Re: Compression call (CMPSC) questions
- From: graeme@xxxxxxxxxx (Graeme Gibson)
- Date: 26 Mar 2007 11:33:34 -0700
Miklos,
While I am not so familiar with the PDF building process, I would assume that the application you mention must use whatever compression algorithms have been defined by Adobe Systems in their specification of the PDF file structure. It's the Adobe PDF Reader that uncompresses this data and whatever it supports is what must be used. The CMPSC instruction does not strictly constitute an algorithm by itself as the dictionary used (suipplied by the invoker) is an essential part of the "algorithm". I believe it is not a requirement to use compression within a PDF file, so perhaps you could just allow your PDF files to be bigger and thus significantly reduce CPU consumption.
If I have misunderstood your original post and you are actually compressing** entire PDF files after they have been built, then you must consider the environment where these files are going to be decompressed. If they are going to be retrieved by web browsers then it is essential to use a file structure (eg zip file) which is understood (ie. supported) by web browsers.
As to algorithm costs, one of the major CPU costs (around 20%) within most implementations of the DEFLATE algorithm is generating the 32 bit CRC value which is a required part of every ZIP file member. The DEFLATE algorithm itself does not require this CRC (it's part of the ZIP file architecture) but I think ZLIB generates it as a matter of course. Having said that, I would not be surprised if the PDF file use of compression requires a CRC value.
It may be that you should consider off-loading the PDF creation process to another platform and then shipping the product (a PDF file?) back to the mainframe, if that is where it is required. While I realise that "off-loading" may be an unpalatable idea to many of us on this list, if you have a cpu-intensive process pushing you towards a capacity increment then it makes sense to at least investigate that approach. Keep in mind that each mainframe capacity increment, with all its associated increased software costs, may be adding ammunition to someone else's case for "moving entirely off the mainframe".
Regards,
Graeme
** It's usually not very useful to try compressing PDF files as they often contain compressed data so that little additional compression can be achieved.
At 01:49 AM 27/03/2007, you wrote:
Hi Greame
Thank you vey much.
I see you are a compresion specialist,
Maybe you have some suggestion for a faster compression ?
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to listserv@xxxxxxxxxxx with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
.
- Prev by Date: VLF vs ISC for catalogs
- Next by Date: RE: ICSF on z890?
- Previous by thread: Re: Compression call (CMPSC) questions
- Next by thread: New real storage management/page stealing in z/OS V1R8
- Index(es):
Relevant Pages
|
