogebra; it has different Views, one where you build calculations/measurements, one for plotting, and so on. Teachers will think of plenty of applications for these that they haven't of course.

Cemetech: Tell me about the multitouch touchscreen. Have you done student testing yet? Will this device be accepted on standardized tests (which currently prohibit touchscreen devices)?
GT: We decided to make it a multitouch device; we stuck to three design principles while creating the device: (1) provide a high degree of mathematical fidelity; (2) give students the touch they've come to expect from their technology; (3) retain a familiar framework for solving mathematical problems. For example, In the Spreadsheet, you can use a multitouch pinch gesture to widen spreadsheet row. I'd describe our attitude towards the Prime as "careful progress with an eye toward the future". It's still a little too early to talk about student testing; we're still in the early days of our understanding of touch technology in the classroom.

The touch features are somewhat restricted: you can't sketch out or write down answers to steal them from an exam, for example. The CollegeBoard will probably not be ones to stifle progress in education technology. We're cautiously optimistic to make it possible for examination boards to see this as a touch-based graphing calculator, not a general touch-based device. It has a very typical graphing calculator keyboard, for example.

The Prime also has an examination mode: teachers can create custom configurations for tests, including disabling the CAS, disable any specific Apps, can start with fresh Apps (which follow a document-like model [like the TI-Nspire]) or just disable saved apps with saved notes/data. There are also three LEDs at the top of the device that will blink those in a random pattern based on the testing configuration. Although the pattern is random, they will blink in sync with other devices with the same configuration. We believe we provide a level of security above that of the competitors.

Cemetech: What connectivity and programming options will the HP Prime provide?
GT: We will provide computer software that allows you to enter data, write user programs, etc directly on a computer. You can attach user programs to Apps; for example, you could write a program to extend the Function App. It's even more extensible because you could eg write a program named PLOT. In the Function App, you could press PLOT and have that PLOT program run, then go to the plot View. The on-calculator language will be similar to the HP-39gII's language; I don't know of a C SDK planned, but I can check.

We decided to extend the connectivity kit: we will provide a wireless module that clips into the microUSB port, with a corresponding accessory that connects to teacher's PC. We don't use 802.11 because we don't want the teacher to have to do a lot of setup. It's more like your wireless keyboard/mouse: you plug it in and it works. Teachers can use it to do formative assessment in class. You could collect results from a poll during an AP Stats class, then send cumulative results back to students' devices for analysis for homework.

As far as connecting to sensors, our competitors do data logging, collecting so many samples per second. Our approach is more like the monitoring systems students would see in professional lives. The calculator automatically detects sensors and starts streaming data from the sensors; you see the stream as a graph. There are four sensors channels, so you could get (say) a thousand samples/second with one sensor, or have the calculator intelligently allocate the available bandwidth among four sensors.

Cemetech: Who would you say your target audience for the HP Prime is?
GT: I suspect there will be a Venn diagram of customers: it will be useful for high school students, but with the Solve App and with "Full Unit Support" and the CAS it will be of interest to engineering and math universities. The programming features will also appeal to college students. The ability to pre-populate data into Apps so that every student starts on the same page during a class discussion will be attractive to high school math teachers/students.

Cemetech: Will RPN be available? What about 3D graphing?
GT: RPN will be a mode setting, and still very much a first-class citizen. This is a departure from the 39gII, where RPN was relegated to an App. The RPN availability will help make it more attractive in a university setting. 3D graphing will not be in the first release. It's in our list of "nice to have", but it's not vital for the first release.

Cemetech: Thank you!

We were able to get an additional note from Jason Smith, the Product Manager for HP's calculator division, about the Prime's name:

Thanks to HP for being forthcoming with these details. Cemetech will soon be getting you screenshots of that new graphing application in action, a full set of technical specifications, and eventually, a hands-on review of the device.

Definitely looking forward to this calculator - though I probably won't be able to afford it, it looks quite admirable, and hopefully will continue to reshape the hardware used in graphing calculators.

Should be "...to 'Geogebra'; it..."?
Also, what is "RPN"?

Great scoop! Look forward to more pictures and details.

Wait, since when did the 39gII have any RPN functionality whatsoever, app or not? (I'm sure you could do something similar to RpN III in the 39gII's programming language, and it'd be plenty fast because 66 MHz ARM9, but still...)

And, what do they mean by, "first-class citizen"? We know that it's not a 100% RPN machine, because programming has already been said to be 39gII-style (which is definitely not the programming-works-exactly-like-normal-use keystroke programming model of the 4-level-stack RPN calcs, or the programming-works-basically-like-normal-use-model (that, for simple stuff, can be treated as keystroke programming) of the RPL calcs). The 49g/49g+/50g can actually do RPL programming in both RPN and algebraic (although the programming model gets a bit weird in algebraic at times, because it clearly wasn't meant for it, but it DOES work (mainly by abusing the fact that an algebraic expression is a form of object in RPL)), whereas this is purely algebraic as far as programming.

