Performed the integral on a TI Voyage 200 which has the same architecture as the TI-89 Titanium with a slightly slower clock speed. About 14 seconds integrating from 0 to 6, and about 15.5 seconds integrating from -6 to 6. Set up the HP-50g as you had and it took over 17.5 minutes to integrate from -6 to 6, a whopping amount of time. Also read through some other comments regarding the HP-50g. The time can be greatly reduced if it's set to approximate and even more important, it's set to engineering (or scientific) with only a handful of digits specified: 0, 1, 2, or 3. Specify a larger number, such as 6 or 8 and the time required skyrockets. It should be pointed out that specifying fewer digits in engineering or scientific presentation of numeric values to reduce run time also reduces precision of the resulting answer, and only a few digits reduces it greatly. I was comparing the results on the 50g with the results on the Voyage and in the video on other other calculators. One can definitely speed up the 50g, but at what cost to precision? The 50g seems an anomaly. Still experimenting with some other settings that wouldn't sacrifice precision. Would be interesting to compare some other HPs in addition to the Prime. In addition, TI now has the Nspire CX II CAS which effectively runs about 3x faster, and HP has the Prime G2 (rev D) that also cranks along at about 3x of the "G1 (rev C). Both were released in 2018. TI-58 Benchmark: Set up the 40 year old TI-58 with f(x) and used Master Library Program 09 which uses a Simpson Approximation to evaluate definite integrals with a specified number of intervals between the lower and upper bounds. Ran it using 100 intervals from 0 to +6 as the lower and upper bounds. As expected, precision in the answer was horrid. Increased intervals to 600 and ran it again. Precision was still lacking with ~2% error. Doubled that to 1200 intervals and ran it again. Took a half-hour and returned an answer with ~0.1% error, about one part in a thousand precision. This is a very wicked function, especially from the interval from ~f(1) to f(6). SwissMicros DM42 Benchmark: The DM42 firmware is based on Free42 which emulates HP-42S functionality. As with the TI-58, it requires programming in the function. One can then use the built-in definite integral function. The approximation used is more sophisticated than the brute force Simpson Approximation and works to a specified accuracy factor instead of a given number of evenly spaced intervals. Set the lower and upper bounds to 0 and 6, and set the accuracy factor to 1 x 10^-11. Took about 14 seconds. Set the lower bound to -6 to run it as had originally been intended in the video before running into problems with the HP-50g and ran it again. Took about 14.5 seconds. This was with its processor clock running on the lower battery power speed. Plugged into USB power and running the processor clock at high speed, it cranked out 0 to +6 in about 5.5 seconds and -6 to +6 in just under 6 seconds.
Just ran this inegral test on a European TI-30X Pro MathPrint - the newest of the TI scientific that replaced the European TI-30X Pro MultiView in 2018. The European 30X Pro MultiView is the same as the TI-36X Pro in North America; there's a significant marketing reason for the model number difference. Its MathPrint replacement significantly faster, doing this integral speed test consistently in ~8 seconds - versus a lethargic 108 seconds. This speed is phenomenal. Ran the same integral from -6 to +6 (which was your original plan) and it only increased to ~9 seconds. Then I cranked up the "epsilon" (error precision) from 1x10^-5 to 1x10^-12, which one would expect to drive up the time required, but it didn't. Nobody should be buying a TI-36X Pro in North America. They should be buying the 30X Pro *MathPrint* ( *not* the MultiView). It looks completely different - with a profile like the Nspire CX II in shape and thickness. They can be easily bought from Amazon UK or Amazon Germany (buying it on Amazon US is much more expensive) using the same login as for Amazon US - which I've been doing for years now. North American buyers want the *"Pro"* , not the cheaper "Plus". The latter one deletes four features banned for tests in two German states (Baden-Wuertemberg and Bavaria) and was created specifically for them. The *Pro* version is only a few dollars more and isn't crippled to satisfy regional German testing restrictions.
Very interesting video. For what it's worth, the HP 50g is an ARM based cpu emulating a Saturn cpu. The Saturn typically ran at about 2 MHz, but was overclocked in the HP 48G series to around 4 MHz. The ARM 9 v 2 in the 50g can run up to 238 MHz before memeory starts to fail, however HP under-clocked the 50g to 75 MHz for the conservation of batteries. An HP 50g typically is about 2.5 times quicker than a 48G series calculator in performing most operations. My 50g (which is plugged in) takes about 1 min, 2 sec to perform the integral problem (from 0 to 6), STD mode (12 significant digits). BTW You can use the TIME function to get the time according to the 50g, and store it in a list just like in the first example.
Dear Martin, I've been watching your series of videos on various scientific/graphing calculators. You've done a very meticulous job. I noticed your surprise at the duration it took the HP50g to evaluate the integral. Being a user and an owner of HP50g, I tried to replicate your results with the same input. The HP50g produced a result almost instantaneously but certainly not as quick as HP Prime or the TI-nSpire but was darn close; about 1 sec. At first I was baffled why it took you so long. Then I realised that your settings were set to STD. So what you must do: 1. Make sure approximate mode is set. Press MODE, then CAS and tick APPROX 2. Change number format from STD to ENG. Press MODE, (second row) and change the number format I hope you can redo the tests again or edit this video by removing the part for the HP50g because leaving the video like this is very unfair to HP50g, which should be the 3rd best among those you're reviewing in this videos. You did not tell the calculator to approximate the value of the integral and so the calculator took a longer time trying to come out with a symbolic answer instead of a numeric one. If you still cannot evaluate the integral under a second, I can post a video response. However, I don't think anything's wrong with your calculator and you shouldn't have any problem. Thanks and cheers.
Medwatt, thanks very much for the feedback and explanation. I have gone and added a couple of annotations to the video about the performance and result. When time allows...life very busy at the moment...I will follow-up with another video to clarify and demonstrate the configuration issue. Cheers, Martin.
Thanks for the info! I wonder if the TI-89 is also being hurt in these scores by performing CAS rather than numerical approximation. Incidentally, I just tested this on my SHARP EL-W516, and it took 14s. However, it's approximation was also slightly off at 7.4662E91.
Kurt H I was thinking about that too, so I set my TI-89 to Approx mode and ran the test. No change in the results, I measured about 14 seconds again. However, the nInt function which is specifically for numeric (non-CAS) integrals return a result in about 8 seconds.
For the HP 50g the desired precision definitely has a very signifcant impact on performance. Setting the number format to ENG with a given number of significant digits makes the HP stop the numeric integration algorithm as soon as the desired precision is reached. If you look at the results given by some of the other calculators in the video, you can see that they only display a handful of digits. I don't know whether they still compute a more precise result and only round for display purposes, or if they take a shortcut during computation. In any case, configuring the calculators to give the same number of digits is really important in this benchmark, and results might look quite differently then. I really enjoy your videos by the way, thanks for doing these tests!
