From: lharr...@gmail.com   
      
   On Friday, December 1, 2023 at 6:23:24 PM UTC-5, lharr...@gmail.com wrote:   
   > On Friday, December 1, 2023 at 4:55:29 PM UTC-5, Kevin Bowling wrote:    
   > > On 12/1/23 04:07, Christian Holzapfel wrote:    
   > > > Here a few benchmarks from my systems:    
   > > >    
   > > > Model 6886 (PC 750)    
   > > > AMD K6-III @ 400 MHz    
   > > > 192 MB RAM    
   > > > Windows 95    
   > > > Burst Mode Rx+Tx    
   > > > Netio 1.32    
   > > >    
   > > > Packet size 1k bytes: 4751.55 KByte/s Tx, 4547.97 KByte/s Rx.    
   > > > Packet size 2k bytes: 6082.19 KByte/s Tx, 4748.86 KByte/s Rx.    
   > > > Packet size 4k bytes: 6772.06 KByte/s Tx, 5595.94 KByte/s Rx.    
   > > > Packet size 8k bytes: 7378.91 KByte/s Tx, 5575.82 KByte/s Rx.    
   > > > Packet size 16k bytes: 7996.92 KByte/s Tx, 6135.80 KByte/s Rx.    
   > > > Packet size 32k bytes: 8226.20 KByte/s Tx, 6186.44 KByte/s Rx.    
   > > >    
   > > >    
   > > >    
   > > > Model 9576i (Lacuna)    
   > > > AMD X5 @ 133 MHz    
   > > > 64 MB RAM    
   > > > No L2 cache :-(    
   > > > Windows 95    
   > > > Burst Mode Rx+Tx    
   > > > Netio 1.32    
   > > >    
   > > > Packet size 1k bytes: 1193.33 KByte/s Tx, 936.21 KByte/s Rx.    
   > > > Packet size 2k bytes: 1290.28 KByte/s Tx, 1114.61 KByte/s Rx.    
   > > > Packet size 4k bytes: 1389.92 KByte/s Tx, 1575.54 KByte/s Rx.    
   > > > Packet size 8k bytes: 1741.82 KByte/s Tx, 1541.64 KByte/s Rx.    
   > > > Packet size 16k bytes: 1969.17 KByte/s Tx, 1173.42 KByte/s Rx.    
   > > > Packet size 32k bytes: 1934.02 KByte/s Tx, 1874.79 KByte/s Rx.    
   > > >    
   > > >    
   > > >    
   > > > Model 7013-59H (RS/6000)    
   > > > POWER2 @ 67 MHz    
   > > > 1,25 GB RAM    
   > > > 1 MB L2 cache    
   > > > AIX 4.3.3 (IBM's driver)    
   > > > Netio 1.32    
   > > >    
   > > > Packet size 1k bytes: 5113.53 KByte/s Tx, 6395.88 KByte/s Rx.    
   > > > Packet size 2k bytes: 6059.71 KByte/s Tx, 7318.23 KByte/s Rx.    
   > > > Packet size 4k bytes: 7213.85 KByte/s Tx, 7667.44 KByte/s Rx.    
   > > > Packet size 8k bytes: 7877.74 KByte/s Tx, 8501.13 KByte/s Rx.    
   > > > Packet size 16k bytes: 8819.90 KByte/s Tx, 8811.09 KByte/s Rx.    
   > > > Packet size 32k bytes: 8627.93 KByte/s Tx, 9067.06 KByte/s Rx.    
   > > >    
   > > >    
   > > >    
   > > > Model 7006-42T (RS/6000)    
   > > > PowerPC @ 120 MHz    
   > > > 192 MB RAM    
   > > > 0.5 MB L2 cache    
   > > > AIX 4.3.3 (IBM's driver)    
   > > > Netio 1.32    
   > > >    
   > > > Packet size 1k bytes: 2558.84 KByte/s Tx, 619.03 KByte/s Rx.    
   > > > Packet size 2k bytes: 2754.74 KByte/s Tx, 1297.03 KByte/s Rx.    
   > > > Packet size 4k bytes: 3340.64 KByte/s Tx, 3816.94 KByte/s Rx.    
   > > > Packet size 8k bytes: 4013.11 KByte/s Tx, 3763.10 KByte/s Rx.    
   > > > Packet size 16k bytes: 4551.11 KByte/s Tx, 3802.62 KByte/s Rx.    
   > > > Packet size 32k bytes: 4562.02 KByte/s Tx, 4275.55 KByte/s Rx.    
   > > >    
   > > >    
   > > > What's interesting is the mixed performance on the RS/6000 systems. The   
   later mid-range workstation machine with a faster CPU performs worse than the   
   early 1992 high-performance server class system. I assume, the 7013 was built   
   specifically for    
   high-throughput and low-latency applications.    
   > > >    
   > > In particular, the memory bandwidth is insane on the POWER2 systems.    
   > > The MHz is also misleading, it is a superscalar processor with 2 integer    
   > > units and has double the cache (10ns I believe). It would be    
   > > interesting to compare the context switch time of the PowerPC.    
   > > > That the Lacuna and Reply perform so low is disappointing, but maybe   
   there's a reason why IBM did not sell this adapter to PS/2 users - or maybe we   
   will find a magic switch to make the card operate faster.    
   > > I suspect you may be able to profit from games with the tx and rx    
   > > interrupt masks.. the Linux pcnet32 driver looks like it has a (limited)    
   > > poll mode which would be preferable to dealing with an interrupt for    
   > > every packet on this old hw.   
   > I ran netio with the 3com 10mbit card and found CPU usage drops when   
   transmitting. As the packet size gets bigger the CPU usage obviously drops   
   with 32k dropping to nearly 60% usage. I'll see if I can try the updated   
   driver tonight.    
   >    
   > As far as the Reply Board, I looked up an old photo of SpeedSys and the   
   POD83 scores 41.68, for memory bandwidth... L1 Cache is 51.69MB/sec, L2 Cache   
   is 31.54MB/sec, and Memory Throughput is 22.85 MB/sec. Not sure if that's   
   decent or not, but I am    
   pretty sure it beats the Valuepoint... and for some reason the POD83 scores   
   higher on that system. I really think it's down to chipset.   
       
   Forgot, cachechk has different numbers than SpeedSys, for the Reply board   
   Cachechk says  L1 is 113.8MB/sec or 9.2ns, L2 is 50.3MB/sec or 20.9ns, and   
   Main memory is 31.6MB/sec or 33.2ns.  I've always been curious how the POD83   
   is gets choked on this    
   system.    
      
   Also, I found the Speedsys/cachechk results for the Valuepoint system,    
   Speedsys CPU scores as 60.41, the L1 cache is  69.40MB/sec, L2 is 38.73MB/sec,   
   and Main memory is 26.06MB/s.   Cachechk says L1 is 114.8MB/sec or 9.1ns, L2   
   is 49.2MB/sec or 21.3ns,    
   and Main memory is 28.1Mb/sec or 37.3ns.  So... I was wrong, the Valuepoint   
   edges out the Reply TurboBoard by a few Mb/sec of memory bandwidth... but   
   would a few MB/sec really make a ~33% dent in CPU performance for these kinds   
   of systems and ranges back    
   then?    
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)