MSX HDMI mutlimedia card

Страница 35/56
28 | 29 | 30 | 31 | 32 | 33 | 34 | | 36 | 37 | 38 | 39 | 40

By maxis

Champion (512)

Аватар пользователя maxis

13-01-2016, 17:15

AxelStone wrote:

@maxis Does we have an estimated launch date and price? Smile

Actually I'm going to discuss this particular topic in Nijmegen.

P.S. The support for the "CRT scanlines" will be added later. In principal everything works right now. This sort of scanline emulation wouldn't require any additional resources from the board. Simply the restriction is in the supported HDTV video modes. Currently I don't use that EDID info.

By PingPong

Prophet (3558)

Аватар пользователя PingPong

13-01-2016, 20:35

@maxis: nice to have v99x8 support!
Few questions: it's possible to configure at which address the HDMI vdp is mapped?
Actually in your vdp model have you took in account the document about the vram timings of the original chip?
How much is ( in your efforts intents ) is the vdp model close to the real one?
(The latter question is justified because most of the time the accuracy needed is far to be perfect, most sw works even with a simplified model. Plus i suspect that it's not so easy to emulate a ancient beast like the v9938 from the perspective of vram timings: you have a completely different architechture nowadays fortunately)
Is the vdp cmd engine fast (sorry slow) as the old one?

thx for your attention

By maxis

Champion (512)

Аватар пользователя maxis

14-01-2016, 00:55

Hello, PingPong,

The base address of the VDP can be software defined. It is the part of the HOST interface design. VDP itself receives the decoded address and the bus transaction of the global system clock (see below).
I have consulted the VDP timing model and made the following approach:
1. The system architecture of the VDP was designed in a way that main video DMA, host, sprites and the command engine have the independent channels in the access to the main RAM. These are the "memory bus masters".
2. Each bus master has its own bus request timer, which CAN be loaded by a different value based on the screen mode, etc. Currently the timers have only 2 different values, one for the pattern modes and the other for the bitmap modes.
3. By having these timers, the architecture can be slowed down to mimic the original V99x8 behavior, but currently the timing model is simplified.
4. By sharing the SDRAM between the main CPU instruction/data accesses and the VDP, the priority is always given to the CPU in order not to slow down the latter.
5. The CMD engine is faster than the original one, but could be slowed down. All the "automatic" commands, not involving the CPU (LMMV, LINE, LMMM, HMMM, YMMM etc) are running much faster, but also can be slowed down by using the "timers".
6. The architecture is very very different from the original chip and from the OCM (Procyon has five fully asyncronous clock domains=> one global and 4 local). The global clock is defined by the SDRAM main clock and the global synthesis timing goals. Currently it is 100 MHz. The other clock domains are for the HDMI video out (mode dependent), digital audio and the MSX slot clock.

Thank you,

By PingPong

Prophet (3558)

Аватар пользователя PingPong

14-01-2016, 19:20

maxis wrote:

Hello, PingPong,

The base address of the VDP can be software defined. It is the part of the HOST interface design. VDP itself receives the decoded address and the bus transaction of the global system clock (see below).
I have consulted the VDP timing model and made the following approach:
1. The system architecture of the VDP was designed in a way that main video DMA, host, sprites and the command engine have the independent channels in the access to the main RAM. These are the "memory bus masters".
2. Each bus master has its own bus request timer, which CAN be loaded by a different value based on the screen mode, etc. Currently the timers have only 2 different values, one for the pattern modes and the other for the bitmap modes.
3. By having these timers, the architecture can be slowed down to mimic the original V99x8 behavior, but currently the timing model is simplified.
4. By sharing the SDRAM between the main CPU instruction/data accesses and the VDP, the priority is always given to the CPU in order not to slow down the latter.
5. The CMD engine is faster than the original one, but could be slowed down. All the "automatic" commands, not involving the CPU (LMMV, LINE, LMMM, HMMM, YMMM etc) are running much faster, but also can be slowed down by using the "timers".
6. The architecture is very very different from the original chip and from the OCM (Procyon has five fully asyncronous clock domains=> one global and 4 local). The global clock is defined by the SDRAM main clock and the global synthesis timing goals. Currently it is 100 MHz. The other clock domains are for the HDMI video out (mode dependent), digital audio and the MSX slot clock.

Thank you,

thank for your explanation, maybe could you differentiate between no blank and active area in speed ?
thx

By maxis

Champion (512)

Аватар пользователя maxis

21-01-2016, 16:12

PingPong wrote:

thank for your explanation, maybe could you differentiate between no blank and active area in speed ?
thx

Well, technically it is doable. The timing refinement will come when I'll complete all the features of V9958. I'll start working on them after the fair.
Thank you!

By maxis

Champion (512)

Аватар пользователя maxis

21-01-2016, 16:17

Here is a couple of video clips about Procyon running MSX2 games!

Zanac-EX

Space Manbow

Interesting fact: at the startup Space Manbow switches the scan mode to 60Hz.

See you at Nijmegen MSX FAIR!

By syn

Paragon (1985)

Аватар пользователя syn

21-01-2016, 20:50

im looking forward to see this device in action during Nijmegen fair, imho one of the most interesting projects recently Smile

By -Neo-

Champion (384)

Аватар пользователя -Neo-

22-01-2016, 01:11

Very interesting. Looking forward too.

You can already count me in for an assembled one.

By maxis

Champion (512)

Аватар пользователя maxis

24-01-2016, 23:23

Thank you guys for being interested and visiting my setup at Nijmegen MSX Fair 2016!!!

The time passed too quickly.
I was so much preoccupied with the visitors and the setup and some testing, that I had no pictures of my booth Tongue

If you have made some photos of the setup or even me, could you, please, share it with me? At least I could show it to my friends/colleagues.

P.S. Great KUDOS to Manuel for helping me in many many ways!!!!! Even for the beamer, he had managed to organize it right on the place! Beamer which I brought wasn't bright enough. But Manuel got the superbright beamer for me, which could provide the necessary experience! And it all worked flawlessly. Thank you so much, Manuel!

PPS. Special thanks to Alexey(RBSC) for kindly providing his game cartridge for the tests. Interesting thing, that some games still use SCREEN 3 mode (which I haven't implemented yet).

By Grauw

Ascended (9397)

Аватар пользователя Grauw

24-01-2016, 23:32

Maxis, it was great to meet you Smile. Thanks for coming all this way!

Страница 35/56
28 | 29 | 30 | 31 | 32 | 33 | 34 | | 36 | 37 | 38 | 39 | 40