Run slots faster then 3.5 mhz on a turbo-r

Page 1/2
| 2

By edoz

Prophet (2465)

edoz's picture

23-10-2014, 10:54

Yesterday I was wondering if it is theoretical possible to speedup the slots on a turbo-r ?
The slots are always running @ 3.5 MHz as this is the MSX standard. On a MSX 2 with 7 Mhz I got a higher true put of disk I/o then on a turbo-r. When I run the MSX 2 on7 MHz the slot is also running @ the same speed. So you could say that if you boosts the slots on the turbo-r it will run even faster than a MSX2 @ 7 Mhz? And what would be the speed benefit in this case ?

Login or register to post comments

By Daemos

Paragon (2044)

Daemos's picture

23-10-2014, 11:16

All the cartridges are designed to run on 3.5 Mhz so speeding matters up will let a whole lot of cartridges go wrong. Those who do not fail at 7Mhz will run faster. In theory twice the speed. Even for a huge ROM this speedincrease is not really a requirement though.

Otherwise boosting the speed of the cartridge port propably won't happen without complications. There is this pin hooked up to the S1990 called SLTCLK. that one is the clockspeed for the cartridgeport. I don't think that cranking that line up to 7Mhz will magicly make your turbo-R's slots run at 7Mhz without problems.

By edoz

Prophet (2465)

edoz's picture

23-10-2014, 13:07

It is true indeed that there are a lot of cartridges out there that supports 7 mhz. Like the moonsound, sunrise IDE, Some SCSI controllers, GFX9000 etc. etc. For example, I think it will boost the turbo-r a lot when using the Sunrise IDE or some other storage devices as the most storage devices supports 7 Mhz by default.

There is a lot of different between my measurements ..:

Turbo-r / BERT : 33 KBytes/sec
Turbo-r / Sunrise CF ATA : 175 KBytes/sec

8280 7Mhz / Sunrise CF ATA: 327 KBytes/sec
8280 3Mhz / Sunrise CF ATA: 161 Kbytes/sec
8280 7Mhz / BERT: 72 KBytes/sec
8280 3Mhz / BERT: 34 KBytes/sec

I heard about that the Novaxis SCSI interface got about 400kB/s on turboR in R800 DRAM mode.. I wonder how fast this is on the turbo-r when the slot is working @ 7 Mhz. I love this idea.

Probably you are right that it is difficult to speed up the slot on a turbo-r .. But is love the idea of a turbo switch on a turbo machine Wink but .... I think I don't want to try it on my turbo-r as it is rare computer.. Wink

By Retrofan

Paragon (1339)

Retrofan's picture

23-10-2014, 13:39

You could use SDSPEED which will use R800 instead of Z80 for storage devices...

By edoz

Prophet (2465)

edoz's picture

23-10-2014, 13:44

Yes.. but even when you use R800 the slot is still @ the same speed as a normal MSX.. as it is always 3.5 Mhz.

By Retrofan

Paragon (1339)

Retrofan's picture

23-10-2014, 13:48

And what's the speed when using this tool? -> Turbo-r / Sunrise CF ATA : 175 KBytes/sec

By edoz

Prophet (2465)

edoz's picture

23-10-2014, 13:53

I will try it out Wink

By edoz

Prophet (2465)

edoz's picture

23-10-2014, 14:45

What is SDSPEED doing ? It seems it was patching something ? But it does not work on the sunrise IDE, only on the padial ?
If I use the util DSPEED I got a 817.42385 Kbit/sec

By Retrofan

Paragon (1339)

Retrofan's picture

23-10-2014, 16:26

Turbo-R normally switches back to Z80 with diskaccess. This tool will take advantage of R800 diskaccess. Also look over here

By Grauw

Ascended (10699)

Grauw's picture

23-10-2014, 19:19

The reason modern IDE interfaces are so fast compared to older ones is because they use memory-mapped I/O. They can read entire 512-byte sectors straight into RAM with a single LDIR or, even faster, 512 LDI instructions. The I/O ports are not as efficient, especially not with 7MHz circuits which slow down on I/O access but not on memory access. The turboR also slows down a bit for external memory access btw.

By edoz

Prophet (2465)

edoz's picture

23-10-2014, 19:35

Grauw wrote:

The reason modern IDE interfaces are so fast compared to older ones is because they use memory-mapped I/O. They can read entire 512-byte sectors straight into RAM with a single LDIR or, even faster, 512 LDI instructions. The I/O ports are not as efficient, especially not with 7MHz circuits which slow down on I/O access but not on memory access. The turboR also slows down a bit for external memory access btw.

Can you explain this a bit more ? Why is the 7 Mhz slowing down the circuits ? As I have the best result on a normal MSX with 7 Mhz compare to the turbo-r...

About the memory access. I found out once while testing SymbOS beta version... Prodatron told me that he use, on the turbo-r, always the internal memory (because this is faster) In some test version this was not the case and I noticed it directly.

Page 1/2
| 2