Threads, performance, response times

[Roger Rines]

This exercise seemed like a good way to learn more about thread performance. From the code provided by Paul & Chris, I created three different code modules, but only created 1 test file with 10,000,000 records so the testing would push the processing time in each section further and each module would be using the same data:

 Record_Count = 10,000,000
 
------------------ Paul's Code #1 - PBCC 4.04
    No Thread = 17.266
  With Thread = 10.406
         Diff =  6.86
     Increase = 39.73%
------------------ Chris' Code #1 - PBWin 8.04
    No Thread = 18.594
  With Thread = 10.688
         Diff =  7.906
     Increase = 42.52%
------------------ Paul's Code #2 - PBCC 4.04
    No Thread = 18.696
  With Thread =  9.064
         Diff =  9.632
     Increase = 51.52%
------------------ END
 
------------------ Computer Hardware & OS - CPU = XEON x 2 :
OS Name	Microsoft Windows XP Professional
Version	5.1.2600 Service Pack 2 Build 2600
OS Manufacturer	Microsoft Corporation
System Name	R9
System Manufacturer	ASUSTeK COMPUTER INC.
System Model	NCCH-DL
System Type	X86-based PC
Processor	x86 Family 15 Model 3 Stepping 4 GenuineIntel ~3200 Mhz
Processor	x86 Family 15 Model 3 Stepping 4 GenuineIntel ~3200 Mhz
Processor	x86 Family 15 Model 3 Stepping 4 GenuineIntel ~3200 Mhz
Processor	x86 Family 15 Model 3 Stepping 4 GenuineIntel ~3200 Mhz
BIOS Version/Date	Phoenix Technologies, LTD ASUS NCCH-DL ACPI BIOS Revision 1003, 10/1/2004
SMBIOS Version	2.3
Windows Directory	C:\WINDOWS
System Directory	C:\WINDOWS\system32
Boot Device	\Device\HarddiskVolume2
Locale	United States
Hardware Abstraction Layer	Version = "5.1.2600.2180 (xpsp_sp2_rtm.040803-2158)"
User Name	R9\Roger
Time Zone	Pacific Daylight Time
Total Physical Memory	2,048.00 MB
Available Physical Memory	1.33 GB
Total Virtual Memory	2.00 GB
Available Virtual Memory	1.96 GB
Page File Space	3.85 GB
Page File	C:\pagefile.sys

It seems that some threading can provide a lot of performance if the machine executing the threads has the capability to handle the work. In the test cases given in this posting and generated on my machine, it sure seems likes threads can save a lot of time because of their ability to better utilize the hardware. This performance is improved even more when the adjustments Paul made in his second code posting are employed.

[David Roberts]

Got it.

With the 2 error very early on in the FOR/NEXT LOOP there is nothing left of the file to read so strNull becomes empty. ClearFileCache gets an empty buffer and so has an early bath. In fact there is no need for ClearFileCache to use the same buffer - my logic is adrift there.

Funny thing is that on introducing error checking on GET I don't get one - should have got an end of file error. Must look into that now. Blast it!

With ClearFileCache2 it uses its own buffer and has no regard for how PBMain collects data so doesn't 'fall over'.

Actually, I think I'll remove ClearFileCache from the Source Code forum - the idea is OK but it was poorly implemented.

[David Roberts]

Breathes sigh of relief.

Still cannot figure out why ClearFilecache came beck so quickly on the second outing with the 2 error. I introduced the same error to ClearFileCache2 and the cache clearing took just as long, that is it behaved as expected. I'll leave it for now and look again - might spot something then. The reason may not be worth knowing but I won't know that until I find it.

[Gösta H. Lovgren-2]

Readings after correcting the code (removing the 2):

 Writing took =  4.38
 
 Clearing First Cache =  2.06
 First Read =  2.52
 Clearing Second Cache =  1.95
 Second Read =  2.53
 Clearing Third Cache =  1.95

All is copesetic again PB Land.

================================================================
"Every time I see an adult on a bicycle,
I no longer despair
for the future of the human race."
H.G. Wells
================================================================

[David Roberts]

Sorry, Gösta.

> Your relative observations may not change, just the absolute timings.

Isn't true. I couldn't figure out why you were getting such small values for second and subsequent filecache clearings. They should be all the same, or thereabouts, since ClearFileCache is reading the file.

When I removed the offending 2 your timings made more sense.

I thought only the GET timings would be affected.

I'll check out that error and see what is going on 'under the bonnet'.

[David Roberts]

I've added a new version of ClearFileCache in the Source Code forum.

