Jump to content
IGNORED

Head to Head JRMC19, Foobar+SACD (and HQ Player) doing Redbook to DSD and native DSD


Recommended Posts

Geoffrey Armstrong of Sound Galleries in Monaco and I did a comparison of players doing Redbook to DSD, Hi Res to DSD and native DSD playback. The results were not what we would have expected a couple of months ago

 

Redbook to DSD conversion - voices and acoustic instruments the big winners, more texture and detail

JRMC 19 Redbook > DSD 128 results are very good (CPU load 13%)

Foobar + SACD in Integer (?) mode > DSD 256 is even better, jaw dropping for me as I have never heard Redbook this good. We almost gave up on Foobar until we switched to non 32 bit Floating Point mode (therefore Integer mode ?) in the SACD plug-in, and then the SQ difference was night and day (CPU load 13%).

 

What was also interesting was comparing Foobar+SACD converting to DSD64 (waste of time, a wet blanket), DSD128 very good, slightly better than JRMC19, and then superb with DSD256 in a class by itself.

HQ Player is good but the CPU load is much higher (40 %) and this might be negatively impacting the sound we could get from HQ Player, the results were not as good as the other two players. We had expected HQ Player to be the best, further investigation and tweaking is required.

 

High Res to DSD conversion - small gains but not always

Here JRMC 19 was the clear winner converting 24/192 (PCM original source) to DSD 128. The improvement was noticeable and worthwhile.

Surprisingly Foobar + SACD did poorly with 24/192 even though it was going up to DSD256, damaging the sound in parts, and making the overall presentation inferior to the 24/192 being played straight which was jolly good, and thoroughly enjoyable already.

What was disastrous was 24/192 material which had been sourced from original DSD 64. The conversion of this material to DSD128 or DSD 256 was strident and unlistenable. It appears that the record company's effort to produce an attractive sound in 24/192 PCM from a DSD64 original including adding "stuff" to make it sound sharper more etched. This "stuff" then becomes poisonous when it's converted to DSD.

 

Native DSD playback - simply surberb, but watch out for the provenance !

We only had native DSD64 material. Played straight with Foobar and JRMC 19 it sounded surprisingly good. Much better than Redbook converted to DSD 64 and in a similar league to a Redbook converted to DSD 128. Then using Foobar to upsample to DSD128, the sound really sparkled. Then we up sampled to DSD256, and the sound was incredible. It's the best I have ever heard. Better than a Light Harmonic DSD setup I heard 3 years ago at the New York Show (I am sure LH have upped their game since then).

 

A further finding is that DSD64 as a recording format may well be good enough. DSD 64 is not good enough as a playback format, but that is easily handled by software upsampling that seems to have no bad effects what so ever, and then sounds fantastic at DSD256. It seems that there are no sound damaging filters perceptible when the DAC is playing DSD256

 

The SACD track of Kind of Blue was sad, close to awful, the technology of converting to DSD was clearly in the dark ages. We don't know if it was done from master tapes or PCM remasters

 

I have been converted from being skeptical about the hassle and inconvenience of DSD recording, to now fully supporting DSD recording at source. Cookie Marenco is on to something, her ears are telling her something that a lot of male recording engineers and electronics wizzes do not want to hear, decimation is damaging !

pink fish media - View Single Post - DSD Dacs Spreadsheet

 

The other thing we tried with good success was doing low frequency room correction (FabFilter) on the latest 24/192 remaster of Kind of Blue and then conversion to DSD128. Room correction seems to improve PRAT in a big way. We could not stop our feet from tapping. DSD 128 was then the icing, the additional clarity, the track sounded pristine. We got to have our cake and eat it too !

