The Memory player

[Telstar]

i.e how to sell an HTPC for 5 times its value.

http://www.novaphysicsgroup.com/MemoryPlayback.html

The only thing that intrigue me a little is to cache the (enitire i guess) cd tracks from the hd into ram (i guess). I think Vista memory management is good enough, but if not, using a ramdisk can be useful.

I'm posting this for Peter and to know more about the data caching performed by XXHE.

[PeterSt]

Hehe ... well, don't worry. XX plays from normal RAM more than any "device" I know of. So, no dependence on cache from Vista (nor XP), just playing from plain RAM here.

Just to give you some insight ... the major challenge starts when you do not want to load the whole CD into RAM in advance but per track (it would otherwise take time, memory, peak cpu cycles) *and* (!) all needs to play gapless at the same time. Performing some FLAC/MP3 conversion on the fly, dealing with Cue files ... it's all just another dimension.

If you then also understand what happens with UnAttended Playback (what that does to SQ and), combined with the control still available ... then indeed you get the grasp of what 5 times the money means. But it's a hobby, right ?

Thanks,
Peter 

[Telstar]

Hehe ... well, don't worry. XX plays from normal RAM more than any "device" I know of. So, no dependence on cache from Vista (nor XP), just playing from plain RAM here.

Just to give you some insight ... the major challenge starts when you do not want to load the whole CD into RAM in advance but per track (it would otherwise take time, memory, peak cpu cycles) *and* (!) all needs to play gapless at the same time. Performing some FLAC/MP3 conversion on the fly, dealing with Cue files ... it's all just another dimension.

If you then also understand what happens with UnAttended Playback (what that does to SQ and), combined with the control still available ... then indeed you get the grasp of what 5 times the money means. But it's a hobby, right ?

Thanks,
Peter 

Considering the ram usage of XXHE I thought that the whole cd did NOT go into cache. Also because it takes a few seconds to start playing (anything) some sort ou caching was undergoing.

I have never tried unattended playback. I had problems loading my whole library and I'm going to play again with the latest release (I stopped at s-1). Because of Intel delaying the 45nm quadcores I'm on wait for about one more full month and therefore not in a hurry to try every single version of the player.

BTW the Memory Player is not exactly like a HTPC, they just took the case of one, it's more like a DAC with a HD and some flash memory. Not worth remotely the pricetag anyway.

Soolos' device was more of a computer, and overpriced like hell too.

[PeterSt]

Considering the ram usage of XXHE I thought that the whole cd did NOT go into cache. Also because it takes a few seconds to start playing (anything) some sort ou caching was undergoing.

Somehow I can't make this consistent. Anyway, it goes into RAM per track, unless with Cue files, then per Album.
A whole CD would take 4-5 seconds (my system), so one track, say, a half of a second max.

Do note that with older versions of XX and diacritical marks in the path/trackname indeed the whole CD would be preloaded. Not anymore since 0.9t-2.

Peter

[Telstar]

Do note that with older versions of XX and diacritical marks in the path/trackname indeed the whole CD would be preloaded. Not anymore since 0.9t-2.

Peter

Cool!
I will try the latest version on my test machine tomorrow and see if the loading is faster

[Telstar]

...and now an open source "memory player":
http://www.audioasylum.com/cgi/vt.mpl?f=pcaudio&m=27041

The cool thing is that his environment can support XXHE. My htpc isn't ready yet, but some guidelines to build a htpc are good.

[PeterSt]

Hahaha, I don't even need to click that link to know what is behind it. Do you want to destroy your PC ? go ahead ! I tried it (three weeks ago or so, when that thread appeared for the first time (it was deleted the first time)).

No. What this does is completely crazy, but in order to ever understand what I mean you must try it (make a backup first).

The cool thing is that his environment can support XXHE

No way it can. Maybe I should have made a remark about that in that thread, but actually I don't care. The guy behind it is, say, a kind of commercial for his own product, and I could do that too. But, looking at "The memory player" as he calls it, this is only a thread to the existing one you referred to earlier, and this is IMHO over the back of others. It may be personal that I don't like that.

