Basic Foobar configuration (1st&2nd posts)

[Deleted]

Chong,

I had high CPU usage.... the 22nd of August!!! that is nearly 2 months ago!! It was version 0.2.4 at the time, later versions corrected the problem and the introduction of SSE instructions eased the CPU load quite a bit. With my Core i7 PC my CPU usage never goes above 4% when playing SACD ISOs.

There is no problem in installing as many SOX versions or mods as you want, that is, as long as you only use one at a time. It'll be pointless to have 2 different SOX simultaneously.

If memmory serves me right, Fidelizer is discussed in the Windows 7 thread and explained way better than could at the fidelizer site (www.windowsxlive.net/fidelizer). In short, it optimizes the computer for audio playback at the expense of making it slower for other purposes.

[XTRProf]

Oct 13, 2011 7:54:56 GMT @javier said:

With my Core i7 PC my CPU usage never goes above 4% when playing SACD ISOs.

Well, I have a i5 PC with SACD decoder 0.3.8 and it's still around 20 to 30% CPU usage without the fidelizer process. You have optimised much for your PC?

Oct 13, 2011 7:54:56 GMT @javier said:

There is no problem in installing as many SOX versions or mods as you want, that is, as long as you only use one at a time. It'll be pointless to have 2 different SOX simultaneously.

Any idea how we know it's working?

Oct 13, 2011 7:54:56 GMT @javier said:

If memmory serves me right, Fidelizer is discussed in the Windows 7 thread and explained way better than could at the fidelizer site (www.windowsxlive.net/fidelizer). In short, it optimizes the computer for audio playback at the expense of making it slower for other purposes.

Ah, that's what I need. So will install Fideliser by looking at the Windows 7 thread. Here, right?

Thanks.

[Deleted]

Oct 13, 2011 8:07:13 GMT XTRProf said:

Well, I have a i5 PC with SACD decoder 0.3.8 and it's still around 20 to 30% CPU usage without the fidelizer process. You have optimised much for your PC?

Try changing DSD2PCM mode to floating point under the SACD options panel, that is the SSE optimized mode, fixed point will be heavier for the CPU.

Oct 13, 2011 8:07:13 GMT XTRProf said:

Any idea how we know it's working?

If your sound card's configuration panel shows incoming signal sampling rate, like my Musiland does, just look there. Else, you'll have to find the command to add to the status bar that shows what Foobar is outputing. SOX works, if you add it to the DSP list it'll do whatever you configure it to do. If you make it upsample everything to, say 24/192, that is exactly what you'll get.
It is not just installing it, you have to add it to active DSPs and then click the "configure selected" to set it up.

Oct 13, 2011 8:07:13 GMT XTRProf said:

Ah, that's what I need. So will install Fideliser by looking at the Windows 7 thread. Here, right?

I would recommend to start by the developer's site then read the comments here rather than the other way around.
BTW, Fidelizer is not installed, you have tu run it everytime you want to use it. That is one of its advantages, its changes ARE NOT PERMANENT

[XTRProf]

Oct 13, 2011 8:43:05 GMT @javier said:

Try changing DSD2PCM mode to floating point under the SACD options panel, that is the SSE optimized mode, fixed point will be heavier for the CPU.

I did as that page setting is the same as yours.

Oct 13, 2011 8:43:05 GMT @javier said:

If your sound card's configuration panel shows incoming signal sampling rate, like my Musiland does, just look there. Else, you'll have to find the command to add to the status bar that shows what Foobar is outputing.

Guess I will have to find the strings for that.

Oct 13, 2011 8:43:05 GMT @javier said:

SOX works, if you add it to the DSP list it'll do whatever you configure it to do. If you make it upsample everything to, say 24/192, that is exactly what you'll get.
It is not just installing it, you have to add it to active DSPs and then click the "configure selected" to set it up.

I did add it thru the DSD manger and configured it to upsample to 88.2Khz except 88200;96000;176400;192000 thru mod as the 96khz will give storage error here:

Unrecoverable playback error: Not enough storage is available to complete this operation. (0x8007000E)

Oct 13, 2011 8:43:05 GMT @javier said:

I would recommend to start by the developer's site then read the comments here rather than the other way around.
BTW, Fidelizer is not installed, you have tu run it everytime you want to use it. That is one of its advantages, its changes ARE NOT PERMANENT

No problem. I will do that when boss don't give me much problems. I hope he is still a mad jack at the project groups all the way until TGIF. Heh, heh ............

