Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

SnakeCoils

macrumors regular
Oct 17, 2018
134
60
Italy
If there is a later version of the Mac firmware for the SCSI card that works better with OS X but lacks OS 9 support we can likely extract the OS 9 'NDRV' from the older firmware and add it to the later.

Assuming there is enough room on the EEPROM.

From what I have seen during my flashing (and re-flashing) experiments the OS9 support for UL4D/UL4S was dropped starting from 1.1.x version, the latest to support OS9 (boot and management) is 1.0.1 and this is visible also looking at size of mac binaries portion that PC flasher will inject in the ATTO card:

From PC archive "epci0101.exe" (OS9 and OSX support, Mac firmware v1.0.1 not "Rev.B" version)
DOS/Utility/EPCI_MAC.U4D ---> 146.944 bytes

From PC archive "epci0102.exe" (OSX support only, Mac firmware v1.1)
DOS/Utility/Epci_mac.u4d (yes, in lowercase) ---> 75.776 bytes

To be honest I have not yet tried to install and boot OSX via the UL4D with 1.0.1 firmware (i will try sooner) but the ATTO Configuration Utility v3.38 (OSX 10.5.8) seems to communicate correctly with the board, the only issues I had was due to SCSI terminators: I have quite a nice collection of terminators of all kind (active and passive) but the only that worked as expected were the ones the seller bundled with the ATTO board and I am very grateful to him for that because they helped me a lot with troubleshooting :)
 

SnakeCoils

macrumors regular
Oct 17, 2018
134
60
Italy
Update: OSX Install results.
Setup: MDD 2003 FW400, dual G4@1.42 GHz, 2x1 Gb RAM CL2, Geforce 4600Ti, Atto UL4D (firmware 1.0.1f0), HDD Seagate ST318404LW Rev.06 (one OSX partition with OS9 drivers installed by OSX Drive Setup)

OSX 10.2.0 ---> Install CD boot fine, SCSI drives recognized, straight install process goes flawlessly to the end.
OSX 10.3.0 ---> Install CD boot fine, SCSI drives NOT recognized, install possible on other HDD drives
OSX 10.4.6 ---> Install CD boot hangs (death curtain on gray Apple logo)
OSX 10.5.6 ---> Install CD boot hangs (death curtain on gray Apple logo)

The 10.4 and 10.5 install DVDs dislikes the presence of the UL4D, they hang even without drives on the SCSI chain but once the drivers are loaded everything is fine so the install process should be the following:

- Remove the UL4D from the Mac
- Do a full 10.4/10.5 install on a spare drive and apply all system updates including ATTO drivers and utilities
- Turn off the Mac and install the UL4D connecting the desired SCSI drive where OSX is wanted
- Boot the Mac from the freshly installed spare drive
- Clone the active partition to the SCSI drive using Carbon Copy Cloner
- Reboot on the SCSI cloned OSX
- If you have on the same drive a OS9 partition too you can now freely go back and forth between the two Systems without troubles, the volumes will both appear as bootable either on OS9 or OSX in their respective boot Preferences panel.
 

ioncannon

macrumors newbie
Apr 6, 2023
2
0
Was looking to throw a Cheetah 15K drive into my QuickSilver, and stumbled on your post SnakeCoils. Does the UL4D work at full speed under OS9 and 10.4 (once installed through the work around)?

Did you try RAID0ing em?
 

SnakeCoils

macrumors regular
Oct 17, 2018
134
60
Italy
From what I've seen it looks like a simple file transfer between HDDs in OS9 is accomplished faster than UL3D but remains about 2-3 times slower if compared to the same task under OSX, on a modern OS the ATTO drivers are far more efficient. Yes, I have raided my two Cheetah, the different firmware does not hurt when in OS9 and the results of ATTO benchmark were very good, their ExpressRAID tool offer specific configurations based on the task the HDD are called for but I assume other freely available RAID software (Macintosh Garden has some) should perform at least equal. About the OSX 10.4 (and 10.5) I can confirm the feel of speed is very good but I've not ran any benchmark there: on OS9 I rememer something like 130 Mb/s in reading and 50 Mb/s in writing with a single drive but the RAID configuration was not much faster, to be honest I have tried both on a single channel, maybe using separate channels would perform better. After the Easter holidays I will reconnect all the drives and I will play a bit with different setups.
 

ioncannon

macrumors newbie
Apr 6, 2023
2
0
Yeah, I was toying with finding the best way to get max read/write out of the computer.