 

Playback hardware and OS notes

Item Audio T1, i5 quad core Ivy Bridge, SSD, Linear power supply > Win 8 ASIO > Exasound E20 > Hypex NCore 400 > KEF Blade

We are just starting to explore the tweaking possibilities with the Item T1. The i5 Quadcore seems to be a good recommendation by Mark Welsh. Powerful enough to handle format conversion, upsampling and room correction on the fly, but still staying comfortably cool and quiet with fanless heat pipe cooling provided by the Streamcon housing. Mark knows his stuff ! Coming up next is trying out Windows Server 2012 Essentials as the OS

 

Software Notes

Foobar 2000 with plug-in “Super Audio CD Decoder” programmed by Maxim Anisiutkin and ASIO Support components

This combination allows conversion of various PCM to DSD64, DSD128, and DSD256 with following a number of options:

SDM Type A through D

and

SDM Type A (FP32) through Type D (FP32)

 

Because Type D (FP32) is the final option and appears to be the most demanding of CPU, we initially tried that expecting it would yield the better results but were severely disappointed. Then we tried SDM Type D, i.e. no Floating Point 32 bit mode, which we think is probably integer mode.

 

DSD up-sampling.

The same Foobar + SACD combo allows DSD64 to be up-sampled to DSD128 or DSD256. The options for this DSD up-sampling are limited to SDM Type A through D integer modes. There are no Floating Point options for DSD up-sampling.

 

Room Correction filters

FabFilter Pro by Clayton Shaw of Spatial Audio,

Dimension EQ

 

Ears and the grey matter between them

Geoffrey is very experienced with over 30 years of serious listening and collecting content in every format that exists. He has his personal PS3 game console to do SACD rips to iso as well as turntables. Geoffrey is big into Pace Rhythm and Timing, and the emotional connection the music can make with the listener. With his previous professional IT career, Geoff is unrelenting in his pursuit of software improving the sonic experience as well as good value to the finest hardware. (He is a Grimm dealer)

 

Me, I am kid that will never grow up. Big into acoustic fireworks and detail, detail, detail. Breathing, fingers sliding over strings, audience whistling float my boat. I am very sensitive to any kind of degradation of high frequencies. I am also pretty timbre sensitive, the harmonics and tones have to sound right.

 

Hope the above is helpful to fellow sound hounds

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment
  • 2 weeks later...

I should update this thread by reporting that after Miska's suggestion on setting's for filters which were less CPU demanding than the one we had been using, we have been getting jaw dropping results with HQPlayer. HQP sound quality is in a class by itself.

 

The poor results we had trying to convert 24/192 to DSD256 we now know we're caused by the CPU getting to highly loaded. Restricting the conversion of 24/192 to DSD128 allows for a splendid sound, the edginess is gone

 

playing around with the noise shaping settings, with find DSD 7 to be amazing on classical performances, you can hear and feel the hall. Amazing stuff !

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment

Very interesting post, EuroDriver.

 

I've wondered a lot about the same issues you raised.

 

I'm sure I missed it but did you use a dac capable of playing DSD256? The T+A DAC8 you show in your equipment list doesn't play DSD256, best as I know.

 

Joel

Link to comment

+1

 

This is an excellent post. Thanks for the time and effort you've put into it.

 

I've been upsampling everything with JRiver to DSD128 and I'm really intrigued to try HQPlayer and DSD256 now. Can you shed some more light on the settings that made the difference between it sounding the worst and then the best?

 

I love JRiver's UI and will most likely continue to use it for library manangement even if I prefer HQPlayer's SQ.

 

HQPlayer's library search and navigation appear (from the screenshots) to be spartan, to put it mildly.

Roon Server: Core i7-3770S, WS2012 + AO => HQP Server: Core, i7-9700K, HQPlayer OS => NAA: Celeron NUC, HQP NAA => ISO Regen with UltraCap LPS 1.2 => Mapleshade USB Cable => Lampizator L4 DSD-Only Balanced DAC Preamp => Blue Jeans Belden Balanced Cables => Mivera PurePower SE Amp => Magnepan 3.7i

Link to comment
Very interesting post, EuroDriver.

 

I've wondered a lot about the same issues you raised.

 

I'm sure I missed it but did you use a dac capable of playing DSD256? The T+A DAC8 you show in your equipment list doesn't play DSD256, best as I know.

