HQ Player

[rickca]

5 minutes ago, JTS said:

I have overclocked two cores of the chip in the bios

 

5 minutes ago, JTS said:

I have a gentle overclock on the cache and ram too.

OK, thanks.  You did say

28 minutes ago, JTS said:

Nothing else fancy going on in the BIOS

 

[Allan F]

Is there an HQPlayer 4 Desktop User Manual and, if so, how can one download it?

[fred_com]

5 hours ago, Hammer said:

 

Hi, I'm running Threadripper 1950X and trying to run the EC modulators to no avail...however, I notice that even when stuttering, my core speeds are around 2200 Mhz, well below the 3400 Mhz speed of the processor and its boost speed of 4000 Mhz.  Is this a Windows issue?  Has anyone tried to run the EC modulators on a Threadripper system on Linux?  Or perhaps HQPlayers internal optimizations are tied to Intel processors?

Which tool are you using to monitor the speed? I've noticed that Windows Task Manager is showing some average clock of all the cores. I'm using a HWInfo64, in Summary mode, and there is a nice bar graph that shows each core speed with fast updates.

 

As I've mentioned earlier - I have drop outs on Linux all the time, while on Windows it sometimes works fine with no dropouts, and sometimes there are dropouts. Just like Miska said - seems to depend on the Moon phases

[JTS]

23 minutes ago, rickca said:

 

OK, thanks.  You did say ...

 

 

Well I guess it depends on the definition of fancy :-).

[fred_com]

@Miska I've tried your kernel, and there were dropouts even on the AMSDM7 512+  into DSD512. So I guess 4.14 is too old probably for Threadripper?

 

Also, as a side note, I've had a hard time building a RAID driver on your kernel, all because of the if.h located in /usr/src/linux-headers-4.14.131-jl+/include/linux. Is this some kind of a leftover? As I don't see this file in 4.14 sources in Linus git.

[StreamFidelity]

2 hours ago, rickca said:

Has anyone with an i9-9900K been able to run the ASDM7EC modulators at DSD256 without stuttering?   

 

Yes 

 

On 7/20/2019 at 1:51 PM, StreamFidelity said:

@brother love

 

Many Thanks. I found the mistake with me. Crucial again is the clock frequency, which must be at least 4GHz for the EC modulator ASDM7EC and DSD 256 for my system. Since I run a fanless PC, I come in the temperature in the critical area. At the moment everything is fine, the continuous operation is crucial.

 

poly-sinc-ext2 + ASDM7EC + 44.1 / 24/2 -> DSD 256
Runs smoothly with a very good sound. The CPU utilization is 16%. The power consumption is 34W, which is not a problem for my HDPLEX 200W Linear Power Supply.

 

poly-sinc-ext2 + ASDM7EC + 96/24/2 -> DSD 256
Sensational sound. Also runs smoothly. The CPU utilization is 21%. The power consumption is 39W.

 

Conclusion

For me the EC modulator ASDM7EC is only really good with DSD 256. With DSD 128 I found it too lean.

[rando]

1 hour ago, Allan F said:

Is there an HQPlayer 4 Desktop User Manual and, if so, how can one download it?

 

At least on W10 it shows up inside the program file after install.  No clue in iOS, but it surely lands somewhere obvious and findable.

 

@JTS 4x RAM many use is the only hardware consideration that stands out.  Those slight (necessary) OC/BIOS/power adjustments may seem sophisticated to some here.  You provided some pretty big clues what you are doing differently.  

 

[JTS]

2 minutes ago, rando said:

@JTSThose slight (necessary) OC/BIOS/power adjustments may seem sophisticated to some here.  You provided some pretty big clues what you are doing differently.  

 

 

Understood. I am coming from a spot of reading about people get pretty crazy granular with their overclocks - mine is a less sophisticated "gently turning everything up until it's fast enough without burning anything and without seeing the Blue Screen of Death". I work in film and the overclock is a major workflow optimizer.

