|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
209.130.140.130
In Reply to: Concern about .DLLs posted by clifff on August 22, 2006 at 13:06:15:
I've seen only one post that actually made this claim (and a number that have repeated it). I noticed that the original person did a number of other things at the same time they removed the unnecessary .dll files, so the claim has no merit, in my opinion.I believe it is just a simple misunderstanding by someone (and now a group of people) that doesn't understand what a .dll file is and how it is used. There are many such claims floating around, none of which have really been substantiated.
Until some of these claims are proven by a basic A/B comparison (preferably blind) in which no other parameters changed, they are about as useful as the claim I will make now:
My system has a lot less jitter when there is a full moon!!!
Follow Ups:
As we all know, DACs are sensitive to even small fluctuations in power supply, EMI etc.And as we know, logic transitions may create small power supply ripples or generate EMI.
Therefore it is theoretically possible that loading and unloading DLLs, or even the sequence in which DLLs are invoked, may have a possible impact on the sound. It is also possible that having too many loaded DLLs and not enough memory may cause additional paging activity which may affect the playback.
However, it is certainly a stretch to suggest that actively manipulating the status of DLLs in memory will make a big difference in the sound quality. At least any more than the zillions of other types of memory activity or logic transitions occuring in a computer.
Then again, Microsoft Windows is a highly complex and non orthogonal operating system - who knows maybe there is more here than it seems.
But I think you are playing a bit of devil's advocate. :)Let's keep in mind that the hard drive is probably being frequently accessed anyway while the music data is streamed off of it. And that these .dll files were probably loaded (if at all) before the user had a chance to hit the Play button. And that the memory, if it were that tight, would probably be swapped out at some point early in the process to make room for the music program and its buffers (causing, perhaps, a one time glitch, if any, not an ongoing problem that affects playback).
I say that such claims are akin to saying that the neighbor's electric toothbrush has an impact on your power. Certainly feasible, but at some point you have to draw the line. I am 100% certain that entertaining the thought that such a thing can be a problem will have a far greater psychnological impact on what you hear than the .dll files themselves. :)
... but I didn't want to dismiss it out of hand, however improbable it may sound. One thing I have learnt in my job is never rule out anything, no matter how improbable. One of my "successes" early on in my career was solving a system performance slowdown that was causing our customer to threaten ligitation - and the cause turned out to be something that everyone else had ruled out as being improbable but I was willing to entertain that it *might* be possible.As an example highlighting the complexity of Windows, one of our inmates discovered a few months ago that you can disable Windows Audio Service and still get sound, and hypothesized that this may be a way of bypassing kmixer.
Unfortunately, it turned out not to be true, after a search revealed the explanation by way of a blog from one of the Microsoft guys working on the audio service.
What actually happens is that disabling Windows audio service disables the user mode DLL that normally intercepts and services Windows multimedia calls, but then these calls then result in setupapi.dll being invoked as a fallback mechanism.
The reason for this redundancy is that setupapi.dll provides a set of services invoked by Windows during initial installation, and obviously there is support for audio even during installation.
setupapi.dll is never meant to be loaded in memory in normal operation as it consumes a fair amount of resources. So what our friend did is actually non-optimal and kmixer (which is a kernel service) is not being bypassed at all.
What has any of this to do with DLLs possibly affecting sound quality? Well, not much, but it does illustrate that the workings of Windows is a lot more complicated than most people think.
I wish I knew who started this dll rumor...do you realize how many different processes run at any given time during an active Windows session? To assume that dll's have an impact on the sound quality of music coming out is absurd, given the fact that there are many more intrusive processes being run by executed code, interdependencies, etc.
And this is what's wrong with Windows - bits thrown together.It is my experience that IT programs go wrong most when 'programers' assemble a lot of different 'library' codes without proper understanding or verification. And there are lots of them!
Call me old fashioned, but I believe, and demand in my work, understanding of what various processes do before accepting that they are 'faultless'.
This is where the EU is fining Microsoft so heavilly for failing to provide transparency in the Media Player side of things, making it difficult for others to provide 'fault free' competition.
Hokay, you've pushed a button here.Microsoft gets pilloried/fined/legislated against because they provide facilities such as browser and media player for free, thus hindering other organisations from screwing revenue out of the general public. These other organisations then lobby their politicos.
If the EU had their way, we would all have to buy a third party media player license as well as Windows. And a browser. And doubtless eventually a GUI.
Personally, I don't like Media Player, but I do appreciate that it's presence makes the alternatives considerably less expensive. Think back to what a WP suite used to cost before Office became the de-facto standard, and consider that the alternatives most worthy of consideration are either free or near as damn it.
Before OS/2 & WIndows, operating systems were far simpler, more expensive, and considerably more flaky. So please don't do the knee-jerk thing about Windows. And don't start wittering on about Apple, the last original idea they had was Lisa.
So you think that the crappy MP10 is better than the free fooobar and others. And you think that MP10 as a video player is acceptable compared to much better low cost alternatives.50 million lines of code is RIDICULUS for an operating system in terms of being able to fully comprehend what is going on.
Actually, if you bothered to read my post, I state that I don't like WMP. For audio I use Foobar, for video Powerdvd.How do you know there are 50 million lines of code in XP - have you counted 'em? Or do you just believe what you hear second hand.
