Nice tools!

On my X5000/20, I am using this CPU fan from Noctua:
https://noctua.at/en/products/fan/nf-a6x25-flx

I've chosen to connect it to the PSU through a speed reduction adapter (included in the box by Noctua).
Hence, it's running at a fixed speed of 1600 RPM (down from its default 3000 RPM).
With this setup, it's close to noiseless.

The computer case is an "Antec 25 Solo II", and there is one 120mm fan in the front (Noctua, sucks air in), and one at the back (blows air out).
Both are running at a heavily reduced speed, and are very quiet (barely audible).

The GFX card is an Asus Radeon R7 240 (2GB version), which isn't generating much heat.

Below are numbers from my system, including from the "x5k_temp_no_serdes_working" tool.
The computer were on for about an hour upfront.


Temperatures:
Motherboard: +37 °C
CPU........: +56 °C
PCIe-Switch: +59 °C

Temperaturen
************
PCB 37ÐÑC CPU 56ÐÑC PCIe 59ÐÑC

Spannungen
**********
CPLD 3350mV
XenaXorroA 3330mV
XenaXorroB 1000mV
PCIe 1000mV
Xena 3330mV
3,3V Schiene 2490mV
2,5V Schiene 1200mV
Ethernet 1030mV
Platform 1090mV
CoreA 1090mV
CoreB 1490mV
DDR3 1810mV
LÐÑfter
******
PWM-Signal 255
Drehzahl/min 0

@apsturk

Thanks for the thorough and informative article!

I am very interested in learning how to program the XMOS on my AmigaOne, so this blog is highly appreciated.

Eh, something strange happended to my post, so here it is once more:

Hello,

I have experienced similar issues regarding use of UDMA on optical drives.
However, UDMA works as expected with a harddrive, significantly offloading the CPU.
I haven't tried the PIO mode.

My opinion is that there has to be a bug somewhere in the OS.
Let's hope Hyperion will fix the issue in a future update of the OS.

To sum it up, here are my experiences with UDMA on my SATA DVD-writer:
- Write speed is limited to 5 MB/sec (using AmiDVD)
- Read speed is limited to 2.5 MB/sec (drag'n'drop copy in workbench), if the files copied is large (multiple MBs each).
- Reading & writing slows down the whole system (if CPUInfo shows only 10% CPU usage, other tasks needs significantly more time to finish compared to without the read/write process. To me, it seems like the multitasking abilities are limited while reading from/writing to a DVD media.
- If I try to access a disk partition while writing, opening the partion involves too high waiting time (a kind of "lockup").

For comparision, here are my experiences by using an external USB DVD-writer instead of the internal SATA:
- Write speed is the same as with SATA
- Read speed is max 6.2 MB/sec

In both cases, the max. speed obtained is far below the speed limits of the DVD drives themselves.


You will find more details of the issues here:
http://forum.hyperion-entertainment.biz/viewtopic.php?f=17&t=328


My system (hardware):
o ACube SAM440EP 667MHz (mini-ITX), with a SATA controller onboard (Silicon Image SII3114). It has Amiga OS4.1 update 4 installed.
o As storage, I use a 30GB OCZ SSD ("Solid series") and a 120GB OCZ SSD ("Vertex 2").
o Internal SATA DVD writer is a Plextor PX-820SA.
o External USB DVD writer is a Samsung WriteMaster SE-S224.

SATA-settings:
DVD-rom: UDMA 5, 100 MB/s
HD (Vertex 2 SSD): UDMA 6, 133 MB/s
HD (Solid series SSD): UDMA 5, 100 MB/s

Hello,

I have experienced similar issues regarding use of UDMA on optical drives.
However, UDMA works as expected with a harddrive, significantly offloading the CPU.
I haven't tried the PIO mode.

My opinion is that there has to be a bug somewhere in the OS.
Let's hope Hyperion will fix the issue in a future update of the OS.

