Help - Search - Members - Calendar
Full Version: Foobar HDD Usage
Hydrogenaudio Forums > Hosted Forums > foobar2000 > General - (fb2k)
Helios
On one of the forums I read the following opinion about Foobar HDD usage:
QUOTE
I used Foobar for a couple of weeks until i decided to see the hard drive usage, and well it was just through the roof, so I decided to change back to Winamp. To put it into prespective, I left foobar running for an hour and in Filemon, There had been well over 900,000 read writes to the drive, at which rate the drive would have failed prematurely.

I've noticed it too! Foobar makes an enormous amount of I\O Reads (can be seen in Task Manager) comparing to other players. Is it normal?
Yirkha
QUOTE(Helios @ Dec 10 2007, 23:33) *

On one of the forums I read the following opinion about Foobar HDD usage:
QUOTE
I used Foobar for a couple of weeks until i decided to see the hard drive usage, and well it was just through the roof, so I decided to change back to Winamp. To put it into prespective, I left foobar running for an hour and in Filemon, There had been well over 900,000 read writes to the drive, at which rate the drive would have failed prematurely.

I've noticed it too! Foobar makes an enormous amount of I\O Reads (can be seen in Task Manager) comparing to other players. Is it normal?


I don't know what is a "read write", but it's year 2007, we have caches at the filesystem and device level, so nobody actually cares about how the application reads the data. (And when I looked, fb2k read my lossless source in ~140 kB chunks, that's perfectly all right.)
But definitely a nice FUD, thanks for sharing!
odyssey
I'd like to add a thing to this.

While foobar2000 is be multithreaded, I've noticed that it's also multithreading the reads on HDD when you're rescanning the library. Edit: Nonsense rolleyes.gif

Here's a thought: If your HDD is fully defragmented with files sorted alphabetically, would a library rescan not be faster if it was read single-threaded instead, lowering the amount of seek requests?
Peter
QUOTE(odyssey @ Dec 11 2007, 01:15) *
I've noticed that it's also multithreading the reads on HDD when you're rescanning the library.
Media Library rescans are not multithreaded, please don't post nonsense/misinformation.

QUOTE(Helios @ Dec 10 2007, 23:33) *

On one of the forums I read the following opinion about Foobar HDD usage:
QUOTE
I used Foobar for a couple of weeks until i decided to see the hard drive usage, and well it was just through the roof, so I decided to change back to Winamp. To put it into prespective, I left foobar running for an hour and in Filemon, There had been well over 900,000 read writes to the drive, at which rate the drive would have failed prematurely.

I've noticed it too! Foobar makes an enormous amount of I\O Reads (can be seen in Task Manager) comparing to other players. Is it normal?
This has been explained before but I guess one more time won't hurt.
The numbers we're talking about here describe read/write requests sent to your operating system, not sent to your device. This has absolutely nothing to do with risk of hardware failure because the operating system optimizes requests that are sent to your device and caches recently accessed data. If this was a real issue, someone would have noticed such increased hardware failure rate by now; all we have is people reporting numbers from Task Manager, Filemon and such, without real life data to back their claims about harm being supposedly done to their hard drives.
We could easily get rid of this "problem" and add a caching layer, but I'm against adding code that serves no purpose other than keeping clueless people comfortable.
odyssey
QUOTE(Peter @ Dec 11 2007, 02:39) *

QUOTE(odyssey @ Dec 11 2007, 01:15) *
I've noticed that it's also multithreading the reads on HDD when you're rescanning the library.
Media Library rescans are not multithreaded, please don't post nonsense/misinformation.

Sorry, it was an assumption (i think library scan on my drive is a little too noisy when it's all defragmented).
shakey_snake
QUOTE(Peter @ Dec 10 2007, 20:39) *

all we have is people reporting numbers from Task Manager, Filemon and such, without real life data to back their claims about harm being supposedly done to their hard drives.
We could easily get rid of this "problem" and add a caching layer, but I'm against adding code that serves no purpose other than keeping clueless people comfortable.

Further evidence to support my assumption that the task manager is nearly useless for anything except killing tasks.
Peter
QUOTE(shakey_snake @ Dec 11 2007, 18:36) *
Further evidence to support my assumption that the task manager is nearly useless for anything except killing tasks.
I disagree. Task Manager is very useful, especially for developers, but also for advanced users trying to figure out why their system is misbehaving - for an example, "GDI Objects" number can be used to pinpoint the offending app leaking resources in certain scenarios.
Unfortunately, people tend to draw conclusions from its reports without having done proper research and learned what exactly each number is trying to tell them, which results in nonsense/FUD posted on internet forums.
Helios
QUOTE
This has been explained before but I guess one more time won't hurt.
The numbers we're talking about here describe read/write requests sent to your operating system, not sent to your device. This has absolutely nothing to do with risk of hardware failure because the operating system optimizes requests that are sent to your device and caches recently accessed data. If this was a real issue, someone would have noticed such increased hardware failure rate by now; all we have is people reporting numbers from Task Manager, Filemon and such, without real life data to back their claims about harm being supposedly done to their hard drives.
We could easily get rid of this "problem" and add a caching layer, but I'm against adding code that serves no purpose other than keeping clueless people comfortable.

Let me quote an article from MS Technet: "I/O Reads - The number of read input/output operations generated by a process, including file, network, and DEVICE I/Os. I/O Reads directed to CONSOLE (console input object) handles are not counted."
So can you tell the number of I\O reads to device HDD the total number of I\O reads shown in TM? Where is the guarantee that these reads are not HDD reads?

It's not about hardware failure, it's about performance. Why if Winamp or other player makes several I\O reads per second (cache, HDD, whatever), Foobar makes hundreds of I\O reads per second? What for? What's the benefit?
FrozenSpoon
QUOTE(Helios @ Dec 13 2007, 17:45) *
So can you tell the number of I\O reads to device HDD the total number of I\O reads shown in TM? Where is the guarantee that these reads are not HDD reads?

Control panel->Admin Tools->Performance->Add Physical Disk, Reads/sec
eevan
Complete nonsense… On my machine, there are on average 1.1 HDD reads per second while foobar is playing a file from a drive on which I keep my collection. I'm wondering if that guy even knew what numbers he were looking at. mad.gif
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.