 

As I understand now, it also has the added amazing benefit of running the EC modulators at DSD256 without any issues.

[Miska]

3 hours ago, Allan F said:

Is there an HQPlayer 4 Desktop User Manual and, if so, how can one download it?

 

It is included in all installation packages. On Windows and Linux you can find it from the Start-menu in same place where the HQPlayer application shortcut is. On macOS it is straight in the DMG together with the HQPlayer & Client application itself.

 

[Allan F]

3 hours ago, Miska said:

 

It is included in all installation packages. On Windows and Linux you can find it from the Start-menu in same place where the HQPlayer application shortcut is. On macOS it is straight in the DMG together with the HQPlayer & Client application itself.

 

 

Thank you, Jussi.

[Hammer]

8 hours ago, fred_com said:

Which tool are you using to monitor the speed? I've noticed that Windows Task Manager is showing some average clock of all the cores. I'm using a HWInfo64, in Summary mode, and there is a nice bar graph that shows each core speed with fast updates.

 

As I've mentioned earlier - I have drop outs on Linux all the time, while on Windows it sometimes works fine with no dropouts, and sometimes there are dropouts. Just like Miska said - seems to depend on the Moon phases

 

Wow!  Is that what you see when running one of the EC modulators?  I installed HWinfo64 and my results are similar to what I got using CPUID HWMonitor...basically 2200 for all 16 cores and on occasion, I'll get one or two cores that jump up to 2700 Mhz.  Also, running ADSM5EC or ADSM7EC doesn't seem to make much of a difference in terms of the clocks shown.

 

Thanks

[fred_com]

12 hours ago, Hammer said:

 

Wow!  Is that what you see when running one of the EC modulators?  I installed HWinfo64 and my results are similar to what I got using CPUID HWMonitor...basically 2200 for all 16 cores and on occasion, I'll get one or two cores that jump up to 2700 Mhz.  Also, running ADSM5EC or ADSM7EC doesn't seem to make much of a difference in terms of the clocks shown.

 

Thanks

No, that's with AMSDM7 512+ and sinc-M into DSD512. EC modulators are too unstable, so I'm waiting for Jussi's optimizations.

 

One thing to note - I have a power plan set to "High performance", so maybe check what is set in your Power options in Control panel.

[rando]

Have to wonder if given the lack of fluency on the subject of OS/BIOS setting.  A new thread for discussing reasonable limits of tuning up settings is in order.  Generally and for HQP.  

 

My sense is at least a few people with top end mb's and fast processors would benefit from understanding the relationship all of the settings in their BIOS have to different aspects of performance.  Along with comprehending the larger picture of overclocking/underclocking voltage and timings impact on SQ in their own system.

[JTS]

7 minutes ago, rando said:

Have to wonder if given the lack of fluency on the subject of OS/BIOS setting.  A new thread for discussing reasonable limits of tuning up settings is in order.  Generally and for HQP.  

 

It's a good idea. It's a huge can of worms! Perhaps something that outlines the principals and refers people to some of the very well documented materials online. In my own experience, I never run anything with stock settings - even if I am not overclocking, the defaults push too much voltage through everything = heat and noise. I'm in the midst of some killer deadlines and travel right now, but would love to jump into this in September.

[rando]

@JTS It is a rather large and unwieldy subject given the cross section of highly technically proficient members mixed in with those a little lost setting up some of the HQP implementations.  Going to approach CC about this being a subject for article(s) on the front page.  Or if another technical thread would be preferable, do that.

 

I don't have any desire to become tech support either.  

 

[Hammer]

Jussi, given the need for higher performing hardware (more heat/energy use) to run the new EC modulators, would it be possible to add a feature in HQPlayer that prevents windows from sleeping if HQPlayer is playing?  Right now, it appears if I set the sleep timer to 5 minutes, the PC will sleep if I don’t touch the mouse or keyboard for 5 minutes regardless if there is load on the system or not...and hence I think somehow HQPlayer needs to prevent sleep if it is playing.  I say “playing” and not open since it will be nice to resume from sleep with HQPlayer already open.  Seems like the Windows 10 sleep algorithm does not check load on the machine.   Thanks!

