Re: Poor USB2 performance on G4 Mac Mini - expected?

[James Haggerty <james.haggerty%gmail.com@localhost>]

On Mon, 19 Oct 2020 at 00:46, Jonathan A. Kollasch
<jakllsch%kollasch.net@localhost> wrote:
>
> On Fri, Oct 16, 2020 at 04:07:57PM +1100, James Haggerty wrote:
> > Hi,
> >
> > I'm finding that I can get at most aound 60mbps from the USB2 ports on
> > my Mac Mini G4. For example:
> >
> > # dd if=/dev/sd0c of=/dev/null bs=1024k count=100
>
> Try the character device /dev/rsd0c instead.
>
> The block device supports byte-oriented access via read/modify/write,
> but accesses the drive in BLKDEV_IOSIZE (usually 2KiB) chunks.
>
> The character device supports logical block-aligned accesses up
> to MAXPHYS bytes (usually 64KiB) in size each.

Thanks very much; that increased speeds from the usb disk to around
150mbps. It's been a long time since I've messed with any BSD system.
That revealed that the bottleneck is not with USB generally, but
unfortunately I'm still seeing my actual problem, which is that my
RTL8153 (ure0) is delivering at most 60mbps (I'm trying to use the
mini as a router).

The only thing that seems different while I'm putting it under load is
that interrupts are much higher:

$ ifconfig ure0
ure0: flags=0x8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
        capabilities=3ff00<IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx,TCP4CSUM_Tx>
        capabilities=3ff00<UDP4CSUM_Rx,UDP4CSUM_Tx,TCP6CSUM_Rx,TCP6CSUM_Tx>
        capabilities=3ff00<UDP6CSUM_Rx,UDP6CSUM_Tx>
        enabled=3ff00<IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx,TCP4CSUM_Tx>
        enabled=3ff00<UDP4CSUM_Rx,UDP4CSUM_Tx,TCP6CSUM_Rx,TCP6CSUM_Tx>
        enabled=3ff00<UDP6CSUM_Rx,UDP6CSUM_Tx>
        ec_capabilities=1<VLAN_MTU>
        ec_enabled=0
        address: 00:e0:4c:68:00:07
        media: Ethernet autoselect (1000baseT full-duplex)
        status: active
        inet 192.168.200.1/24 broadcast 192.168.200.255 flags 0x0
        inet6 fe80::2e0:4cff:fe68:7%ure0/64 flags 0x0 scopeid 0x3

$ vmstat -w 1
 procs    memory      page                       disks   faults      cpu
 r b      avm    fre  flt  re  pi   po   fr   sr w0 s0   in   sy  cs us sy id
 1 0   668184  54100   12   0   0    0    0    0  0  0  277    0  46  0  0 100
 0 0   668488  54096   79   0   0    0    0    0  0  0 4147    0 4804 0 21 79
 0 0   668816  54096   78   0   0    0    0    0  0  0 4149    0 4918 0 14 86
 0 0   669148  54096   79   0   0    0    0    0  0  0 4152    0 4973 0 11 89
 0 0   669564  54096   76   0   0    0    0    0  0  0 4152    0 4967 0 10 90
 0 0   669864  54096  326   0   0    0    0    0  0  0 4149    0 4995 1 15 84
 1 0   670240  54096   76   0   0    0    0    0  0  0 4133    0 4934 0 14 86
 1 0   670684  54096   75   0   0    0    0    0  0  0 4192    0 4932 0 16 84
 0 0   671068  54096   76   0   0    0    0    0  0  0 4170    0 4881 0 14 86

Compared to when I'm doing dd if=/dev/rsd0c of=/dev/null bs=1024k count=100 :

$ vmstat -w 1
 procs    memory      page                       disks   faults      cpu
 r b      avm    fre  flt  re  pi   po   fr   sr w0 s0   in   sy  cs us sy id
 0 1   676956  53172   13   0   0    0    0    0  0  0  278    0  47  0  0 100
 0 1   676512  53168 3585   0   0    0    0    0  0 217 776    0 708  0 11 89
 0 1   677024  53168 3328   0   0    0    0    0  0 214 764    0 698  0 15 85
 0 1   676704  53168 3584   0   0    0    0    0  0 217 770    0 716  0 10 90
 0 1   676448  53168 3584   0   0    0    0    0  0 218 778    0 735  0  9 91

Perhaps I should just be trying other ethernet adapters.