That being said, the 50g is certainly not the fastest calculator out there. You would expect it to be blazingly fast given its 200 MHz ARM CPU, but then it's only clocked at 75 MHz (to reduce power consumption) and more importantly loads of cycles are lost because - with a few exceptions (memory copy routines afaik) - all the code runs inside an emulator for HP's previous Saturn architecture. The Prime has the algorithms running natively on its ARM CPU, which makes it so fast.
I hit the Advanced manual for the HP 50g to verify that to get a numeric result for your benchmark integral, you need to use "numeric results mode" (with flag -3 set). The calculator consults the displayed digits setting to set the accuracy of the result, so fewer displayed digits would mean faster (but less accurate) results. So ∫ in numeric results mode on the 50g is the same as nInt on the TI-89 / TI-92 / Voyage 200 (possibly also the NSpire CAS -- I don't know). To compare timings, we have to know how many digits were displayed on those posters whose 50g's got the answer in 16 seconds. But all in all, 8 seconds in pretty good for 11 digits of precision on a machine with an older 12 Mhz processor, compared to the 75 Mhz ARM processor on the 50g. A tribute to the programmers of Derive (richly deserved, IMHO).
Dear Martin, I tested two of my calculators which can perform the numerical integration as in your benchmark.Perhaps you can add these into your list: Casio fx-570ES 02:12 Casio fx-991EX Classwiz 00:24 According to their manuals, both calculators use Gauss-Kronrod quadrature formula to evaluate the integral numerically. Cheers, Andreas.
My FX 880BTG compares to FX 991CW Aww man , it was so painful for me . Arrggh my calculator took at least 50 secs to give the answer. 😢😢😢😢😢 Thats a totall failure of a calculator speed test.
∫ is not necessarily ∫ on the different calculators. On my TI-92+ (last ROM revision, same as TI-89 titanium) ∫(e^(x^3),x,0,6) tries to do a symbolic integral. It doesn't succeed so cranks out 5.9639380919e91 in 14 seconds. If you want the numeric answer (the only answer a non-CAS calculator can provide) you use the nInt function: nInt(e^(x^3),x,0,6) to get the same answer in 8 seconds. If you use -6 to +6 for the limits you get the same answer with both functions in 15 and 9 seconds respectively. To show that ∫ and nInt are really different functions, try ∫(tan(x),π/4,π/3) in radian mode to see the symbolic answer, ln(2)/2 . with nInt(tan(x),π/4,π/3) you get only the numeric equivalent, .34657359028 . Again, the simplistic numeric algorithm is the only algorithm available on the TI-36X Pro, fx-9860GII, fx-CG10, and TI-84 Plus SE, so ∫ is actually nInt on those calculators. The processor is a little faster on the TI-89 titanium than the TI-92+, but the functions are the same. The HP 50g is probably trying much harder to find a symbolic answer, which is not necessarily bad nor good.
Rick do you know of any explanation for the 'wrong' answer on the Sharp EL-W516X, in the above comment from Carlos Calderon? --copied below: " I have the sharp el w 516x and I got 7.46621... in about 15 seconds. Does anyone have an idea as to this result? I use the calculator with other integrals and I get right answers." /Thanks in advance
George Obregon Yup, I can explain that. My EL-W516, as well as several other Sharp models, uses Simpson's method, and allows you to enter "subintervals." Since e^(x^3) grows very rapidly as x increases, the default interval of 100 is far too small. You enter a subinterval by putting a comma after the integrand, then entering a number such as 2000 or 2000. Calculation time is directly proportional to the number of intervals chosen. On my EL-W516, with: interval = 100, I get 7.466216848 e91 in just over 17 seconds interval = 200, I get 6.1365636 e91 in about 35 seconds interval = 1000, I get 5.964300819 e91 (4 digit accuracy) in about 175 seconds interval = 10000, I get 5.963938129 e91 (8 digit accuracy) in about 29 minutes Apparently, most calculators use a more sophisticated algorithm. Using e^(x*x*x) didn't speed things up at all. - calcideemus
The TI-82 STATS needed 16,5 secs for the integral of e^x^3 (0-6). However you have to keep in mind that you are only testing integrating here. Speed of that always depends on the implementation in the calculator, so the hp-50g could be much faster in some other tests.
My FX991EX got 25 seconds as well Sharp EL-W516X got 15 seconds HP 35s got 1:47 HP 49G got 1:02 Casio fx-9750GII got 7.5 seconds (It has the 9850GII firmware loaded on it, looks like the processor is the same)
My Casio classwiz took about 25 sec. regardless of the speed, it’s display resolution blows the others away in resolution considering the 19 dollars I bought it new for. It doesn’t graph but it’s a real pleasure to use. BTW when doing equations in fraction form, press the fraction button first so you won’t need to put the numerator in parentheses.
It is interesting to do comparisons of a single feature/function like this, but I have found that knowing how to best approach each problem, gives you an edge over other users, no matter what calculator. When at Polytech many moons ago, the young guys would start the semester with the latest greatest this or that calculator, and on most single functions, they would outperform my trusty Hp15c. In class, however, I ran circles around them; mostly because of my insight in order of calculating and RPN. I have never seen the Casios you showcased here, in my country.
Two results: My goto calc, the Casio fx-CG-10 when run at "full speed" using the publicly available Ptune2 app solves the integral in about 1.5s, still in 3rd place behind the two behemoth models but a much closer race that way. I don't know how many of the other calculators are over-clickable but the Prizm calc when running over-speed is quite a workhorse. It can't graph nearly as quickly as the HP Prime One (that is drool worthy) but it is a quite nice balance My other Casio, fx-115ES Plus ($20 at the local office supply store) took over 3:45 but did get the correct answer.
these calculators are likely calculating this to different levels of accuracy which will either slow them down or speed them up. the titanium only had like 6 decimal places. some calculators had different answers. also this calculation may not be sensitive to accuracy whereas something like a tan function could be affected by accuracy of calculation
I had a think about why the HP Prime and the TI-nSpire were so much faster than the others. The reason is that they're using CAS. All the other calculators are evaluating the function at every point and adding up each slice of the area under the graph to calculate the integral. The HP and the TI use CAS to symbolically solve the integral, and they they simply evaluate the resulting equation. Kinda clever.
They have far more powerful and efficient processors. That's really all it comes down to. The TI-89 has a CAS too, one of the first to have it(second to the TI-92 in regard to Texas Instruments calculators), but has an older and more limited processor compared to the Nspire and Prime; so it'll be noticeably slower.
