Summation based benchmark for calculators

Summation based benchmark for calculators - Printable Version

+- HP Forums (https://www.hpmuseum.org/forum)
+-- Forum: HP Calculators (and very old HP Computers) (/forum-3.html)
+--- Forum: General Forum (/forum-4.html)
+--- Thread: Summation based benchmark for calculators (/thread-9750.html)

Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13

RE: Summation based benchmark for calculators - Guenter Schink - 04-22-2021 06:33 PM

I think we have new champion -- by quite a margin

Prime G2 Beta 2.1.14549 using Python

Code:

#PYTHON name

from math import *

import sys

import hpprime

t=hpprime.eval("time")

c=int(sys.argv[0])

print()

print("Exponent = ",c)

print("Iterations = ", 10**c)

a=0

for i in range(1, 10**c+1):

    a=a+pow((e**(sin(atan(i)))),1/3) 

print("Time (sec) = ",round(hpprime.eval("time")-t,4))

print("Sum = ",a)

#end

Export sums(c)

Begin

  PYTHON(name, c);

End;

From the Home screen invoke this Python program with sums(?)

where "?" question mark has to be replaced by the exponent which denotes the magnitude

3 for 1000
4 for 10 000
5 for 100 000
6 for 1 000 000

Output will be the exponent, the resulting number, the time and the sum. E.g.
3
1000
0.036 sec
1395.3462877433426

Code:

Results:

3       1 000  0.036 sec     1395.3462877433426

4      10 000  0.159 sec    13955.857904429155

5     100 000  1.425 sec   139560.9761410521

6   1 000 000 14.046 sec  1395612.158725383

I think that's quite impressive specifically when compared with the values for native PPL.

Günter

Edit: put the timing into the Python code rather than using TEVAL

RE: Summation based benchmark for calculators - OlidaBel - 05-01-2021 09:18 AM

Something new to me, maybe a lack of Prime knowledge.
My Prime is configured in RPN (for stack Home computations).

I tried the sum "benchmark" proposed here, then Pick (copy) the first formula to modify it in the command line, I want to run a bigger one, from a 1000 to 10000 sum.
I only added a zero behind "1000", the Editor refuses to validate the formula. Why ?

[Image: wgZJbrL]

https://ibb.co/wgZJbrL
[Image: VD2Wzp6]

https://ibb.co/VD2Wzp6
[Image: WfpDSGv]

https://ibb.co/WfpDSGv

RE: Summation based benchmark for calculators - pier4r - 08-05-2021 04:06 PM

Finally updated up to post #221 (yay after years! Well priorities...)

It took more than 2 hours, so it wasn't that trivial.

Updated also here: http://www.wiki4hp.com/doku.php?id=benchmarks:sum_trig_exp_root
and here: https://osdn.net/users/pier4r/pf/various_works_only_code/scm/blobs/master/rpl_hp48-50/wiki_notes_backup/summation_based_benchmark.txt

I am not so sure about some results regarding 500k with the prime G2. Now you all need to add more entries so that I have to catch up again Big Grin

RE: Summation based benchmark for calculators - Guenter Schink - 08-08-2021 10:59 AM

(08-05-2021 04:06 PM)pier4r Wrote: Finally updated up to post #221 (yay after years! Well priorities...)

It took more than 2 hours, so it wasn't that trivial.

Updated also here: http://www.wiki4hp.com/doku.php?id=benchmarks:sum_trig_exp_root
and here: https://osdn.net/users/pier4r/pf/various_works_only_code/scm/blobs/master/rpl_hp48-50/wiki_notes_backup/summation_based_benchmark.txt

I am not so sure about some results regarding 500k with the prime G2. Now you all need to add more entries so that I have to catch up again

The HP Expander (not XP Expander) is a calculator from HP, not a mobile device. It never saw the light of the markets. I've got a prototype at one of the HHCCs.

Günter

RE: Summation based benchmark for calculators - Massimo Gnerucci - 08-08-2021 11:03 AM

(08-08-2021 10:59 AM)Guenter Schink Wrote: The HP Expander (not XP Expander) is a calculator from HP, not a mobile device. But I think it never saw the light of the markets. I've got a prototype at one of the HHCCs.

It actually was "HP Xpander".
Have a couple, from different development stages.

RE: Summation based benchmark for calculators - Guenter Schink - 08-08-2021 11:44 AM

(08-08-2021 11:03 AM)Massimo Gnerucci Wrote:
(08-08-2021 10:59 AM)Guenter Schink Wrote: The HP Expander (not XP Expander) is a calculator from HP, not a mobile device. But I think it never saw the light of the markets. I've got a prototype at one of the HHCCs.

It actually was "HP Xpander".
Have a couple, from different development stages.

Thanks for the correction. I could have had a closer look either to the device or the Wiki article Smile

Günter

RE: Summation based benchmark for calculators - Felix Stehli - 08-12-2021 08:00 AM

ThetaCalc, an RPN calculator app running on iOS which can be programmed in JavaScript, produced the following result for n = 1000000 on an iPhone 12:

HP museum benchmark (n = 1000000)
Sum = 1395612.1587253837
Time = 0.189 seconds

I don't know, however, whether the app fits into one of the given categories.