There is no single one way that XXHighEnd could drive another sound producing kernel because of the way it works. Besides that, think of it : where XX creates an environment for itself (Core Appointment, Priorities) the first it would do is destroy the environment of that "memory player" which it so carefully created for itself. No ... I don't think he ever tried.
Btw, might it be the idea that his player makes the sound of XX work (so, the other way around) ... that would be the very last possible.

On another note, I think what he is doing in the base is right. But also, it is the opposite of what XX presents : the most simple handling for the best sound (anyway that's the idea ). On that matter, please keep in mind that what's happening so far is the provision of tools so we can all find out what's best, later resulting in a "one setting" player (maybe with "tweak" button).
If you look at UnAttended Playback you can see what it is going to lead to : nothing but a wallpaper and only those services running necessary to play sound. You will visually see no difference with what happens today (like all the desktop icons disappearing), but what will be running is near nothing. This "AA memory player" implies similar. But now try and look how it does it. .

[Telstar]

Hahaha, I don't even need to click that link to know what is behind it. Do you want to destroy your PC ? go ahead ! I tried it (three weeks ago or so, when that thread appeared for the first time (it was deleted the first time)).

No. What this does is completely crazy, but in order to ever understand what I mean you must try it (make a backup first).

The cool thing is that his environment can support XXHE

No way it can. Maybe I should have made a remark about that in that thread, but actually I don't care. The guy behind it is, say, a kind of commercial for his own product, and I could do that too. But, looking at "The memory player" as he calls it, this is only a thread to the existing one you referred to earlier, and this is IMHO over the back of others. It may be personal that I don't like that.

There is no single one way that XXHighEnd could drive another sound producing kernel because of the way it works. Besides that, think of it : where XX creates an environment for itself (Core Appointment, Priorities) the first it would do is destroy the environment of that "memory player" which it so carefully created for itself. No ... I don't think he ever tried.
Btw, might it be the idea that his player makes the sound of XX work (so, the other way around) ... that would be the very last possible.

On another note, I think what he is doing in the base is right. But also, it is the opposite of what XX presents : the most simple handling for the best sound (anyway that's the idea ).

I posted is as a bit or provocation, as it was this thread for starters

I had read his posts about hardware settings influencing the SQ and found it between ridicolous and silly.
I think that using an external soundcard, USB, or even better Firewire would eliminate all interfecences (like the Fireface, but today I found also a new ESI that retails for under 200€).

Then one can build its htpc using the true and proven cooling and noise directions (that can be found on silentpcreview, not on audioasylum).

But if someone do NOT want to use an external soundcard nor interface their dac to the USB channel, some of those advices can be useful, because they can reduce noise pollution. I prefer to build my HTPC with fans and to leave it on 24/7 (suspend mode) and I would never trust it to be fanless. Besides, his underclocked E2x core can work fanless safely, but not a E6x or Qx. Etc, etc.

Regarding XXHE I think that it is the other way around, i.e. his program would launch it, rather than foobar or winamp. I have some doubts that it can work properly on vista.

In conclusion the only thing that I like is that he managed to use the touchscreen of the zalman case (which is different but not too different than the one used in my soon-to-be silverstone cw03).

I do need to do a lot of trys and I will bother u much more next month when the QX9450 should be in stock.

[edward]

Hey guys, please don't put "cics" down. I have emailed with him many times and he is actually a really good guy. We collaborated on a few ideas when he was putting together his "Art of Building a Computer Transport" manual and it is definitely not ridiculous or silly. The hardware settings most definitely influence SQ. Please do not make fun of it, especially if you have not tried it. I used that "manual" as a guideline when I recently built a new computer, and the difference between my old computer and this one was night and day. (And that includes using the same software, settings, and DAC) The improvements really come in the form of much "cleaner" power supply. For those that are interested, these are the specs of the two computers:

NEW ONE:
Zalman TNN-300 (Case with fanless CPU heatsink and fanless PSU)
Biostar P4M900-M7 SE (Low power motherboard)
No video card! (VGA on motherboard)
Intel E2160 (Low power dual-core CPU; 1.8GHz; 800MHz FSB)
Kingston 1 stick of 2GB (Low power PC2 3200 DDR2 400 RAM)
(Both CPU and RAM run at 200MHz frequency - 1:1 ratio is very important for good SQ and using only 1 stick is also important for low power consumption)
Seagate Momentus 5400.3 80GB laptop HDD (low power, low vibration)
(HDD is powered externally with Granite Digital external PS - leaving only the CPU & Motherboard power connectors being used by the Zalman PSU)