I got a brand new in the box HP Prime about 2 weeks ago, it is the latest revision hardware version and I updated the firmware the moment I opened it. I tried to enter the integral as per the video and it does not return the expected result. Instead of that, I get :00.000.?>: when I enter it in the Home environment. If I enter it in the CAS environment, I get a warning reading: "Error while checking exact value with approximate value, returning both!¨ I then hit Enter again and I get a very long answer which starts with 1/3*(Gamma(1/3,-216) and so on, and at the end of that I get the result shown in the video. Any idea why my results differ so much from yours? I would think it has to do with the latest firmware version I have but would like an opinion about that
On my Casio fx-115 MS (which I used through school and college): Time for the same integral : 4 min 41 sec Answer it gave : 6 * 10^91 (Yes, only 1 significant figure)
Hey, something bugged me about your HP 50g test, so I checked it out on mine. For some reason, when you evaluate an expression on the stack it's faster than when you do it in the equations writer. That being said, my times were never as long as yours, not even close, I used the same function as you did e^(x^3) and I used the same boundary conditions. I had about 17s in the equation writer and 16s on the stack, I repeated this test several times, 4x, (thinking the discrepancy was due to reaction time, but the difference was consistent). Then I went and checked the boundary conditions -6 to 6, and it doubled the calculation time to about 33 & 30s in both the equation writer and on the stack, respectively. Weird isn't it? Love the Calculators And Math(s) series, tried to purchase the HP Prime through your Amazon store but alas, it doesn't ship to Canada. Let me know if that changes I'd love to be able to support this series in some small way. Keep up the great work!
Hello John, thanks for the post and support...yes...Amazon shipping out of NA is a pain... Others have also had faster results on their 50g..I'm going to check firmware and configuration on mine. Cheers, Martin.
on HP calculators, try approx(. . .), where in place of ". . ." put that integral. Execute this in CAS mode. It should work even in range -6 to 6. I've tried it Android version of HP Prime. I'v got answer in form of matrix where first result is analitic to numeric [Undef] and second is numeric solution. Maybe you will have to install latest HP firmware.
I really wish you would upload more calculator videos. I just found this channel and binged them all. You do everything so meticulously and it's wonderful. I would love to know what you think about the Casio fx-CG50 (the ability to find and shade the integral between the intersection of two graphs is incredibly useful as a calculus student and I'm definitely considering buying it). Casio also offers a 90 day free trial of the emulator if you or anyone want to try it out before purchasing it.
Been waiting for a while for this video. Love your work and look forward to your videos on this topic. I have most of these calculators, and others as well. The Casio FX991ES Plus did the integral correctly in 2:03. The SHARP EL-W516X does the calculation in 15 seconds but gives the wrong answer of 7.466216848 E91 Several things that you should note about your testing. Each calculator may implement the algorithm to solve the integral quite differently ( Ie. an intelligent vs. a non intelligent quadrature ) thus it is hard to truly compare apples to apples even when you are doing the same integral on them all. A further issue, is that if you write the program in machine language instead of the native Basic language built into these calculators, the times would be quite a bit faster. All this being said, it seems quite obvious in many tests that the PRIME clearly is the fastest calculator in most everything, as the processor alone is much faster than the processors in the others. Keep going with your videos, Nice Work !!!
I watched the video and was curious to see how my SHARP EL-W516X did. As Bard mentioned, it gives the wrong answer. Argh! I've lost confidence in it (at least for calculus). I predict a Prime be sitting on my desk in the future. UPDATE: By default, the EL-W516X uses 100 subdivisions but you can specify any number say 1000. When that is done the SHARP gives the expected answer but it takes 10x longer. So which other calculators this test have the ability to specify the subdivisions? Would this comparison be more valid if all calculators under review used the same number of subdivisions?
I have the sharp el w 516x and I got 7.46621... in about 15 seconds. Does anyone have an idea as to this result? I use the calculator with other integrals and I get right answers.
Carlos Calderon The Sharp uses a different algorithm to do the calculation. It has trouble with the larger numbers. You have to enter along with the integral the subintervals. If you use 2000 subintervals, the calculation takes 5:10 and you get the answer 5.963960979 E91 still not correct, but a lot closer.
As a few more point of reference, I tested on my Casio fx-115 ES and it took 2:17. On my iPad 3, PocketCAS and TI-Nspire CAS are basically instant, and m48+ with the 49+ rom took 5 seconds seconds.
In some cases if you integrate is prefer use exact mode not aprox. This way the calculator ran faster the convert the result to numerical value. That is a software issue only solved in hp prime or higher rom version.
While I like the practicality of the integral test, I'd like to see these calculators attempt to factor the Mersenne Prime: (2^67)-1. I've tried it on my Casio Classpad and the machine can't do it, I'd be interested to see if the HP Prime can.
I ran your calculator benchmark 'e^x^3' on my 'HP 35s' calculator, which has been my favorite calculator (other than the new HP Prime). I got a time of 56 sec, which was surprisingly slow to me.
Using the Android app "Droid 48" (HP-48SX/GX sim) on my HTC One it completed the equation in ~ 7seconds. I wonder how long it would take an actual HP-48SX. (Mine's boxed up so I didn't try it.) BTW the HP-48 sim was able to use -6 to 6 limits.
I tried what was said in the comments on my HP-48g. I set it to Fix 0. If I did everything correctly, it only took a few seconds, but the answer I got was "6.E91".
I just loved the video! I still think that is not quite fair to compare CAS vs non-CAS calculators to do the same task as they probably will have very different approaches to the problem. But still I love your videos, thanks!
I tested the first integral on the programmable calculators I own : 1) on the CASIO fx-9860GII SD overclocked to 265.42Mhz with both ROM and RAM wait access time reduced to 1: it needs 1.40s 2) on the TI-nspire cx CAS overclocked to 234Mhz with AHB at 58Mhz it needs 0.81s
What ROM version are you using on the HP 50g? In 2.15 which I'm using, even in approx mode, the equation writer uses integers rather than reals by default, and that makes the calculation 2x faster.
TI-84 isn't equipped with CAS, so it takes any integral by using numerical method ONLY. TI-89 Titanium first of all tries to take the integral in an analytic way. By using *nInt* function instead of *S* You can force it to use numerical method immediately and You can get ~8 sec with the same accuracy.
A bit late here, but I'm a little confused by the results I got when I did the test. I did indeed confirm your results, but the the ti 89 isn't able to symbolically integrate the function. Doesn't the symbolic integrator on the ti89 switch to numerical integration if symbolic doesn't work? Why is there a difference between the two?
Excellent comparison test. Would like to see how you program the calculators. And I bet if you programmed the ti 89 and 84 using c/asm that would be the great equaliser.
All it does is start a loop where it is adding one to a number over and over again. You can also scale it... You might do 1000 on a TI-83, but you can do 10,000 on a faster calculator, or 100,000... and the best part is you can type it into the calculator in like... 1 minute as it is only 7 lines of code.