[XTRProf]

Spoken too soon, now resampler mod having problem with 96khz files with the same error message. May have to take resampler mod out to enable play.

[Deleted]

Are you using a desktop PC or a laptop? 20-30% is quite high for a Core i5.

[XTRProf]

Oct 13, 2011 9:33:05 GMT @javier said:

Are you using a desktop PC or a laptop? 20-30% is quite high for a Core i5.

PC and not laptop as in office. Only home laptop is i5 2nd generation and PC i7 1st generation.

Oh, ya, forgotten to mention as this is a comopnay's PC there are about 70 over processes running in it as well if you care to see the performance Win Task Manager window. So is this still normal? Ya, no least processes here just for audio as the IT guys will come after me if I uninstall the programmes in the PC without administrator rights.

Home ones no problem as those are mine and not company's assets.

[Deleted]

So you meant 20-30% total CPU usage and not just the Foobar process?
It is easy to see Foobar usage if you check the Processes tab of the Task Manager. You can always use affinity and task priority to have Foobar smoothly. You can access those by right clicking individual processes in that tab.

[XTRProf]

I again spoke too soon. Now after 6pm local time, the PC CPU all of a sudden drop to below 10% usage with Foobar running Jazz Side of the Moon hirez and the same about 70 processes running in the background.

Strange?

[XTRProf]

See ...........

[Deleted]

In that pic only agregate CPU usage is shown, unless you click on the processes tab (2nd starting from left) you won't see what Foobar is actually using. High usage you saw before could be anything, an anti-virus background scan, an update being installed, disk defragmentation....

[XTRProf]

Oct 13, 2011 13:05:03 GMT @javier said:

In that pic only agregate CPU usage is shown, unless you click on the processes tab (2nd starting from left) you won't see what Foobar is actually using. High usage you saw before could be anything, an anti-virus background scan, an update being installed, disk defragmentation....

Yes, you are right that it's aggregate CPU usage and I thought you meant that when you say 20-30% CPU usage.

Here for my home laptop, the aggregate is again below 10% as shown here playing DXD 352 Khz 24 bits file downsampled to 88.2khz 24bits with about 90 background processes:





Just Foobar 2000 CPU usage is below 3% for the same home laptop playing the same DXD file as shown here:







Hope I got the correct screenshots.

[Deleted]

Wav files are easy on the CPU even if they are 24/352, SOX can increase the payload but downsampling to a multiple shouldn't be to hard. Try the same but decoding a SACD at 88.2KHz and see those same graphs. It should hover around 5-10% CPU usage for the Foobar process.

[Deleted]

Hi Javier/Chong

interesting stuff guys

take care

[XTRProf]

Oct 13, 2011 16:43:13 GMT @javier said:

Wav files are easy on the CPU even if they are 24/352, SOX can increase the payload but downsampling to a multiple shouldn't be to hard. Try the same but decoding a SACD at 88.2KHz and see those same graphs. It should hover around 5-10% CPU usage for the Foobar process.

Yes, you are right but can't get your 5-10% CPU result for the Foobar. See pictures. Have you optimised your i7 with anything? Anyway to improve to your setup? I'm sure everybody active or on the sideline are eager to know. Shaun for sure. My i5 2nd gen is quite far from yours at about 20-25% CPU for Foobar playing Hotel California dsd iso although I allocated 1.1GB buffer for the largest size track.


OVERALL
JUST FOOBAR

[XTRProf]

Have some time to play with ABX for the first time and so the score for the Eagles PCM and DSD comparison downsampled to 88.2khz for both samples are:







The "errors" are because I'm not familiar with the ABX program at first.

[Deleted]

I haven't optimized anything for my i7, in fact quite the opposite. I've disabled 2 cores in the BIOS so instead of 8 graphs (4 cores + hyperthreading = 8 graphs or virtual CPUs). I'll try to post some pics of it decoding DSD this evening.