OLD ONE:
Asus P5W DH Deluxe
Intel E6600 (Core2Duo 2.4GHz)
Zalman CNPS9700 LED CPU fan
nVidia 8600 GT
Corsair XMS2 (PC2 6400 DDR2800)
Seagate 80GB Desktop HDD
Seasonic S12 550 PSU

As I said, I used the same software settings (with XXHighEnd) and on both I aggressively disabled many processes/services and other little tweaks and used the same DAC. So in comparing, the only difference was the hardware. Trust me, it was not subtle. What a great improvement in SQ.

Anyway, as for the cMP (memory player), I did try it and indeed it does not work with XXHighEnd. "cics" is dead set on using XP and foobar 0.9.5 and this is where we have a difference of opinion. But that does not make it any less of an innovation. Indeed he never tested it with Vista nor XXHighEnd and I told him I don't think it will be very effective with XXHighEnd (because XXHighEnd already effectively does most of what he is trying to accomplish). The one thing he accomplishes with the "memory player" that I like (and which is most controversial and I'm sure to hear about it from you guys) is something I experimented with: After I begin a playlist with XXHighEnd (unattended), with Process Explorer open, I suspend all the svchost and lsass and explorer. This renders your PC virtually dead, with only the music playing. What a nice improvement in SQ! When my playlist is done playing, I can resume all those services and return to normal. Mind you, this is a truly dedicated audio only PC and I do not do anything else with it, so if it's dead while I listen to music, I don't care.

[PeterSt]

On another note, I think what he is doing in the base is right.

Edward, at writing this I wanted to eloborate and emphasize more on his hardware ideas, but forgot. Also, I wanted that piece of my text softening my earlier remarks, but since I lacked in mentioning the examples, that did not happen. Things are out of balance now in a direction not justified.

That this caused Telstar to be challenged to call things ridiculous and silly is just my fault because I created the atmosphere, though unintentionally as said. If I need to take back more I will, but not including the way your PC gets more or less destroyed if you only try this software like you would try any software. If you're not the best in computers it would render your PC worthless, just because not knowing, or not paying all that much attention on what to do in order to go back to normal once you'e in "cMP mode". But let's say I don't even know that, noticing my cookies disappeared and decoders got unregistered or not working otherwise.

On could call it "silly" (and I don't think this english word necessarily is a negative) to go through all the trouble on the hardware side from the perspective of better sound. However, once you dive into the matter and judge the findings (which seem to be cics' findings for a great deal ?), it may be nothing less than the future route to the best playback possible for SQ, including memory sims using less Watts to name one. But me spending thousands of hours on "memory playback" is as silly the least. Some people are just enthusiasts ...

Maybe worthy to tell, is the funny thing of me being the first responding to that thread when it was posted the first time, where I was rather politely telling that maybe there should be a big warning sign for people with not all that much computer knowledge. I started my response with "Maybe out of all it should not be me responding here as the first", and I ended with "If I could not see your dedication and enthusiasm I would think this is a hoax. But I do see, so I know it is not". That got deleted with the thread ...

[Telstar]

Hey guys, please don't put "cics" down. I have emailed with him many times and he is actually a really good guy. We collaborated on a few ideas when he was putting together his "Art of Building a Computer Transport" manual and it is definitely not ridiculous or silly. The hardware settings most definitely influence SQ. Please do not make fun of it, especially if you have not tried it. I used that "manual" as a guideline when I recently built a new computer, and the difference between my old computer and this one was night and day. (And that includes using the same software, settings, and DAC) The improvements really come in the form of much "cleaner" power supply.

As I said, I used the same software settings (with XXHighEnd) and on both I aggressively disabled many processes/services and other little tweaks and used the same DAC. So in comparing, the only difference was the hardware. Trust me, it was not subtle. What a great improvement in SQ.