I collect scientific calculators. It took 2:52 for my hp 48sx. Just ordered me a 50g off ebay. I love hp calculators. They might not be the fastest (except for the prime g2 , witch I also own). The prime g2 smokes all the calculators. I just love rpn. And I already know how to program them. Go team hp! I am such a nerd. Lol
In regards to the queens program... I copied the program to my TI-83... The program is incorrect the answer for 2 should be 0 and the answer for 3 should be 0 as well. I get 6 and 18, just like you. With 4 I get 26... but the correct answer is 2.
I think that the answer for 3x3 should be 3, because that's the maximum numbers of Queens in the field that can't attack each other... Sorry for my bad English BTW
No solution for 3X3 board. Queen cannot be the center, otherwise second queen cannot be added. If 2 queens on the corner, they attack each other. If 2 queens on the side (not corner), they attack each other. To insert 3 queens, either of above must be possible.
(NOTE:Precision doesn't affected timings, tested on FLOAT 6 and FLOAT 12, results are the same.. Well, I run same test on TI-89 Titanium Black and on TI-89 Titanium Gray(classic), Black is double faster than classic. Here it is: TI-89 Titanium Black - 6.8 sec TI-89 Titanium(classic) - 14.6 sec The reason for that is because Black edition of Titanium TI-89 has newer hardware revision, HW4 in comparison to TI-89 Titanium(classic) which has HW3. Also, late production of TI-89 Titanium(classic) is HW4
Yeah it's about 4 MHz faster oh HW4 compared to HW2 and HW3. It can make a pretty big difference in calculations like this. Also, didn't know there was a black Titanium? Is the entire thing black or is it like black accents or something?
@blue_glowstick yes it exist. I think TI initially only released black edition in France but now I see them everywhere on Ebay. Black is my favourite tho, regarding looks but also performance. I sold classic.
I would like to use an HP Prime but unfortunately I'm stuck with a crappy TI-84SEC that is (Expensive, underpowered, lacks CAS) required in my high-school senior class. :-(
Is the source for your benchmarks available somewhere? It would be cool to be able to run those to all my new calculators (with appropriate conversion of course).
In fact it doesn't, TI-89 calculates it more precisely and with hardware revision 4 it's even more faster. Like double faster than with hardware revision 3.
i tried to follow along with HP's Prime Android app, but unfortunately this integral crashes the app every time. :-( Good news, the Ti-89 android app ran the integral with nearly the exact same time as your actual calc. A good emulator!
The new casio 991EX takes 15 seconds (emulated, early next month I will try on an actual calculator). Quite impressive for a 20$ non programmable calculator.
Little late to the party, but the ti-30X pro can do the integration in 8 seconds.....which is shocking as the ti-36x pro is much much slower and basically the same model calculator
Your timer doesn't work as expected. The timer isn't fixed by a crystal. It's voltage controlled and temperature sensitive. Nice try though. Use a stopwatch over increasing difficulty to see the timer isn't fixed. I used a counter. 200, 500, 1000, and so forth. It's easy enough to do the math. 1000 takes more than twice as long to count as 500.
Ti89 is really not that good. I bought a used one from eBay and used it for 2 weeks. It really sucked in almost every aspect compared to my TI nspire CAS touchpad and Casio Classpad 330. I sold it after 2 weeks
If Queens on a chess board cannot take each other on a 2 by 2 chess board there are only 4 possible positions. So only 1 Queen could be placed on this board. Your program is definitely wrong! :(
Performed the integral on a TI Voyage 200 which has the same architecture as the TI-89 Titanium with a slightly slower clock speed. About 14 seconds integrating from 0 to 6, and about 15.5 seconds integrating from -6 to 6. Set up the HP-50g as you had and it took over 17.5 minutes to integrate from -6 to 6, a whopping amount of time. Also read through some other comments regarding the HP-50g. The time can be greatly reduced if it's set to approximate and even more important, it's set to engineering (or scientific) with only a handful of digits specified: 0, 1, 2, or 3. Specify a larger number, such as 6 or 8 and the time required skyrockets. It should be pointed out that specifying fewer digits in engineering or scientific presentation of numeric values to reduce run time also reduces precision of the resulting answer, and only a few digits reduces it greatly. I was comparing the results on the 50g with the results on the Voyage and in the video on other other calculators. One can definitely speed up the 50g, but at what cost to precision? The 50g seems an anomaly. Still experimenting with some other settings that wouldn't sacrifice precision. Would be interesting to compare some other HPs in addition to the Prime. In addition, TI now has the Nspire CX II CAS which effectively runs about 3x faster, and HP has the Prime G2 (rev D) that also cranks along at about 3x of the "G1 (rev C). Both were released in 2018.
TI-58 Benchmark:
Set up the 40 year old TI-58 with f(x) and used Master Library Program 09 which uses a Simpson Approximation to evaluate definite integrals with a specified number of intervals between the lower and upper bounds. Ran it using 100 intervals from 0 to +6 as the lower and upper bounds. As expected, precision in the answer was horrid. Increased intervals to 600 and ran it again. Precision was still lacking with ~2% error. Doubled that to 1200 intervals and ran it again. Took a half-hour and returned an answer with ~0.1% error, about one part in a thousand precision. This is a very wicked function, especially from the interval from ~f(1) to f(6).
SwissMicros DM42 Benchmark:
The DM42 firmware is based on Free42 which emulates HP-42S functionality. As with the TI-58, it requires programming in the function. One can then use the built-in definite integral function. The approximation used is more sophisticated than the brute force Simpson Approximation and works to a specified accuracy factor instead of a given number of evenly spaced intervals. Set the lower and upper bounds to 0 and 6, and set the accuracy factor to 1 x 10^-11. Took about 14 seconds. Set the lower bound to -6 to run it as had originally been intended in the video before running into problems with the HP-50g and ran it again. Took about 14.5 seconds. This was with its processor clock running on the lower battery power speed. Plugged into USB power and running the processor clock at high speed, it cranked out 0 to +6 in about 5.5 seconds and -6 to +6 in just under 6 seconds.
HP don't use Gauss Kronrod to numerical integration. Casio use Gauss Kronrod method. I don't know what method use TI.
Just ran this inegral test on a European TI-30X Pro MathPrint - the newest of the TI scientific that replaced the European TI-30X Pro MultiView in 2018. The European 30X Pro MultiView is the same as the TI-36X Pro in North America; there's a significant marketing reason for the model number difference. Its MathPrint replacement significantly faster, doing this integral speed test consistently in ~8 seconds - versus a lethargic 108 seconds. This speed is phenomenal. Ran the same integral from -6 to +6 (which was your original plan) and it only increased to ~9 seconds. Then I cranked up the "epsilon" (error precision) from 1x10^-5 to 1x10^-12, which one would expect to drive up the time required, but it didn't. Nobody should be buying a TI-36X Pro in North America. They should be buying the 30X Pro *MathPrint* ( *not* the MultiView). It looks completely different - with a profile like the Nspire CX II in shape and thickness. They can be easily bought from Amazon UK or Amazon Germany (buying it on Amazon US is much more expensive) using the same login as for Amazon US - which I've been doing for years now. North American buyers want the *"Pro"* , not the cheaper "Plus". The latter one deletes four features banned for tests in two German states (Baden-Wuertemberg and Bavaria) and was created specifically for them. The *Pro* version is only a few dollars more and isn't crippled to satisfy regional German testing restrictions.
