Math Question re: TIMER and hz

[StanHelton]

My math is fuzzy beyond VERY basic Calculus and I've never been an electrician. Can anybody tell me how to convert intervals retrieved from the TIMER function (startTime@, elapsedTime@) and executed lines per second into a hz rate through a parallel port?

I'm using PBCC 4.04 trying to benchmark the WinIo driver. The program generates this output (extraneous digits removed):

InstallWinIoDriver() = 1
InitializeWinIo() = 1
ShutdownWinIo() = 1
RemoveWinIoDriver = 1
Elapsed time = 1.08 for 2400 iterations.
Lines per second = 11131.7

How does that translate into hz?

#COMPILE EXE
#DIM ALL

#INCLUDE "f:\PBWin80\WinIO\WinIO.bas"

FUNCTION PBMAIN () AS LONG
   LOCAL lResult AS LONG
   LOCAL startTime AS DOUBLE
   LOCAL elapsedTime AS DOUBLE
   LOCAL tempStr AS STRING
   LOCAL iteration AS LONG
   LOCAL PortAddr%
   LOCAL PortVal&
   LOCAL bSize?
   LOCAL PhysAddr&
   LOCAL PhysSize&
   LOCAL PhysMemHandle&
   LOCAL PhysVal&
   LOCAL LinAddr&


   lResult = InstallWinIoDriver("f:\PBCC40\My Source Code\", 0)
   PRINT "InstallWinIoDriver() = " lResult
   lResult = InitializeWinIo()
   PRINT "InitializeWinIo() = " lResult
   tempStr = WAITKEY$

   startTime = TIMER
   FOR iteration = 1 TO 2400
      lResult = GetPortVal(PortAddr%, PortVal&, bSize?)
      lResult = MapPhysToLin(PhysAddr&, PhysSize&, PhysMemHandle&)
      lResult = GetPortVal(PortAddr%, PortVal&, bSize?)
      lResult = GetPhysLong(PhysAddr&, PhysVal&)
      lResult = UnmapPhysicalMemory(PhysMemHandle&, LinAddr&)
   NEXT interation
   elapsedTime = TIMER - startTime
   linesPerSecond = 2400 * 5 / elapsedTime

   lResult = ShutdownWinIo()
   PRINT "ShutdownWinIo() = " lResult
   RemoveWinIoDriver()
   PRINT "RemoveWinIoDriver = " lResult

   PRINT "Elapsed time = " elapsedTime " for 2400 iterations."
   PRINT "Lines per second = " linesPerSecond
   tempStr = WAITKEY$

END FUNCTION

Thanks,
Stan

[Michael Mattias]

A "hertz" is an inverse second.

So your "hertz" is simply (number of whatevers)/ (time in seconds).

[StanHelton]

A "hertz" is an inverse second.

So your "hertz" is simply (number of whatevers)/ (time in seconds).

Thanks Michael. I thought I was on the right track.

[Paul Dixon]

Stan,
what are you trying to do? That code makes no sense. You don't need to be mapping memory in order to read an I/O port.
For speed, use direct port access, not the GetPortVal call.

If you're trying to find out how fast you can read an I/O port it'll probably turn out to be around 1 to 4MHz, depending on your hardware.

More generally, to get hertz it's just:
(number Of Times Thing Is Done)/(Time in seconds) = frequency in Hz

So if you do 2400 things in 1.08 seconds then you do those things at a frequency of 2,400/1.08=2,222Hz.

Paul.

[Gary Barnes]

Hi Stan,
You might need to put a Sleep 0 in your loop somewhere to let windows do other stuff.
Strictly speaking, you are after samples per seconds, but none the less you divide the number of iterations by the time.
Either 2400/1.08 or 5*2400/1.08 will get you there.

Oops - I just noticed that MCM beat me to it.

Anyway, If you want some precision in your timing, look at David Roberts code in the "Source" forum.
Why settle for milliseconds when you can have micro second resolution?

Cheers

[Michael Mattias]

>but none the less you divide the number of iterations...

...or number of bytes, or the number of reads, or number of lollipops... "hertz" by itself is only a denominator: a numerator is required.

MCM

[StanHelton]

Thanks Paul. I didn't realize I wasn't actually doing anything here. Hmmmm.
:kick:

Stan,
what are you trying to do? That code makes no sense. You don't need to be mapping memory in order to read an I/O port.
For speed, use direct port access, not the GetPortVal call.

If you're trying to find out how fast you can read an I/O port it'll probably turn out to be around 1 to 4MHz, depending on your hardware.

More generally, to get hertz it's just:
(number Of Times Thing Is Done)/(Time in seconds) = frequency in Hz

So if you do 2400 things in 1.08 seconds then you do those things at a frequency of 2,400/1.08=2,222Hz.

Paul.