Hollywood way to benchmarks...?

Discuss any general programming issues here
HelmutH
Posts: 198
Joined: Fri Feb 19, 2010 1:41 pm
Location: Oberhausen / Germany
Contact:

Re: Hollywood way to benchmarks...?

Post by HelmutH »

Hi Tuxedo
Nice tool. :)
Here my test result with HBT12 on my AOne XE 800, OS4.1 FE and Radeon 9200.

CPU 1: 3276 CPU 2 : 20679 CPU 3: Nil CPU 4: 10219
Graphics 1: 18578 Graphics 2 : 84 Graphics 3: 6229 Graphics 4: 6687
greeting Helmut
User avatar
Tuxedo
Posts: 338
Joined: Sun Feb 14, 2010 12:41 pm

Re: Hollywood way to benchmarks...? (HELP needed!)

Post by Tuxedo »

@Andreas

Ok, I've been updated to AmigaOS4.1FE and got really WEIRD results, here the score:

AmigaOS 4.1 Final Edition:

CPU 1: 2294 CPU 2 : 14542 CPU 3: Nil CPU 4: 7205
Graphics 1: 13639 Graphics 2 : 68 Graphics 3: 4250 Graphics 4: 4324

AmigaOS4.1.6 with A1FE kernel:

CPU 1: 2375 CPU 2 : 14738 CPU 3: Nil CPU 4: 7236
Graphics 1: 13714 Graphics 2 : 69 Graphics 3: 4307 Graphics 4: 4376

AmigaOS4.1.6 with 4.1.6 Kernel:

CPU 1: 1553 CPU 2 : 9843 CPU 3: Nil CPU 4: 4789
Graphics 1: 9335 Graphics 2 : 59 Graphics 3: 4112 Graphics 4: 4188

It seems that something in kernel will slowdon that tests significatively...any idea om what/why?

I've also do some test with my LoView image viewer and noticed instead some sort of 0,1 secs faster loading in every picture dont matter the size of the pic...so 1,25 secs will load roughly in 1,15 and 1,75 in 1,65, so up results was also more weird...

Plz help me!

PS plx also reply to my question in above reply about applets on AmigaOS-like Hollywood players...

PPS that was the HBT source code:

Code: Select all

/* ************** Hollywood made benchmarks tool *********** */

@DISPLAY 1, {Hidden = FALSE, Width = 800, Height = 600, Sizeable = TRUE, NoModeSwitch = TRUE, Color = $444444}	; definisco il Display 1(quello primario) e lo nascondo così da cambiarlo e poi aprirlo in base alla necessità...

Box (0, 0, 800, 600, $ff0000)

LOCAL y = 0, x1 = 0

; ***** TEST 1 ***** semplice ciclo For Next per vedere il tempo di calcolo

StartTimer(1)

	For LOCAL i = 1 TO 300000
	Next

LOCAL t1 = GetTimer(1)

StopTimer(1)

; ***** TEST 2 ***** test "misto" per impegnare la cpu un po'...

StartTimer(2)

LOCAL yy = 1

For LOCAL xx = 1 to 300000
	yy = yy * xx
	yy = xx * xx
	yy = xx / yy
	yy = xx *(xx/yy)
	yy = xx * (XX^2) * (xx^3) * (xx^4) * (xx^5)
Next
LOCAL t2 = GetTimer(2)
StopTimer(2)

/*
; ***** TEST 3 ***** allocazione e riepimento della memoria

StartTimer(3)

	AllocMem(1, 102400000)
	FillMem(1, 0, 102400000)

LOCAL t3 = GetTimer(3)
StopTimer(3)

; ***** TEST 4 ***** allocazione e riepimento della memoria

LOCAL l = 0, logval = 0
*/

StartTimer(4)

	For LOCAL i = 1 TO 300000
	logval = i * 3,4
	l = Ln(logval)
	Next

LOCAL t4 = GetTimer(4)
StopTimer(4)




/* NB quì sotto iniziano i test grafici che sono numericamente sulla seconda decina per semplicità di 
organizzazione e modifica  */

; ***** TEST 11 ***** Disegno on screen di un pixel alla volta 

StartTimer(11)

BeginRefresh()
	For LOCAL x = 1 TO 480000
		Plot(x - x1, y, #RED)
		IF Frac(x/800) = 0
			y = y + 1
			x1 = (800 * Int(x/800))
		EndIF
	Next
EndRefresh()

LOCAL t11 = GetTimer(11)
StopTimer(11)



; ***** TEST 12 ***** Disegno on screen di quadrati concentrici sempre più grandi

StartTimer(12)

LOCAL bl, bh = 0, 0

	For LOCAL x = 1 TO 800
		BeginRefresh()
			Box(#CENTER, #CENTER, bl + x ,bh + x, #GREEN)
		EndRefresh()
	Next

LOCAL t12 = GetTimer(12)
StopTimer(12)

; ***** TEST 13 ***** Disegno on screen di cerchi concentrici sempre più grandi

StartTimer(13)

LOCAL centro = 0

	For LOCAL x = 1 TO 400
		BeginRefresh()
			Circle(#CENTER, #CENTER, centro + x, #BLACK)
		EndRefresh()
	Next

LOCAL t13 = GetTimer(13)
StopTimer(13)

; ***** TEST 14 ***** Disegno on screen di ellissi concentriche sempre più grandi

StartTimer(14)