Very interesting video.
For what it's worth, the HP 50g is an ARM based cpu emulating a Saturn cpu. The Saturn typically ran at about 2 MHz, but was overclocked in the HP 48G series to around 4 MHz. The ARM 9 v 2 in the 50g can run up to 238 MHz before memeory starts to fail, however HP under-clocked the 50g to 75 MHz for the conservation of batteries. An HP 50g typically is about 2.5 times quicker than a 48G series calculator in performing most operations.
My 50g (which is plugged in) takes about 1 min, 2 sec to perform the integral problem (from 0 to 6), STD mode (12 significant digits).
BTW You can use the TIME function to get the time according to the 50g, and store it in a list just like in the first example.
Dear Martin, I've been watching your series of videos on various scientific/graphing calculators. You've done a very meticulous job. I noticed your surprise at the duration it took the HP50g to evaluate the integral. Being a user and an owner of HP50g, I tried to replicate your results with the same input. The HP50g produced a result almost instantaneously but certainly not as quick as HP Prime or the TI-nSpire but was darn close; about 1 sec.
At first I was baffled why it took you so long. Then I realised that your settings were set to STD. So what you must do:
1. Make sure approximate mode is set. Press MODE, then CAS and tick APPROX
2. Change number format from STD to ENG. Press MODE, (second row) and change the number format
I hope you can redo the tests again or edit this video by removing the part for the HP50g because leaving the video like this is very unfair to HP50g, which should be the 3rd best among those you're reviewing in this videos. You did not tell the calculator to approximate the value of the integral and so the calculator took a longer time trying to come out with a symbolic answer instead of a numeric one.
If you still cannot evaluate the integral under a second, I can post a video response. However, I don't think anything's wrong with your calculator and you shouldn't have any problem.
Thanks and cheers.
Medwatt, thanks very much for the feedback and explanation. I have gone and added a couple of annotations to the video about the performance and result. When time allows...life very busy at the moment...I will follow-up with another video to clarify and demonstrate the configuration issue. Cheers, Martin.
Thanks for the info! I wonder if the TI-89 is also being hurt in these scores by performing CAS rather than numerical approximation.
Incidentally, I just tested this on my SHARP EL-W516, and it took 14s. However, it's approximation was also slightly off at 7.4662E91.
Kurt H I was thinking about that too, so I set my TI-89 to Approx mode and ran the test. No change in the results, I measured about 14 seconds again.
However, the nInt function which is specifically for numeric (non-CAS) integrals return a result in about 8 seconds.
For the HP 50g the desired precision definitely has a very signifcant impact on performance.
Setting the number format to ENG with a given number of significant digits makes the HP stop the numeric integration algorithm as soon as the desired precision is reached.
If you look at the results given by some of the other calculators in the video, you can see that they only display a handful of digits. I don't know whether they still compute a more precise result and only round for display purposes, or if they take a shortcut during computation. In any case, configuring the calculators to give the same number of digits is really important in this benchmark, and results might look quite differently then.
I really enjoy your videos by the way, thanks for doing these tests!
That being said, the 50g is certainly not the fastest calculator out there. You would expect it to be blazingly fast given its 200 MHz ARM CPU, but then it's only clocked at 75 MHz (to reduce power consumption) and more importantly loads of cycles are lost because - with a few exceptions (memory copy routines afaik) - all the code runs inside an emulator for HP's previous Saturn architecture.
The Prime has the algorithms running natively on its ARM CPU, which makes it so fast.
I hit the Advanced manual for the HP 50g to verify that to get a numeric result for your benchmark integral, you need to use "numeric results mode" (with flag -3 set). The calculator consults the displayed digits setting to set the accuracy of the result, so fewer displayed digits would mean faster (but less accurate) results.
So ∫ in numeric results mode on the 50g is the same as nInt on the TI-89 / TI-92 / Voyage 200 (possibly also the NSpire CAS -- I don't know). To compare timings, we have to know how many digits were displayed on those posters whose 50g's got the answer in 16 seconds. But all in all, 8 seconds in pretty good for 11 digits of precision on a machine with an older 12 Mhz processor, compared to the 75 Mhz ARM processor on the 50g. A tribute to the programmers of Derive (richly deserved, IMHO).
Dear Martin, I tested two of my calculators which can perform the numerical integration as in your benchmark.Perhaps you can add these into your list:
Casio fx-570ES 02:12
Casio fx-991EX Classwiz 00:24
According to their manuals, both calculators use Gauss-Kronrod quadrature formula to evaluate the integral numerically.
Cheers, Andreas.
My FX 880BTG compares to
FX 991CW
Aww man , it was so painful for me . Arrggh my calculator took at least 50 secs to give the answer.
😢😢😢😢😢
Thats a totall failure of a calculator speed test.
You might find that the TI-89 was slower because it is calculating the answer to more digits even if the displayed answer has been rounded.
∫ is not necessarily ∫ on the different calculators. On my TI-92+ (last ROM revision, same as TI-89 titanium) ∫(e^(x^3),x,0,6) tries to do a symbolic integral. It doesn't succeed so cranks out 5.9639380919e91 in 14 seconds. If you want the numeric answer (the only answer a non-CAS calculator can provide) you use the nInt function: nInt(e^(x^3),x,0,6) to get the same answer in 8 seconds. If you use -6 to +6 for the limits you get the same answer with both functions in 15 and 9 seconds respectively.
To show that ∫ and nInt are really different functions, try ∫(tan(x),π/4,π/3) in radian mode to see the symbolic answer, ln(2)/2 . with nInt(tan(x),π/4,π/3) you get only the numeric equivalent, .34657359028 . Again, the simplistic numeric algorithm is the only algorithm available on the TI-36X Pro, fx-9860GII, fx-CG10, and TI-84 Plus SE, so ∫ is actually nInt on those calculators.
The processor is a little faster on the TI-89 titanium than the TI-92+, but the functions are the same. The HP 50g is probably trying much harder to find a symbolic answer, which is not necessarily bad nor good.
Rick do you know of any explanation for the 'wrong' answer on the Sharp EL-W516X, in the above comment from Carlos Calderon? --copied below:
" I have the sharp el w 516x and I got 7.46621... in about 15 seconds. Does anyone have an idea as to this result? I use the calculator with other integrals and I get right answers."