G4 PCI bus is 33.33MHz * 64bits -> 2133.12 Mb/s or 266.64 MB/s if only doing one operation (read or write). This is our hard cap. Mind you the PCI bus will be shared with other cards so that number can go down. Ultra ATA/66 is maxed out at 66.6 MB/s.

The UL4D is U320 based and it maxes out at 320 MB/s so the single channel shouldn't be the issue (unless you are using an older Cheetah).

Finally a Seagate Cheetah 15.5k has a sustained write time of 73-125 MB/s. Datasheet

I dunno if you can simply double the sustained time for a theoretical RAID0 number. Was your value Mb/s or MB/s? Bits or bytes?

Oh and I assume both operating systems were bootable while stored on the raid volume?
 
Last edited:

Ivan Shpak

macrumors member
Feb 19, 2021
32
7
Success! :) I was able not only to bring my UL4D back to life but also to flash the OS9 firmware 1.00U that previously bricked the board. I can confirm that ExpressPro Tools 2.8.2 now see and scan both SCSI channels but I have not yet tested the OS9 bootability.
It was a very tricky and annoying process, a PC with PCI slots and XP installed was also needed along the ATTO Configuration Tools, Drivers and flash files for Windows, plus others that were on the PC session of the ATTO ExpressRAID install CD... tomorrow I will detail all the steps, now I really need to go to sleep ;-)
Hey!
Unfortunately, UL4S also died after the firmware, half a year ago, I went to this topic and saw that you have progress here)) I install on two PCs of different motherboards, and there is no boot, tell me in more detail. You soldered the ROM?, and which there was a config on your PC computer with Windows XP?
 

Ivan Shpak

macrumors member
Feb 19, 2021
32
7
Yeah, I was toying with finding the best way to get max read/write out of the computer.

G4 PCI bus is 33.33MHz * 64bits -> 2133.12 Mb/s or 266.64 MB/s if only doing one operation (read or write). This is our hard cap. Mind you the PCI bus will be shared with other cards so that number can go down. Ultra ATA/66 is maxed out at 66.6 MB/s.

The UL4D is U320 based and it maxes out at 320 MB/s so the single channel shouldn't be the issue (unless you are using an older Cheetah).

Finally a Seagate Cheetah 15.5k has a sustained write time of 73-125 MB/s. Datasheet

I dunno if you can simply double the sustained time for a theoretical RAID0 number. Was your value Mb/s or MB/s? Bits or bytes?

Oh and I assume both operating systems were bootable while stored on the raid volume?
As for the data transfer speed UL4D (last firmware), I have a 1.33 dual, MDD FW800 that works with 4 Cheetah HDD (ST373453LW), 2 HDD (channel 1) + 2 HDD (channel 2) which in theory gives 640 MB of speed, bootable Apple RAID0 (OSX 10.5.8). Speeds around 200-250mb
 

Attachments

  • photo_2023-04-07_23-21-05.jpg
    photo_2023-04-07_23-21-05.jpg
    197.8 KB · Views: 58

SnakeCoils

macrumors regular
Oct 17, 2018
134
60
Italy
Hey!
Unfortunately, UL4S also died after the firmware, half a year ago, I went to this topic and saw that you have progress here)) I install on two PCs of different motherboards, and there is no boot, tell me in more detail. You soldered the ROM?, and which there was a config on your PC computer with Windows XP?

Well, I was quite desperate about the non-booting Mac or PC where my UL4D was installed so I tried a dangerous move but in the end it was enough to allow the machine to boot: using a paperclip or a screwdriver I shorted two pins of the UL4D EEPROM and then I've power up the computer. Luckily I did not fried anything but I cannot guarantee the same will work for you, I mean, if the result would be a dead card (or a dead PC/Mac) don't blame me :)

If you need the Config Utilities, the firmware and ATTO drivers for Windows Xp I can PM them to you.
 