Also, wow, this thing is really educationally focused, even more so than I thought - splitting the CAS out entirely (whereas the 50g, the CAS is always there to an extent) is something that only makes sense if you're trying to reassure test boards that the CAS is a discrete feature that can be locked out. It's rather telling that, IIRC, HP people have actually said that the 50g isn't going away any time soon. (Then again, the 41 family kept running AFTER the 28 line was discontinued, and even a few months into the 48's run.)

I hope HP succeeds with this and this can finance more professional machines, but as I said in the last thread, I suspect that their prospects in the US are poor due to TI's dominance. (Outside the US, it could get interesting.) And, this isn't the kind of calc I want (although I'll certainly play with one in a store if I see one).

As far as third-party development goes, what usually happens is, HP will release a new platform with a decent on-board language (at least for its time). Then, as people start exploring the calculator, they figure out the bugs/features that let them go deeper into the existing language ("synthetic programming" on the HP-41, System RPL (the HP internal variant of RPL used for higher performance but without input validation) on the HP-28/48/49/50). Then, HP ends up releasing docs on how to exploit that. Of course, that's not enough, people want to program on the metal, so they figure out how to do that, and then HP releases docs on THAT. And, in the case of the RPL machines, the 50g *SHIPS* with tools to program in System RPL, Saturn assembly (Saturn being the HP custom CPU that most RPL calcs use, and the 50g emulates), and ARM assembly - two of which were never meant to be touched by the user, and one of which didn't even exist when the OS was designed.

As far as what RPN, or Reverse Polish Notation, is... it's a different way of performing calculations, that's much closer to what the calculator actually does. You put numbers into a stack, and then perform operations on them. So, the simplest example is, 2 + 3 ENTER. On an HP, it's 2 ENTER 3 +, enter serving a completely different purpose.

When you put in 2 + 3 ENTER on an algebraic machine, the calculator looks through the expression. It sees the first 2, and sends it onto the stack. It then sees the +, and sets it aside, waiting to see what it's adding. Then, it sees the 3, and puts that on the stack. Once it sees that everything is done, it goes ahead and pushes the + onto the stack, and executes the stack as-is.

When you put 2 ENTER 3 + on an HP, the 2 ENTER puts the 2 onto the stack (although there's differences in behavior between various HP models, I'm oversimplifying here, but any model introduced nowadays behaves like this). Then, the 3 is lined up to go onto the stack, then the + consumes both the 2 and 3, returning 5 onto the stack. It's computationally more efficient, and it removes all ambiguity (variations in how the calculator implements order of operations never come into play). On the downside, it does require you to think about and disassemble an equation, rather than just blindly enter it. On the upside, it does require you to think about and disassemble an equation.

And, for completeness, RPL is a sort of Forth/Lisp hybrid that HP uses as the environment that all of their graphing calculators (depends on how you define graphing, but if your definition requires that it be on the calc with no additional hardware, that works) run under. It was designed as a successor to the 4-level-stack RPN machines, and it uses an RPN syntax, although it's different enough in behavior that RPN vs. RPL is the vi vs. emacs battle of the HP community.

Interesting interview. It confirms what we (or most) suspected. It boils down to a price in my view. Most of my calcs are from eBay because I conside the retail price of TI's machines to be inflated.
I enjoy the hacking side of a small device and as BrandonW has stated the past, he knows everything there is to know about the TI-8x which is a joy to behold. This would not be possible with a say, a modern PC or a more sophisticated device. There is a certain "Venn diagram" effect with users and I feel this also applies to platforms too. I would probably fall back to Excel on my PC rather than do something similar on calc. I would also use MathCAD for complex modelling rather than attempt the same on calc. So it's an educational device really, as professionals will use the means listed above. However, there is no denying that it looks great and will appeal to the iPhone generation. It does make the TI-8* line look severely dated which will be a concern.

BTW, looks like the clock speed is 400 MHz, if that hadn't been posted: http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/archv021.cgi?read=242415#242415

And, just to be sure it's not underclocked...

http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/forum.cgi?read=242907#242907 is a post with a benchmark result, compared to the 39gII, and that indicates ~385 MHz for the Prime (well within the timing error to be 400 MHz).

Bhtooefr, thanks for linking that thread. 400Mhz is quite a respectable speed, at least compared to the currently-available Casio Prizm and TI-Nspire calculators. Since it would seem to be running a 400MHz ARM processor, I hope there will be opportunities to find a way to run native ARM code on it.

ti83head, I completely agree: one of the great characteristics of the TI-8x platform is that one person can get a handle on how everything in it works, from the hardware to the software. That's not possible with almost every complex modern device, which uses hardware and software from a hundred different sources. I also agree that this is squarely aimed at the advanced high schooler / engineering college or graduate student, and not as much at the professional, although I see no reason why with the proper programming features and applications it could easily be made a professional calculator as well.