 

Joel

 

About 3 months ago Geoff Armstrong played me Redbook tracks converted on the fly to DSD128 which was sent to an Antelope Zodiac Platinum. My Jaw dropped, I ordered a T1 from Item Audio, and an Exasound E20 Mk III. Both of thses units are sitting at Sound Galleries for extensive software testing and experimenting

 

The T+A DAC 8 does a nice job of umpsampling to 24/384 has has good filters but the level of enjoyment is significantly behind the E20 doing DSD. I have not done any A/B testing, but the sound we get out of the Exasound doing DSD256 is simply incredible. Geoff has heard a lot of gear, but he says this is almost better than anything he can remember. The AZP cannot do DSD256 input at the present time, but internally it does a very good job of upsampling DSD 64 and 128 to DSD256

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment
+1

 

This is an excellent post. Thanks for the time and effort you've put into it.

 

I've been upsampling everything with JRiver to DSD128 and I'm really intrigued to try HQPlayer and DSD256 now. Can you shed some more light on the settings that made the difference between it sounding the worst and then the best?

 

I love JRiver's UI and will most likely continue to use it for library manangement even if I prefer HQPlayer's SQ.

 

HQPlayer's library search and navigation appear (from the screenshots) to be spartan, to put it mildly.

 

I will let Geoff come back to you on HQ Player settings, but you need to tell us your PC setup and which DAC you are using.

 

HQP Player filtering choices are very CPU specific (and OS influenced)

 

There is a fairly steep learning curve but the benefit / hassle ratio is way higher than vinyl ;-)

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment
I will let Geoff come back to you on HQ Player settings, but you need to tell us your PC setup and which DAC you are using.

 

HQP Player filtering choices are very CPU specific (and OS influenced)

 

There is a fairly steep learning curve but the benefit / hassle ratio is way higher than vinyl ;-)

 

My DAC is the Lampizator DSD DAC, which I uses the Amanero Combo384 USB board and which I understand is DSD256 capable.

 

My server is a Shuttle XPC running a variant of the Core i7 (I forgot which, but it's at about the middle of the line) and I have 8GB of RAM and an SSD. I have the Paul Pang USB 3.0 card. I'm currently running Windows Server 2012 and AudioPhil's Optimizer with JRiver. My system is dual boot in case I need to run Windows 7/8 instead.

 

Let me know if more granular information is needed, but the system (hopefully) has the computing power for DSD256 upsampling. If not, I'll consider some upgrades to get the improvements you've described.

 

Thanks,

Keith

Roon Server: Core i7-3770S, WS2012 + AO => HQP Server: Core, i7-9700K, HQPlayer OS => NAA: Celeron NUC, HQP NAA => ISO Regen with UltraCap LPS 1.2 => Mapleshade USB Cable => Lampizator L4 DSD-Only Balanced DAC Preamp => Blue Jeans Belden Balanced Cables => Mivera PurePower SE Amp => Magnepan 3.7i

Link to comment
My DAC is the Lampizator DSD DAC, which I uses the Amanero Combo384 USB board and which I understand is DSD256 capable.

 

My server is a Shuttle XPC running a variant of the Core i7 (I forgot which, but it's at about the middle of the line) and I have 8GB of RAM and an SSD. I have the Paul Pang USB 3.0 card. I'm currently running Windows Server 2012 and AudioPhil's Optimizer with JRiver. My system is dual boot in case I need to run Windows 7/8 instead.

 

Let me know if more granular information is needed, but the system (hopefully) has the computing power for DSD256 upsampling. If not, I'll consider some upgrades to get the improvements you've described.

 

Thanks,

Keith

 

We don't know if HQP Player will run happily under Win 2012 Server, so my suggestion is to play around with HQP installed under Win 8 first. I assume that the Core i7 is a quad core, so my suggestion is to turn Hypethreading off in the BIOS of the Shuttle XPC.

 

The tree structure of the library management take a bit of getting used to, I could not get anything to load until Geoff showed me how.

 

It will be interesting to see how HQP > Lampi DSD will compare with the DirectStream on a) Redbook, b) 24/192, c) DSD64 and DSD128

 

Don't hesitate to give us a shout if you run in to a problem

 

A bit green with envy ;-)

Edward

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment
We don't know if HQP Player will run happily under Win 2012 Server, so my suggestion is to play around with HQP installed under Win 8 first. I assume that the Core i7 is a quad core, so my suggestion is to turn Hypethreading off in the BIOS of the Shuttle XPC.

 

The tree structure of the library management take a bit of getting used to, I could not get anything to load until Geoff showed me how.

