I have the same issue with my Mac Mini 1.5Ghz with 64Mb vram...

Pretty unscientific, but with the CNet Bandwidth test:

490 kbps with MorphOS

862 kbps with Mac OS X 10.5.8

(same machine, same hardware)

I'd be glad to provide any other details as needed.

IKE

@IKE

Are these bytes or bits?

Anyway, it's unlikely to be the same issue. In your case, it has more to do with the dating AmiTCP/Sana2 design, not really suited for higher inet speeds (but locally, you probably still get >8MB/s speed, i guess, you could check that).

The real slowness problem originally exposed in this thread is about getting unstable and very low speed, like less than 10kB/s (which may be related to some autonegociation issue).

@Fab

They are bytes. I'll check the local speed.

OK thanks, that makes sense about AmiTCP/Sana2 because my A1200/060 get's about the same transfer rates when connected to the same line as my MorphOS machine.

IKE

Hi,

IMO there must be something going wrong (perhaps autonegotiation) with the sungem_eth.device. I connected a macmini together with some Peagsos2's using Envoy3 and when accessing a shared partition it takes everytime several seconds before the content of the partition becomes available. Transfers from or to these partitions are much slower then when doing the same between 2 Pegasos2...

If testing is needed please let me know what excately to test, I will be glad to help.

Quote:

Hmh... I don't believe it's the issue still. Here are some results I've got earlier with different AmigaOS/MorphOS systems and they don't indicate such slowness because of AmiTCP/Sana2...

And one online speed meter with OWB says currently 8.82 Mbps (1.1 MB/sec) with Mac mini 1.5GHz & MorphOS.

Heh.. I made some tests with OSX on 1.42GHz mini too... with that online web based speed test. Seems to give different results with different browsers. Firefox 9.60 Mbps (1.2 MB/sec), Safari 9.25 Mbps (1.16 MB/sec), Opera 547.50 kbps (68.44 KB/sec) :). I repeated the test many times and switched to different browsers many times in different order. Results were always similar.

Sooo, I wouldn't trust to those online tests too much. Better to download something yourself from fast and reliable source instead.

[ Edited by jPV on 2010/2/11 19:47 ]

As I said, I tried with wget & owb and I get the same speed...

Quote:

If you read all of this thread you will see what my suggestions are in my earlier posts and which sites I used to test the speed of my 1.5GHz G4 MacMini running MorphOS2.4 compared to my PC with Windows Vista Ultimate using the same connection hardware to access the Internet. As you can see from my earlier post, when I removed a Netgear switch and moved my MacMini so it could be connected directly to my router and it seems to have fixed my problem (most of the time).

Others have suggested different testing, so you might want to look at those suggestions as well, specially anything suggested by Fab, or other members of the MorphOS development team. The more info you can provide to the team, the better the chance they can figure out what is causing the problem.

Hi,

this ethernet problem with the MacMini is driving me crazy! Doing a lot more testing with the MacMini I found the following out: sometimes there is no problem, it works fast and fine, no stops during e. g. download, or not stops during copying from or to another LAN partition and sometimes eg. after a reset the problem is there! And I don't know why or from what this could come from... The speed I get is about 5.5 MB/s during file-copy from one LAN partition to another with Envoy3 when there is no problem, which is basically the same I get between two Pegasos2 though when the problem exists I experience stops during copying, transfer rates are much slower either LAN or internet...

On both, the MacMini and the Pegasos2 no dhcp is used but a fixed ip instead (of course the ip's are different on the Pegasos2 it's 192.168.1.153 and on the MacMini 192.168.1.159, mask on both is 255.255.255.0 and gateway on both is 192.168.1.1

Just to show this problem I made 2 small videos where I open MyMorphOS, where all the mounted partitions then are shown. As you can see doing this on the Pegasos2 is instantly, doing this on the MacMini, when there is the problem, the same takes about 8 secs... Excately the same mounts are available on the Pegasos2 and the MacMini.

