Re: MI SONIC Ethernet driver for mac68k

thorpejshagadelic.org wrote:

> > I'm still waiting reports on nubus based one or 16
bit mode one,
> > but anyway I'll commit it next weekend if there is
no objection.
> 
> Do you have a performance comparison for the old vs.
the MI one?

Unfortunately, MI one is slower (currently).


---
old MD driver:

# dmesg|grep sn0
sn0 at obio0: integrated Ethernet adapter
sn0: Ethernet address 08:00:07:9f:07:c6
# ./ttcp -rs
ttcp-r: buflen=8192, nbuf=2048, align=16384/0, port=5001 
tcp
ttcp-r: socket
ttcp-r: accept from 192.168.20.1
ttcp-r: 16777216 bytes in 19.31 real seconds = 848.59 KB/sec
+++
ttcp-r: 2049 I/O calls, msec/call = 9.65, calls/sec =
106.13
ttcp-r: 0.0user 19.2sys 0:19real 99% 0i+0d 0maxrss 0+2pf
0+0csw
# ./ttcp -ts 192.168.20.1
ttcp-t: buflen=8192, nbuf=2048, align=16384/0, port=5001 
tcp  -> 192.168.20.1
ttcp-t: socket
ttcp-t: connect
ttcp-t: 16777216 bytes in 15.93 real seconds = 1028.29
KB/sec +++
ttcp-t: 2048 I/O calls, msec/call = 7.97, calls/sec =
128.54
ttcp-t: 0.1user 15.4sys 0:15real 98% 0i+0d 0maxrss 0+4098pf
0+0csw
# 

---
MI driver:

# dmesg|grep sn0
sn0 at obio0: integrated SONIC Ethernet adapter
sn0: Ethernet address 08:00:07:9f:07:c6
# ./ttcp -rs
ttcp-r: buflen=8192, nbuf=2048, align=16384/0, port=5001 
tcp
ttcp-r: socket
ttcp-r: accept from 192.168.20.1
ttcp-r: 16777216 bytes in 23.85 real seconds = 686.84 KB/sec
+++
ttcp-r: 2049 I/O calls, msec/call = 11.92, calls/sec =
85.90
ttcp-r: 0.1user 23.7sys 0:23real 99% 0i+0d 0maxrss 0+2pf
0+0csw
# ./ttcp -ts 192.168.20.1
ttcp-t: buflen=8192, nbuf=2048, align=16384/0, port=5001 
tcp  -> 192.168.20.1
ttcp-t: socket
ttcp-t: connect
ttcp-t: 16777216 bytes in 23.18 real seconds = 706.90 KB/sec
+++
ttcp-t: 2048 I/O calls, msec/call = 11.59, calls/sec =
88.36
ttcp-t: 0.1user 22.9sys 0:23real 99% 0i+0d 0maxrss 0+4098pf
0+0csw
# 
---

I wonder if BUS_DMA_COHERENT support or DMA threshold
settings
(DCR_RFT/DCR_TFT) could improve this..

(MI one still has better stability on heavy traffic though)
---
Izumi Tsutsui

At 9:05 Uhr +0900 5.6.2007, Izumi Tsutsui wrote:
>> Do you have a performance comparison for the old
vs. the MI one?
>
>Unfortunately, MI one is slower (currently).

Can you time the transfers from the other (I assume,
non-mac68k) machine
for comparison? Since, as I understand, the MD driver does
buffer-to-memory
transfers by cpu, it may well lock out timer interrupts and
lose clock
ticks, possibly skewing your timing results.

	hauke

--
"It's never straight up and down"     (DEVO)