Anyway, as for the cMP (memory player), I did try it and indeed it does not work with XXHighEnd. "cics" is dead set on using XP and foobar 0.9.5 and this is where we have a difference of opinion. But that does not make it any less of an innovation. Indeed he never tested it with Vista nor XXHighEnd and I told him I don't think it will be very effective with XXHighEnd (because XXHighEnd already effectively does most of what he is trying to accomplish). The one thing he accomplishes with the "memory player" that I like (and which is most controversial and I'm sure to hear about it from you guys) is something I experimented with: After I begin a playlist with XXHighEnd (unattended), with Process Explorer open, I suspend all the svchost and lsass and explorer. This renders your PC virtually dead, with only the music playing. What a nice improvement in SQ! When my playlist is done playing, I can resume all those services and return to normal. Mind you, this is a truly dedicated audio only PC and I do not do anything else with it, so if it's dead while I listen to music, I don't care.

Hi Edward.
I criticized cics and I will explain you why.

I didnt want to rebut point to point what cics affirm that I believe it cannot effect the SQ. So i didnt on Asylum. I read the original thread of his recommended transport and after awhile i couldn't continue because my brain told me that most of it was snake oil.

First and foremost, are you using an external soundcard? I'm expressly referring to a firewire one, because the usb protocol is quite flawed and therefore providing a cleaner power to (some parts of) the computer can attain an improvement to the SQ. The new ESI is quite cheap, around 200€, just a bit more than the Juli@ he is recommending and would IMVHO remove all the problems about cleaning the power source.

I build computers for myself and friends for many years. Here is my reply point to point to cics recommendations.
-RAM: it consumes so few power that it cannot interfere with voltages and pollute the power delivered to the pci bus (where is the soundcard in cics build) Did you try changing this ALONE in your new setup? 1:1 vs other ratios depend on the chipset and the motherboard, it could be that some aggressive setting could make the system less stable, but not the brand, please - Kingston or Corsair with the same specs cannot make a computer transport sound differently.
-All other hardware tweaks (discrete graphic card, fans, HDD powered externally) do influence the SQ only for internal soundcards and MAYBE at a much lesser extent usb. The improvements comes from three things, not only the clearer power: (1) noise, (2) vibrations and (3) polluting the pci power. I'm also pretty confident that an internal pci-e soundcard would be much more resilient to such problems (i.e. asus xonar, that is also enclosed in an isolating box).
-Software recommendations: those are correct, although Vista is a much better OS regarding memory management. Disabling indexing service and automatic defrag is enough for me. Diskeeper is much smart and it does not interfere for me with sound reproduction.

P.S. You can also try using an UPS that also filters the power line, that should help the PSU to have more stable voltages.

P.P.S. Out of curiosity yesterday I went on granite website and i didn't find any psu.

[edward]

...I believe it cannot effect the SQ...my brain told me that most of it was snake oil.

Well, Telstar, your criticisms are invalid then. Just because you cannot believe something you "read" does not give you the right to criticize. If, however, you have tried specific examples of "tweaks" and have conclusions to the contrary, then that is a conversation I will have with you. But please do not comment on things you haven't tried. And BTW, you are not the only one who has built computers for many years.

-I did in fact compare my Corsair RAM to my Kingston RAM, both with the same manual latency settings and also comparing with each using the settings from their SPD. In all cases the Kingston sounded better and the difference was not subtle. I even performed blind testing with my wife, and she defintely heard a difference as well. I'm sorry you don't "believe" it.
-I am in fact using a USB DAC (see my sig), and I feel that because it is such an entry-level product and probably very sensitive to jitter, that is the reason I hear a difference with every little "tweak" whereas most may find the "tweak" on their setup (with a "higher-end" DAC) to be too subtle.

BTW, I do use a UPS. And FWIW, here is the information for the Granite Digital PS I'm using (item #7130):
http://www.granitedigital.com/index.asp?PageAction=PRODSEARCH&txtSearch=7130&Page=1

And please, Telstar, I will apologize now if this sounds like I'm putting you down - I do not intend to. I want to have an open discussion with you. When you have tried the same hardware things that I have, then I would love to discuss whether or not we hear the same things. I am in no way trying to imply that I'm the expert who has found the perfect solution. This is a learning process for me as well and I am completely open to all suggestions. But one thing I've already learned definitively is that no matter how much you believe that things shouldn't theoretically make a difference - when you talk about SQ on a PC, EVERYTHING makes a difference.

