HQ Player

[Miska]

10 minutes ago, mirekti said:

@Miska Could you comment on NVIDA Quadro P series, please?
My plan was to add P2000 or P4000, and use them for CUDA offload for i7-7700 when upsampling to DSD512 and non -2 filters. Are these cards good candidates, any comparison to GTX in terms of CUDA?

I would recommend to look at performance test results for double precision FP, if there is no big difference you just end up spending lot of extra money for no gain. At least it seems like one needs P6000 to match GTX 1080 Ti on that front and likely P6000 costs almost 10x more...

P2000 has basically same GPU as GTX 1060 and P4000 has basically same GPU as GTX 1070/1080. But Quadro has larger RAM sizes. However, HQPlayer cannot utilize that much RAM in most typical configurations. Only Quadro GP100 is vastly different in terms of performance, but probably in terms of cost too...

[zorntel]

13 hours ago, DSD256 said:

I think the differences you are hearing are not only due to better mastering. Some sacd feature a complete new multitrack mixing that makes the big difference.

However If i compare the identically rebook layer and sacd layer i'm not able to tell differences when blind listening. We did a really nice blind listening journey with a few music lovers. In the end, the Sacd was more rarely chosen as better sounding. We used ripped SACD's in dsf format.

 

13 hours ago, DSD256 said:

I have somewhere in the neighborhood of 500 SACDs all ripped.  I have both the redbook and DSD rips of all of them and I can't every remember having an instance where I preferred the redbook. Likewise I have lots of high-res PCM and DSD files from the same mastering and likewise almost always prefer the DSD files. Many of these comparisons's were made while I was only upsampling to DSD128 (Auralic Vega) or DSD256 (Mytek Brooklyn), but now I'm upsampling to DSD512 (IFI micro iDSD Black Label) the difference still holds.  Yes, redbook sounds astonishingly good at DSD512...but the DSD files still win...and require a lot less processing.

Robert

 

13 hours ago, DSD256 said:

 

 

 

[mirekti]

28 minutes ago, Miska said:

I would recommend to look at performance test results for double precision FP, if there is no big difference you just end up spending lot of extra money for no gain...

 

Thank you very much. I was aiming P series as it requires less power, but might end up using GTX card.
I understand the more the better for DPFP, but what would be your guesstimate of minimum given the i7-7700 processor so it would make sense adding it? 
For example, if one opts for passively cooled 1050 Ti, how much load it would take off of the CPU? It's DPFP is 62 (91) GFLOPS where 91 is boost.

[Miska]

26 minutes ago, mirekti said:

 

Thank you very much. I was aiming P series as it requires less power, but might end up using GTX card.
I understand the more the better for DPFP, but what would be your guesstimate of minimum given the i7-7700 processor so it would make sense adding it? 
For example, if one opts for passively cooled 1050 Ti, how much load it would take off of the CPU? It's DPFP is 62 (91) GFLOPS where 91 is boost.

I certainly wouldn't go below GTX 1060, the performance drops so much.

I have ASUS Strix GTX 1080 on the "big machine" the card is huge so it needs very spacious computer case, but it is also quiet due to big heat sinks and smart fans. There are other Strix-series models too.

On my Linux workstation I used to have Maxwell-generation Quadro, but switched it to a GTX 1060.

[mirekti]

53 minutes ago, Miska said:

I certainly wouldn't go below GTX 1060, the performance drops so much.

 

Ok, I think all is clear now. 
I will wait for my DSD512 capable DAC and see how will CPU perform on its own. 
Given my case size and power supply, the only option would be P4000 which would put me somewhere between 1060 and 1070 (I tried to make correlation between GTX and P series by comparing single precision floating point values and estimated DPFP values).
It will be interesting if it is possible to mod it to use Streacom's case side heatsink and put Noctua 120mm ULN fan under it, but that's for another topic.

Thanks!!!

[Le Concombre Masqué]

20 hours ago, Miska said:

It is best to use the format used for the original recording, when possible. When it is not possible, using something as close as possible is preferable.

In any case, HQPlayer has been designed to deliver best possible sound from any source format to any kind of DAC.

I understand you recommend that no or as little as possible alteration might have been applied ; very sound

 

Regarding SACD sourced material, taking into account that I apply convolution via a 32/192 .wav, SAVE engine, that the cpu load is #19% RB to 128 (xtr 2s) and #38% for 64 to 128, which layer do you recommend ?

