> I find it hard to believe there is nothing that can be borrowed and ported from
> the open sorce world. But I am not a programmer, just an frustrated end user.

I too am just a layman when it comes to this, but according to my understanding, simply porting a recent stack wouldn't suffice as you'd also need a revised Amiga-specific (or MorphOS-specific) interface between the stack and the driver (i.e. "SANA-III") so that the new driver can make use of the modern features of the stack.

Quote:

Ok, I should have had more emphasis on the word "quick test" in my original post. What I meant with that was that I just made quick tests from random sites and got pretty easily faster values than those 500kB/s than other saying the max was for them. I didn't try to find the fastest location without bottlenecks elsewhere in that case.

After MorphOS 3.0 release I have no more problems with network speed. I'm getting 100% speed - the same as on my pc.

> I just made quick tests from random sites and got pretty easily faster
> values than those 500kB/s than other saying the max was for them.

I'm baffled as to why this is the case. As said, I can confirm their 500 kB/s limit with my setup here. Maybe this is one of those mysteries that will never be resolved ;-)

> After MorphOS 3.0 release I have no more problems with network speed.
> I'm getting 100% speed - the same as on my pc.

Interesting. How many kByte/s is "100% speed" with your connection?

Just wondering, if my friend goes ahead and buys a Mac Mini for MOS3 on it... what sort of speeds will he get? Or there is no way to know this until he buys it?

Thanks!

> if my friend goes ahead and buys a Mac Mini for MOS3 on it... what sort
> of speeds will he get? Or there is no way to know this until he buys it?

If you read this thread you'll see that some people (including me) say that about 500 kByte/s is the maximum there is to get with HTTP downloads, but you'll also see that there are some users who for whatever reason experience a significantly higher download rate with their setup. The reasons for this discrepancy are yet to be determined.

Just to confirm, using Powerbook G4 5,6 (100Mbps wired) with Mac-Mini on the same LAN (different subnet) via 54Mbps wireless:

Tests were performed on the Powerbook, with the Mac-min as Samba server and Mint (Maya) ISO mirrored at Kent University (UK) as HTTP fileserver. HTTP transfer was using OWB, SMB transfers using SMBFS to mount shares.

HTTP over wired 100Mbps LAN > 20Mbps ADSL (with an actual Tx rate around 10Mbps).
SMB over wired LAN @100Mbps, wireless subnet @56Mbps)

HTTP maxes out around 800kBps (around 6Mbps) on download (upload not tested)
SMB maxes out around 1.1MBps (around 9Mbps) on download (around 250kBps on upload).

SMBFS is v1.74, Samba is the MorphOS v3.0.4 (updated) from Aminet.

Samba is started from INetD.

SMBFS is run with command line:

Quote:

(adding cache increased SMB transfer max from ~800kBps to ~1.1MBps)
(large stack value was just there for testing, haven't found a reason to reduce this yet)

Setting up Samba was more difficult than it should have been, but it's now working across subnets on all machines (albeit a little slowly)

----------------

On final thing which annoys me with Samba - after a while I seem to have multiple netbios-ssn tasks running taking up cpu usage to around 75% (splitting cpu usage evenly between tasks). Even after all transfers have stopped (even with the smb client disconnected completely), the tasks keep running.
Does anyone else experience this? Any workarounds/fixes?

[ Edited by boot_wb 10.11.2012 - 15:29 ]