Page 4 of 6
Re: Figuring out the causes of slow disk access on the TF536
Posted: 20 Jul 2020 15:04
by alenppc
I don't notice any difference but I never had any speed issues here...
Re: Figuring out the causes of slow disk access on the TF536
Posted: 20 Jul 2020 20:08
by arkadiusz.makarenko
With new firmware plus combination of all tricks I could find. I managed to get to 12 seconds boot time. Classic WB 68k on 8 GB CF card.
It is without MMULibs, but with cpu fastrom very high in startup-sequence.
I have noticed that initial startup is much faster with CF card than SD2IDE (more than 12 seconds just to start loading anything).
Re: Figuring out the causes of slow disk access on the TF536
Posted: 20 Jul 2020 20:49
by terriblefire
arkadiusz.makarenko wrote: 20 Jul 2020 20:08
With new firmware plus combination of all tricks I could find. I managed to get to 12 seconds boot time. Classic WB 68k on 8 GB CF card.
It is without MMULibs, but with cpu fastrom very high in startup-sequence.
I have noticed that initial startup is much faster with CF card than SD2IDE (more than 12 seconds just to start loading anything).
Have absolutely no clue if this will work (may wipe your drive) but this firmware accesses IDE at 25Mhz instead of 7mhz.
tf536r2_2020_07_20b_alpha.zip
Re: Figuring out the causes of slow disk access on the TF536
Posted: 21 Jul 2020 19:46
by arkadiusz.makarenko
It didn't wipe my cards, but I didn't do any writes. It didn't have impact on speed of loading workbench or fake news sysinfo numbers. I didn't perform any more tests.
If there is even slight more chance of corrupted card on this firmware personally I wouldn't risk it.
PS
With 4096 block size on my SD cards system starts to load much quicker as well. It was about 10-14s after power on, now it is about 7s (almost like cf card) after power on, which makes 13s to load workbench vs CF 10s on CF card.
Conclusion: If your system loads very long on TF536, test firmware says everything is OK then this means you are doing something wrong!
PS2.
Do you think it is worth to write like tutorial how to set CF up?
Re: Figuring out the causes of slow disk access on the TF536
Posted: 21 Jul 2020 19:51
by terriblefire
arkadiusz.makarenko wrote: 21 Jul 2020 19:46
It didn't wipe my cards, but I didn't do any writes. It didn't have impact on speed of loading workbench or fake news sysinfo numbers. I didn't perform any more tests.
If there is even slight more chance of corrupted card on this firmware personally I wouldn't risk it.
PS
With 4096 block size on my SD cards system starts to load much quicker as well. It was about 10-14s after power on, now it is about 7s (almost like cf card) after power on, which makes 13s to load workbench vs CF 10s on CF card.
Conclusion: If your system loads very long on TF536, test firmware says everything is OK then this means you are doing something wrong!
PS2.
Do you think it is worth to write like tutorial how to set CF up?
If that firmware works at all write will be fine.
Re: Figuring out the causes of slow disk access on the TF536
Posted: 21 Jul 2020 22:25
by Smyckers
I haven't had any problems with the boot time on my TF536 rev 2b but I've been following this thread out of interest.
I've run a few tests on WB2.1, 3.0, 3.1, and 3.9 base installs with the TF536 (no IDE buffers installed) using the TF330 as reference.
- MC68030RC50 in both cards
- In all cases the 2.1/3.0/3.1 Workbench base installs are approximately the same so I've just included 3.1 and 3.9.
- I tried the original 2b firmware and the test firmware from TF (534 timings), it didn't appear to make any difference to boot time or IDE speed.
- SD2IDE adapter with Verbatim Premium 16GB SD card (the 330 is on a 15cm cable, the 536 is on a 5cm cable)
- PFS3AIO PDS/03 formatted single 100MB partition image with 100 buffers and a 512B block size in all tests.
| System |
OS |
Start time* |
Sys info 4.3 disk speed |
| TF536 |
3.1 |
19 seconds |
702KB/s |
| TF330 |
3.1 |
10 seconds |
2.1MB/s |
| TF536 |
3.1 with MMU Libs and MUFastRom |
19 seconds |
1.97MB/s |
| TF330 |
3.1 with MMU Libs and MUFastRom |
10 seconds |
3.9MB/s |
| TF536 |
3.9 |
44 seconds |
1.8MB/s |
| TF330 |
3.9 |
30 seconds |
3.9MB/s |
| TF536 |
3.9 with MMU Libs and MUFastRom |
43 seconds |
2.8MB/s |
| TF330 |
3.9 with MMU Libs and MUFastRom |
30 seconds |
3.4MB/s |
* Start time is measured from switch on to the Workbench drive icon appearing on the screen.
Re: Figuring out the causes of slow disk access on the TF536
Posted: 21 Jul 2020 22:31
by terriblefire
Thanks for that. I'd say that this entirely acceptable for the way we've avoided people needing to solder down an INT2 wire.
Remember the CD32 has an INT2 signal out on the edge connector. The TF536 emulates that .. so its slower. Its also 7Mhz vs 14Mhz.
Re: Figuring out the causes of slow disk access on the TF536
Posted: 22 Jul 2020 19:52
by Maximilian
Would it be possible and acceptable to make a firmware with and without INT2 emulation?
Re: Figuring out the causes of slow disk access on the TF536
Posted: 22 Jul 2020 20:23
by 8 Bit Dreams
Unlike TF534 there is no INT2 Pin on TF536
Re: Figuring out the causes of slow disk access on the TF536
Posted: 23 Jul 2020 09:39
by terriblefire
Maximilian wrote: 22 Jul 2020 19:52
Would it be possible and acceptable to make a firmware with and without INT2 emulation?
It’s literally the entire design goal of this card. I really invite people to read up before buying.
Can people who want fast IDE not buy a TF536