View Single Post
  #7  
Old December 28th 08, 02:05 PM posted to comp.os.linux.hardware,alt.comp.hardware.overclocking
Aragorn
external usenet poster
 
Posts: 17
Default New release of sys_basher

On Sunday 28 December 2008 14:10, someone identifying as *General
Schvantzkoph* wrote in /comp.os.linux.hardwa/

On Sun, 28 Dec 2008 07:22:40 +0100, Aragorn wrote:

On Sunday 28 December 2008 05:24, someone identifying as *General
Schvantzkoph* wrote in /comp.os.linux.hardwa/

I've put a new release of sys_basher on the web,

http://www.polybus.com/sys_basher_web/

sys_basher is a multi-threaded hardware exerciser, memory tester and
benchmarking tool. It will run on any Linux or Unix.


Does it also do diagnostics of other possibly failing hardware
components than memory? I'm just asking because I've got a machine
sitting here idly for quite some time now due to strange lockups and
BIOS ECC error log entries.


I use the term exerciser rather than diagnostic because it can't identify
the individual component that's failing, all it can do is tell you that
there is a problem with a CPU, RAM or disk. Sys_basher is an ordinary
user space program, and as far as I can tell there is no way to get Linux
to do a logical address to physical address translation or to determine
which physical core a thread is running on for a user space program. If
there is a way I'd appreciate someone telling me how to do it and I'd add
it to sys_basher so that it could give more precise error messages.

Sys_basher is designed to maximize the stress on the system. In addition
to memory tests, which operate at the maximum read and write bandwidth of
each level of the memory hierarchy, there are integer and floating point
arithmetic tests which are unrolled loops that can use up to eight
functional units (4 adders and 4 multipliers) per cycle which should push
the core temperatures to their maximum. All of the arithmetic tests are
done register to register, memory to register and memory to memory. I
step through array sizes from 1K, which fits entirely in the primary
cache, to 64M which operates almost entirely in DRAM. The RAM and disk
tests also step through a range of array sizes, from 1K to 64M for disk I/
O and from 1K to the full virtual memory size for the memory tests (there
is a switch to set the amount of RAM tested, I find that the physical
size - 300M is a good choice, more than that will cause the system to
swap which will kill the performance).

Sys_basher also reports memory bandwidth for each array size, disk
bandwidth for each of the supported I/O modes and file sizes, OPS and
MFLOPS for each array size.

Give it a try and see if it helps you identify your problem. I'd
appreciate any feedback and suggestions that you have.


I will, thank you. It'll be some time before I actually can, though. The
machine is currently sitting disassembled - i.e. monitor, powercord,
keyboard and mouse disconnected - in my living room as I needed and at the
moment still need the space for maintenance on my rather vast collection of
guitars...

Yet I will run the test and I will report back to you. ;-)

--
*Aragorn*
(registered GNU/Linux user #223157)