To sum it up, here are my experiences with UDMA on my SATA DVD-writer:
- Write speed is limited to 5 MB/sec (using AmiDVD)
- Read speed is limited to 2.5 MB/sec (drag'n'drop copy in workbench), if the files copied is large (multiple MBs each).
- Reading & writing slows down the whole system (if CPUInfo shows only 10% CPU usage, other tasks needs significantly more time to finish compared to without the read/write process. To me, it seems like the multitasking abilities are limited while reading from/writing to a DVD media.
- If I try to access a disk partition while writing, opening the partion involves too high waiting time (a kind of "lockup").

For comparision, here are my experiences by using an external USB DVD-writer instead of the internal SATA:
- Write speed is the same as with SATA
- Read speed is max 6.2 MB/sec

In both cases, the max. speed obtained is far below the speed limits of the DVD drives themselves.


You will find more details of the issues here:
http://forum.hyperion-entertainment.biz/viewtopic.php?f=17&t=328


My system (hardware):
o ACube SAM440EP 667MHz (mini-ITX), with a SATA controller onboard (Silicon Image SII3114). It has Amiga OS4.1 update 4 installed.
o As storage, I use a 30GB OCZ SSD ("Solid series") and a 120GB OCZ SSD ("Vertex 2").
o Internal SATA DVD writer is a Plextor PX-820SA.
o External USB DVD writer is a Samsung WriteMaster SE-S224.

SATA-settings:
DVD-rom: UDMA 5, 100 MB/s
HD (Vertex 2 SSD): UDMA 6, 133 MB/s
HD (Solid series SSD): UDMA 5, 100 MB/s

@Snuffy

Thanks for your feedback.

Ok, I see.
No wonder my DVD-rom's reading/writing speed isn't optimal.

However, I still don't understand why my realworld writing speed is 2x faster than the reading speed, and that I am far away from reaching 16MB/sec.
(maybe the OS chooses a slower PIO-mode than reported by IDETOOL?)

I am really looking forward to getting this fixed.

@colinw

I have done some more tests, and the error occur on the RAM-disk only.

On these filesystems, renaming works ok:
- FastFileSystem 53.2 (DOS1)
- SmartFilesystem 1.286 (SFS0)
- SmartFilesystem 1.286 (SFS2)
- JXFileSystem 4.34 (JXF4)


My tests below were done using the standard CLI-command "RENAME".
When running the same tests on a classic Amiga, the result is ok.


Test 1)
old filename on disk: readme.txt
old filename in parameter list: README.TXT
new filename in parameter list: README.TXT
=>Result (RAM): Wrong. Filename was kept unchanged (should have been README.TXT)
=>Result (other filesystems): Ok. Filename was changed to README.TXT

This test was done typing in the following from my shell:
rename from=README.TXT to=README.TXT



Test 2)
old filename on disk: readme.txt
old filename in parameter list: README.TXt
new filename in parameter list: README.TXT
=>Result: Ok (all filesystems). Filename was changed to README.TXT

This test was done typing in the following from my shell:
rename from=README.TXt to=README.TXT

Overall, I think the OS4.1 update 1 is a good package.
Although, there still are some critical (and less critical) issues that need to be fixed (see below).
I really wish Hyperion have some time to have a look at them.

My computer is a Sam440EP (mini-ITX, 667MHz, 512MB RAM) with a clean Amiga OS 4.1 update 1 install.
To measure the CPU usage, I have used CPUInfo.


Summary:

1) Multitasking
a) Limited resources left for GUI-interactions
b) Delayed access to partitions in certain circumstances

2) DVD-rom read/write speed
a) Slow write speed
b) Slow read speed
c) Adjusting read speed of DVD-roms using IdeTool

3) (Partly) non-working USB devices
a) Razer Copperhead gaming mouse (laser)
b) Corruption of flash memory cards (formatting issue?)

4) Kickstart-function "Rename" in DOS.library




Details:

1) Multitasking

a) Limited resources left for GUI-interactions

When the CPU usage is 100% when running several programs in parallell, I get problems moving the mouse pointer; The mouse pointer lags behind.

All tasks have priority 0.

My opinion is that tasks related to mouse pointer movements and keyboard keypresses should always have the highest priority, and/or the user should have the opportunity to choose the priority himself.
I think this is important in order to avoid problems when/if programs "hang".

