Figuring out the causes of slow disk access on the TF536

[terriblefire]

Has anyone tried the 25Mhz firmware? I really need feedback if i am going to improve this.

[PaulJ_2.0]

Has anyone tried the 25Mhz firmware? I really need feedback if i am going to improve this.

My SD adapter failed to init and my CF card worked the same transfer speeds as 7mhz.

[alenppc]

Has anyone tried the 25Mhz firmware? I really need feedback if i am going to improve this.

I tried it, no difference for me. It works the same as the other one, but I never really had any speed issues.

[arkadiusz.makarenko]

I have done some more comparison tests.

Amiga 500 Rev6 1MB Chip
I used this image https://drive.google.com/file/d/1Lzhb5M ... sp=sharing
SD2IDE on 2GB SD card - class 4

Boot time exactly the same. approximately 15 seconds

SysInfo
Standard firmware - 2,259 - 2,437 - 2,042
Experimental firmware - 2,097 - 2,215 - 2,069

I did run series of test using this : http://aminet.net/package/disk/moni/DiskSpeed42
Directory access - the same
experimental:
File Create: 2 files/sec | CPU Available: 97%
File Open: < 1 files/sec | CPU Available: 98%
Directory Scan: 13 files/sec | CPU Available: 96%
File Delete: 12 files/sec | CPU Available: 95%

standard:
File Create: 2 files/sec | CPU Available: 98%
File Open: < 1 files/sec | CPU Available: 98%
Directory Scan: 11 files/sec | CPU Available: 96%
File Delete: 13 files/sec | CPU Available: 95%

Seek time - the same
experimental:
Seek/Read: 1007 seeks/sec | CPU Available: 21%

standard:
Seek/Read: 1020 seeks/sec | CPU Available: 21%

Read/Write/Create on FAST - Long aligned buffers - very similar results
standard:
Testing with a 512 byte, MEMF_FAST, LONG-aligned buffer.
Create file: 24413 bytes/sec | CPU Available: 96%
Write to file: 20721 bytes/sec | CPU Available: 96%
Read from file: 51 bytes/sec | CPU Available: 98%

Testing with a 4096 byte, MEMF_FAST, LONG-aligned buffer.
Create file: 11126 bytes/sec | CPU Available: 98%
Write to file: 11084 bytes/sec | CPU Available: 98%
Read from file: 50448 bytes/sec | CPU Available: 97%

Testing with a 32768 byte, MEMF_FAST, LONG-aligned buffer.
Create file: 11152 bytes/sec | CPU Available: 99%
Write to file: 11103 bytes/sec | CPU Available: 99%
Read from file: 435874 bytes/sec | CPU Available: 88%

Testing with a 262144 byte, MEMF_FAST, LONG-aligned buffer.
Create file: 11129 bytes/sec | CPU Available: 99%
Write to file: 11134 bytes/sec | CPU Available: 98%
Read from file: 2412768 bytes/sec | CPU Available: 39%

experimental:
Testing with a 512 byte, MEMF_FAST, LONG-aligned buffer.
Create file: 24924 bytes/sec | CPU Available: 96%
Write to file: 21259 bytes/sec | CPU Available: 96%
Read from file: 52 bytes/sec | CPU Available: 98%

Testing with a 4096 byte, MEMF_FAST, LONG-aligned buffer.
Create file: 11150 bytes/sec | CPU Available: 98%
Write to file: 11084 bytes/sec | CPU Available: 98%
Read from file: 49430 bytes/sec | CPU Available: 96%

Testing with a 32768 byte, MEMF_FAST, LONG-aligned buffer.
Create file: 11154 bytes/sec | CPU Available: 98%
Write to file: 11201 bytes/sec | CPU Available: 98%
Read from file: 472478 bytes/sec | CPU Available: 86%

Testing with a 262144 byte, MEMF_FAST, LONG-aligned buffer.
Create file: 10476 bytes/sec | CPU Available: 98%
Write to file: 11084 bytes/sec | CPU Available: 98%
Read from file: 2095862 bytes/sec | CPU Available: 47%


My testing show that there is not real differences, on this particular tests which I have done it would even suggest that experimental firmware is slower, but I bet if I run it several times averages would be the same.

[go0se]

Tried the 25MHz firmware along with the 534 timings version on an unbuffered and a buffered card. The IDE speed and boot time were the same as the original Rev2b firmware.

I'm firmly in the "don't have any problem with the 536 IDE speed" camp.

[terriblefire]

I'm firmly in the "don't have any problem with the 536 IDE speed" camp.

I've never seen IDE speed problems either. So unless i can reproduce somehow i've no way to fix it. And its not like i only have a handful of Amiga's I've tested this on 2 different A500s, a A2000 (although i did skip the 2b on this one) and a CDTV. All worked fine. :/

[GadgetUK164]

Tried the 25MHz firmware along with the 534 timings version on an unbuffered and a buffered card. The IDE speed and boot time were the same as the original Rev2b firmware.

I'm firmly in the "don't have any problem with the 536 IDE speed" camp.

I don't think there is a speed problem. I think maybe its people testing on the A500 and assuming they should be getting 1MB/s or more (when not using an MMU).

[arkadiusz.makarenko]

Tried the 25MHz firmware along with the 534 timings version on an unbuffered and a buffered card. The IDE speed and boot time were the same as the original Rev2b firmware.

I'm firmly in the "don't have any problem with the 536 IDE speed" camp.

I don't think there is a speed problem. I think maybe its people testing on the A500 and assuming they should be getting 1MB/s or more (when not using an MMU).

I think it red herring.

If there are complaints those might only be from people who had extremely overblown expectations.
Realistically 030 cpu, 64mb fast ram + 2,5mb/s ide is what you ever need on Amiga 500.

[8 Bit Dreams]

I don't think there is a speed problem. I think maybe its people testing on the A500 and assuming they should be getting 1MB/s or more (when not using an MMU).

That's right, it was about long booting time, amiga reads whole bitmap of the drive, so if Your drive is 8 or 16GB (yes, You can use large drives with KS 3.1.4) it will take ages to boot unless You make bigger block size as default..

[snifferman]

Till now I have soldered 3 cards. All runs at 108Mhz main clock, and no ide problems (no ide buffers)
Im using firmware tf536r2_2020_04_22
Amiga 500+ soj-mod 2MB chip Kick 3.1
I tested few SD cards.
WB3.1, Load Module/scsi.device 46.0 (Don Adan from http://wt.exotica.org.uk/test.html),MMUlib

SD 64GB(samsung evo plus)
0 System partition 0,5GB SFS Block size 4096
1 soft 1,5GB SFS Block size 4096
2 soft 2GB SFS Block size 4096
3 Soft 53,5GB PFS3AIO Block size 4096

BOOT time-Power on (one reset for load moudle included) 34s
Keyboard Reset time 19s
Speed sysinfo 4.0 around 3.3MB/s for all partitions
-----------------------------------------------------------------------------
Without LoadModule scsi.device
Boot time-power on 22s
keyboard reset 22s
sysinfo 4.0 disk speed 2.9MB

On small cards or big ones( but with small partitions only) i had boot time under 15s (clean WB3.1 + MMUlib)

SD 128GB was not recognized In system (not visible)