[David Roberts]

Gösta, I found a typo in my code. A 2 crept in on reading the files. The first was a typo and then the error was duplicated on a 'Copy and Paste'. The 10248576 should read 1048576.

The moral here, of course, and I should know better after goodness knows how long is not to use use numbers in code but use a declaration of some sort. Of course when testing something it is very easy to slip into bad practice. We shouldn't, but I guess that I may not be the only one. Your relative observations may not change, just the absolute timings. I'll have a look at what you are doing.

[Gösta H. Lovgren-2]

Does the Cache have a Cache?

Been playing around this morning with the code. I put a Kill in before the first write and added a couple more timers. Here are the results:

Writing took = 3.95

Clearing First Cache = 1.98
First Read = 9.50
Clearing Second Cache = 0.02
Second Read = 10.08
Clearing Third Cache = 0.02

Clearing the cache the first time takes pretty consistently 100 times longer than subsequent Clear Caches.

Here's a run without the Kill:

Writing took = 0.00

Clearing First Cache = 1.97
First Read = 9.50
Clearing Second Cache = 0.00
Second Read = 10.25
Clearing Third Cache = 0.00

Is there a Catch to the Cache?

Okay, as I'm writing this it occurs to me the compiler is setting up internal values (or clears memory to work) when it first runs ClearCache and that's what takes the time and it isn't necessary subsequently. Does that make sense?

Okay, it doesn't. Here's another run with the first ClearCache remmed:

Writing took = 0.00

Clearing First Cache = 0.00
First Read = 10.17
Clearing Second Cache = 0.02
Second Read = 9.81
Clearing Third Cache = 0.02

Really curious. Anyway, if any want to try it themselves here's my David's code with my junk added:


'

'http://www.powerbasic.com/support/pbforums/showthread.php?t=35260
#Dim      All
#Compile  Exe 
#Register None
#Tools    Off
%NOMMIDS = 1
%NOGDI = 1
#Include "WIN32API.inc" ' 1 November 2006
'#Include "C:\Power Basic\Includes\Clear_File_Cache.inc"
'#Include "C:\Power Basic\Includes\clipboard.inc"
$sFile = "c:\test"
'         
'Clipboard stuff gotten from Poffs but I forget from whom
Function Clipboard_Set_Text Alias "Clipboard_Set_Text" _
              (ByVal sText As String) Export As Long
  Local hData As Long, hGlob As Long
' ** Create a global memory object and copy the data into it
  hData = GlobalAlloc(%GMEM_MOVEABLE Or %GMEM_DDESHARE, Len(sText) + 1)
  hGlob = GlobalLock(hData)
  Poke$ hGlob, sText + Chr$(0)
  GlobalUnlock hData
' ** Open the clipboard
  If IsFalse(OpenClipboard(%Null)) Then
    GlobalFree hData
    Exit Function
  End If
' ** Paste the data into the clipboard
  EmptyClipboard  'WinAPI
  Function = SetClipboardData(%CF_TEXT, hData) 'WinAPI
  CloseClipboard  'WinAPI
End Function
Function Clipboard_Get_Text Alias "Clipboard_Get_Text"() Export As String
    Local zPtr As Asciiz Ptr
    OpenClipboard 0 'WinAPI
    zPtr = GetClipboardData(%CF_TEXT)
    If zPtr <> 0 Then Function = @zPtr
    CloseClipboard   'WinAPI
End Function
'                   
'         
' '             
Sub ClearFileCache(sFile As String , Buffer As String)
   Local hFile, Chunk, nbLoaded As Long, ptrBuffer As Dword Ptr
   Dim zFile As Asciiz * 256
 
   zFile = sFile
   hFile = CreateFile(zFile, %GENERIC_READ, 0, ByVal 0, %OPEN_EXISTING, %FILE_FLAG_SEQUENTIAL_SCAN Or %FILE_FLAG_NO_BUFFERING, ByVal 0)
   Chunk = Len(Buffer)
   ptrBuffer = StrPtr(Buffer)
   Do
     ReadFile( hFile, ByVal ptrBuffer, ByVal Chunk, nbLoaded, ByVal 0 )
   Loop Until nbLoaded = 0
   CloseHandle(hFile)
End Sub
 '                   
Function PBMain( ) As Long
  Dim strNull As String
  Local i As Long, t1, t2 As Ext
  Local Clr_Cache01, Clr_Cache02, Clr_Cache03, Write_01 As Ext
  strNull = Nul$(1048576) ' 1 MB
 
