TurboR OCM

Pagina 1/2
| 2

Door MichelM

Expert (80)

afbeelding van MichelM

28-10-2014, 04:37

I'm curious about something: Why isn't there a turboR OCM yet? We've seen MSX2 and MSX2+ one chip implementations, but what is keeping us from having a turboR OCM?

Aangemeld of registreer om reacties te plaatsen

Van Latok

msx guru (3925)

afbeelding van Latok

28-10-2014, 06:58

The gates size of the FPGA used...

Van Retrofan

Paragon (1339)

afbeelding van Retrofan

28-10-2014, 07:45

Ofcourse that's very cool to have, but unfortunately there's not a lot of official software for MSX turbo R as for MSX1/MSX2 with a lot of titles. What do you miss of MSX turbo R? PCM? MSX turbo R is actually a super charged MSX2+. I think it's better to work on compatibility for the current MSX2+ firmware (VDP engine), before looking at next possibilities...

Van meits

Scribe (6530)

afbeelding van meits

28-10-2014, 08:24

I like the speed of my turbo R... Haven't seen my OCM go as fast as that...

Van Retrofan

Paragon (1339)

afbeelding van Retrofan

28-10-2014, 08:55

Me too, but do we need that for the MSX1/MSX2 software? Did you try 10 MHz mode of OCM? It's quite fast as well and switchable with F12... Wink

Van meits

Scribe (6530)

afbeelding van meits

28-10-2014, 09:21

I did... My turbo R outruns that with ease...

Van AxelStone

Prophet (3189)

afbeelding van AxelStone

28-10-2014, 11:33

Meits wrote:

I did... My turbo R outruns that with ease...

The R800 runs at 7.16Mhz but can runs Z80 instructions 4xtimes faster, so it's like a Z80 at 28.64Mhz, almost three times OCM!

Really the MSX turbo R is a great piece of hardware, sad not to see exclusive software Sad

Van KdL

Paragon (1450)

afbeelding van KdL

28-10-2014, 12:47

MichelM wrote:

I'm curious about something: Why isn't there a turboR OCM yet? We've seen MSX2 and MSX2+ one chip implementations, but what is keeping us from having a turboR OCM?

I do not think really a problem to do but there are not many people interested into VHDL code.
However I confirm for the high gates size of R800 --> 10.000 (R800) vs 3.000 (Z80)
http://www.hat.hi-ho.ne.jp/tujikawa/esepld/intro.html

Van Manuel

Ascended (19270)

afbeelding van Manuel

28-10-2014, 22:46

Who so many more? The R800 isn't *that* spectacularly different is it? Just a few more instructions and loads faster...

Van hit9918

Prophet (2927)

afbeelding van hit9918

29-10-2014, 00:38

The R800 must be very different inside.
When an instruction goes in one cycle instead some neat 4 cycles microprogram.
And then a 16bit add too goes in 1 cycle while on z80 it takes more than twice an 8bit add.
That shows a huge difference even if it were secretly clocked 4x Mhz inside.
Unfortunately it's not the same picture for the 16bit add of making the address of (ix+offset), relative to simple instructions it is even slower than on z80.
It smells like chicken and egg, IX register was used little on z80, and then R800 is cheap on it.
Still coding whatever style you feel comfortable with runs factors faster than z80 Smile

Van darkschneider666

Supporter (8)

afbeelding van darkschneider666

29-10-2014, 23:54

Sorry, but i want to put my 0,02 cents in this.

The OCM is manufactured with an FPGA that it's almost used in all capacity, have some space of course, but don't for a Turbo R implementation. Of course, if someone make a new one with another FPGA that has more capacity, there will be no problem.
Some another issue is that what Retrofan talked about, has no such many things more to justify the work.

hit9918 every enterprise that manufactured the Z80 fabricated a different chip inside, the difference in the speed can be translated by many factors, for example, the technology, Z80 originally was manufactured in NMOS process while R800 probably was manufactured in CMOS process, in fact, i'm don't know exactly in NMOS process but in CMOS process i can guarantee that in every foundry the speed is very little different, as well as the architectural differences. This is only 2 factors that make two same IC very different.
But this isn't something that a end user should not worry, in this specific case, the problem is count time with the execution time of the instructions, this will create mismatch problems for obvious reasons. And this problem will remain forever unfortunately.

So, in my point of view, OCM like with Turbo R configurations, it's possible. The major problem is software that count time with execution time of the instructions, this will prevent any MSX go faster.

By the way, amazing work of HRA and KdL in the upgrades.

Pagina 1/2
| 2