The videos are avilable here:

http://www.rmsvideo.ch/morphos/

User: rms14@dmz.local
PW: rms4234

in the directory mos_24

- macmini_mymorphos_open.avi
- pegasos2_mymorphos_open.avi

I also noticed that in the ramdebug log after a reset there are "context failed" entries, is this normal? Here is a log I get after reset the MacMini:

Setting Mac MAC
Setting Mac Serial
Setting up Keylargo ATA 3
Setting Mac EDID
SYS_HWEnableInt: IntID 0x0
ABox 1.26 (17.11.2007) ? 1999-2007 by Ralph Schmidt, Emmanuel Lesueur, Teemu Suikki, Harry Sintonen
SYS_HWEnableInt: IntID 0x4
SYS_HWEnableInt: IntID 0x400001d
keylargo_Init: ResBase 0x200b8c58 SysBase: 0x20001948
keylargo_Init: done
PMac_Init:
SYS_HWEnableInt: IntID 0x4000027
ide.device@0: ATA-HD WDC WD1600BEVE-0 <DIRECT_ACCESS>
LIB_Open from task $200b0138 "? AsyncInit - <ide.device> ?" context failed!
LIB_Open from task $200b0138 "? AsyncInit - <ide.device> ?" context failed!
LIB_Open from task $200b0138 "? AsyncInit - <ide.device> ?" context failed!
ide.device@0: Mount <M9System>
ide.device@0: Partition <M9System> DosType 0x53465300 BootPri 0
LIB_Open from task $200b0138 "? AsyncInit - <ide.device> ?" context failed!
ide.device@0: Mount <M9System1>
ide.device@0: Partition <M9System1> DosType 0x53465300
LIB_Open from task $200b0138 "? AsyncInit - <ide.device> ?" context failed!
ide.device@0: Mount <M9Work>
ide.device@0: Partition <M9Work> DosType 0x53465300
LIB_Open from task $200b0138 "? AsyncInit - <ide.device> ?" context failed!
ide.device@0: Mount <M9AV>
ide.device@0: Partition <M9AV> DosType 0x53465300
LIB_Open from task $200b0138 "? AsyncInit - <ide.device> ?" context failed!
ide.device@0: Mount <M9MM>
ide.device@0: Partition <M9MM> DosType 0x53465300
ide.device@1: MATSHITA CD-RW CW-8124 <CDROM>
SYS_HWEnableInt: IntID 0x400003f
SYS_HWEnableInt: IntID 0x4000030
UserLibOpen : $VER: muimaster.library 20.6848 (10.10.09) ? 2008 Stefan Stuntz, PPC, MorphOS, DEBUG
SYS_HWEnableInt: IntID 0x4000001
SYS_HWEnableInt: IntID 0x4000029
Brynhildr [Aboutbox.mcc 20.14 (10.10.09)]

Any idea what could go wrong here?

I could make further tests if I would know what excately to do...

Regards

Same here with the eth problem on 1.25Ghz Mini 512MB Ram. Works good for about 10 minutes straight to my zyxel 660, then lag between network activity with varying download rates. Average rate is around 45 kB/s. I now usually use my PC to download at 350kB/s since it's much faster. I also see other issues with task management, but still, this is the first bit of awesomeness I've seen in a long time!

I also have problems with network on MacMini. It's not related to speed (didn't check download rates at all), but I think it's still related to the topic of this thread.

I use MacMini G4/1.4 Ghz, 1 GB Ram, 64 MB VRAM.
MorphOS 2.4, OWB 1.6.1

Whenever I start OWB it stalls for short time, system seems to freeze, cannot even move the mouse. After some seconds everything works normal again. The same always, when I change the site. I click on a link -> system freezes for some seconds -> new site is shown.

Until now I only got known to freezes when a program does something illegal (freeing memory not reserve before etc) and MOS need some time to do the debug output. But in this case LogTool shows nothing.

OS-X + Safari on the same machine works fine.

The same MOS/OWB-configuration on my PEG2/G3 works fine, too.

So it has probably to do with the ethernet driver for the Mini.

It doesn't make a difference if I use my router (DLink DIR-615) or a direct ethernet connection to the cable modem. Well, the modem itself also has kind of router in side. Otherwise MOS wouldn't be able to go online at all, as it only supports LAN connections but not direct internet connections.

I found out sth. interesting. When I use the the IP-adress instead of the hostname to access a site, the problem does not occur. Of course, if the site loaded contains absolute paths and thus requires dns-lookups, there will be problems again after the main page has been loaded.
If I use IP-adress and the site only contains relative references (for images etc), no problems occur.

For me it seems to be a DNS-Lockup problem.

Hope, this helps...

Err, do you really mean "system freeze" or just "application freeze", because the first one is really weird. If it's the latter, then just enable "use network thread" option in OWB network settings. By default, OWB (or rather all webkit implementations using CURL) resolves hosts in main task context, leading to that kind of latency, especially if the lookup is long, but it's limited to the application, of course

I never have experienced network problems on MorphOS. All the downloads i made were in the order of 100kbs. Of course, i have 10mb adsl line.