'  Kill $sfile 'just to test
  Write_01 = Timer
  If Dir$($SFile) = "" Then
    ' Write a 100MB file
    Open $sFile For Binary As #1
    For i = 1 To 100
      Put$ #1, strNull
    Next
    Close #1                   
    ?"Wrote " & $sfile
  End If          
  Write_01 = Timer - Write_01
 
  ' Clear filecache after creation, if done, 
  ' And before before first Read
 
?"Clearing original Cache"
 Clr_Cache01 = Timer
  ClearFileCache($sFile, strNull)  
'? Using$("Clearing any existing cache took #.## seconds", Timer - t1)  
 Clr_Cache01 = Timer - Clr_Cache01
 
  ' Read file
  t1 = Timer
  t1 = Timer
  Open $sFile For Binary As #1
  For i = 1 To 100
    Get$ #1, 10248576, strNull
  Next
  Close #1
  t1 = Timer - t1
 ?"Clearing Read cache"
 
  ' Clear file cache after first reading and before second read
 Clr_Cache02 = Timer
  ClearFileCache($sFile, strNull)
  Clr_Cache02 = Timer - Clr_Cache02
'? Using$("Clearing Read Cache took #.## seconds", Timer - t3)  
 
  ' Read file again
  t2 = Timer
  Open $sFile For Binary As #1
  For i = 1 To 100
    Get$ #1, 10248576, strNull
  Next
  Close #1
  t2 = Timer - t2
 Clr_Cache03 = Timer
  ClearFileCache($sFile, strNull)
  Clr_Cache03 = Timer - Clr_Cache03
 
  Local m$
 
    M$ = Using$(" Writing took = ##.##", Write_01) + $CrLf + $CrLf  + _
         Using$(" Clearing First Cache = ##.##", Clr_Cache01) & $CrLf & _
         Using$(" First Read = ##.##",t1) +  $CrLf  + _
         Using$(" Clearing Second Cache = ##.##", Clr_Cache02) & $CrLf & _
         Using$(" Second Read = ##.##",t2) & $CrLf & _ 
         Using$(" Clearing Third Cache = ##.##", Clr_Cache03)
    ? m$
    Clipboard_Set_Text(m$)     
 End Function
 '

[David Roberts]

Blimey. Thanks Gösta.

[Gösta H. Lovgren-2]

Dave, your comment sparked a memory so I looked Paul Squires JF forum and found this:

http://planetsquires.com/support/ind...b&topic=1836.0

It's a quirk in JFP. Can't have a comment on the same line as Compile:

Compile Exe ' 8.04 '<== won't run after compile


Compile Exe '<== will run after compile

================================================================

I do not feel obliged to believe
the same God who has endowed us
with sense, reason, and intellect
intended us to forgo their use.
~ Galileo Galilei
================================================================

[David Roberts]

I don't believe this - I'm sure I've done it before - I must check.

A comment following #COMPILE.EXE is getting treated as a compile only in JellyFish when Build and Execute is chosen. I added the comment just before posting.

[Gösta H. Lovgren-2]

>
That is exactly what my GetFileSize does but if the system handle is available why use a sledgehammer when three lines of PB does the job?

Because, .... errrr.... ahhh .... hrrrrummmmph..... REAL men use the API.

================================================================
"Some editors are failed writers,
but so are most writers."
T. S. Eliot (1888-1965)
================================================================

[David Roberts]

> Real Men Don't Need No Stinkin' Handle...Write once, use many....The Power of MACROs.....<insert platitude of choice here>...

That is exactly what my GetFileSize does but if the system handle is available why use a sledgehammer when three lines of PB does the job?

[David Roberts]

> Must be something in your code that causes JellyFish to not run the code after compiling.

I have seen this with JellyFish on a few occasions leaving me to run the exe directly. It used to happen on my old Win98 machine as well as my XP Pro machine. If you load the bas into the PB IDE then it should work with 'Run>Compile and Execute'. I can only think that on editing a non-printable character gets embedded that throws JellyFish but not the PB IDE.

Sometimes, I have rewritten sections of code and deleted the section copied to find that all is well again, removing the offending character without knowing what it was.

It happens so rarely that I've just accepted it especially since, to my knowledge, nobody had reported this behaviour.

> At any rate neat code. (Contest material?)

No, its just a trick and I cannot even prove why it works.

[Gösta H. Lovgren-2]

In the meantime you guys may like to have a look at this to avoid restarting your machines or writing massive files, Gösta.