/Thanks in advance
George Obregon
Yup, I can explain that. My EL-W516, as well as several other Sharp models, uses Simpson's method, and allows you to enter "subintervals." Since e^(x^3) grows very rapidly as x increases, the default interval of 100 is far too small.
You enter a subinterval by putting a comma after the integrand, then entering a number such as 2000 or 2000. Calculation time is directly proportional to the number of intervals chosen. On my EL-W516, with:
interval = 100, I get 7.466216848 e91 in just over 17 seconds
interval = 200, I get 6.1365636 e91 in about 35 seconds
interval = 1000, I get 5.964300819 e91 (4 digit accuracy) in about 175 seconds
interval = 10000, I get 5.963938129 e91 (8 digit accuracy) in about 29 minutes
Apparently, most calculators use a more sophisticated algorithm. Using e^(x*x*x) didn't speed things up at all.
- calcideemus
rickideemus
Sorry for all the typos in the above. Hopefully the idea got across.
The TI-82 STATS needed 16,5 secs for the integral of e^x^3 (0-6). However you have to keep in mind that you are only testing integrating here. Speed of that always depends on the implementation in the calculator, so the hp-50g could be much faster in some other tests.
Thanks for your figures...and yes you are correct, we'll see how well the HP 50G does in other areas.
Here are my results:
HP-48SX: 2:53
Casio FX-115 ES: 2:26
TI-84: 7.3
Dániel Tóth my FX991EX got 25 seconds!
My FX991EX got 25 seconds as well
Sharp EL-W516X got 15 seconds
HP 35s got 1:47
HP 49G got 1:02
Casio fx-9750GII got 7.5 seconds (It has the 9850GII firmware loaded on it, looks like the processor is the same)
Casio fx-CG50:
29 MHz: 4.5 s
58 MHz: 2.5 s
93.4 MHz: 1.5 s
Tried it on my Numworks. It's been over 5 minutes...
My Casio classwiz took about 25 sec. regardless of the speed, it’s display resolution blows the others away in resolution considering the 19 dollars I bought it new for. It doesn’t graph but it’s a real pleasure to use. BTW when doing equations in fraction form, press the fraction button first so you won’t need to put the numerator in parentheses.
It is interesting to do comparisons of a single feature/function like this, but I have found that knowing how to best approach each problem, gives you an edge over other users, no matter what calculator. When at Polytech many moons ago, the young guys would start the semester with the latest greatest this or that calculator, and on most single functions, they would outperform my trusty Hp15c. In class, however, I ran circles around them; mostly because of my insight in order of calculating and RPN.
I have never seen the Casios you showcased here, in my country.
I did the exercise on an HP-50g and got 8.8 seconds, not 1'08" as you got at 11:40. Perhaps it has to do with the CAS settings?
Two results:
My goto calc, the Casio fx-CG-10 when run at "full speed" using the publicly available Ptune2 app solves the integral in about 1.5s, still in 3rd place behind the two behemoth models but a much closer race that way.
I don't know how many of the other calculators are over-clickable but the Prizm calc when running over-speed is quite a workhorse. It can't graph nearly as quickly as the HP Prime One (that is drool worthy) but it is a quite nice balance
My other Casio, fx-115ES Plus ($20 at the local office supply store) took over 3:45 but did get the correct answer.
these calculators are likely calculating this to different levels of accuracy which will either slow them down or speed them up.
the titanium only had like 6 decimal places. some calculators had different answers.
also this calculation may not be sensitive to accuracy whereas something like a tan function could be affected by accuracy of calculation
Isn’t this a non-factor, though? This gives the results in standard usage
I tried this on the newer version of the fx-CG10, the fx-CG50, and it was about 2.5 seconds.
I had a think about why the HP Prime and the TI-nSpire were so much faster than the others. The reason is that they're using CAS. All the other calculators are evaluating the function at every point and adding up each slice of the area under the graph to calculate the integral. The HP and the TI use CAS to symbolically solve the integral, and they they simply evaluate the resulting equation. Kinda clever.
As does the TI-89
They have far more powerful and efficient processors. That's really all it comes down to. The TI-89 has a CAS too, one of the first to have it(second to the TI-92 in regard to Texas Instruments calculators), but has an older and more limited processor compared to the Nspire and Prime; so it'll be noticeably slower.
I got a brand new in the box HP Prime about 2 weeks ago, it is the latest revision hardware version and I updated the firmware the moment I opened it. I tried to enter the integral as per the video and it does not return the expected result. Instead of that, I get :00.000.?>: when I enter it in the Home environment. If I enter it in the CAS environment, I get a warning reading: "Error while checking exact value with approximate value, returning both!¨ I then hit Enter again and I get a very long answer which starts with 1/3*(Gamma(1/3,-216) and so on, and at the end of that I get the result shown in the video. Any idea why my results differ so much from yours? I would think it has to do with the latest firmware version I have but would like an opinion about that
On my Casio fx-115 MS (which I used through school and college):
Time for the same integral : 4 min 41 sec
Answer it gave : 6 * 10^91 (Yes, only 1 significant figure)
I tested my HP35s with the integral you used in your test. It took 1.75 minutes to obtain the result!
Interesting...thanks for posting.
Hey, something bugged me about your HP 50g test, so I checked it out on mine. For some reason, when you evaluate an expression on the stack it's faster than when you do it in the equations writer.
That being said, my times were never as long as yours, not even close, I used the same function as you did e^(x^3) and I used the same boundary conditions. I had about 17s in the equation writer and 16s on the stack, I repeated this test several times, 4x, (thinking the discrepancy was due to reaction time, but the difference was consistent). Then I went and checked the boundary conditions -6 to 6, and it doubled the calculation time to about 33 & 30s in both the equation writer and on the stack, respectively. Weird isn't it?
Love the Calculators And Math(s) series, tried to purchase the HP Prime through your Amazon store but alas, it doesn't ship to Canada. Let me know if that changes I'd love to be able to support this series in some small way. Keep up the great work!
Hello John, thanks for the post and support...yes...Amazon shipping out of NA is a pain...
Others have also had faster results on their 50g..I'm going to check firmware and configuration on mine.
Cheers, Martin.
on HP calculators, try approx(. . .), where in place of ". . ." put that integral. Execute this in CAS mode. It should work even in range -6 to 6. I've tried it Android version of HP Prime. I'v got answer in form of matrix where first result is analitic to numeric [Undef] and second is numeric solution. Maybe you will have to install latest HP firmware.
I really wish you would upload more calculator videos. I just found this channel and binged them all. You do everything so meticulously and it's wonderful. I would love to know what you think about the Casio fx-CG50 (the ability to find and shade the integral between the intersection of two graphs is incredibly useful as a calculus student and I'm definitely considering buying it). Casio also offers a 90 day free trial of the emulator if you or anyone want to try it out before purchasing it.