And the bottom line is, if you cannot hear a difference and you feel only certain things are "enough" for you, then that's perfectly acceptable if you are content. But that does not mean that someone else won't hear a difference also (with their ears and their setup).

[Telstar]

...I believe it cannot effect the SQ...my brain told me that most of it was snake oil.

Well, Telstar, your criticisms are invalid then. Just because you cannot believe something you "read" does not give you the right to criticize.

You quoted something I wrote at the beginning of my post, referring to the first cics guide on audio asylum, not to your experience.

If, however, you have tried specific examples of "tweaks" and have conclusions to the contrary, then that is a conversation I will have with you. But please do not comment on things you haven't tried. And BTW, you are not the only one who has built computers for many years.

I was commenting on a scientific basis. We know that music reproduction sometimes does NOT follow the physical laws. So I give you and I gave cics the benefit of doubt.

-I did in fact compare my Corsair RAM to my Kingston RAM, both with the same manual latency settings and also comparing with each using the settings from their SPD. In all cases the Kingston sounded better and the difference was not subtle. I even performed blind testing with my wife, and she defintely heard a difference as well. I'm sorry you don't "believe" it.

It is indeed very difficult to believe.
Excuse me, did the module have the same rating or you put the settings in the bios? i.e. kingston was a cas5 module and corsair was cas4 and you put both at cas5 (obviously u could hardly push the slower one). I'm assuming you did the test keeping all other conditions the same.
BTW you speak of Kingston. Do you know the exact chips that they mount? Within the same model from a big manifacturer like kingston they can use different chips with different rating and they would sound differently like happened with corsair.
Also, how many different motherboards have you tried? Maybe the corsair would sound better with a different chipset. I do not know the biostar mobo that you use, but it should be a G33 or older intel chipset. And generally intel chipsets are very good and stable. Anyway I can assume that your motherboard did not like the Corsair memory module that you used.

I'm assuming that you used XXHE in Vista for this test.

-I am in fact using a USB DAC (see my sig), and I feel that because it is such an entry-level product and probably very sensitive to jitter, that is the reason I hear a difference with every little "tweak" whereas most may find the "tweak" on their setup (with a "higher-end" DAC) to be too subtle.

Quite much has been told about the problems of the usb protocol and its suitability to transport the music signal. Your experiences and cics ones seem to sustain the thesis that USB dacs are very sensitive to. I think you know that the m-audio external soundcard is of quite low quality and probably does nothing to correct the jitter caused by the usb bus.

Oh I have another question: it is powered by the usb cable itself or has a separate psu?

Which is the rest of the audio chain?

And FWIW, here is the information for the Granite Digital PS I'm using (item #7130):
http://www.granitedigital.com/index.asp?PageAction=PRODSEARCH&txtSearch=7130&Page=1

Thanks, I will check them out. I have two HDD and a dvd player to power in my HTPC so I could use one of those.

Lately, I cannot do all the tests that you did, because for my needs I cannot have a fanless low-end system for starters and I dont want two machines, one for music and one for video. I need both the onboard and a discrete videocard. But I could buy a kingston module and another one and test it.
I have a TwinDAC+ which has a good reclock and it is battery powered, but I'm tempted to get either an ESI or rme Fireface (both firewire) and use the digital spdif out instead of the USB. Peter prefers this setting. I do not know. Reduced jitter can be the reason for this.

Some people on audioasylum says that the Fireface 800 sounds more airy and open than the FF400 (my pocket wouldnt like it), so in conclusion anything COULD impact SQ, but we are here to analyze what on a case-by case basis before saying like cics that everyone should build its computer transport that way.

[Edit: corrected typo: soundcard for videocard]

[edward]

Thank you, Telstar, for your open discussion (and the benefit of the doubt). You've made some valid points.

It is indeed very difficult to believe.
Excuse me, did the module have the same rating or you put the settings in the bios? i.e. kingston was a cas5 module and corsair was cas4 and you put both at cas5 (obviously u could hardly push the slower one). I'm assuming you did the test keeping all other conditions the same.
BTW you speak of Kingston. Do you know the exact chips that they mount? Within the same model from a big manifacturer like kingston they can use different chips with different rating and they would sound differently like happened with corsair.
Also, how many different motherboards have you tried? Maybe the corsair would sound better with a different chipset. I do not know the biostar mobo that you use, but it should be a G33 or older intel chipset. And generally intel chipsets are very good and stable. Anyway I can assume that your motherboard did not like the Corsair memory module that you used.

