T+a dac 8 dsd

[tsaett]

Reading this I'm now thinking of upgrading my Lampi with the 512 capable board.

My question to the cpu-versed people here, is my xeon e3 1265lv3 equipped CAPSv4 (cpubenchmark rating of 8770, similar to Ted's 3770S) capable of doing 512 upsampling via HQP? I'm currently upsampling everything to 128 including room correction with 12% average load.

I simulated this by just selecting 512 in HQP which gave me a average CPU load of <50%, is this ok for stutter-free playback?

 

Anyway, I'm looking forward to the Sound Galleries Event in Munich next month where I can finally hear the T+A with the SGPC in action.

(Edward, did you have a chance in the meantime to compare Lampi 512 vs the T+A?)

 

Christoph

[tsaett]

Having talked with Lothar Wiemann of T+A at the SGM Event, he explained to me that the main difference between the DAC8DSD and the PDP3000 is - besides the CD slot - that the PDP3000 actually has 2 different individually optimized output stages for DSD and PCM, whereas the DAC8DSD only has one.

From what I heard at BMW Welt, the output stage of the DAC8DSD seems then optimized for DSD, rather than PCM.

[tsaett]

The volume control works in the analog domain, the quality of the signal therefor should be independent from the volume.

 

I also have a DAC8 for testing at home right now, driving my Gryphon Antileon Signature directly.

Initial impression is stunning, clarity and resolution is top notch. I am only capable of using the -2s filters for 512 at the moment but am planning to upgrade my CAPS with more cpu power.

Comparing PCM with DSD took less than a minute, no contest at all. This confirms what I heard at the tech session at the SGM event in Munich.

I'm sure the SGM would take things even further.

 

Comparing DAC is a DSD128 Lampi with preamp, so far I like the T+A better overall.

[tsaett]

Al,

 

yes I should have been more clear.

I meant the Lampi DSD Dac - which basically is a Level 4 without PCM - with built in analog volume control, so a fair game.

Both are fully balanced.

[tsaett]

Al,

to be fair to the Lampi, it was not a A/B test as a day prior to the arrival of the T+A I sent it to Poland for service.

But the time was short enough to rely on my aural memory.

 

In a comparison between Lampi 128 and T+A 128 it would probably come to a draw. The Lampi might have the edge on density, colour and sense of flow, the T+A on dynamic, space and noisefloor (so classic tube - ss differences).

With the T+A doing 512 things change. I would say it equals out where the Lampi had the edge before but intensifies it's core strengths to an impressive degree.

And that with -2s, I'm sure the non 2s filters would take things further still.

 

Christoph

[tsaett]

Chrille,

I think you missed the point here.

 

First, I think it's snobbery to state that only with large scale classical a sq judgement can be made. That is just nonsense.

At the SG event you mentioned, pure acoustic music was part of the menu, although no classical.

For me this was perfect as I do not listen to classical and therefor would not have been able to adequately judge the sound quality (which was excellent). For you it might not have been perfect, granted. For me it was.

 

Second, the PCM - DSD debate, especially between Chord and Jussi of HQPlayer is already worn out.

Both have their reasonable arguments, both are right in a way.

Of course there is no native 512 material, the trick is in the post processing (as does Rob Watts in his FPGA btw.)

 

Your Dave is optimized for PCM, no secret here.

The T+A is good with PCM, I tested it in my system. But it really shines with DSD, even DSD64. No contest at all.

 

The point is: There is no right or wrong here.

Get HQPlayer, get a T+A and compare with the Chord.

[tsaett]

But you are not the only ones that fail to play music that really matters in highend HI FI, basically everybody else does the same.

I have heard some of the theoretically best systems on the market like Gryphon Pendragon speakers with appropriate amping demoed with hip hop and low res muzak.

I don't go to HI FI shows because virtually nobody plays acoustic music anyway and particuarly not large scale classical . It scares away the pop crowd is an excuse I have heard more than once in the past.

A sad state of affairs imo .

 

I believe folks play pop music at shows to scare away people, that think their music is the only music that really matters in hifi.

 

(Couldn't resist)

Cheers, Christoph

[tsaett]

My T+A loaner is gone again, but I am happy to report that Edward aka EuroDriver is visiting me at the end of this week.

He will bring not only his T+A, but also a SGM along. It will be extremely exciting to compare the SGM to my CAPS Pipeline and see what the T+A is really capable of.

 

Unfortunately it seems that my Lampi is not coming back from service on time, so a direct A-B test is probably not possible.

 

I will report.

[tsaett]

Shahed,

your i7 seems identical to my xeon 1265L, so here is my experience:

First you should enable hyper threading and turbo. It is not much but it helps a bit. Also enable the pipeline feature in HQP.

The turbo on all 4 cores maxes out at about 3,1 Ghz which is definitely enough for upsampling everything (including high res PCM) to DSD512 with the -2s family.

I even got 16/44 PCM to 512 using non -2s, but the CPU was totally maxed out then and even opening the explorer caused playback skips.

So for the better non -2s filters, you want something faster. But it will work with the -2s filters which are not bad either.

[tsaett]

You will definitely be absolutely fine with the -2s filters, best for me was polysinc shrt mp-2s with ASDM7.

I was even able to use matrix room correction in HQP while upsampling, so even more demanding.

Yes I used roonserver as the frontend, it added almost nothing to the total cpu load in my testing.

 

It truly would be great to be able to simulate the cpu load....