Been waiting for a while for this video. Love your work and look forward to your videos on this topic. I have most of these calculators, and others as well. The Casio FX991ES Plus did the integral correctly in 2:03. The SHARP EL-W516X does the calculation in 15 seconds but gives the wrong answer of 7.466216848 E91
Several things that you should note about your testing. Each calculator may implement the algorithm to solve the integral quite differently ( Ie. an intelligent vs. a non intelligent quadrature ) thus it is hard to truly compare apples to apples even when you are doing the same integral on them all.
A further issue, is that if you write the program in machine language instead of the native Basic language built into these calculators, the times would be quite a bit faster.
All this being said, it seems quite obvious in many tests that the PRIME clearly is the fastest calculator in most everything, as the processor alone is much faster than the processors in the others.
Keep going with your videos, Nice Work !!!
Hi Brad, thanks very much for the information and feedback. I've noted what you've said and may chat about this in a future video. Cheers, Martin.
I watched the video and was curious to see how my SHARP EL-W516X did. As Bard mentioned, it gives the wrong answer. Argh! I've lost confidence in it (at least for calculus). I predict a Prime be sitting on my desk in the future.
UPDATE: By default, the EL-W516X uses 100 subdivisions but you can specify any number say 1000. When that is done the SHARP gives the expected answer but it takes 10x longer.
So which other calculators this test have the ability to specify the subdivisions? Would this comparison be more valid if all calculators under review used the same number of subdivisions?
My Casio fx-991ES took 2:15
So I guess the Plus version has a slight performance increase.
I have the sharp el w 516x and I got 7.46621... in about 15 seconds. Does anyone have an idea as to this result? I use the calculator with other integrals and I get right answers.
Carlos Calderon The Sharp uses a different algorithm to do the calculation. It has trouble with the larger numbers. You have to enter along with the integral the subintervals. If you use 2000 subintervals, the calculation takes 5:10 and you get the answer 5.963960979 E91 still not correct, but a lot closer.
As a few more point of reference, I tested on my Casio fx-115 ES and it took 2:17.
On my iPad 3, PocketCAS and TI-Nspire CAS are basically instant, and m48+ with the 49+ rom took 5 seconds seconds.
Thanks for the post Steve.
There might be a configuration issue with the HP50G, mine can do the integral in 16s
Aaah...thanks for that feedback. I will look into that as I thought the time I had was far too long.
I agree,
I solved the integral in my HP50G in 2 seconds, you should configure your calculator.
In some cases if you integrate is prefer use exact mode not aprox. This way the calculator ran faster the convert the result to numerical value.
That is a software issue only solved in hp prime or higher rom version.
While I like the practicality of the integral test, I'd like to see these calculators attempt to factor the Mersenne Prime: (2^67)-1. I've tried it on my Casio Classpad and the machine can't do it, I'd be interested to see if the HP Prime can.
The HP Prime factors it in less than a second.
I got ~25 seconds on the Casio fx-991ex.
I'm actually impressed by this $35 calculator!
It's easily the best calculator at its level, shame it can't do graphing though.
My calculator FX 880 BTG is a failure.
It took 51.6 secs to give the ans .
My calculator at that time calculate things
🤒💫💫💫💔💔💔
where did u find about the time computimg
I ran your calculator benchmark 'e^x^3' on my 'HP 35s' calculator, which has been my favorite calculator (other than the new HP Prime).
I got a time of 56 sec, which was surprisingly slow to me.
Have you considered doing a video about the TI Voyage 200? There's not a lot of videos about it and it looks interesting.
Having run the calculation already under the Casio machines, wouldn't the cache of the CPUs have been primed, helping them perform faster?
Using the Android app "Droid 48" (HP-48SX/GX sim) on my HTC One it completed the equation in ~ 7seconds. I wonder how long it would take an actual HP-48SX. (Mine's boxed up so I didn't try it.) BTW the HP-48 sim was able to use -6 to 6 limits.
Very interesting info, I was hoping to include a phone / Android app but didn't find a good one in time. Thanks for the post.
I just tried on my HP 48s: 21 seconds (in Eng 2 mode)
Also, my HP 49g took 16 seconds (in Eng 2 mode) and 31 seconds (in Eng 4 mode).
My HP 48GX> 13.8s (fix 2 mode). My HP 48GX Emulator in Android phone > 1s (fix)
Проверил эту формулу на HP-48GX, там вышло время около 2 мин 15 сек. Ti конечно значительно шустрее.
the 2nd gen HP Prime (G2) is a couple of times faster than the older G1, uses a faster ARM processor aye caramba
I tried what was said in the comments on my HP-48g. I set it to Fix 0. If I did everything correctly, it only took a few seconds, but the answer I got was "6.E91".
I just loved the video! I still think that is not quite fair to compare CAS vs non-CAS calculators to do the same task as they probably will have very different approaches to the problem. But still I love your videos, thanks!
Also the TI-89t is using a 30 year old processor. It's advantage isn't speed. It's how the processor is used.
I tested the first integral on the programmable calculators I own :
1) on the CASIO fx-9860GII SD overclocked to 265.42Mhz with both ROM and RAM wait access time reduced to 1: it needs 1.40s
2) on the TI-nspire cx CAS overclocked to 234Mhz with AHB at 58Mhz it needs 0.81s
What ROM version are you using on the HP 50g? In 2.15 which I'm using, even in approx mode, the equation writer uses integers rather than reals by default, and that makes the calculation 2x faster.
Ok...thanks, I'll check and look into that.
TI-84 isn't equipped with CAS, so it takes any integral by using numerical method ONLY. TI-89 Titanium first of all tries to take the integral in an analytic way. By using *nInt* function instead of *S* You can force it to use numerical method immediately and You can get ~8 sec with the same accuracy.
A bit late here, but I'm a little confused by the results I got when I did the test. I did indeed confirm your results, but the the ti 89 isn't able to symbolically integrate the function. Doesn't the symbolic integrator on the ti89 switch to numerical integration if symbolic doesn't work? Why is there a difference between the two?
Same calculation on my old hp 48GX 1:42.62 and a Cifra SC-2022G 1:15.72 faster but less accurate result.
Thanks for the post David.
David Fernando Bellsolá I got 2:53 on my HP-48SX.
Excellent comparison test. Would like to see how you program the calculators. And I bet if you programmed the ti 89 and 84 using c/asm that would be the great equaliser.
I made a VERY SIMPLE benchmark program... On my TI-83 the code is like this
:ClrHome
:1->A
:Input "Enter Cycles: ",B
:While AA
:End
:Disp A
All it does is start a loop where it is adding one to a number over and over again. You can also scale it... You might do 1000 on a TI-83, but you can do 10,000 on a faster calculator, or 100,000... and the best part is you can type it into the calculator in like... 1 minute as it is only 7 lines of code.
