Stream numbers from Tyan Tiger 133 with VIA Pro 133a chipset ...

Brian D. Haymore brian at chpc.utah.edu
Wed Jun 14 13:55:23 PDT 2000


Here are our results from our new AMD 950Mhz boxes using the ABIT KA7
Motherboard and the latest Beta "RY" bios.

-------------------------------------------------------------
This system uses 8 bytes per DOUBLE PRECISION word.
-------------------------------------------------------------
Array size = 10000000, Offset = 0
Total memory required = 228.9 MB.
Each test is run 10 times, but only
the *best* time for each is used.
-------------------------------------------------------------
Your clock granularity/precision appears to be 1 microseconds.
Each test below will take on the order of 268355 microseconds.
   (= 268355 clock ticks)
Increase the size of the arrays if this shows that
you are not getting at least 20 clock ticks per test.
-------------------------------------------------------------
WARNING -- The above is only a rough guideline.
For best results, please be sure you know the
precision of your system timer.
-------------------------------------------------------------
Function      Rate (MB/s)   RMS time     Min time     Max time
Copy:         492.3896       0.3263       0.3249       0.3280
Scale:        492.1473       0.3265       0.3251       0.3296
Add:          557.8127       0.4317       0.4303       0.4348
Triad:        570.6840       0.4240       0.4205      
0.4256                                                                                               





The numbers you got are close to what we saw before we started using the
"memory interleaving" features in this beta bios.

Here are the pre-interleaving numbers:

-------------------------------------------------------------
This system uses 8 bytes per DOUBLE PRECISION word.
-------------------------------------------------------------
Array size = 10000000, Offset = 0
Total memory required = 228.9 MB.
Each test is run 10 times, but only
the *best* time for each is used.
-------------------------------------------------------------
Your clock granularity/precision appears to be 1 microseconds.
Each test below will take on the order of 283765 microseconds.
   (= 283765 clock ticks)
Increase the size of the arrays if this shows that
you are not getting at least 20 clock ticks per test.
-------------------------------------------------------------
WARNING -- The above is only a rough guideline.
For best results, please be sure you know the
precision of your system timer.
-------------------------------------------------------------
Function      Rate (MB/s)   RMS time     Min time     Max time
Copy:         380.1740       0.4215       0.4209       0.4224
Scale:        381.4227       0.4219       0.4195       0.4243
Add:          446.3248       0.5391       0.5377       0.5406
Triad:        441.2017       0.5460       0.5440      
0.5472                                                                                               



I'm not sure but it seems that some if not many of the KX133 based
boards are still working on stability and have not put all the
performance options in yet.


We have been working without issues on this latest beta bios for about a
3 weeks now.  





Richard Walsh wrote:
> 
> All,
> 
> I wanted to get some feedback from others on the
> reasonableness of the following stream benchmark
> for a Tyan Tiger 133 based cluster we have assembled.
> They are substantially less than those reported for
> a PIIIEB_600 on the stream site and are about the
> same as those reported for the 440_BX chipset with
> a Katmai 600. I was expecting better numbers.  Perhaps
> others have run the benchmark as well with a different
> compiler or options.
> 
> Interested in your comments ...
> 
> rbw
> 
> System Specifications:
> 
> 1. Dual PIII/EB 667 CPUs
> 2. 256 MB PC133 Memory (Micron no ECC)
> 3. VIA Pro 133a chipset with 133 FSB.
> 
> Compilation:
> 
> gcc -O3 stream_d.c second_wall.c -lm
> (under RHAT Linux 6.2/ kernel 2.2.15)
> 
> Results:
> 
> -------------------------------------------------------------
> This system uses 8 bytes per DOUBLE PRECISION word.
> -------------------------------------------------------------
> Array size = 5000000, Offset = 0
> Total memory required = 114.4 MB.
> Each test is run 10 times, but only
> the *best* time for each is used.
> -------------------------------------------------------------
> Your clock granularity/precision appears to be 1 microseconds.
> Each test below will take on the order of 217494 microseconds.
>    (= 217494 clock ticks)
> Increase the size of the arrays if this shows that
> you are not getting at least 20 clock ticks per test.
> -------------------------------------------------------------
> WARNING -- The above is only a rough guideline.
> For best results, please be sure you know the
> precision of your system timer.
> -------------------------------------------------------------
> Function      Rate (MB/s)   RMS time     Min time     Max time
> Copy:         335.6732       0.2385       0.2383       0.2391
> Scale:        336.2616       0.2380       0.2379       0.2380
> Add:          436.9053       0.2747       0.2747       0.2748
> Triad:        309.8261       0.3877       0.3873       0.3882
> 
> #---------------------------------------------------
> #
> # Richard Walsh
> # NetAPSx, Inc.
> # 1200 Washington Ave. So.
> # Minneapolis, MN 55415
> # VOX:    612-337-3467
> # FAX:    612-337-3400
> # EMAIL:  rbw at networkcs.com
> #
> #---------------------------------------------------
> # "What you can do, or dream you can, begin it;
> #  Boldness has genius, power, and magic in it."
> #                                       -Goethe
> #---------------------------------------------------
> 
> _______________________________________________
> Beowulf mailing list
> Beowulf at beowulf.org
> http://www.beowulf.org/mailman/listinfo/beowulf

--
Brian D. Haymore
University of Utah
Center for High Performance Computing
155 South 1452 East RM 405
Salt Lake City, Ut 84112-0190

Email: brian at chpc.utah.edu - Phone: (801) 585-1755 - Fax: (801) 585-5366




More information about the Beowulf mailing list