PPL vs others!

by **Guest** » Jan 4, 2005 @ 11:54pm

Curious.. any benchmarks from Java for comparison?

by **Smok Wawelski** » Mar 20, 2005 @ 3:53am

Hi,

I've done a small test in... Macromedia Flash MX

(AKA Flash 6) (interpreted bytecode).

Athlon XP 2500+

code:

1 
2 
3 
4 
5 
6 
t = getTimer();
while ((getTimer() - t) <= 1000) {
  i = i + 1;
}
trace("Iterations per second : "+ i);
i = 0;
6 lines; 1 keywds; 3 nums; 24 ops; 1 strs; 0 coms    Syntactic Coloring v0.4 - Dan East

about 84000 iterations

with i++: 95000

Tomorrow I'll check Java 1.5 (Tiger) and send results and flash 6 player on 400Mhz ARM

by **Smok Wawelski** » Mar 20, 2005 @ 2:52pm

Java benchmark, console application, JDK 1.5.0.01:
(on Athlon 2500+ XP)

Code: Select all

1 
2 
3 
4 
5 
6 
7 
8 
9 
10 
11 
12 
13 
14 
15 
16 
17 
import java.awt.*;
import java.util.*;

class javatestspeed
{
        public static void main(String args[])
        {
    int i = 0;
    long t = System.currentTimeMillis();
    while (System.currentTimeMillis() - t <= 10000)
    {
    i = i + 1;
    //i++;
    }
     System.out.println("Iterations per second " + i / 10);
        }
}
17 lines; 8 keywds; 4 nums; 43 ops; 1 strs; 1 coms    Syntactic Coloring v0.4 - Dan East

with i = i + 1: 12,372,875
with i++ : 12,348,473
with i = i + 2 : 12,281,583

I'm afraid there's one thing we did not think about: function getTime() - it may interfere with pure incrementations. Let's say we have a very fast environment but a slow getTime function and slower environment with very fast getTime... how to intepret such a result?

by **Smok Wawelski** » Mar 20, 2005 @ 2:58pm

Qtek 2020, ARM 263 @400Mhz, flash player 6 with ZINC V2:

Code: Select all

1 
2 
3 
4 
5 
6 
t = getTimer();
while ((getTimer() - t) <= 1000) {
  i = i + 1;
}
trace("Iterations per second : "+ i);
i = 0;
6 lines; 1 keywds; 3 nums; 24 ops; 1 strs; 0 coms    Syntactic Coloring v0.4 - Dan East

The result: 5000-6000 (ONLY!!!).

by **kornalius** » Mar 20, 2005 @ 5:31pm

Java is quite fast! Wow. Does anyone know if this is JIT compiler?

I would also like to know if there are any java source code anywhere?

by **Smok Wawelski** » Mar 20, 2005 @ 7:33pm

JDK 1.5 (Tiger) comes with highly optimized JIT compiler called Hotspot. There's also a bunch of other JIT:

http://www.geocities.com/marcoschmidt.g ... ilers.html

or try http://www.google.pl/search?hl=pl&q=jav ... gle%21&lr=

I'm afraid, there's no Java source available (yet). But look at Waba/Superwaba sources, they come with source code (waba for sure).

Do you want any other benchmarks? I could do BlitzBasic (PC) and Superwaba (PC & PPC) tests

by **kornalius** » Mar 20, 2005 @ 10:08pm

Why not, more is better!

by **kornalius** » Apr 5, 2005 @ 4:46pm

Updated the numbers from the latest version of PPL. The PPC version of PPL is quite a bit faster.

by **kornalius** » Apr 14, 2005 @ 11:43pm

Updated the PPC values again. These tests were done using the latest release of PPL with the new integreated optimizer.

by **mmtbb** » Apr 30, 2005 @ 7:36am

I just noticed that my mother's 2.8 Ghz celeron runs PPL games at 1300 fps. My AMD 2800+ does 630 fps. It seems like it was faster in the past. any sugestions as to the slowness?

by **mmtbb** » Apr 30, 2005 @ 7:42am

Ahh haaa! I figured it out! I switched from 32 bit color to 16 bit on the desktop. It went from 630 fps to 5333 fps. That's more like it!

by **sponge** » Apr 30, 2005 @ 6:35pm

Can the runtime switch to 16 bit colour if it detects 32bit on the desktop? This was a problem that lots of older Win games ran into, not sure what causes it though.

by **kornalius** » May 1, 2005 @ 10:08pm

Right now the runtime is 16bits by default. I will take a look to see why it is not switching to 16bits while changing resolution.

by **Zibx** » Jan 15, 2006 @ 11:14am

I'va got an idea — not to use gettime function so often, but use it in start, after that use code like:

Code: Select all

1 
2 
for(a$=0;a$<1000000;a$++)
  i$=i$+1;
2 lines; 1 keywds; 3 nums; 11 ops; 0 strs; 0 coms    Syntactic Coloring v0.4 - Dan East

and after that display the difference betwen current time, and first one.

P.S. sorry, i don't know the syntax of cycles in your language (I here only the first day).
P.P.S. sorry for my english, it's not my first language =)
P.P.P.S. I can help to write documentation on russian

by **kornalius** » Jan 15, 2006 @ 5:08pm

Yep this would be a better test.

I will try this:

Code: Select all

1 
2 
3 
4 
5 
int i = 0;
unsigned long t = GetTickCount();
while (i <= 1000000)
  i = i + 1;
printf("Iterations per second : %f", 1000000.0 / ((GetTickCount() - t) / 1000));
5 lines; 4 keywds; 6 nums; 27 ops; 1 strs; 0 coms    Syntactic Coloring v0.4 - Dan East

I will give an update soon.

PPL vs others!

Idea

Sort

Forum Description

Moderator:

Forum permissions

`1 2 3 4 5 6`	`t = getTimer(); while ((getTimer() - t) <= 1000) { i = i + 1; } trace("Iterations per second : "+ i); i = 0;`
	6 lines; 1 keywds; 3 nums; 24 ops; 1 strs; 0 coms Syntactic Coloring v0.4 - Dan East

`1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17`	`import java.awt.; import java.util.; class javatestspeed { public static void main(String args[]) { int i = 0; long t = System.currentTimeMillis(); while (System.currentTimeMillis() - t <= 10000) { i = i + 1; //i++; } System.out.println("Iterations per second " + i / 10); } }`
	17 lines; 8 keywds; 4 nums; 43 ops; 1 strs; 1 coms Syntactic Coloring v0.4 - Dan East

`1 2`	`for(a$=0;a$<1000000;a$++) i$=i$+1;`
	2 lines; 1 keywds; 3 nums; 11 ops; 0 strs; 0 coms Syntactic Coloring v0.4 - Dan East

`1 2 3 4 5`	`int i = 0; unsigned long t = GetTickCount(); while (i <= 1000000) i = i + 1; printf("Iterations per second : %f", 1000000.0 / ((GetTickCount() - t) / 1000));`
	5 lines; 4 keywds; 6 nums; 27 ops; 1 strs; 0 coms Syntactic Coloring v0.4 - Dan East