Am having a peculiar problem with your code, David. I C&P'ed, and while it compiles fine, it doesn't display anything on the screen, it doesn't appear to run, just compiles. I tried sprinkling the code with MsgBoxes but none display.

{just occurred to me} Just chkd the folder and the .exe is there, and runs as advertised. Must be something in your code that causes JellyFish to not run the code after compiling. Have another prog loaded into JF at the same time which compiles and runs as normal. Strange.

I've been playing around with the code a little and found that clearing the cache the first time takes about 2 seconds but clearing the cache the second time (after reading the file) is effectively instantaneous (.02 if at all).

At any rate neat code. (Contest material?)

[Michael Mattias]

Real Men Don't Need No Stinkin' Handle...Write once, use many....The Power of MACROs.....<insert platitude of choice here>...

MACRO EQ32(a,b) = (BITS???(a)=BITS???(b))

MACRO FUNCTION ANY_FILESIZE (szFile) 

   MACROTEMP 32, hSearch , fsize 

   DIM w32 AS WIN32_Find_Data, hSearch AS LONG, fsize AS LONG
   hSearch       = FindFirstFile (szFile, W32)
   IF eq32(hsearch, %INVALID_HANDLE_VALUE) THEN
      fsize = -1&   ' "not found" 
   ELSE
      FindClose hSearch
      fSize =  w32.nFilesizeLow
   END IF
END MACRO = fSize

...

   fsize = ANY_FILESIZE (szFile)

(Fails if file open and size is being changed since directory is not updated until file is CLOSEd)
(Also, not written for filesize > 2 Gb)

[David Roberts]

Paul Purvis asked me a little while back why I used a function called GetFileSize instaed of LOF and I told him that the handle returned by CREATEFILE didn't work.

Now I realise that I could have simply written

Open Handle hSys As #1
Filelength = Lof(#1)
Close #1

Live and learn.

[Michael Mattias]

While OPEN HANDLE can be used to effect an OPEN with 'non-default' options otherwise not (yet?) controllable thru PB syntax, it's more common use would be passing an open file handle to a function which resides in another code module.

Since PB internal handles (e g "#12" or whatever is returned by FREEFILE) are only valid in the code module in which that handle was obtained, you can't pass it to such a function and expect it to be valid. Instead you do something like...

' MAIN.BAS
#COMPILE EXE 

DECLARE FUNCTION SupportFunction LIB "MyDLL.DLL" (BYVAL h AS LONG) AS LONG

...
  hFile = FREEFILE
  OPEN   "byfile"  FOR OUTPUT as hFile
  hSys  = FILEATTR(hFIle, 2)    ' get system handle
  CALL    SupportFunction ( hSys)
  PRINT #hFile, "End of Report"
  CLOSE hFile

' =================================================
' 
#COMPILE DLL

FUNCTION SupportFunction (BYVAL hSys AS LONG) EXPORT AS LONG

    hFile = FREEFILE
    OPEN   HANDLE hSys FOR OUTPUT AS hFile 

    PRINT #hFile, "Hello World"    ' <<< PB syntax not "Writefile"

    CLOSE hFile     ' does not close underlying system handle

    FUNCTION = %TRUE
END FUNCTION

MCM

[David Roberts]

That is an interesting aspect of RTFM. It is a long time since I read OPEN in detail. It was read on first getting PBWin but I probably glossed over any reference to the API as it meant little too me at the time. It has occurred to me of late that some aspects of the manual may be worth reading again as a very much higher percentage may be absorbed now compared to the initial read. I tend to use the manual when using an aspect of PB that I have not ventured into much, or at all.

On the issue of "buffersize and alignment requirements" I adhere exactly to the requirements in a working app but the idea used in the link above totally ignores any such requirements. Perhaps I should look at it again to see if the results differ any by adherence. Whilst on this subject I have not been able to find any comments about possible mishaps if such requirements are not adhered to.

[Michael Mattias]

Just for the record....

You don't need to use ReadFile and WriteFile to use the FILE_FLAG_NO_BUFFERING (or any other "non standard" 'open' flags)...

 hSys = CreateFile (......  special flags....)
 hFile  = FREEFILE
 OPEN HANDLE hSys for <PB syntax mode> AS Hfile
 ' -------------------------------------
 ' GET/PUT here 
 ' --------------------------------------
 CLOSE hFile
 CloseHandle hSys

This does not eliminate the buffersize and alignment requirements for the "var" in the PUT or GETwhen using the FILE_FLAG_NO_BUFFERING option , but some might find PUT/GET a bit easier - or at least a bit more familiar - to work with.

MCM