[Account Closed]

38 minutes ago, Hammer said:

Jussi, given the need for higher performing hardware (more heat/energy use) to run the new EC modulators, would it be possible to add a feature in HQPlayer that prevents windows from sleeping if HQPlayer is playing?  Right now, it appears if I set the sleep timer to 5 minutes, the PC will sleep if I don’t touch the mouse or keyboard for 5 minutes regardless if there is load on the system or not...and hence I think somehow HQPlayer needs to prevent sleep if it is playing.  I say “playing” and not open since it will be nice to resume from sleep with HQPlayer already open.  Seems like the Windows 10 sleep algorithm does not check load on the machine.   Thanks!

Download and try the free version of Process Lasso from Bitsum. You can set it to prevent sleep in a number of ways and I find that it really helps with workload management. There is a paid version that does more but try the free version first. there is also another tool they have called Park Control which keeps all cores active. It is also free. I have been using both for years.

[Le Concombre Masqué]

1 hour ago, bobflood said:

Download and try the free version of Process Lasso from Bitsum. You can set it to prevent sleep in a number of ways and I find that it really helps with workload management. There is a paid version that does more but try the free version first. there is also another tool they have called Park Control which keeps all cores active. It is also free. I have been using both for years.

Jussi (will correct me if I'm wrong/things changed) advocated against PL when I mentioned it a few years back for we shouldn't mess with HQP's management of priorities . However I can't recall he wrote anything about preventing the parking of cores and I believe I read he sets his machines to high profiles.

 

Bottom line : I use Park Control, not PL

[k6davis]

Barring some kind of overclock that forces the chip to run at its maximum all of the time, the chip loafs when it's not being used and temps and power consumption decrease commensurately. Or at least that's what I thought.

[Account Closed]

2 hours ago, Le Concombre Masqué said:

Jussi (will correct me if I'm wrong/things changed) advocated against PL when I mentioned it a few years back for we shouldn't mess with HQP's management of priorities . However I can't recall he wrote anything about preventing the parking of cores and I believe I read he sets his machines to high profiles.

 

Bottom line : I use Park Control, not PL

I have had no problems using it and as long as you don't go crazy with the priority stuff it does a nice job of restraining a lot of the other junk that win 10 has going on all the time. But, to each his own. YMMV

[Le Concombre Masqué]

2 hours ago, bobflood said:

I have had no problems using it and as long as you don't go crazy with the priority stuff it does a nice job of restraining a lot of the other junk that win 10 has going on all the time. But, to each his own. YMMV

I use Windows Server in minimum  config : about 40 processes only...

[k6davis]

Prior to the release of the (great sounding) EC modulators, I had been running an AMD Ryzen 2700x, listening to poly-sinc-ext2 & DSD7 256fs+ at DSD512 x 48. Unfortunately, that CPU left me unable to listen to any EC modultor at anything above DSD128, so I went shopping. I found a good deal on a i7-9700K/RTX 2080 machine. I also bought a much less expensive GTX 1650 video card to try.

The following are my findings. When I state that something worked, I mean with 24-192 and DSD128 files, at DSD x 48 and for hours straight without high CPU utilization and with low CPU temperatures. I attempted some overclocking but I'm a novice at that, so I ran the CPUs stock, using their automatic clock boost. My testing was done with Windows 10.

Others here are more knowledgeable than I. I welcome their comments and corrections.

• Any kind of EC at DSD512 is currently unattainable (with the hardware I have, at least).

• The i7-9700K/RTX 2080 is capable of any combination of filters & modulators up to DSD256 (EC) or DSD512 (non EC).

• The i7-9700K is capable of poly-sinc-ext2 (Jussi's favorite) and any EC modulator at DSD256 - without a video card.