100kb/S on a 10mb line?

You should be getting ~1mb/sec in a 10 mbit line...

Just purchased a mac mini just for the purpose of running morphos.Download speed is 10-40 kbytes,and from rapidshare premium only 150kbytes while with macos 10.4.2 i can get 4x,6x times the speed.Youtube also is slow.Although browsing the web seems to be fine.The problem is with owb.I have a 1.33 gz mac mini with 512mbytes ram,and a ati 9200 32mbyte video ram.The ethernet cable is connected directly to my router which is a sagem fast 2444 isdn router.
I was planning to purchase morhpos 2.4 but seeing this thing really made me unhappy.Is there a fix ?Is there something else i should post here?Please help me,if it is fixed i will buy a morphos license at once.
Kind regards
Nikos Drogosis

@nikodr

About download speed, if you get 150kB/s with rapidshare, the other 10-40kB figures just mean these sites are slower.
There are a couple things you can do to improve download speed: in envarc:sys/net/netstack.config, add
TCP_RECVSPACE=32768
TCP_SENDSPACE=32768

Regarding youtube, keep in mind there's no official support from adobe and we have to rely on opensource reimplementations that are far from being optimal speed-wise (gnash, swfdec, ...). And even the official adobe plugin is slow for youtube on mac mini anyway. :)

The good news is youtube and dailymotion recently moved to HTML5 video/audio, and it gives much better results. Next OWB version will be able to play them.

Fab thank you for your reply.

Regarding rapidshare,150 is not enough if you consider that as a premium user i can get double or triple the speed with macos.Actually anything i try is usally double the speed or triple with macos.So there must be some problem with morphos.

Kind regards

Yes, 150 isn't clearly not enough, and I also get more here with "fast" site (i max' at about 600kB/s ). So add the two lines i wrote earlier in netstack.config. Additionally, you could also type in a shell "setenv save OWB_POLLTIME 1" (default is 10ms). I doubt this will help in your case, but with local browsing, this variable is neeed to reach 10MB/s (CURL buffers are too small, so they must be emptied more often).

Try with sputnik, my test with owb and sputnik are both slow =(

Quote:
Well, from my point of view you should consider yourself lucky The problem a few of us encountered and that originated this thread is that sometimes for us the downloads come to a complete halt (or they continue at something like 100 bytes/s). For this reason I am mainly using the Peg2, while for many things the mini would effectively be a better choice.

Kind regards,
Andrea

Amigadave,

I wrote: "The Mac Mini is directly connected" (to the router).

BTW more and more people are reporting that there are no problems under OSX (the macmini at the same place) and problems under MOS...

Just did an ftp test... get wine-1.1.39.tar.bz2 using morphos cli and win7 command prompt in native ftp clients... results were: 1) morphos 16262746 bytes received in 59.2 seconds (274816 bytes/s) -- 268.38 Kbytes/sec 2) win7 16263746 bytes received in 47.14 seconds 345.00Kbytes/sec. -- (353280 bytes/s). Not a big difference. In OWB the transfer maxed at approximately 303 Kbytes/s, but did take a dive to 57 Kbytes/sec. Those were all ftp speed tests... I did an http speedtest in OWB also at the same site with erratic results. The http speedtests seem to drop out and restart often with an average of 45 Kbytes/sec. I'd say with further investigation, it may be that OWB has problems with the http protocol for file transfers. I'm personally going to look for a gui ftp client.

[ Edited by frotzen on 2010/2/19 19:54 ]

@guruman
yes, this is the point, I think the same, I wait the update MorphOS 2.5 :)

[ Edited by Divinity on 2010/2/20 12:07 ]

Quote:

That would be somewhat problematic for a *browser* if it couldn't deal with http protocol. :)
Anyway that's not the issue there. It's using CURL to handle that. If your speed is random it's either the network or the network driver.

@Fab:

i have the same speed problem as others with the macmini so i try the settings you propose in netstack.config.
i made a backup of the file first and i type in the end of the file the settings: TCP_RECVSPACE=32768 and TCP_SENDSPACE=32768.
after the reboot, when morphos coming up i get a startup message: need version 0 of bsdsocket.library
it's not clear from what application it's coming from. the worst is owb is not running any more.
i reboot a fiew times, the problem was remaining so i restore the original netstack.config back, but with no luck. now, i don't have a network and owb is out of order...
when i ping to router or the pc or the pegasos i get the same message about bsdsocket...