[tsaett]

I'll think about adding a dummy "freerunning" backend to HQPlayer at some future version for measuring playback speed without actual output....

 

Awesome Jussi, that would be great!

Could you also consider a feature like "when source material is PCM, upsample to DSD512 and when source material is DSD, upsample to DSD256"?

I imagine this to be a great solution for users to keep their existing computers and use the max output possible for each resolution.

Just a thought.....

[tsaett]

What would be the point of doing so? I would rather not add too much more complexity to HQPlayer settings, it is already quite complex. Also testing all the possible combinations is becoming challenging.

 

As an example, my xeon 1265L is fast enough to upsample all PCM resolutions with -2s to dsd512 while simultaneously using matrix processing. It is not fast enough for doing the same thing with dsd source material. However, dsd to dsd 256 does work in this configuration. With this feature I could take advantage of the biggest output mode for every filetype the cpu is capable of and it would probably save you much time when people ask you if their cpu is capable of this or that.

Of course I understand that this adds another layer of complexity and it's just a suggestion for you.

 

(Sorry for OT but this came to my mind while using the T+A)

[tsaett]

[ATTACH=CONFIG]26892[/ATTACH]

 

listen to 'Chocolate Caramel Angel' , Bootsy is slaping in the living room unbelievable !!

large scale classical music will have to wait ...

 

I had the SGM server with the T+A over the weekend and took advantage of Edwards TIDAL account, this track was truly awesome, thanks for that :-)

 

here is the link to my report

http://www.computeraudiophile.com/f10-music-servers/sound-galleries-music-server-25805/index8.html#post551690

[tsaett]

I would like an honest opinion about T+a dac 8 dsd. Would it be better than Mytek Manhattan? How would it compare to Lampi DAC?

 

The T+A was better than my Level 4 in pretty much every aspect.

I didn't compare it to the Mytek but I assume the T+A is ahead of it as well.

[tsaett]

And the way I read the price list, the 3.900€ are before tax, so add roughly 20% on top and you're at 4.700€ in Europe.

[tsaett]

Hi,

I have also a server with an I7 6700 "non K" and I can up-sample all to DSD512 using HQPlayer using the -2 filters. With the normal "non -2" filters (poly-sinc for example), I have stuttering. It is strange because the CPU load is around 45%

 

Hi,

as Miska explained elsewhere, HQPlayer can't use the virtual cores to full extend.

That means, 45% load equal almost 90% of load, so you are right on the edge with your CPU.

As a HQ rule of thumb, roughly multiply the percentage by 2 and you arrive at the actual load.

 

FYI, in my CAPS I use a 1271v3 with 3,6Ghz, 3,9Ghz turbo and 80W TDP. I can upsample everything except DSD to DSD512 with the non -2s filters including room correction.

Caveat is, my HDPlex PSU gets very hot as it's operating on it's limit.

[tsaett]

Great news Bibo, would love to hear a T+A with a sotm sms200 or mR.

Finally, I'm getting my T+A next week. I sold my Lampi, get better sound and even have some cash left over. That's new to me.

 

By the way, do you still want to see a photo of the inside of my CAPS with the additional heatpipes to the left? I'm back home and could make a photo now.

(I visited Ascoli Piceno 2 weeks before the earthquake, so sad what happened there...)

[tsaett]

There you go, not pretty but it does DSD512 fanless and without a 6700k (but with a xeon e3 1271v3).

Without the extra two, the CPU would eventually heat up, now it can play 24/7.

 

Ascoli is a beautiful city, on the square there were 3 weddings going on at the same time!

Watched the scene while drinking my Anisetta and eating Migliori Olive Ascolana.

The whole region was fantastic, especially Castelluccio was out of this world.

[tsaett]

I didn't, that's really awful. Hope they are able to rebuild it, fingers crossed.

[tsaett]

The heat in my CAPS is under control now, but I might try this out in order to reduce stressing my power supply which operates at it's limit and does get very hot.

Does this have a direct effect on the CPU performance? I need every GHz I can get.

 

 

Edit: Nevermind, just looked it up myself, I'm going to try this out, hopefully my HDPlex will thank me for that. Thanks!

[tsaett]

I made my decision based on listening with my CAPS Pipeline without NAA.

Although I had a SGM in my system for a weekend and therefor know what it can do with the T+A, it already works brilliantly with a computer built to normal audiophile standards (e.g. linear power, audio USB card, isolation devices, etc).

 

I have exactly the same expectations for a DAC as Keith, therefor my Lampi was balanced and with volume control as well (although with Mundorf silver/gold/oil caps instead of Duelund).

The T+A did outperform it and did so at a considerably lower price (in Europe) and until then the L4 was always the price/performance champ in the DAC market for me.

The money I gained from the switch I spent on some room acoustic treatment that should be arriving soon.

 

The Atlantic does seem like a great DAC, although with comparable features (balanced, with volume control and PCM/DSD512) and after EU/US taxes, it costs almost 3 times as much as the T+A (based on list price, actual price may be lower).

 

This makes the T+A for me the steal in the current DAC market.

[tsaett]

If you have a standard CAPS-like computer with linear power and audio grade USB card, Intona is the way to go, imho.

I compared the usual AQJB/Regen combo against Intona and in my system the Intona was the winner.

The T+A is less affected by poor sources, but not immune. Only if you bring on the big boys like an SGM, no additional devices are needed.

As a side note, Intona, AQJB and Regen together was not better than Intona alone, in my system at least.