 

 in case complexity matters, convolution setting file Left channel : 

Equaliser: Generic
mai 21 11:15:57
Filter  1: ON  PK       Fc    38,6 Hz  Gain  -9,3 dB  Q  2,18
Filter  2: ON  LS       Fc    70,0 Hz  Gain   3,0 dB
Filter  3: ON  PK       Fc     155 Hz  Gain  -5,8 dB  Q  5,80
Filter  4: ON  PK       Fc     245 Hz  Gain  -2,8 dB  Q  5,00
Filter  5: ON  PK       Fc     637 Hz  Gain  -8,1 dB  Q  3,06
Filter  6: ON  PK       Fc     891 Hz  Gain  -4,7 dB  Q  5,00
Filter  7: ON  PK       Fc     963 Hz  Gain   3,0 dB  Q  5,00
Filter  8: ON  PK       Fc   1 097 Hz  Gain  -5,7 dB  Q  5,00
Filter  9: ON  PK       Fc   1 235 Hz  Gain  -4,0 dB  Q  5,00
Filter 10: ON  PK       Fc   1 464 Hz  Gain   0,6 dB  Q  5,00
Filter 11: ON  PK       Fc   1 903 Hz  Gain  -6,4 dB  Q  5,00
Filter 12: ON  PK       Fc   2 764 Hz  Gain  -5,8 dB  Q  1,60
Filter 13: ON  PK       Fc   4 575 Hz  Gain  -4,1 dB  Q  5,00
Filter 14: ON  PK       Fc   8 908 Hz  Gain  -6,7 dB  Q  1,00
Filter 15: ON  PK       Fc  10 022 Hz  Gain   2,4 dB  Q  5,00
Filter 16: ON  None   
Filter 17: ON  None   
Filter 18: ON  None   
Filter 19: ON  None   
Filter 20: ON  None)

[Miska]

3 hours ago, Le Concombre Masqué said:

Regarding SACD sourced material, taking into account that I apply convolution via a 32/192 .wav, SAVE engine, that the cpu load is #19% RB to 128 (xtr 2s) and #38% for 64 to 128, which layer do you recommend ?

I would certainly stick to DSD->DSD processing as long as you have enough power available to perform it... GPU can help on that front.

[SwissBear]

Hi Miska,

Trying to plan for my next move:

• NAA will be on a small fanless computer, with a very low noise/low impedance LPSU
• server will be outside the room : how important is it to go fanless for the server ? Is it better to have a powerfull server with a good GPU card or to put more money in a fanless one with expensive LPSU ?

[Le Concombre Masqué]

1 hour ago, Miska said:

I would certainly stick to DSD->DSD processing as long as you have enough power available to perform it... GPU can help on that front.

Thank you, it's good to know that I made the right choices collecting hires files and SACD ; once again, kudos, since, especially trough xtr, rb files sound so impressive that I was on the verge of becoming a priest in the 16/44 religion.

 

From your others recent posts, I understand that even though I would lean more towards pcm files, the absolute best dacs I should consider are sigma delta ; am I correct?

[Theobetley]

'A priest in the 16/44 religion' beautiful use of a metaphor to describe your reaction to xtr filters. Bravo!

[DSD256]

15 hours ago, Miska said:

I have commented about the ADI-2 Pro in the corresponding thread on this forum.

I'm not going to spend money on Yggdrasil because it's input is limited to 192k PCM, it's DSP is not bypassable and it has cross-over distortion problems, pretty high correlated jitter, etc. You can check the Stereophile's measurements...

 

Now, after Miska does not want to answer, I must be clear.

It is simply unfair to make a statement like: "Those multi-bit ladders don't achieve the price/performance of delta-sigma DACs though. And even best ones don't come close to best delta-sigma DACs in terms of performance.",

if you do not know what you're talking about. If I allow myself such a disgraceful judgment, this judgment should already be based on much better arguments than I would like to put it: doubtful measurements of a controversial author of a booklet. See thread in superbestaudiofriends.

 

Just because the Yggdrasil got and received such extremely good reviews worldwide, among other things also on computeraudiophile, is extremely well sold and very popular, Miska's proffessional ambition should have been aroused, to clarify what is with the Yggdrasil on itself. Miska reacts defiantly like a child: the DAC can "only" 192khz, the filters can not be circumvented ... etc. That means yes in reverse Miska's filter must be clearly better than the closedform in the Ygg. All this culminates in the campaign that Mike Moffat is not supposed to master the basic structure of a DAC. Alone the assumption is something of silly and bordering on reputation.

A little more respect can be expected and a less arrogance.

 

I sum up:  

Miska's  statement "Those multi-bit ladders don't achieve the price/performance of delta-sigma DACs though. And even best ones don't come close to best delta-sigma DACs in terms of performance."is simply nonsense!!

Because he can not prove this statement if he does not even know the "DAC" of the last years. 
Besides Miska can not simply ignore the overwhelming amount of raving positive reviews from users and professionals about the Yggdrasil.

 

Miska's submissions only allow the conclusion that he is only one person that creates the best filters for DAC's and is the only one who listened properly. Otherwise, he would give the Yggdrasil a chance and listen him.