• If you're listening in 2 channel with no convolution, and want to experiment with heavier filters, the inexpensive GTX 1650 GPU is capable of DSD512 (non EC, of course) with all but poly-sinc-xtr. As Jussi has stated, the GPU helps with filters, but not with modulators. So for my use case, the RTX 2080 (at 4x the price) was no better for upsampling to EC DSD than the GTX 1650. The GTX did have to work a little harder, but it was still very comfortable. I had never tried Cuda. I discovered that it's as easy as installing the video card and clicking the checkbox in HQP. It can be inexpensive and it works very well.

• The AMD Ryzen 2700x was not able to upsample any EC modulator to DSD256. Even with RedBook. Even with the help of a video card. Even when I overclocked it. Whatever AMD's strengths are, their chips are really not good for HQP EC. New and forthcoming AMD chips and Jussi's coding optimizations may change that in the future - or not.

• My current preference is poly-sync-ext2 and DSD5EC, so all I really need for now is the i7-9700K and an Intel motherboard, which I can get for a modest price. The same is true if you prefer ASDM7EC. So I'm leaning toward returning everything I bought and going in that direction.

• When HQP code optimization and/or new, more powerful CPU's make EC DSD512 an option, it will be time to reevaluate.

TL:DR - EC DSD256 is the best sound available now, IMO. Last gen Intel chips do it easily. Last gen AMD chips don't do it at all. If your system can't do EC DSD256 without a GPU, it probably can't do it with one either.

[rickca]

@k6davis thanks those are very useful results.  I had been considering i7-9700K and Ryzen 3700X.  The 3700X is particularly attractive because it's 65W, but I'm not sure whether 3700X will outperform 2700X enough to change the Intel/AMD decision in your use case.  I've looked at the published benchmarks, but they may not be representative of HQPlayer performance.  Ryzen 3700x has twice the L3 cache vs 2700X, but I'm not sure that will do the trick.

[k6davis]

14 minutes ago, rickca said:

@k6davis thanks those are very useful results.  I had been considering i7-9700K and Ryzen 3700X.  The 3700X is particularly attractive because it's 65W, but I'm not sure whether 3700X will outperform 2700X enough to change the Intel/AMD decision in your use case.  I've looked at the published benchmarks, but they may not be representative of HQPlayer performance.  Ryzen 3700x has twice the L3 cache vs 2700X, but I'm not sure that will do the trick.

 

Thanks rickca. It's all very interesting. And not quite what it might seem when you look at it on the surface. There's several angles from which to look at it. 

 

I started out hoping for EC DSD512, at least with RedBook. I was ready to spend a decent amount to get there. When I realized that wasn't happening, my goal switched to a machine that can handle EC DSD256 easily and inexpensively. Swapping out the motherboard & CPU in my current machine in favor of the i7-9700K will do that for me. Adding a modest GPU would get me the more processor intensive filters too. Ultimately, I will do that, but I'm happy with poly-sinc-ext2 for now.

 

As for AMD and their new chips, time will tell. My guess, and that's all it is, is that at best their new high end chips will offer better HQP performance than an i7-9900K, but they will cost more and still not get you to EC DSD512. The efficiency aspect is interesting. We'll have to wait and see.

 

Only thing we can know for sure is that it all remains a moving target. 😂

[k6davis]

There's a simple idea that I didn't get across as clearly as I wanted to in my long post from earlier today. My experimentation broke down to three tiers of servers:

1. i7-9900K: Full EC DSD256 with poly-sinc-ext2 (and other lighter filters)
2. i7-9900K + GTX 1650: Full EC DSD256 with all filters except poly-sinc-xtr
3. i7-9900K + RTX 2080: Full EC DSD256 with all filters including poly-sinc-xtr

Again, I am using 2-channels with no convolution. What I found is that while a GPU is not mandatory, an inexpensive one like the GTX 1650 offers a lot of bang for the buck.

Conversely, spending more money than Tier 2, in my use case, gains me almost nothing. Not until the equation changes with different HQP code or newer chips on the market.