Casio CFX-9950GB Plus:
Simpson integration - 0:30 & the answer is 6.0e91
Gauss integration - 4:40 & Math error, no answer
Casio FX-795P:
Err 0.001 (default) - 3:06 & Impossible, no answer
Err 1e90 - 3:06 & Impossible, no answer
Err 2e90 - 3:06 & the answer is 9.2e91
Casio fx-5800P: 48:38
HP 50 is not very fast at numeric integrals you has to use exact mode this way is faster then convert result to a numerical value.
Ho50g look in hp manual fibonacci , programs have timer in it.
I collect scientific calculators.
It took 2:52 for my hp 48sx.
Just ordered me a 50g off ebay.
I love hp calculators.
They might not be the fastest (except for the prime g2 , witch I also own). The prime g2 smokes all the calculators.
I just love rpn.
And I already know how to program them. Go team hp!
I am such a nerd. Lol
My Casio fx-115ES got 3:48.5 on that integral benchmark.
In regards to the queens program... I copied the program to my TI-83... The program is incorrect the answer for 2 should be 0 and the answer for 3 should be 0 as well. I get 6 and 18, just like you. With 4 I get 26... but the correct answer is 2.
I think that the answer for 3x3 should be 3, because that's the maximum numbers of Queens in the field that can't attack each other...
Sorry for my bad English BTW
No solution for 3X3 board.
Queen cannot be the center, otherwise second queen cannot be added.
If 2 queens on the corner, they attack each other.
If 2 queens on the side (not corner), they attack each other.
To insert 3 queens, either of above must be possible.
Hi sir, what kind of the hardware revision of your HP Prime?
(NOTE:Precision doesn't affected timings, tested on FLOAT 6 and FLOAT 12, results are the same..
Well, I run same test on TI-89 Titanium Black and on TI-89 Titanium Gray(classic), Black is double faster than classic. Here it is:
TI-89 Titanium Black - 6.8 sec
TI-89 Titanium(classic) - 14.6 sec
The reason for that is because Black edition of Titanium TI-89 has newer hardware revision, HW4 in comparison to TI-89 Titanium(classic) which has HW3. Also, late production of TI-89 Titanium(classic) is HW4
Yeah it's about 4 MHz faster oh HW4 compared to HW2 and HW3. It can make a pretty big difference in calculations like this. Also, didn't know there was a black Titanium? Is the entire thing black or is it like black accents or something?
@blue_glowstick yes it exist. I think TI initially only released black edition in France but now I see them everywhere on Ebay. Black is my favourite tho, regarding looks but also performance. I sold classic.
I know the Casio fx9860GII can but how many of the others can play Space Invaders and Super Mario World
I would be interested in a video about how to create the programs in the computer and then send it to the calculator.
Noted...I'll try get to that in the future. Cheers, Martin.
My casio fx 5800p needs 00:47 to calculate the same
Tried it on the CXII CAS and it instantly gave me the answer😂
Calculator Speed Benchmark using the N-Queens Problem
www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/articles.cgi?read=700
I would like to use an HP Prime but unfortunately I'm stuck with a crappy TI-84SEC that is (Expensive, underpowered, lacks CAS) required in my high-school senior class. :-(
Is the source for your benchmarks available somewhere? It would be cool to be able to run those to all my new calculators (with appropriate conversion of course).
TI-89 is dope and superior in everyway. Skip 83s/84s all together.
Hey, Out of curiosity I pulled out my old CASIO fx-5800P witch managed to complete this test in 49.08 seconds ...
With integration Result: 5,963938092 x 10^91
Casio fx-570SPX about 25s
HP 50g 16sec
Casio ClassPad II 4sec
Why was the to 84 faster ten the 89 titanium?
In fact it doesn't, TI-89 calculates it more precisely and with hardware revision 4 it's even more faster. Like double faster than with hardware revision 3.
i have the TI-89 Titanium and HP 50G, i got the same times for the TI, the HP 50G is just a little over 4 seconds!..
i tried to follow along with HP's Prime Android app, but unfortunately this integral crashes the app every time. :-( Good news, the Ti-89 android app ran the integral with nearly the exact same time as your actual calc. A good emulator!
Got my 9860GII for only £20. What a good deal!
HP Prime G2
tested on casio fx-991 es : 230 Second
991-es is cheap non graphing calc
Wow the hp50g was that slow! And it is so bragged about high performance . I'm very disappointed in the hp50g. Very good testing!
Result for Casio fx-9750G: 6.e+91 in 30,76 seconds
The new casio 991EX takes 15 seconds (emulated, early next month I will try on an actual calculator). Quite impressive for a 20$ non programmable calculator.
+Amit Sandler (pootis spencering) 27 Secs
Amit Sandler My fx-991ex last 41sec to perform (real calculator not a emulator)
I performed it in my hp-50g and took 15.72 secs.
Genial!, Saludos desde Bolivia
Little late to the party, but the ti-30X pro can do the integration in 8 seconds.....which is shocking as the ti-36x pro is much much slower and basically the same model calculator
ti83 plus, took about 22 seconds.
Your timer doesn't work as expected. The timer isn't fixed by a crystal. It's voltage controlled and temperature sensitive. Nice try though. Use a stopwatch over increasing difficulty to see the timer isn't fixed. I used a counter. 200, 500, 1000, and so forth. It's easy enough to do the math. 1000 takes more than twice as long to count as 500.
I found in later testing that the voltage and circuit is not well regulated. 1000 might take longer than twice as long as 500 to count.
I have bought TI-Nspire CX for180 dollar a week ago..I really want HP Prime...
+shadowtsunami Maybe nothing-- I would like to have both myself. :)
Ti89 is really not that good. I bought a used one from eBay and used it for 2 weeks. It really sucked in almost every aspect compared to my TI nspire CAS touchpad and Casio Classpad 330. I sold it after 2 weeks
Did this on my overclocked INTEL XEON 8C Proccessor with 4 Nvidia titans in sli, took only 4 seconds.
Life on the edge
Prof. Killionare I'm sure you will sort the problem out.
Mathematica on a quad core i7 does it in 52 milliseconds. Which is actually slower than I was expecting...
Yeah that was using its symbolic solve. Using a numeric integrate takes only 3.9ms. Quite a bit faster.
I tested on my Casio fx-991 CW and it took 51 seconds.
My Fx-991ES take 2 m 15 sec......
+Matthew Wang My exact same model took 3m 22s? wut?
@@johnr001 Congrats, you need to change your calculator batteries
are you south african
If Queens on a chess board cannot take each other on a 2 by 2 chess board there are only 4 possible positions. So only 1 Queen could be placed on this board. Your program is definitely wrong! :(
You are correct. There are no solutions for a 2x2 or a 3x3 chess board. Surprised nobody else noticed ;-)