If outstanding measurements always go with corresponding good sound, RME's Adi-2 Pro would have to be at the very top. But he is not, quite the opposite. Thus the argument of Miska with the dubious measurements of the Yggy only from this aspect is flimsy. Ultimately, it depends on the sound. And with the Yggdrasil, Shiit seems to be doing quite well when Bob Katz comes in blind listening tests with the delta-sigma Antelope Audio Zodiac Platinum DSD DAC Voltikus Power Supply 10M Rubidium Atomic Clock ($ 13,000) to a result of 50:50.

 

 

 

 

[orgel]

20 minutes ago, DSD256 said:

It is simply unfair to make a statement like: "Those multi-bit ladders don't achieve the price/performance of delta-sigma DACs though. And even best ones don't come close to best delta-sigma DACs in terms of performance.",

if you do not know what you're talking about.

 

It's fine if you like the Yggdrasil; a lot of people do. But you really haven't established your credentials to allege that @Miska doesn't know what he's talking about. Here's what I know:

 

DSD256 — 5 posts

 

Miska — 8,361 posts and, oh yeah, he writes what I and many, many others consider to be the best playback software available.

 

So maybe just ease up for a few thousand posts?

 

--David

[Jud]

21 minutes ago, DSD256 said:

 

Now, after Miska does not want to answer, I must be clear....

 

 

It is fine for you to disagree with Miska about the sound of the Yggdrasil, but I would suggest that since this is a thread about Miska's software it would be more appropriate to post further comments about Yggdrasil in a thread where the sound of that DAC is more on topic.

[louisxiawei]

@DSD256 Someone is acting like a child now. 

 

Miska is just expressing his own opinion regarding ygg. Just like everyone can. 

 

If you haven't experienced the advantage of  HQplayer SDM's filter with correct DAC such as the ones listed on the HQplayer.  and obsessed with your own Ygg. Then stick to it, don't bother to broaden your horizon and use ygg for the rest of your life. 

 

As I'm aware, Zodiac Platinum DSD DAC  uses its own dsp upsampling and filter instead of HQplayer's one, so you are not making any sense.

 

Hold your tongue, you come to the wrong place and make the very wrong comment. 

 

I would suggest you educate yourself a bit before making any ridiculous judgment.

 

 

[DSD256]

13 minutes ago, louisxiawei said:

@DSD256 Someone is acting like a child now. 

 

Miska is just expressing his own opinion regarding ygg. Just like everyone can. 

 

If you haven't experienced the advantage of  HQplayer SDM's filter with correct DAC such as the ones listed on the HQplayer.  and obsessed with your own Ygg. Then stick to it, don't bother to broaden your horizon and use ygg for the rest of your life. 

 

As I'm aware, Zodiac Platinum DSD DAC  uses its own dsp upsampling and filter instead of HQplayer's one, so you are not making any sense.

 

Hold your tongue, you come to the wrong place and make the very wrong comment. 

 

I would suggest you educate yourself a bit before making any ridiculous judgment.

 

 

You are mixing up some things. 

Miska makes comment on a DAC he does not know! Simply! And a very bad comment! An opinion is not believing..its about knowing. 

[louisxiawei]

16 hours ago, Miska said:

I would recommend to look at performance test results for double precision FP, if there is no big difference you just end up spending lot of extra money for no gain. At least it seems like one needs P6000 to match GTX 1080 Ti on that front and likely P6000 costs almost 10x more...

P2000 has basically same GPU as GTX 1060 and P4000 has basically same GPU as GTX 1070/1080. But Quadro has larger RAM sizes. However, HQPlayer cannot utilize that much RAM in most typical configurations. Only Quadro GP100 is vastly different in terms of performance, but probably in terms of cost too...

 

 

 

Hey Miska,

 

Got a noob question. Does CUDA offload in HQplayer use single-precision FP or double-precision FP?

 

If double-precision FP matters a lot, does it mean that by adding K20 tesla accelerating card, boosting HQplayer using Nsight hub driver can run HQplayer's most heaviest filters and upsampling better? 

 

Since k20 has much more double precision floating point than 1080Ti.

[DSD256]

18 minutes ago, Jud said:

 

It is fine for you to disagree with Miska about the sound of the Yggdrasil, but I would suggest that since this is a thread about Miska's software it would be more appropriate to post further comments about Yggdrasil in a thread where the sound of that DAC is more on topic.

You are right...regarding the thread. But Miska does not talk about the sound of the Ygg...because obviously he does not know the sound. As he claims, his judgement is based on measurement he did not do.

[Jud]

 

3 minutes ago, DSD256 said:

An opinion is not believing..its about knowing.

