Open-source scsi device

[_mandark_]

Running flawlessly.

[Bamiga2002]

Running good! Getting 3.4Mb/s with my IDEfix Express

[tom256]

Speed test on bare A1200 ROM 3.0 + Kingston 4GB CF card, (no turbo or memory cards):
scsi.device from Kick 3.0 - 1.1MB/s
scsi.device from Don Adan 1.3MB/s

speedup is about 20%.

Working for months without any problems

[NovaCoder]

Originally Posted by Bamiga2002

Running good! Getting 3.4Mb/s with my IDEfix Express

Cool, is that in SysInfo?

Are you running any other patches like SpeedyIDE and are you using the IDEfix software?

I'm getting about 3.8 with my Blizzard 1260 and managed 4.2 with my Apollo 1260 (IdeFix Express and SS HD). The best I've ever seen is 4.6 with my ACA 1230 @ 56 Mhz (great IDE speed thanks to ACATune), I don't think it's possible to get the full 5.1 out of if

I'll try this open source scsi.device again and report back here.

[Don_Adan]

Here is test version with removed multiple mode limit (16). Seems average data transfer is over 20% fastest for some HD's. Thanks to Tumo007 for inspiration.

[Tumo007]

Thanks Don_Adan for the credits

It's funny because I'm also toying with your source to get more speed. Tried little tweaks they seem to work, got faster speed with lower cpu usage but only under WinUAE since I don't have my A4000 nearby. As soon as I tested it on real hardware, I'll release my tweaks. Still got some debugging to do...

[jbenam]

Originally Posted by Don_Adan

Here is test version with removed multiple mode limit (16). Seems average data transfer is over 20% fastest for some HD's. Thanks to Tumo007 for inspiration.

On my Amiga the test version is slower than the latest (which I've downloaded from the WT's page).

Config is unexpanded A1200 with 2,5" 600MB IDE drive, I get ~1.48MB/s with the latest, and ~1.31MB/s with the test version.

[tom256]

Originally Posted by jbenam

On my Amiga the test version is slower than the latest (which I've downloaded from the WT's page).

Config is unexpanded A1200 with 2,5" 600MB IDE drive, I get ~1.48MB/s with the latest, and ~1.31MB/s with the test version.

First of all last optimization will benefit only in real life performance and will not provide Sysinfo speed changes.
Try copy/move big file and check time. Let us know the results also

Second reason could be that You hard drive is very old and does not support multiple mode above 16. Future versions will read this parameter during startup and set it according to disk parameter.

In my case I notice speed up when AmiQuake is loading levels, and with copying or deleting big files, but in Sysinfo I didnt have any performance increase. Tested with filesystem PFS3AIO.

[bladecgn]

There is a SpeedyIDE BlizKick module available. Is there any benefit using it with this scsi.device? Or is it not recommended to use the two together?

[mfilos]

I never saw any speed increase with using SpeedyIDE module with all the newer scsi.devices.

[Don_Adan]

Originally Posted by bladecgn

There is a SpeedyIDE BlizKick module available. Is there any benefit using it with this scsi.device? Or is it not recommended to use the two together?

SpeedyIDE is useless for this version of scsi device, this version has build-in 32 data transfer.

[Bamiga2002]

Originally Posted by NovaCoder

Cool, is that in SysInfo?

Are you running any other patches like SpeedyIDE and are you using the IDEfix software?

Measured with SysInfo. No SpeedyIDE or other patches. Using Loadmodule to get this scsi.device resident. Somehow there are still references to IDE-Max with Scout-Residents but this scsi.device should be in use.

[amiwolf]

I'd like to thank Don Adan for this scsi.device, it's really great! Was the only one I could get working on a real, 68000 powered A600. Anyone know if 46.0 is the final version?

[HanSolo]

Which scsi.device is better - cosmo's scsi.device 43.47 or don_adan open source scsi.device ???

[tom256]

cosmos scsi.device is faster (at least in benchmarks), but in my opinion scsi.device from Don_Adan is more reliable.

[voxel]

+999999 for don's version ! ^^)

[tom256]

Originally Posted by jbenam

On my Amiga the test version is slower than the latest (which I've downloaded from the WT's page).

Config is unexpanded A1200 with 2,5" 600MB IDE drive, I get ~1.48MB/s with the latest, and ~1.31MB/s with the test version.

Post is very old but I know many people are looking there, so to explain performance degradation with Your hard disk.

In this version with remove multiple write/read limit,
this value was set to 32 if I remember correctly(standard is 16 - can be check with hdparm on Windows or UNIX), and if disk didn't support this multiple read/write value disk will disable this feature, what means worse performance. So people with newer hard disk should try version with multiple read/write 32 to get better performance, for people with old hard disk it can degrade performance. With this value also should benefit users with CF. I'm not sure how does it affect optical drives.

[Rochabian]

I've tested the latest Don Aran's scsi.device and i've found that it works great on HDD but i've some problems with CF or SD. Are the access differents with CF or SD ? . I use them on a sx-1 with PFS3. CF or SD works with other scsi.device, 43.46 for example

[Don_Adan]

V46.1 was experimental version, with removed multiple mode limit, maybe this can be problem, i dont know. You can get V46.0 version from WT page:

http://wt.exotica.org.uk/test.html

BTW. Interesting thing is that A600 (68000) version can be fastest than A1200 (68020+) version on some A1200 configs, even if in theory must be slowest (more commands in loop).

[Rochabian]

You're right, just tested 46.0 and it's ok with SD and CF card.

Thanks Don