The Kingston I'm using is KVR400D2N3/2G. The SPD rating is PC2-3200 (DDR2400) 3-3-3-8. The Corsair (CM2X1024-6400C4), on the other hand, is rated at as PC2-6400 (DDR2 800) 4-4-4-12. My testing was done comparing both at DDR2 400 3-3-3-8, with voltage at 1.8 (So, yes, I manually changed the settings in the BIOS for the Corsair). I wouldn't think under-clocking the Corsair would make it "less stable" and hence the sound difference, but who knows. And, yes, ALL other conditions were identical. As far as the motherboards, I tried each of them on the ASUS board (Intel 975X chipset) and the Biostar board (VIA chipset), and in both scenarios the Kingston sounded better. Here's a conversation I had with "cics" about chipset power consumption:
http://www.audioasylum.com/forums/pcaudio/messages/27186.html
And, no, I do not know what chips are used on my Kingston. I shudder to think of how they all sound different. I really wanted to use only 1 stick and I wanted 2GB because I'm using a RAMDisk, and that would allow me to load 2 CDs into the RAMDisk at a time. I also wanted to run the RAM at 200MHz (DDR2 400) to match the CPU, with the lowest latency, so this Kingston model was the only one I found matching those specs.

I'm assuming that you used XXHE in Vista for this test.

Yes, of course.

Quite much has been told about the problems of the usb protocol and its suitability to transport the music signal. Your experiences and cics ones seem to sustain the thesis that USB dacs are very sensitive to. I think you know that the m-audio external soundcard is of quite low quality and probably does nothing to correct the jitter caused by the usb bus.

I don't think it is so much the USB protocol that is the problem as it is the inferior USB to SPDIF converter chip inside the DAC that is usually the problem. USB in itself contains no clock so I'm not sure what you mean about jitter caused by the usb bus. Perhaps it is intrinsic jitter?

Oh I have another question: it is powered by the usb cable itself or has a separate psu?

The M-Audio I have is not bus-powered. It has a separate PS.

...so in conclusion anything COULD impact SQ, but we are here to analyze what on a case-by case basis before saying like cics that everyone should build its computer transport that way.

Well, first of all, he is not charging any money for his information - it is open source, and I don't think he is outright saying everyone should build a computer his way. He is just trying to help. And he is very smart. And he has spent A LOT of time testing this and that. So he just wants to give people a turnkey solution benefiting from his R&D without having to do all the testing themselves. He has also been open to mine and others suggestions and does not insist it has to be his way. For people like you and I, who can do all these tests ourselves and like to "tweak", then this is just a starting block. And in reality, he is just trying to achieve with foobar what we are achieving with XXHighEnd. And I bet you, things like suspending processes while the music playing, would be "easy" for Peter to implement in XXHighEnd. 

[Telstar]

Thank you, Telstar, for your open discussion (and the benefit of the doubt). You've made some valid points.

Hi Edward,

I dont have much to reply. About the ram, it could have been the voltage you used that was too low for the Corsair modules.
Even if rated higher, sometimes the voltage that a module requires to be completely stable is higher as well. It depends on the chips used. For overclockers some series of micron chips were the best last year. I think to remember that cics suggested to raise the ram voltage of 0.1v in one of his posts to improve SQ - that can actually improve the stability.

About USB, in short usb to spdif has intrinsic jitter. Good DACs like the TwinDAC reclock it to minimize the jitter, bad DACs like m-audio do that poorly and they are very susceptible to. Another good option is to use USB to I2S, which if well implemented it is superior to usb-to-spdif, no matter how good is the jitter correction in the DAC.
This is why Peter uses the Fireface 800 (and not only him) to transport the digital signal to the DAC. I have come to the conclusion that the improved SQ in this config is due to the lack of jitter from the USB bus.
I wonder if the cheap ESI Duafire (200€) would be good enough for just this task...

To sum up, you made me think more of power optimization for my HTPC that I dint want to consider (i.e. double psu).
You should really try a firewire DAC, though