 

Very good.  Then in a thread relevant to the topic (not this one, please, which is dedicated to discussion of Miska's software), it would be good for you to discuss what you know about Yggdrasil.

[DSD256]

25 minutes ago, orgel said:

 

It's fine if you like the Yggdrasil; a lot of people do. But you really haven't established your credentials to allege that @Miska doesn't know what he's talking about. Here's what I know:

 

DSD256 — 5 posts

 

Miska — 8,361 posts and, oh yeah, he writes what I and many, many others consider to be the best playback software available.

 

So maybe just ease up for a few thousand posts?

 

--David

What does this have to do with the amount of posts? Please deal with the content. I think I have made my criticism of Miska's statement clearly enough.

[DSD256]

5 minutes ago, Jud said:

 

 

Very good.  Then in a thread relevant to the topic (not this one, please, which is dedicated to discussion of Miska's software), it would be good for you to discuss what you know about Yggdrasil.

Did you read Miska's statements in this thread?

[DSD256]

I'm pulling out of here now ... it happens what was to be expected: who criticizes the Guru here ...
It is as simple as this: I can not judge as a proven expert about a product that I obviously do not know. Furthermore, I do not want to know this product, because it does not correspond to what I would like. This is simply silly and not professional.
Then to judge the multibit Dac's do not come close to the quality of delta-sigma Wanndler is simply nonsense.

 

 

[Bob Stern]

@DSD256:

Miska's criticism of the Yggy was merely an explanation of why it did not appeal to him. Why should he feel obligated to spend his limited time and money to evaluate a product of no interest to him, regardless of how many other audiophiles like it?

If you're as confident in the Yggy's virtues as your extensive defense indicates, you should be sufficiently confident to not need your preference validated by Miska. It's bizarre and childish that many audiophiles get upset when a reviewer or other authority doesn’t share their taste or priorities.

[richard kimber]

Miska:

 

Before I re-install 3.15.1, I wonder if you have any view on the problem I raised with 3.16 here:

(I'm not sure I've mastered the embedded link facility yet )

[Jud]

18 minutes ago, DSD256 said:

I'm pulling out of here now ...

 

Thank you.  This was extensively off-topic.

 

18 minutes ago, DSD256 said:

it happens what was to be expected: who criticizes the Guru here ...

 

Of course this has nothing to do with any "guru."  As Bob just put it quite well, Miska simply described, in terms of measurements and design characteristics - things that are simply factual - why the Yggdrasil does not appeal to him. 

 

I have heard Yggy and liked it; I owned DACs Mike Moffat designed for more than 20 years, so I would guess I have been listening to his designs for considerably longer than you have.  I think he is obviously a very smart man, and I have spoken with him personally at length and like him.  Miska is also obviously very knowledgeable. 

 

Therefore I take what Mike says at Head-Fi and what Miska says here and try to learn from both.  I suggest you try to do the same.  I think you will be rewarded with increased understanding and enjoyment of our hobby.

 

Of course Mike and Miska disagree on some things.  Have all the people you ever learned from always agreed with each other?  If you try to learn from each of them, I don't think this is a problem.

[simonklp]

On Friday, March 17, 2017 at 5:15 AM, louisxiawei said:

 

Hey Bogi,

 

Although I find 3.16 is much better than 3.15.1 in terms of SQ. Less harshness. I still feel the 3.15 has more natural and analog sounding.

 

The explanation of Multicore DSP 3 levels is provided in the manual.

 

Blank~None.

Grey box ~ Auto detection

Tick ~ Multicore DSP enabled.

 

However, I do feel the Multicore DSP difference. Without ticking multicore DSP, DSD512 cannot be played because of serious stutter. (my CPU is quite weak, i7 3770). Fortunately, I have my GTX780 to ease my CPU load.

 

 

On the manual, if the cuda offload is enabled, the multicore dsp is recommended to be swithced on.

 

On Friday, March 17, 2017 at 5:15 AM, louisxiawei said:

 

Hey Bogi,

 

Although I find 3.16 is much better than 3.15.1 in terms of SQ. Less harshness. I still feel the 3.15 has more natural and analog sounding.

 

The explanation of Multicore DSP 3 levels is provided in the manual.

 

Blank~None.

Grey box ~ Auto detection

Tick ~ Multicore DSP enabled.

 

However, I do feel the Multicore DSP difference. Without ticking multicore DSP, DSD512 cannot be played because of serious stutter. (my CPU is quite weak, i7 3770). Fortunately, I have my GTX780 to ease my CPU load.

 

 

On the manual, if the cuda offload is enabled, the multicore dsp is recommended to be swithced on.

I concur with the saything that HQP 3.15.1 has a more natural sound than 3.16, although the sound of 3.16 is more smooth, i.e. less harshness.