LOCAL ex, ey = 1, 1

	For LOCAL x = 1 TO 400
		BeginRefresh()
			Ellipse(#CENTER, #CENTER, ex + x, ey + x/2, #YELLOW)
			Ellipse(#CENTER, #CENTER, ex + x/2, ey + x, #YELLOW)
		EndRefresh()
	Next

LOCAL t14 = GetTimer(14)
StopTimer(14)

Cls()

TextOut(10, 150, "CPU 1: " .. t1 .. ", CPU 2 : " .. t2 .. ", CPU 3: " .. t4) ;  .. ", CPU 4: " .. t4)

TextOut(10, 200, "Graphics 1: " .. t11 .. ", Graphics 2 : " .. t12 .. ", Graphics 3: " .. t13 .. ", Graphics 4: " .. t14, #RED)

; display dei risultati delle funzioni "matematiche"

DebugPrint("CPU 1: ", t1, "CPU 2 : ", t2, "CPU 3: ", t3, "CPU 4: ", t4)

; display dei risultati delle funzioni grafiche

DebugPrint("Graphics 1: ", t11, "Graphics 2 : ", t12, "Graphics 3: ", t13, "Graphics 4: ", t14)

WaitLeftMouse()
Simone"Tuxedo"Monsignori, Perugia, ITALY.
User avatar
airsoftsoftwair
Posts: 5433
Joined: Fri Feb 12, 2010 2:33 pm
Location: Germany
Contact:

Re: Hollywood way to benchmarks...?

Post by airsoftsoftwair »

Why If I compile an applet with Hollywood 5.3 GUI the hollywood player under AmigaOS like systems says that it wasnt a compressed applet and so wasnt playable but same applet on Android hollywood player works without problem?
The Android player also works with uncompressed applets. The next update of the Hollywood Player will support uncompressed applets for all platforms, though.
It seems that something in kernel will slowdon that tests significatively...any idea om what/why?
Sorry, I don't know why you get these results because I don't know anything about changes in the OS4 kernel...
User avatar
Tuxedo
Posts: 338
Joined: Sun Feb 14, 2010 12:41 pm

Re: Hollywood way to benchmarks...?

Post by Tuxedo »

Ok,
I'll try to post test result on AW and see what happen than...

Regarding compressed applets why was generated automatically and not works on AmigaOS? I dont press anything ceating it...s lo I toughts to create an applet that works everywere...I made somethint wrong?
Simone"Tuxedo"Monsignori, Perugia, ITALY.
User avatar
airsoftsoftwair
Posts: 5433
Joined: Fri Feb 12, 2010 2:33 pm
Location: Germany
Contact:

Re: Hollywood way to benchmarks...?

Post by airsoftsoftwair »

You need to compile applets using the -compress argument, i.e.

Code: Select all

Hollywood test.hws -compile ram:test.hwa -exetype applet -compress
But as I said, the next version of the player will also supported uncompressed applets.
User avatar
Tuxedo
Posts: 338
Joined: Sun Feb 14, 2010 12:41 pm

Re: Hollywood way to benchmarks...?

Post by Tuxedo »

Ok,
I asked because from the gui that seems not possible...
And I always compile my scripts from gui...
Simone"Tuxedo"Monsignori, Perugia, ITALY.
User avatar
Tuxedo
Posts: 338
Joined: Sun Feb 14, 2010 12:41 pm

Re: Hollywood way to benchmarks...?

Post by Tuxedo »

airsoftsoftwair wrote:
Hi again, it seems you miss a my post about some weird differences from OS4.1FE and MOS3.7 on cpu that seems to me really not fine since on cpu dunno why we can get on same machine big differences...

That's the results:

CPU 1: 1547 CPU 2 : 9860 CPU 3: 4780
Graphics 1: 9403 Graphics 2 : 50 Graphics 3: 4039 Graphics 4: 4120

Results related to MOS 3.7 with Pegasos2 G4@1131 and Radeon9000 pro:

CPU 1: 654 CPU 2 : 4758 CPU 3: CPU3: 2195
Graphics 1: 6815 Graphics 2 : 50 Graphics 3: 5001 Graphics 4: 5243

Wasnt possible in theory?

I miss something?
Simone"Tuxedo"Monsignori, Perugia, ITALY.
User avatar
airsoftsoftwair
Posts: 5433
Joined: Fri Feb 12, 2010 2:33 pm
Location: Germany
Contact:

Re: Hollywood way to benchmarks...?

Post by airsoftsoftwair »

Hi again, it seems you miss a my post about some weird differences from OS4.1FE and MOS3.7 on cpu that seems to me really not fine since on cpu dunno why we can get on same machine big differences.
Of course that's possible because OS4 and MorphOS are implemented in entirely different ways so there can be huge differences... Hollywood's code is pretty much the same on OS4 and MorphOS but that's not decisive. Memory management, gfx drivers and threading are completely different between OS4 and MorphOS...
User avatar
Tuxedo
Posts: 338
Joined: Sun Feb 14, 2010 12:41 pm

Re: Hollywood way to benchmarks...?

Post by Tuxedo »

I see...
however that big differences may occur in complete different program behaviours in OS4.x and MOS and maybe gets a thing nice to do on MOS and sluggish on OS4.x....

I'll investigate deeper on that with some more tests to look effective performance behaviours on the 2 systems, just after completing studying MUIRoyale...

Thank you for reply.
Simone"Tuxedo"Monsignori, Perugia, ITALY.
Post Reply