 

It will be interesting to see how HQP > Lampi DSD will compare with the DirectStream on a) Redbook, b) 24/192, c) DSD64 and DSD128

 

Don't hesitate to give us a shout if you run in to a problem

 

A bit green with envy ;-)

Edward

 

Thanks Edward,

 

It'll be next week when I'm able to get started with it all, but I'll certainly keep you posted. I'm excited to hear HQPlayer for myself as there seems to be a consensus that it offers improved SQ.

 

I really hope that it does work with Windows Server 2012, because that, combined with Audiophil's script, is something I'd not want to do without.

Roon Server: Core i7-3770S, WS2012 + AO => HQP Server: Core, i7-9700K, HQPlayer OS => NAA: Celeron NUC, HQP NAA => ISO Regen with UltraCap LPS 1.2 => Mapleshade USB Cable => Lampizator L4 DSD-Only Balanced DAC Preamp => Blue Jeans Belden Balanced Cables => Mivera PurePower SE Amp => Magnepan 3.7i

Link to comment
I really hope that it does work with Windows Server 2012, because that, combined with Audiophil's script, is something I'd not want to do without.

 

System optimization tools are known to cause problems with HQPlayer because the tools conflict with what HQPlayer is trying to do. So YMMV... My recommendation is to use standard unmodified desktop install of Windows 7 or Windows 8 / 8.1.

 

Server versions of OS are optimized for maximum network throughput which is in direct conflict with multimedia requirements. This is the reason why there are separate server and desktop OS versions.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment

FWIW, I was not as impressed with DSD 128 or 256 via JRiver. I drag and drop from JRiver onto HQPlayer. One extra step for some, but I was used to it coming from Fidelia and then XXHighend.

+1

 

This is an excellent post. Thanks for the time and effort you've put into it.

 

I've been upsampling everything with JRiver to DSD128 and I'm really intrigued to try HQPlayer and DSD256 now. Can you shed some more light on the settings that made the difference between it sounding the worst and then the best?

 

I love JRiver's UI and will most likely continue to use it for library manangement even if I prefer HQPlayer's SQ.

 

HQPlayer's library search and navigation appear (from the screenshots) to be spartan, to put it mildly.

Forrest:

Win10 i9 9900KS/GTX1060 HQPlayer4>Win10 NAA

DSD>Pavel's DSC2.6>Bent Audio TAP>

Parasound JC1>"Naked" Quad ESL63/Tannoy PS350B subs<100Hz

Link to comment

Why would you want to turn off hyper threading? It just brings more resources into play for the OS, and the chip is optimized to use hyper threading.

 

-Paul

 

 

We don't know if HQP Player will run happily under Win 2012 Server, so my suggestion is to play around with HQP installed under Win 8 first. I assume that the Core i7 is a quad core, so my suggestion is to turn Hypethreading off in the BIOS of the Shuttle XPC.

 

The tree structure of the library management take a bit of getting used to, I could not get anything to load until Geoff showed me how.

 

It will be interesting to see how HQP > Lampi DSD will compare with the DirectStream on a) Redbook, b) 24/192, c) DSD64 and DSD128

 

Don't hesitate to give us a shout if you run in to a problem

 

A bit green with envy ;-)

Edward

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment
Why would you want to turn off hyper threading? It just brings more resources into play for the OS, and the chip is optimized to use hyper threading.

 

-Paul

 

The thinking behind the suggestion to turn off Hyperthreading goes something like this.

 

On my Quad core T1, the sound gets very screechy when the CPU load goes over about 35% to 45%. We see that with 4 physical cores available HQ Player is running 2 threads with very high load, and about 8 threads with low load. This results in 2 physical cores highly loaded, and 2 physical cores lightly loaded.

 

The thinking is that the screechy sound is caused by high CPU load delaying the delivery of data to the USB transmitter after a USB data request has been received. To allow prompt delivery of data into the DAC's relatively small sized buffer, it's important to have some cores which are lightly loaded and have idle cycles.

