[HP35s] Benchmark of some operations

[fred_76]

Hello,

Here is a benchmark of some of the HP35s operations/functions. It shows the execution time in milliseconds.

Code:


Function             Time (ms)
1                         28.1
1.1                       37.4
1.11                      40.5
1.111                     43.7
12                        31.2
123                       34.3
1234                      37.5
-                         13.7
*                         16.3
/                         18.3
+                         13.8
+/-                        8.4
1/x                       14.7
CLx                        2.3
cos                       37.8
DES                         31
ENTER                      4.4
GTO                        2.8
ISG                       29.4
LASTx                      4.6
LBL N                      1.6
OR                         0.7
pi                         9.7
R^                         4.3
RCL x                      5.1
RCL+ D                     9.7
RCL-D                      9.7
REGY                      31.7
RMDR                      10.3
Rv                         4.3
SF                         1.8
sin                       21.9
sqrt                      14.3
STO x                      1.9
STOx U                    11.7
tan                         28
x<=0?                      6.5
x<>A                       4.4
x<>y                       3.2
x<0?                       3.7
x=0?                         3
x>=y?                      3.5
x²                          11
XEQ                        6.8

You will notice that the more digits a number has, the longer it takes for execution.

Take these figures as indicative only !

Fred

[Leviset]

Out of interest - How are you measuring these?

[fred_76]

(02-13-2019 10:57 AM)Leviset Wrote:  Out of interest - How are you measuring these?

I programmed a loop, and first manually measure the time taken by the empty loop (t0). I do this manually as the HP35s does'nt have internal timer.


The code is like that, nnn being the number of iterations (usually 1000) :

Code:


L001 LBL L
L002 nnn
L003 STO I
L004 (set of) instruction(s) to test
L005 DSE I
L006 GTO L004
L007 RTN

Then I just have to insert the instruction(s) to test in line 4, restart the test program and measure the time t1.

The time per (set of) instruction(s) is then 1000*(t1-t0)/nnn in ms.

I repeat this several times (especialy when t1-t0 < 10s) and take the shortest time measured.

It is quite precise, I calculated the time for the calculation of the prime number 99 999 999 977 using the algorithm shown here and found a theoretical value of 14'40", to be compared to the real time of 14'37".

Fred