Ivan Shpak

macrumors member
Feb 19, 2021
32
7
Well, I was quite desperate about the non-booting Mac or PC where my UL4D was installed so I tried a dangerous move but in the end it was enough to allow the machine to boot: using a paperclip or a screwdriver I shorted two pins of the UL4D EEPROM and then I've power up the computer. Luckily I did not fried anything but I cannot guarantee the same will work for you, I mean, if the result would be a dead card (or a dead PC/Mac) don't blame me :)

If you need the Config Utilities, the firmware and ATTO drivers for Windows Xp I can PM them to you.
I like the solution, but what contacts?))))

If they are not in this topic, then please send, I will be very grateful
 

SnakeCoils

macrumors regular
Oct 17, 2018
134
60
Italy
Here is the Flash EEPROM soldered on my UL4D, I simply shorted some of the bottom pins, I can't say for sure what couple it was, but it was in the range DQ1-DQ6, at a certain point the PC booted and after I was able to install the ATTO tools and the drivers to flash it back. I repeat myself: do it at your own risk, I don't carry any reponsability if this trick will blow up your PC!

Screenshot 2023-04-10 at 18-24-14 DATASHEET SEARCH SITE WWW.ALLDATASHEET.COM - 8926.pdf.png
 
Last edited:

Ivan Shpak

macrumors member
Feb 19, 2021
32
7
Here is the Flash EEPROM soldered on my UL4D, I simply shorted some of the bottom pins, I can't say for sure what couple it was, but it was in the range DQ1-DQ6, at a certain point the PC booted and I was able after installing the ATTO tools and the drivers to flash it back. I repeat myself: do it at your own risk, I don't carry any reponsability if this trick will blow up your PC!

View attachment 2186999
Don't worry, I was actually on the verge of doing something like this or doing a PCI hot swap.

Thanks!
 
  • Like
Reactions: SnakeCoils

DearthnVader

Suspended
Dec 17, 2015
2,207
6,392
Red Springs, NC
Don't worry, I was actually on the verge of doing something like this or doing a PCI hot swap.

Thanks!
I did an AGP hot swap in my G5 to bring back a PC flashed Geforce 2 MX400 that I put a bad Mac firmware on.

It was pulling the invisible act in the Mac and my PC AGP motherboard took a **** on me many years back.

I used a OEM Geforce 4MX in the AGP slot with no monitor connected, and a PCI card for video. Then I put the Mac to sleep and did the swap while it was sleeping under OS X. Woke the G5 and flashed it with OS X's nvflash.

Worked like a charm.

YMMV!
 

Ivan Shpak

macrumors member
Feb 19, 2021
32
7
Friends, I report to you and share the result of the resurrection of UL4S

I needed an old motherboard Gigabyte (GA-8PE667 Ultra) Pentium 4 (2.66ghz) and 1gb DDR, as well as a monstrous ATI 9800 Pro:eek:

Windows 7 operating system (I had HDD with it and I decided to try it so as not to waste time installing Windows XP)

In advance, I installed ConfigTool and drivers kindly given to me by SnakeCoils

I closed two contacts DQ4 and DQ5 with a paper clip to boot.

Further, the computer hung all the time searching for a driver in the system.

I removed the ATTO Card from the PCI slot, turned on the hibernation mode, the computer seemed to turn off, but the session remained. Next, I put the card in the slot, closed the contacts and entered my session, the drivers were successfully initialized, the ConfigTool saw the card, but the firmware window was not active, in the end I rebooted with a paper clip, went into the ConfigTool and tried to flash it, the mosfets on the card squealed 🥶, and there was a Death Blue Screen, after that the computer itself started up without a paper clip, in the firmware window there was some kind of year 77656564) Then I flashed the card again, it was successfully flashed and WALA!

Friends, thank you)
 
  • Like
Reactions: SnakeCoils

Ivan Shpak

macrumors member
Feb 19, 2021
32
7
If that fails there are things you can do with Open Firmware to present the PCI card as the flashing utility expects so that it can R/W the EEPROM. We don't have this ability with BIOS/PC's.
Can you please tell me where I can read more about this?
 

DearthnVader

Suspended
Dec 17, 2015
2,207
6,392
Red Springs, NC
Can you please tell me where I can read more about this?
A basic overview:



It's just basically getting the current instance of the PCI card and executing the FCode ROM.

There is also a way to write a script to build properties for the device the ROM would build without the use of them ROM itself.

If you have specific questions or there is something you don't understand ask in the thread I linked and we'll help you sort it out for your needs.
 
  • Like
Reactions: Ivan Shpak
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.