 

According to the documentation HQ Player has been written to distribute it's processing over multiple cores up to 27 in number. If Hyperthreading were enabled, the 4 core CPU would function as 8 virtual cores and HQ Player could well then increase the number of computationally intense threads to 4 or even 6. With 6 highly loaded virtual cores there would only be 2 lightly loaded virtual cores and physically this would be actually only 2 half physical cores. The concern is that the prompt delivery of data after a USB request would be adversly affected

 

We have had great SQ results with Hyperthreading disabled, so I made the suggestion to try Hyperthreading disabled first. The difference in SQ between filters in the middle of the list and the next filter up in CPU load is small, whereas the screeching one gets when the CPU load crosses the USB functional limit is pretty awful.

 

It will need some free time to enable Hyperthreading in BIOS, and then see how many high threads HQ Player is running when working with 8 virtual cores.

 

Item Audio on their Windows PC's always disable Hyperthreading, as in their experience it is beneficial, but they have limited experience with CPU intensive software players.

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment

"Item Audio on their Windows PC's always disable Hyperthreading, as in their experience it is beneficial, but they have limited experience with CPU intensive software players."

 

In all my testing hyperthreading being turned off is a clear improvement. I believe JPlay is actually optimized to use one core. Have not used HQPlayer as of yet though.

Link to comment

"Software Notes

Foobar 2000 with plug-in “Super Audio CD Decoder” programmed by Maxim Anisiutkin and ASIO Support components

This combination allows conversion of various PCM to DSD64, DSD128, and DSD256 with following a number of options:

SDM Type A through D

and

SDM Type A (FP32) through Type D (FP32)"

 

 

 

Thanks for the informative post.

 

In foobar which of the eight settings did you settle on as the best? I read somewhere, not sure though, that B might actually have the best sound quality.

 

regards

Bob

Link to comment

In foobar which of the eight settings did you settle on as the best? I read somewhere, not sure though, that B might actually have the best sound quality.

 

We have to confess since the time 2 weeks ago when we got HQ Player singing, we have spent nearly all of our listening time either with HQ Player, or with JRMC 19 doing PCM room correction first and then conversion to DSD.

 

Room correction does wonders for PRAT. We can not stop our feet from tapping :-)

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment

I am not quite able to get a clear picture of why disabling the superscalar processing that HTT in Intel chips enables is a good thing. Seems rather counter-intuitive.

 

The total amount of work to be done does not change, and if a processor is stalled waiting on resources, using that processor for other work seems like a good idea.

 

-Paul

 

 

The thinking behind the suggestion to turn off Hyperthreading goes something like this.

 

On my Quad core T1, the sound gets very screechy when the CPU load goes over about 35% to 45%. We see that with 4 physical cores available HQ Player is running 2 threads with very high load, and about 8 threads with low load. This results in 2 physical cores highly loaded, and 2 physical cores lightly loaded.

 

The thinking is that the screechy sound is caused by high CPU load delaying the delivery of data to the USB transmitter after a USB data request has been received. To allow prompt delivery of data into the DAC's relatively small sized buffer, it's important to have some cores which are lightly loaded and have idle cycles.

 

According to the documentation HQ Player has been written to distribute it's processing over multiple cores up to 27 in number. If Hyperthreading were enabled, the 4 core CPU would function as 8 virtual cores and HQ Player could well then increase the number of computationally intense threads to 4 or even 6. With 6 highly loaded virtual cores there would only be 2 lightly loaded virtual cores and physically this would be actually only 2 half physical cores. The concern is that the prompt delivery of data after a USB request would be adversly affected

 

We have had great SQ results with Hyperthreading disabled, so I made the suggestion to try Hyperthreading disabled first. The difference in SQ between filters in the middle of the list and the next filter up in CPU load is small, whereas the screeching one gets when the CPU load crosses the USB functional limit is pretty awful.

 

It will need some free time to enable Hyperthreading in BIOS, and then see how many high threads HQ Player is running when working with 8 virtual cores.

 

Item Audio on their Windows PC's always disable Hyperthreading, as in their experience it is beneficial, but they have limited experience with CPU intensive software players.

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment
I am not quite able to get a clear picture of why disabling the superscalar processing that HTT in Intel chips enables is a good thing. Seems rather counter-intuitive.

 

The total amount of work to be done does not change, and if a processor is stalled waiting on resources, using that processor for other work seems like a good idea.

