Hyperion Entertainment Message Boards

A full format doesnt always mean Zeroing out(regarding Windows etc). Just means bad blocks havnt been noted

xenic wrote:

Christopher Follett wrote:Might sound stupid, but how many buffers do you have set?

As I would get random errors on the Micro A1, until I thought about buffers. Set them to 2000 on the USB and it solved my issues and more than trebled the speed of data transfer.

How did you change the buffers on a USB drive? I don't see any way to do it in the "USB Massstorage" commodity and the "Edit partitions and filesystems" gadget in Media Toolbox is disabled for "usbdisk.device".

If its a normal usb disk. i.e. NOT PARTITIONED using Toolbox as a HDD. I was going round in circles until I decided to try buffers.
If I recall, USB devices get mounted as UMSD0: so addbuffers UMSD0: 1970 works great.

However this only works when you put the usb device in. if you try doing it before you insert device, it just guru's.
If you have set it up as a HDD using toolbox, then you set the buffers in toolbox.

daveyw wrote:I tried a full format last night. After about an hour, it had managed 1% of a 500 GB partition, and I got the same error come up, although this time the "io_Offset" was a positive number 1864358912 as well as a dialogue box stating "Unable to format cylinder 72384475" . I rebooted and tried again to see if it stopped at the same place. Left it formatting over night, come morning it had done 4% but Workbench had frozen. No error messages, and the drive light was still going (although I've found in the past that it flickers even after an error message), but Workbench had locked up.

That does seem suggestive (but not conclusive) of some kind of data-transfer problem (or possibly memory corruption). At the very least it is different behaviour to what I saw with a problematic harddrive (which I blamed on bad blocks).

Maybe AmigaOS4 simply doesn't play well with this particular model of harddrive? While unlikely in general, it wouldn't be unheard of, and at this point it seems the most likely explanation to me... ALSO, since this is a new harddrive(?), it is very likely an "Advanced Format" aka "4K Sector" harddrive. I'm not sure if AmigaOS4 is compatible with those (should be but who knows?).

I've discovered something odd. While running RageMem, I found the benchmarks for my machine way off the standard:



While it may not be connected to my current problem (RageMem on the Sam shows it to be very similar to its standard), it makes me worry about my board. L2 cache is only 256 KB instead of 512? (I don't pretend to know exactly what that means, but it sounds odd).

That looks ok, my G4 has 256 L2 cache, mips and video bus speed seem right too.

Well, it looks like I have beaten this problem on my A1XE.

Firstly, I replaced the PSU with this: http://www.elive.co.nz/antec-520w-power ... ac1602.php and added a slot-mounted fan under my GFX card.

This seemed to stabilise CPU temperature to 59 degrees, although I was still getting drive errors.

Secondly, I added an 80mm Xilence fan on the rear, and installed a Deepcool Rockman fan controller. The Xilence doesn't seem to add too much, the breeze coming out of it is nothing compared to the blizzard coming out of the slot fan.

Finally I replaced the stock Thermaltake Crystal Orb CPU HSF with the recommended Zalman VF900-CU (or rather, a friend replaced it). It too is connected to the Rockman fan controller.

The temperature dropped from 59 degrees to 31, although after a while it creeps up to 35.

The day after the operation, I was able to successfully back up 100 GB of data from a partition. Initially I had another dreaded drive-error, but I had the CPU fan speed on low and after cranking it up to maximum, I had no problems.

Next, I was able to do a full format on that partition. The first attempt crashed after 2-hours (32%) but the second go went without a hitch, and after a marathon 6-hour format (!) I was able to copy the 100 GB back again with no errors.

The hard drive was still set to UDMA 0, so the next step was to raise the UDMA and see if it can handle it. I tested it by loading up a video with DVDPlayer and randomly skipping around. If it freezes up, I know there's a problem.

UDMA 2 went fine, but when I went to UDMA 4, my video-test froze up. Same with UDMA 3. So I'm back to UDMA 2, running well below the drive's maximum of UDMA 6, but at least an improvement.

I still need to have a go running memtester again. As part of my manic attempts to solve this problem (the above probably came to about $100 PSU + $16 fan + $40 Zalman + $8 fan + $30 fan controller) I also ordered a different brand of 1 GB RAM, but it hasn't arrived yet.

But otherwise, the problem is solved. Since Monday I have copied literally 100s of GB across different drives (USB and PATA), and, as long as the CPU fan is on max, I don't get any errors.

Not sure why my master drive goes OK with its maximum UDMA 5 but not the slave drive. Cable length maybe? It's 60 cm (I think the EIDE spec max is only 42). Perhaps the slave cannot be faster than the master?

Maybe the years of high temperature have permanently affected the CPU? I could try upping VCORE, but right now I'm not inclined to fiddle (I forgot to plug the HDD LED back in and have been putting even that simple job off).

I haven't delved into why the problem was also occuring on the Sam460, but given that memtester freezes when I try and run it, it might be memory.

Could the thermal paste between the CPU & the fan have failed? Did installing the new fan require new thermal paste (like it would on a PC mobo)?

Could the problem handling (perfectly normal) CPU temperatures indicate the CPU voltage is wrong?

ChrisH wrote:Could the thermal paste between the CPU & the fan have failed? Did installing the new fan require new thermal paste (like it would on a PC mobo)?

Could the problem handling (perfectly normal) CPU temperatures indicate the CPU voltage is wrong?

Yes, we used new thermal paste on the fan - the packaged Zalman paste. Probably not quite as good as Arctic Silver, but should suffice.

Well, the CPU voltage is wrong - according to Stefane G, but not according to EyeTech, i.e., it's 1.59 instead of 1.79. I might try upping it to 1.79 as some point, but everything with the exception of my 2nd hard drive is working fine.