Looks very cool! We need specs though!

Specs are coming soon; I'll be posting an article next week with full specs and some screenshots of that Advanced Graphing App in action.

The HP-Museum threads are interesting, but sadly, contain posts from a well-known attention whore self-advertising his own site, a site which routinely boasts world exclusivity on flat out wrong information (wrong specs for the Prime, a so-called internal WiFi for the Prime, etc.)...

I would have to disagree on that. One of the largest complaints about the 50g is that the CAS caused so much headache when not desired. You can't ever work purely numerically on it because once you get past basic operations the CAS can start sticking itself in when not desired. I can think of plenty of "engineers" and "professionals" that have made that claim over the years.

Think about all of the "well, just set this flag/that flag/this other, put it in this mode" that has to be done on the 50g to effectively use the CAS. And then all of the times that STILL has to be done on the 50g when working 'numerically'.

You can still call CAS commands from anywhere in the system. You just have to consciously choose to do so. It is definitely different then the 50g (or nspire, or classpad), but actually works very well in practice. After all, most of the time simple number crunching is what it is all about. Having the CAS stick itself in and screw something up happened to me all the time on the 49/50 during my engineering courses in school.

When you switch to the CAS, everything is now purely symbolic and exact. You have to manually choose to start working numerically. It is SOOOOO much easier to do things with this clear separation.

If by keypad, you mean the directional pad... this is because the calculator is so darn thin that putting it on the edge feels awkward. It was very surprising to me too.

If by something else, please elaborate and I'll explain why it is how it is if I can.

Tim, I wouldn't put too much weight in Willwac's post. I suspect what he really meant was, "as a TI calculator user, I find the key layout unfamiliar". That's not really a valid complaint in my book. I had to get used to the Casio Prizm's odd keyboard layout after a decade of using TI calculators; I would expect to have to get used to an HP calculator's different layout as well. To be fair, the TI-Nspire has the directional pad in the middle too, so it's not a completely radical change. Also, that's great to hear that it's a very thin calculator!

Yup, I suspect that is the case.

However, there are plenty of people that I've talked with that have complaints about something and then after hearing how the device operates, and the reasoning behind it, it is much less worrisome.

Besides, I think we can all agree that the Casio keyboard layouts are ... less then optimal...

Umm... Did my post get deleted?

Anyways, I think that having the enter key where it is isn't a good placement. Kerm, I think that you are right, I've used TI's calculators, and I've never used an HP, or CASIO.

Also, when I read what you wrote, Kerm, I took it offensively at first. .

Alright, now that makes perfect sense. I usually don't do things that trigger the 50g spit out something symbolic when I don't want it, (usually just keeping it in approximate mode is enough for what I do), but can totally see that happening and being annoying.



HP historically had the enter key above the operations (on the left side, though. That changed with the Pioneer series, IIRC, when the Enter key stayed, but operations moved to the right side. Then, with the 49G, the enter key moved to a TI-like layout, and now, the enter key moved above the operations but still on the right.)

This is definitely one nice looking calculator, and planned to be added to my collection, along with getting the 50g, at some point

Eager to get some more of those juicy details on specs and programmability!

If I'm not mistaken, I think this is also supported in OS 3.2 and upwards for the TI-Nspire. But anyway, all those features sure look neat

Sounds interesting, though it sounds like it will go by the same trends as the other educationally-oriented TI and Casio color calcs. I could be wrong, though; we'll see what happens.

In terms of on-calc programmability, I consider the 50g the ultimate calculator I've seen so far in terms of on-board development tools (though some of the programming documentation leaves quite a bit to be desired). It would be difficult for me to be interested in any other calculator model unless it matches or exceeds this. I'd miss the RPL programming model, though it's a bit weird and has disadvantages when it comes to deciphering code you wrote in the past, there's a certain degree of “fun” and appeal that I have with a stack-based approach, as well as the performance gains you can get if you know what you're doing. Though, with a more powerful processor and/or a native OS rather than emulation, the performance aspects become less important, though in a way it loses the admiration and appeal of working with a system that was historically designed by necessity for maximum efficiency on very limited hardware.

Interesting how the Enter key's position and shape suggests the RPN-style layout of the 48 series. Too bad the layout appears not to allow simultaneous entry of digits and letters without having to toggle Alpha mode like the 49/50, though it appears there are fewer keys, so that wouldn't have been possible. Perhaps I'm a bit picky and spoiled, though—the 50g allows entry of quite a range of special characters and symbols without having to use the CHARS picker, if you care to memorize the key combinations. I'm definitely also spoiled by the ability to customize almost every key combo on the keyboard—for the few deficiencies that exist in Alpha mode, I can just add keys for additional symbols or whatever functions I want. It's gotten to the point that I have so many custom keys that I don't even remember half of them.