 

-Paul

 

We need free idle cores to respond promptly USB data requests. With Hyperthreading enabled, multi-thread jobs will be completed faster, but it means that during the excution of the job, the cores will be busier, and therefore less able to quickly respond to a USB data request.

 

Bob's experience above supports this theory of the importance of idle / lightly loaded cores

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment

Have you instrumented to verify that this is what is really happening?

 

If the core is busy, it is busy. That won't really change with hyper threading, except you will have more cores to choose from.

 

For example, if a core is waiting on some information to be transferred to memory, a second thread can use the execution engine to service a USB request. In a Quad Core machine, that means there are eight chances at least one core will be able to service the thread, with HTT. Without, it is one in 4, and the workload has not diminished.

 

I'm not arguing your results, just trying to understand why. In other words, if 4 cores handles the workload with plenty of time to service USB requests, then eight cores will be even more likely to do so. The workload isn't going to increase, just flow smoother.

 

Yes, those are "virtual" cores, but a CPU - almost any CPU - spends most of it's time waiting for data or a buss to become available. HTT simply takes advantae of that waiting time that is there anyway.

Now obviously, your observations are in conflict with that reasoning, I am just looking for how to reconcile them. :)

 

Yours,

-Paul

 

 

We need free idle cores to respond promptly USB data requests. With Hyperthreading enabled, multi-thread jobs will be completed faster, but it means that during the excution of the job, the cores will be busier, and therefore less able to quickly respond to a USB data request.

 

Bob's experience above supports this theory of the importance of idle / lightly loaded cores

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment
""Have you instrumented to verify that this is what is really happening?"

 

Ears"

 

I suspect I speak for more than myself when I say, please, not again.

 

Respectfully and hopefully.

 

Joel

 

I hope not also?

 

When we hear differences, or prefer say one cable to another... Are we asking for measurements? :)

Link to comment
Have you instrumented to verify that this is what is really happening?

 

If the core is busy, it is busy. That won't really change with hyper threading, except you will have more cores to choose from.

 

For example, if a core is waiting on some information to be transferred to memory, a second thread can use the execution engine to service a USB request. In a Quad Core machine, that means there are eight chances at least one core will be able to service the thread, with HTT. Without, it is one in 4, and the workload has not diminished.

 

I'm not arguing your results, just trying to understand why. In other words, if 4 cores handles the workload with plenty of time to service USB requests, then eight cores will be even more likely to do so. The workload isn't going to increase, just flow smoother.

 

Yes, those are "virtual" cores, but a CPU - almost any CPU - spends most of it's time waiting for data or a buss to become available. HTT simply takes advantae of that waiting time that is there anyway.

Now obviously, your observations are in conflict with that reasoning, I am just looking for how to reconcile them. :)

 

Yours,

-Paul

 

lets try try to formulate the thinking this way.

 

if it takes 10 milliseconds to complete a heavy multi threaded processing job with Hyperthreading off, it will take just 7 milliseconds to complete the job with Hyperthreading on. During those 7 milliseconds, the the CPU will about 30 % more loaded than they were during the 10 milliseconds when Hyperthreading was off. During the next 3 millseconds the CPU will be close to idle

 

lets assume that there is a USB data request every millisecond, then the first 7 requests are not going to be treated as nicely as the next 3 requests which will be responded to quickly by an idle core or two.

 

this is all conjecture, triggered by the observation that awfully screech sound was heard as the the CPU load increased due to the conversion of a sound file with a higher sample rate which then involved more computation (24/192 > DSD256)

 

This is all speculative, so please take with a pinch of salt ;-) but I am willing to bet a beer that screeching will occur at lower processing loads with Hyperthreading on

 

Mind you, if the USB requests were handled with a higher priority than they seem to be, Hyperthreading on or off would make much less difference and the CPU load could be much higher.

 

Another possibility I can imagine is that Core streaming and similar direct data delivery methods do not work well when a core is switching between two threads

 

The sad state of affairs is that PC chips and Win and MAC OS were not designed with audio timing precision in mind

 

apologies for this rambling post !

Sound Test, Monaco

Consultant to Sound Galleries Monaco, and Taiko Audio Holland

e-mail [email protected]

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...