VDP speed curiousity

Page 1/12
| 2 | 3 | 4 | 5 | 6

By Daemos

Paragon (1664)

Daemos's picture

15-02-2014, 22:24

I just wonder. What kind of numbers people will get on different machines when running this benchmark tool I found on the net.

I get 146000 on the first test on my nms8255 machine (V9958)

Strange is that most of the numbers I read on the net basicly point out that the VDP runs about 4-8% slower on MSX2+ and turbo-R machines. I wonder if this is true. I also wonder if there is a difference between V9958 and V9938.

This is just pure curiousity.

Login or register to post comments

By Manuel

Ascended (14673)

Manuel's picture

15-02-2014, 23:15

146122 on Philips NMS 8250 on openMSX Smile
2nd test: 94578
3rd test: 8693
4th test: 30993
5th test: 22474
6th test: 10781

Can someone run it on real 8250 and report the results?

By mars2000you

Enlighted (5372)

mars2000you's picture

15-02-2014, 23:29

Quick tests made on blueMSX with boosted generic machines under DOS1 :

MSX 2 European - test 1 : 146136 - test 2 : 94588 - test 3 : 8694 - test 4 : 30996 - test 5 : 22476 - test 6 : 10782

MSX2 Japanese - test 1 : 122217 - test 2 : 79093 - test 3 : 7284 - test 4 : 25919 - test 5 : 18830 - test 6 : 9033

MSX2+ Japanese and TurboR Japanese - test 1 : 120155 - test 2 : 77711 - test 3 : 6917 - test 4 : 25009 - test 5 : 18049 - test 6 : 8848

By hit9918

Prophet (2777)

hit9918's picture

15-02-2014, 23:33

I set breakpoint 0x38 while it draws and debugger halts.
so it didnt disable interrupts, again disk ROM and extensions change results.

By mars2000you

Enlighted (5372)

mars2000you's picture

15-02-2014, 23:37

You can notice that 146136/122217 = 1.2, and 60/50 = 1.2.

So there's an inversal relation with the 50 or 60 Hz mode.

By Daemos

Paragon (1664)

Daemos's picture

15-02-2014, 23:42

Quote:

Can someone run it on real 8250 and report the results?

146105
94568
8707
30990
22509
10798

By hit9918

Prophet (2777)

hit9918's picture

15-02-2014, 23:39

much more likely a 2+ has more extension software ado than 9958 having different timings as 9938.

By Daemos

Paragon (1664)

Daemos's picture

15-02-2014, 23:52

@hit:

So this test is basicly unreliable? It doesn't disable ints so you will get wierd results if the machines are different. The sourcecode was supplied with the program. Lets see if we can disable the ints.

EDIT: the program has custom int handler so extensions cannot change results.

Quote:

Quick tests made on blueMSX with boosted generic machines under DOS1 :

MSX 2 European - test 1 : 146136 - test 2 : 94588 - test 3 : 8694 - test 4 : 30996 - test 5 : 22476 - test 6 : 10782

MSX2 Japanese - test 1 : 122217 - test 2 : 79093 - test 3 : 7284 - test 4 : 25919 - test 5 : 18830 - test 6 : 9033

MSX2+ Japanese and TurboR Japanese - test 1 : 120155 - test 2 : 77711 - test 3 : 6917 - test 4 : 25009 - test 5 : 18049 - test 6 : 8848

interesting. I guess the T9769 inserts some extra wait states?

By hit9918

Prophet (2777)

hit9918's picture

15-02-2014, 23:53

mhm but then all machines should do same speed.
where did the 8% difference occur.

By mars2000you

Enlighted (5372)

mars2000you's picture

15-02-2014, 23:57

Daemos : I don't remember exactly the cause, but Daniel Vik, when he was making some very special MSX1 demos, had indicated me that the compatibility was actually broken on MSX2+/turboR for some parts of these demos using special VDP tricks. That 's why these parts are skipped on MSX2+/turboR.

By hit9918

Prophet (2777)

hit9918's picture

16-02-2014, 00:04

The benchmark names sound like doing blitter.
But I rather see it doing cpu. as filename "port 98" is saying.
Then there is no possibility that VDP slows down the cpu.
Except that 9958 WAIT feature was enabled.
But that would have no meaning to games performance. And no meaning to bios performance.

Page 1/12
| 2 | 3 | 4 | 5 | 6
My MSX profile