Example:
Several programs together uses 100% of the CPU (and maybe the GPU), and it's impossible to close the application (because the mouse pointer can't be moved).
If the mouse pointer always had top priority, I could have closed the application to free resources (or reduce the application's task priority).



b) Delayed access to partitions in certain circumstances

While copying between/to/from flash memory cards (via USB) or from a DVD-rom drive, other tasks don't get the CPU resources they need.
This prevents efficient use of my computer.
(PS: As soon as the copy is finished, everything works as normal again).

In my tests, some of the files being copied were large (up to 800MB each).

Examples:
- When running other tasks from the shell at the same time as the copy, they don't get the CPU-resources they need. According to CPUInfo, the CPU usage is unusual low (<30-35%). In my testcase I used OpenSFV (which normally uses much CPU).
When the copy is finished, the tasks runs as normal again.

- When burning files to DVD, it's almost impossible to access directories and files on the same partition as the burning software is reading from (0,5-1,5 minutes waiting time). Accessing other partitions works better.
I get the delay no matter which program I use; When browsing files directly from my Workbench, from the shell (cd and list commands), from Directory Opus 4, so this is most likely an OS issue. As soon as the burning is finished, the response is quick again.




2) DVD-rom read/write speed
First some general info:

My DVD-burner is a Plextor PX-820SA.
Output from idetool (my current settings for my DVD-rom (unit 0)):

Flags : $0000011D - present, supports DMA, removable media, ATAPI, interrupts used,
Xfer mode : best pio 12 (PIO 4, 16 MB/s) / best dma 69 (UDMA 5, 100 MB/s) / current 69 (UDMA 5, 100 MB/s)
Total blocks : 0
Blocksize : 0
SCSI devtyp : 5
Packet size : 12
Current medium read speed : x 65535 ('65535' stands for max.)
Current medium write speed : x 65535 (idem)
Power mode : 2 / idle (ready for operation)
IO1 / IO2 / BMCR @ : $1100 / $110A / $1120

Here is the description of the issues I have with my DVD-rom:


a) Slow write speed

On my system, a write speed of 4xDVD (5.4MB/sec) seems to be max, though the drive itself should handle up to 20xDVD and the media used supports 16x writing speeds.
The DVD burning software I use is AmiDVD.

I have tried to change the maximum speed with IdeTool, but the result doesn't get any better.

CPUinfo shows me very little CPU load, so it's not the CPU which is the bottleneck here.



b) Slow read speed

The maximum read speed seems to be 2xDVD (2.7MB/sec), though the drive should handle up to 16xDVD. It's a lot of noise from my DVD-rom, so the rotating speed is most likely at maximum.
The strange thing here, is that the reading speed is only 1/2 of the writing speed. Something strange is going on here..

I have tried to change the maximum speed with IdeTool, but the result doesn't get any better.

CPUinfo shows me very little CPU load, so it's not the CPU which is the bottleneck here.



c) Adjusting read speed of DVD-roms using IdeTool

To avoid unnecessary delays and noise from my DVD-rom, I use idetool to reduce the reading speed.
This doesn't work as intended.
It is not possible to reduce the DVD-rom's MAX read speed for DVD-R and DVD-RW media.
However, CD-R and CD-RW media works fine.

Details:
Idetool reports it has changed the max speed to the requested value, but the DVD-rom acts like this:
* DVD-R media: Drive always spins up to MAX rotating speed, even if IdeTool has reported it has changed it to 1x.
* DVD-RW media: Same here. Although with the previous version of the OS4.1, 4xDVD RW media were always read with a low speed (4x?).
* CD-R & CD-RW media: The drive spins up/down to the requested speed when reading (ok).




3) (Partly) non-working USB devices
Below is a list of devices that doesn't work correctly with OS4.1:


a) Razer Copperhead gaming mouse (laser)

When moving the mouse slowly to the right or down, the mouse pointer doesn't move at all.
(Left and up works fine).



b) Corruption of flash memory cards

Copying to/from memory flash cards sometimes leads to corrupt directories and/or files (CRC error and/or cryptical directory names).
The biggest problem with this, is that no error messages are given, so the user doesn't know which files are ok and which aren't.