Here is the program:

Code:

function benchmark(n) {

  let s = 0;

  for (let i = 1; i <= n; i++) {

    s += Math.pow(

      Math.exp(

        Math.sin(

          Math.atan(i))), 1/3);

  }

  return s;

}

const n = 1000000;

startTimer();

let s = benchmark(n);

stopTimer();

println(`HP museum benchmark (n = ${n})`);

println(`Sum = ${s}`);

printlnElapsed();

RE: Summation based benchmark for calculators - Marc van Lemmen - 08-17-2021 11:04 PM

I recently received a Sharp PC-1201 and was currious how this nice little machine would perform, here are the results :

N=10 ~25s result 13,318529402
N=100 ~285s result 138,90159786
N=1000 ~2844s result 1394,95067416

RE: Summation based benchmark for calculators - toml_12953 - 08-17-2021 11:23 PM

(04-22-2021 06:33 PM)Guenter Schink Wrote: I think we have new champion -- by quite a margin

Prime G2 Beta 2.1.14549 using Python

When I copy and paste that code into the emulator, I get

Code:

Traceback (most recent call last):  File "<stdin>", line 5, in <module>

ValueError: invalid syntax for integer with base 10: '10\xef\xb9\x90000'

Traceback (most recent call last):

  File "<stdin>", line 5, in <module>

ValueError: invalid syntax for integer with base 

10: '1\xef\xb9\x90000'

Traceback (most recent call last):

  File "<stdin>", line 5, in <module>

ValueError: invalid syntax for integer with base 

10: '1\xef\xb9\x90000'

Traceback (most recent call last):

  File "<stdin>", line 5, in <module>

ValueError: invalid syntax for integer with base 

10: '1\xef\xb9\x90000'

RE: Summation based benchmark for calculators - Guenter Schink - 08-18-2021 03:09 PM

(08-17-2021 11:23 PM)toml_12953 Wrote:
(04-22-2021 06:33 PM)Guenter Schink Wrote: I think we have new champion -- by quite a margin

Prime G2 Beta 2.1.14549 using Python

When I copy and paste that code into the emulator, I get

Code:

Traceback (most recent call last): File "<stdin>", line 5, in <module> ValueError: invalid syntax for integer with base 10: '10\xef\xb9\x90000' Traceback (most recent call last): File "<stdin>", line 5, in <module> ValueError: invalid syntax for integer with base 10: '1\xef\xb9\x90000' Traceback (most recent call last): File "<stdin>", line 5, in <module> ValueError: invalid syntax for integer with base 10: '1\xef\xb9\x90000' Traceback (most recent call last): File "<stdin>", line 5, in <module> ValueError: invalid syntax for integer with base 10: '1\xef\xb9\x90000'

I just also copied it to the virtual Prime. works flawlessly. (?)

The error actually points to line 6 "c=int(sys.argv[0])" Can't assume what's going wrong.

Günter

RE: Summation based benchmark for calculators - toml_12953 - 08-18-2021 04:01 PM

(08-18-2021 03:09 PM)Guenter Schink Wrote:
(08-17-2021 11:23 PM)toml_12953 Wrote: When I copy and paste that code into the emulator, I get

Code:

Traceback (most recent call last): File "<stdin>", line 5, in <module> ValueError: invalid syntax for integer with base 10: '10\xef\xb9\x90000' Traceback (most recent call last): File "<stdin>", line 5, in <module> ValueError: invalid syntax for integer with base 10: '1\xef\xb9\x90000' Traceback (most recent call last): File "<stdin>", line 5, in <module> ValueError: invalid syntax for integer with base 10: '1\xef\xb9\x90000' Traceback (most recent call last): File "<stdin>", line 5, in <module> ValueError: invalid syntax for integer with base 10: '1\xef\xb9\x90000'

I just also copied it to the virtual Prime. works flawlessly. (?)

The error actually points to line 6 "c=int(sys.argv[0])" Can't assume what's going wrong.

Günter

I had digit grouping set to 123,456.789. When I change it to 123456.789, I don't get an error but I get a lot of zeroes on the screen and what may be an infinite loop.

RE: Summation based benchmark for calculators - Guenter Schink - 08-18-2021 06:56 PM

(08-18-2021 04:01 PM)toml_12953 Wrote: I had digit grouping set to 123,456.789. When I change it to 123456.789, I don't get an error but I get a lot of zeroes on the screen and what may be an infinite loop.

o.K. I think I found the problem. It wasn't clearly explained. You have to enter the exponent rather then the number. E.g. enter 3 to get 1000 iterations. Have a look at the code and now it should become obvious.

I chose this because "sys.argv[0]" doesn't seem to work reliably with numbers greater 999. Consequently if it works somehow, the first output is a digit with as many zeros as requested, and that output takes a while.