One thing to be careful when ABXing files is different mastering and, in the case of SACD ISOs, different output level. Differences in loudness are the easiest to detect. You should try to match the SACD decoder to PCM output level (higher) using the +dB drop dwon list. But it is a fun to play with plugin. Try comparing original masters to a recent "remaster" of the same CD, you'll hear loudness war in its full glory (easy to check also with the Dynamic Range Meter plugin). Another fun test is comparing a CD ripped to wav to those very same wavs compressed to say 320Kbps MP3.

[Deleted]

A new version of the SACD plugin (0.3.10) is available.

Change Log:

10/14/11:
Version 0.3.10 - Gapless playback (of ISO images) control added.

10/02/11:
Version 0.3.9 - Replaygain metadata dependence of "PCM Volume" control fixed.


As I didn't comment on previous one I'll do so now.

Version 0.3.9 brought automatic replay gain based on a SACD metabase that should make the different ISOs have a very similar output level (loudness). Obviously, for this to work Replay Gain has to be enabled and set to track or album. This metabase can be downloaded from the same site as the plugin and should be installed as follows:

1. Copy all files to %appdata%\Roaming\foobar2000\sacd_metabase
2. Refresh the tag info:
Add all your SACD ISOs to a playlist, select them, right-click and do "Reload info from file(s)" under "Tagging".


To download the 0.3.10 plugin:

sourceforge.net/projects/sacddecoder/files/foo_input_sacd/foo_input_sacd-0.3.10.zip/download?_test=goal

To download SACD metabase:

sourceforge.net/projects/sacddecoder/files/sacd_metabase/sacd_metabase_2011.10.09.zip/download

[XTRProf]

Oct 14, 2011 11:19:15 GMT @javier said:

I haven't optimized anything for my i7, in fact quite the opposite. I've disabled 2 cores in the BIOS so instead of 8 graphs (4 cores + hyperthreading = 8 graphs or virtual CPUs). I'll try to post some pics of it decoding DSD this evening.

Yes, please do post to advance the state of understanding of this as we sure like to get to your state. Shaun and me for sure.

Oct 14, 2011 11:19:15 GMT @javier said:

One thing to be careful when ABXing files is different mastering and, in the case of SACD ISOs, different output level. Differences in loudness are the easiest to detect. You should try to match the SACD decoder to PCM output level (higher) using the +dB drop dwon list. But it is a fun to play with plugin. Try comparing original masters to a recent "remaster" of the same CD, you'll hear loudness war in its full glory (easy to check also with the Dynamic Range Meter plugin). Another fun test is comparing a CD ripped to wav to those very same wavs compressed to say 320Kbps MP3.

Yes, I do understand what you mean. But the Eagles dsd one should be from the same 192khz 24bits master as it's from Warner Bros and not from some audiophile label. Usually, the case for these type of projects to reduce cost. Also, I have put the Foobar to play at max out before clipping and so the vol for both should be about the same. There are so many adjustments in Foobar that I also don't know where I had adjusted the last said.

Yeah, my next trial will be to compare ripped wav with the original cd when I have the time. This can be done with foobar right? No, I wouldn't compare mp3 with wav. No fight and challenge at all thru a high enough hirez system. Just imagine comparing higher rez already can hear the difference and mp3? OMG, don't waste my time.

[Deleted]

Ok here are the screen captures of my PC decoding 24/96 FLAC and DSD ISO:


FLAC:




DSD:





I'm suprised by the results, I hadn't checked CPU usage when decoding high resolution FLAC and I didn't excpect it to be higher than decoding DSD. With 24/96 FLAC CPU usage ranged from 4 to 6% while DSD was 1 to 3%.

Even if the Eagles SACD and DVD-A come from the same master they can be encoded (by Warner) and decoded (by Foobar's plugin) at different levels and in fact, as I have them both, I can tell you they do have different levels. When ABX'ed if this is not corrected increasing the SACD plugin's output to match the PCM source you'll get them right 100% of the time with ease.

It shouldn't take long to test a wav and its MP3 version, not much time wasted and you may be in for a surprise, ABX doesn't lie. If it is so easy, at most you'll have wasted 2 or 3 min. Just make sure you use a good MP3 encoder a select 320Kbps CBR as output.