To detect this error, I have to calculate checksums manually before and recalculate/check it after the file copy (otherwise I don't discover the file errors until it's too late).
The copy in my test was done between 2 cards (SD and Compact Flash) with a 8-in-1 flash card reader.
I got the error above with a Kingston 128MB SD Card. The flashcard was formatted with FAT16 from OS4.1.
After formatting the SD-card on my digicam, everything worked fine.
My best guess is that this bug has something to do with the formatting-routine in OS4.1 + maybe a lack of checksum checks.

Another strange behaviour:
Deleting files sometimes leads to corrupt files.
First I get a "Seek error" message, then the file was unreadable.
Tries deleting the file again, and the file gets deleted as normal.




4) Kickstart-function "Rename" in DOS.library
This function has a different behaviour compared to OS3.x.
The consequences of the bug, is that some files will not be renamed.

Details:
If the parameter which specifies the old filename has exactly the same case'ing as the new one, rename isn't done, even if the file on disk has different case'ing.
This behaviour differ from that in OS3.x, where renaming always is done.

My opinion:
I think this function should have similar behaviour as in OS3.x, rather than trusting the calling program always "knows" the correct case'ing of the old filename.

It's nice being here.

For me, the Amiga is everything.
I am relatively new to this forum, but have been using Amigas since the "good old days".
My first Amiga was an A500 (bought back in 1990).
I was (and still am) very fascinated by what this computer could do. Especially regarding multitasking, perfect scrolling, user interface etc.

Later, in the mid-90's, I got an A1200 with a Blizzard 1230-IV accelerator card, which I still use today.

Last year, I got my long-awaited Sam440EP (mini-ITX, 667MHz).
You don't know how excited I was when I powered on my Sam the first time.
My system's case & PSU are fanless, and a 30GB Solid State Disk (OCZ) is installed. The only thing that makes some noise, is the DVD-rom.
Sam works great (...although it should have a XENA chip installed onto it...).

My main interest is programming, mainly assembly language (Motorola 680xx/68882).
And of course: I love playing around with the classic Amiga's custom chipset. Parallell processing is great!
I like tweaking to get the most out of my computer, and the assembly language is perfect to achieve that goal.
I still think you don't need to have the highest spec'ed hardware to do things fast and smooth; It's all about how the computer, the OS and the programs are designed that matters.

For the moment, I am learning how to program on my Sam/OS4.


Ongoing projects:
- JustName (A CLI file renaming utility. A new version will be released on Aminet soon).
- JustComm (A CLI filenote utility. A new version will be released on Aminet soon).
- An "oldschool" demo for classic Amigas.


Other classic computers I have:
- Amstrad CPC464 (My first computer. 1985).
- Amstrad CPC464+ (bought a few years ago)
- Amstrad CPC6128 (bought a few years ago)
- A1200 (my spare A1200. Bought a few years ago)
- A500+ (bought recently)



JOS

On my 667MHz Sam440EP, there is 1 command in the Startup-Sequence that delays the bootup time by a significant amount of time (about 8 seconds). It's the AddNetInterface command.
I found this by checking the execution time of every command in the Startup-Sequence.
The other commands in the Startup-Sequence doesn't consume much time.


Here is what you need to do to remove the delay:

1) Open the file S:Startup-Sequence in a text editor

2) Find the line where the AddNetInterface command is located:
C:AddNetInterface QUIET DEVS:NetInterfaces/~(#?.info)

3) Replace the line above with the one below. This will make the command run as a background process, which lets the remaining commands in the Startup-Sequence run in parallell with AddNetInterface:

Run >NIL: C:AddNetInterface QUIET DEVS:NetInterfaces/~(#?.info)

4) ...and voila!
When the Workbench screen is visible, you can surf the net as usual.



After this tuning + the latest U-Boot update from ACube, it should be possible to boot with less than 30 seconds (my Sam uses 29 seconds).



Happy New Year to you all!