Günter

edit: I just realized that I can't run the program from the command line I had to go to the program catalog <SHIFT> <1> and run it fro there

RE: Summation based benchmark for calculators - toml_12953 - 08-18-2021 07:18 PM

(08-18-2021 06:56 PM)Guenter Schink Wrote:
(08-18-2021 04:01 PM)toml_12953 Wrote: I had digit grouping set to 123,456.789. When I change it to 123456.789, I don't get an error but I get a lot of zeroes on the screen and what may be an infinite loop.

o.K. I think I found the problem. It wasn't clearly explained. You have to enter the exponent rather then the number. E.g. enter 3 to get 1000 iterations. Have a look at the code and now it should become obvious.

I chose this because "sys.argv[0]" doesn't seem to work reliably with numbers greater 999. Consequently if it works somehow, the first output is a digit with as many zeros as requested, and that output takes a while.

Günter

edit: I just realized that I can't run the program from the command line I had to go to the program catalog <SHIFT> <1> and run it fro there

Oof! Es tuit mir leid! Yes, an exponent of 1000 would take a long time! Thank you for replying.

RE: Summation based benchmark for calculators - pier4r - 11-27-2022 10:34 AM

Updated up to post #233
Also the HP Xpander name and category fixed. Hopefully I didn't miss anything.

RE: Summation based benchmark for calculators - BINUBALL - 01-24-2023 01:12 PM

I have bought newer casio model, fx-570CW several months ago and reminded this thread. Test results below:

max = 10000
~ 581s - casio fx-570CW, 13955.8579 (sum function)
max = 1000
~ 59s - casio fx-570CW, 1395.346288 (sum function)
max = 100
~ 7s - casio fx-570CW, 139.297187 (sum function)
max = 10
~ 1s - casio fx-570CW, 13.71183502 (sum function)

fx-570CW is significantly faster than fx-570EX. Let's compare them.

1s / 2s (max = 10)
7s / 12s (max = 100)
59s / 109s (max = 1000)
581s / 1073s (max = 10000)

RE: Summation based benchmark for calculators - BINUBALL - 01-28-2023 09:22 AM

Tested with HP-200LX too.

max = 10000
~ 257s - HP-200LX turbo c 2.01, 13955.857904
max = 1000
~ 27s - HP-200LX turbo c 2.01, 1395.346288
max = 100
~ 4s - HP-200LX turbo c 2.01, 139.297187
max = 10
~ 2s - HP-200LX turbo c 2.01, 13.711835

Code:

#include <math.h>

main()

{

double i,sum;

i=1;

sum=0;

for(i=1;i<=10000;i++)

{

sum+=pow(exp(sin(atan(i))),1/3.);

}

printf("%lf\n",sum);

}

RE: Summation based benchmark for calculators - Eddie W. Shore - 04-05-2023 12:53 PM

Σ( ³√[e^(sin(tan^-1(x)))], x=1 to 1000)

Casio fx-991CW:
Result: 1395.346288
Time: approx. 57 seconds

TI-30X Pro MathPrint (Europe):
Result: 1395.346288
Time: approx. 1 minute, 44 seconds

RE: Summation based benchmark for calculators - Guenter Schink - 04-05-2023 08:01 PM

Once again Python on PRIME

This time using "Generator"

Code:

from math import *

from hpprime import *

Number=input("Enter number:")

Number=int(Number)

print(" {:,d}".format(Number))

t=eval('time')

Sum = sum(pow(e**(sin(atan(i))),1/3) for i in range(1,Number+1)) 

t=eval('time')-t

print("Sum = {:}".format(Sum),"\nTime = {:}".format(round(t,8)))

significantly faster than the previous published Python code (Top of this page). The calculation is done in this one-liner:
>>Sum = sum(pow(e**(sin(atan(i))),1/3) for i in range(1,Number+1)) <<

Code:

10,000      ->   0.101 sec     13955.8579042916

100,000     ->   1.038 sec    139560.9761411052

1,000,000   ->  10.381 sec   1395612.15872538

10,000,000  -> 103.724 sec  13956123.9815476

Günter

RE: Summation based benchmark for calculators - gehakte_bits - 04-10-2023 02:13 PM

HP Prime G2 V.14603 Python
==========================
100.000 -> 0.84s
1.000.000 -> 8.4s
10.000.000 -> 84.1s

optimized post #238:

x=1/3
Sum = sum(exp(sin(atan(i)))**x for i in range(1,Number+1))

RE: Summation based benchmark for calculators - Mike T. - 04-10-2023 10:21 PM

(01-28-2023 09:22 AM)BINUBALL Wrote: Tested with HP-200LX too.

Couldn't resist trying this on a modern PC...

Code:

#include <math.h>

#include <stdio.h>

void main()

{

  double i,sum;

  i=1;

  sum=0;

  for(i=1;i<=10000;i++)

  {

  sum+=pow(exp(sin(atan(i))),1/3.);

  }

  printf("%lf\n",sum);

}

$ gcc sum.c -o sum -lm

$ time ./sum 

13955.857904

real    0m0.005s

user    0m0.005s

sys    0m0.001s

$