[XTRProf]

Oct 14, 2011 18:07:25 GMT @javier said:

I'm suprised by the results, I hadn't checked CPU usage when decoding high resolution FLAC and I didn't excpect it to be higher than decoding DSD. With 24/96 FLAC CPU usage ranged from 4 to 6% while DSD was 1 to 3%.

Wow, very good result. I (We. Shaun and others interested may also have the same scenario) must study why you can achieve that and I'm still in the 20-30% Foobar CPU range. Call it whatever, voodoo, objective, etc, this will surely advance the state of understand to all interested. Btw, how many background processes you having in the background for that? At least have some idea although not related to Foobar and this type of "voodoo" can can help explain things more clearly. PCM seems to be on track for both of us at about below 4%. Only DSD I (We) can improve.

Btw, the screen shots you had shown were not just the Foobar process but overall atlhough the bulk of the CPU usage should be taken up by the Foobar process.

Oct 14, 2011 18:07:25 GMT @javier said:

Even if the Eagles SACD and DVD-A come from the same master they can be encoded (by Warner) and decoded (by Foobar's plugin) at different levels and in fact, as I have them both, I can tell you they do have different levels. When ABX'ed if this is not corrected increasing the SACD plugin's output to match the PCM source you'll get them right 100% of the time with ease.

Well, like I had say earlier, I had programmed Foobar to play back at max before clipping. But I just can't remember where I did that in Foobar as too many configuration options.

Great! Thanks for sharing to advacne the state of understanding for all.

[philippe]

Hi,

About the foo_input_sacd plugin :

I'd like to know which soundcard and asio driver is able to output the DSD without converting to PCM in order to let my denon 3808 amp decode DSD to analog (DSD pure direct Mode) without any PCM conversion.

Thank's

Philippe

[XTRProf]

Well, tonight have some time before calling it a day to do some more ABX between same CD track and ripped track as said earlier. Although feeling tire and sleepy, I wanted to test my human limit for ABX. Here are the results using just 44 16 music tracks.


FIRST

SECOND




The first result is better as don't feel too much confusion from "data overload" at 6/10 thru repeated comparison non stop.

The second result fare worser as now you can feel the confusion and pressure from "data overload" from too much comparison at 6/11.

Not too surprising were the scores only started to drop after the intial <5.

Of course, these comparisons are thru only a low end HP and laptop internal sound card. If thru my main speaker rig, the results will be different as it's higher rez. I'm very confident it will be much better thru a really hirez system.

CONCLUSION: Tireness and comparsion overload do affect how sharp our ears can determine things. But that doesn't mean we cannot hear the differences. Also, higher rez music files are also easier to hear the difference. All these points to what had been known all along by the subjectivists that we can hear the differences when you are properly experienced and atuned to listening for differences. The higher the rez, to John Atkinson's defination of highend hirez, the better. Also, actually how much a difference between the media counts. Here we are talking of about 5% difference as observed by me between ripped CD and the original CD. To reduce confusion and fatigue further, we must also live with whatever for some time. That will be the real "feeling" we have towards the whatever.

So I had already done my part for ABX, can the rest also do their parts by actually using the ABX in the Foobar and report back as ABX is claimed to be the ultimate test for "correctness". Yeah, post your screen shots here for proof. Don't be afraid as this is just to satisfy your egos. ;D There is no right or wrong or correcter but only discovery about yourself limitations.

Have fun as I do! Goodnight to me! ;D

[Deleted]

Oct 18, 2011 9:16:12 GMT philippe said:

Hi,

About the foo_input_sacd plugin :

I'd like to know which soundcard and asio driver is able to output the DSD without converting to PCM in order to let my denon 3808 amp decode DSD to analog (DSD pure direct Mode) without any PCM conversion.

Thank's

Philippe

Hi Philippe,

I don't know of any such product. It won't be easy or even impossible to find a PCI or USB device outputting DSD thorugh HDMI which is what I guess your Denon wants.

[philippe]

Oct 18, 2011 15:17:01 GMT @javier said:

Hi Philippe,

I don't know of any such product. It won't be easy or even impossible to find a PCI or USB device outputting DSD thorugh HDMI which is what I guess your Denon wants.

Hi Javier,

I'm agree with you.
But I ask this because of this sentence in the "readme.txt" file that's come with the "foo_input_sacd" plugin :

"For direct DSD playback ASIO driver with DSD support is required..."