Speed comparison for the C1541

[neutrino]

Found this neat transfer speed comparison for the C1541:

https://www.obliterator918.com/dtb/

1541                             403 byte/s     1x

SD2IEC                           502 byte/s     1x

1581                             507 byte/s     1x

1541 w/ Epyx Fastload           2121 byte/s     5x

1541 w/ Warp Speed              3017 byte/s     8x

1541-II w/ JiffyDOS             3746 byte/s     9x

1541 w/ Super Snapshot 5        4163 byte/s     10x

1541 w/ Action Replay 5         5352 byte/s     13x

1541 w/ Retro Replay PAL 3.8q   5352 byte/s     13x

1581 w/ JiffyDOS                6515 byte/s     16x

SD2IEC w/ Epyx Fastload         7136 byte/s     18x

SD2IEC w/ JiffyDOS              9175 byte/s     23x

1541 w/ DolphinDOS (U64)        9366 byte/s     23x

Parallel IEEE-488 on the Commodore PET seems to perform at circa 1 kB/s asfair.

 

[BruceMcF]

Warp speed was the one I had in grad school in the early 1980s. I originally got it as Warp Speed 128 for the C128D that I fried by plugging in a power tap to the cassette port upside down, but it had a switch between C128 and C64 mode, for when the C128 was in C64 mode, so I switched it to C64 mode and used it with my C64.

Reiterating previous discussions on this topic: C128 Fast Mode on the 1571 is around 1600byte/s and burst mode on the order of 4000bytes/s. Since it is an asynchronous protocol that acknowledges each bit received, it runs as fast as the slower of the two sides of the transfer supports, so an X16 burst mode with a SD2IEC modified to support burst mode should be able to beat the SD2IEC w/fastloader numbers above.

[neutrino]

What's the difference between Fast mode and Burst mode? only one of the modes use the CIA hardware shift register?

 

[BruceMcF]

On 11/2/2022 at 1:02 PM, neutrino said:

What's the difference between Fast mode and Burst mode? only one of the modes use the CIA hardware shift register?

The VIC-20 / C64 / C128 KERNAL operations are done a byte at a time. Both fast mode and burst mode have to first get the other side into burst mode, but fast mode does it for every byte, while burst mode does it once for a series of bytes. LOAD could be rewritten to use burst mode instead of fast mode, but in the C128 KERNAL it is not.

[neutrino]

Ok so it works like this?

Fast mode: Need to request the other CBM-488 node to go fast mode for ONE byte each time.

Burst mode: Need to request the other CBM-488 node to go burst mode for any number of bytes.

Almost like commodore thought.. hey this is too fast, lets fix it! ?

 

[BruceMcF]

On 11/3/2022 at 1:08 PM, neutrino said:

Ok so it works like this?



Fast mode: Need to request the other CBM-488 node to go fast mode for ONE byte each time.



Burst mode: Need to request the other CBM-488 node to go burst mode for any number of bytes.



Almost like commodore thought.. hey this is too fast, lets fix it! ?

Like they thought, "hey, this is a noticeable speed up with only a need to change the code in two very low lying KERNAL routines, and then we can fit other things into the KERNAL ROM." Tramgineering.

But the X16 has the MACPTR to justify implementing the bulk burst mode transfer, and then that transfer can be used by LOAD if the device supports burst mode.

[neutrino]

On 11/3/2022 at 7:45 PM, BruceMcF said:

MACPTR

What's MACPTR ?

 

[TomXP411]

On 11/3/2022 at 5:45 PM, neutrino said:

What's MACPTR ?



 

[BruceMcF]

Look in the X16 programmers reference guide Kernel section.

[neutrino]

On 11/4/2022 at 2:13 AM, TomXP411 said:

I did try that, but it gave mostly posts that were unclear on context and purpose. Mostly along the lines that it read N number of bytes into a buffer, but not so much more.

On 11/4/2022 at 2:31 AM, BruceMcF said:

Look in the X16 programmers reference guide Kernel section.

Thanks, it specified what it is and what for.