On 954100635 seconds since the Beginning of the UNIX epoch
Simon Burge wrote:
>
>There was a short thread called "PC164 memory bus speeed" around August
>year where I had the same concerns about the ``slowness'' of a pc164.
>There was no answer on the (lack of) memory speed performance.  Here's
>part of my lmbench summary:
>
>*Local* Communication bandwidths in MB/s - bigger is better
>-----------------------------------------------------------
>Host                OS  Pipe AF    TCP  File   Mmap  Bcopy  Bcopy  Mem   Mem
>                             UNIX      reread reread (libc) (hand) read write
>--------- ------------- ---- ---- ---- ------ ------ ------ ------ ---- -----
>i386-netb  NetBSD 1.4.1   39   39   36     29    201     92     68  201    92
>alpha-net   NetBSD 1.4J   93  103          92    148     79     77  149   129
>alpha-dec     OSF1 T5.0  411  366         438   1158    527    293 1153   486
>
>The first is my _old_ DEC Prioris PPro 166 (to show that DEC sometimes
>got memory bandwidth right in the past), the middle is a PC164 (whose
>number match up with hbench) and the last is a DS20 (to show that they
>still can sometimes get memory bandwidth right ;).

Here are the results that I found with my AS200 4/233 and my PWS433a:

*Local* Communication bandwidths in MB/s - bigger is better
-----------------------------------------------------------
Host                OS  Pipe AF    TCP  File   Mmap  Bcopy  Bcopy  Mem   Mem
                             UNIX      reread reread (libc) (hand) read write
--------- ------------- ---- ---- ---- ------ ------ ------ ------ ---- -----
AS200.233  NetBSD 1.4.1   30   30   13     35     58     31     31   59    72
PWS433a     NetBSD 1.4V   77   87   -1     81    263    125    124  263   204

The PWS433a has no L3 cache.

 == Roland Dowdeswell                      http://www.Imrryr.ORG/~elric/  ==
 == The Unofficial NetBSD Web Pages        http://www.Imrryr.ORG/NetBSD/  ==
 == The NetBSD Project                            http://www.NetBSD.ORG/  ==