Or perhaps mis-read.
Have you any idea of the overheads in providing out of the box support for the range of peripherals that XP handles?
Sorry, I don't usually get this aggressive - just don't like it when people mount a populist hobby horse. Badly.
here...
- http://www.businessweek.com/technology/content/apr2003/tc20030429_6540_tc047.htm (Open in New Window)
This only illustrates my point.1. It must be true 'cos it was in an up-market rag. Presumably written by some clueless hack with an eye for the sensational, and a qualification in English, media studies, or marketing.
2. If you actually READ the article it's about WinServer 2003, not XP.
BTW - please define what a line of code is in this context. Does it include comments, scripts, squiggly brackets, data definitions, picture definitions, help, etc. etc.. Real chaps talk about instructions.
AND, given the broad support for platforms, peripherals, and interfaces, how much of the code is actually relevent at any one time.
Tell you, a fully featured Linux distro makes XP look small. XP is even more scaleable. I have it running on a P133 laptop with 48MB memory & 2GB HD - including office. Not exactly fast, but it does run. Try and do this with KDE or Gnome.
that's the end of it... ;-)
Yet somehow WMP is half the size of Powerdvd 7.Bloatware is the name of the game nowadays, chiefly to make copying/txing as painfull as possible. All suppliers, not just Microsoft.
The MS corporate pack is even bundled so that it requires dual layer dvds for copying.
Look, I'm not saying that Microsoft is the best thing since sliced bread - I don't believe that they are any better, worse, smaller, or more bloated than the viable alternatives. I'm just trying to get people to think, rather than just paraphrasing populist media.
Or, to put it another way: "40 billion flies can't be wrong - eat shit".
... if you are familiar with fmak's posts, he has very fixed views on a lot of things and seldom change them no matter what other people say.I would agree that Microsoft does not necessarily produce code any more bloated than other developers. Though I could possibly suggest I have been less than impressed with some of their code.
Still, I do remember the days where you could comfortably fit a full Unix System V distribution on less than 20MB. And MS-DOS for many years can be installed on a single floppy.
When MSDOS first came out it had a slow take up in the UK 'cos people thought an OS that required 128KB ludicrous.
I remember the Apple II had an operating system (the Apple Monitor) that fits into 2K.By contrast, I rewrote the operating system for the Commodore 64 when I was a kid, borrowing a lot of the concepts from the Apple II, and it fits into 4K.
Here's a bit of interesting trivia: the current Apple MacOS is based on the Mach "micro-kernel" (there was quite a lot of research into distributed operating systems in the 80s, and micro-kernels came out of there) - but the Mach "micro-kernel" is far larger than the completely monolithic statically linked Unix V7 kernel :-)
A lot of the bloat is intentional though, because cheap memory allows you to have algorithms that are memory intensive but computationally cheap. Eg. my paper on Unix password decryption, rather than than doing the DES algorithm step by step, it's far easier to look up a 4MB precomputed table.
A lot of implementations these days do something similar, precompute, and use a lookup table. However, this causes large code.
instead they have bloated backwards compatible crap in there that creates a nightmare for interdependent processes and all the rest of the nonsense...
It's because of the "Microsoft way" - the majority of Microsoft development is done by very bright but recent university graduates. In other words, "mini Bill Gates"They have great productivity, and occasionally fantastic innovation, but it does mean Microsoft tends to have a "not invented here" syndrome and often the same mistakes are repeated again and again by each new team. Also the design tends to be long on "kruftiness" but sometimes short on the bigger architecture picture and longer term maintenance issues are sometimes not well thought through.
Take DLLs for example. Microsoft made the same mistake as Sun in not considering versioning - hence the issues we all face today. At least Sun was able to rectify their mistake - Microsoft took too long and the flawed design became entrenched.
I can understand "not invented here" because one believes one can do a better job, but jeez at least try and learn from your competitors' mistakes! :-)
that's why...
Inertia - changing policy in a large corporate is like turning a supertanker, so with Microsoft it must be like having several supertankers strapped together.Problem is the decision makers are too far away from the workface.
So you now agree with me?No, I do not have fixed views. I respect relevant views and learn from them, not comments that onlt computer codes and bits matter in audio.
Christine, if I have a view on band width and audio, it is because in 30 years of listening and measuring, limited band audio can sound nice, but never as good as wide band audio done right.
No I don't agree with you - it seems you filter every post to suite your requirements. Try reading all the words out loud (under your breath if this embarrasses you) it may help.How do you decide what is relevent?
What has longevity got to do with it - in fact, it could be argued that your training & premises are obsolete.
Certainly you should cultivate listening skills, open-mindedness, tolerance and flexibility. I can point you at suitable training material if you wish.
Just to show you how infantile your arguments are ...For the record, I have been "listening and measuring" for over 35 years. So you have 5 years of experience to catch up on.
Come and and talk to us then!
I made no arguments; if what I post is infantile, then you remarkds are stupid.
Christine,Fmak knows all. Remember this.
40 years in my case. Course this just means we are all hf deaf....
This post is made possible by the generous support of people like you and our sponsors: