Jump to content
IGNORED

Best CPU for hqplayer


sbenyo

Recommended Posts

Hi @Miska

 

Do you think even more performance can come from Apple's M1 SoC with even bigger heatsink and bigger fan?

 

Or do you think Mac Mini M1 has probably gotten the most out of M1 possible ?

 

I can't wait to see them go all out on Mac Pro, bigger and loaded SoC and loads of copper cooling and multiple fans

 

First machine to do DSD512 with ASDM7EC ? Even if not, I have a feeling something special is coming with their Mac Pro with Apple Silicon.

 

I use both Win10 and macOS

Link to comment

Aha, I see I really have to upgrade to version 4, 3 can't do this.

 

Does it matter (much) what signal is used for testing?

I can imagine a few minutes of music is enough to generate a reliable weighing of the average speed and if the signal doesn't matter much it's even reliable to compare between different systems.

Didn't expect this to be so easy :-)

 

Thanks a lot Miska, excellent stuff!

Link to comment

To complete and better understand the EC filters.
With an Intel architecture (and Win10), only 2 cores (1 per channel) are used.
We can assume that a processor with a higher base frequency will be more efficient (no turbo).
Example: an Intel Core i5-7500 Kaby Lake Quad-Core 3.4 GHz (LGA 1151 65W) could be better than an i7-
9700T 2.9 Ghz 6 core ?
I am targeting maximum 65w models to keep fanless hope.

Of course, if we can do EC in DSD 128 (I do it with an i5-3470T), we can hope to do DSD256 but never 512 (i9-9900k necessary).
But anyway my RME ADI-2 is limited to 256.

ROON + HQP / Hdplex H3-i5 + 400ATX >Gustard A26 (NAA twk) > SQM > Benchmark AHB2 / Recital Audio Illumine HEFA

Link to comment
12 hours ago, asdf1000 said:

Do you think even more performance can come from Apple's M1 SoC with even bigger heatsink and bigger fan?

 

Or do you think Mac Mini M1 has probably gotten the most out of M1 possible ?

 

I think Mac Mini and likely MacBook Pro make all out of the CPU. Only MacBook Air is likely thermal throttled.

 

12 hours ago, asdf1000 said:

I can't wait to see them go all out on Mac Pro, bigger and loaded SoC and loads of copper cooling and multiple fans

 

First machine to do DSD512 with ASDM7EC ? Even if not, I have a feeling something special is coming with their Mac Pro with Apple Silicon.

 

Only good if they increase the clock frequencies instead of cutting the clock frequencies when adding more cores...

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
4 hours ago, Marco said:

Does it matter (much) what signal is used for testing?

I can imagine a few minutes of music is enough to generate a reliable weighing of the average speed and if the signal doesn't matter much it's even reliable to compare between different systems.

 

No, just always use the same track, since length of the track of course affects how long it'll take to process.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
4 hours ago, Zauurx said:

To complete and better understand the EC filters.
With an Intel architecture (and Win10), only 2 cores (1 per channel) are used.

 

EC modulators can use one or two cores per modulator (channel). But usually on Intel/AMD performance will just suffer if you use more than one core for the modulator due to extra costs of multi-core work split.

 

4 hours ago, Zauurx said:

We can assume that a processor with a higher base frequency will be more efficient (no turbo).
Example: an Intel Core i5-7500 Kaby Lake Quad-Core 3.4 GHz (LGA 1151 65W) could be better than an i7-
9700T 2.9 Ghz 6 core ?

 

What practically matters is what clock speed, with turbo, those two cores end up running at.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
4 hours ago, Miska said:

What practically matters is what clock speed, with turbo, those two cores end up running at.

Thx Miska.
If i understood, an i5-9600 : 3,10 GHz / 4,60 GHz (6 core / 6 threads) and 65w... max performance for this TDP (and my motherboard) is a good choice (perhaps fanless).

ROON + HQP / Hdplex H3-i5 + 400ATX >Gustard A26 (NAA twk) > SQM > Benchmark AHB2 / Recital Audio Illumine HEFA

Link to comment
2 hours ago, Zauurx said:

If i understood, an i5-9600 : 3,10 GHz / 4,60 GHz (6 core / 6 threads) and 65w... max performance for this TDP (and my motherboard) is a good choice (perhaps fanless).

 

The only way to know is to test it out. Usually the smaller chips have for example less cache, so it may impose some additional performance penalties.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
4 minutes ago, salaryman said:

It is 9700K + Z390 itx MB. Magically, it works smoothly in the HDplex case. There is no drop out no overheat, and I am happy to keep it running day and night.

 

Are you monitoring CPU temp (not case temp) ?

 

What is the CPU temp after running for 1 hour, DSD256 with EC7?

 

Link to comment
17 minutes ago, asdf1000 said:

 

Are you monitoring CPU temp (not case temp) ?

 

What is the CPU temp after running for 1 hour, DSD256 with EC7?

 

For 5600X, it is the CPU Temp shown in Ryzen Master.

 

For 9700K, I have not taken CPU temp. It just run smooth, the heatsink of Hdplex H3 is warm only not hot.

There is no overclock needed, everything runs in default, audiolinux + hqplayerembeded + roonserver in Ramboot.

hqp is poly_sinc_ext2 + ASDM7EC

Link to comment
2 hours ago, salaryman said:

For 9700K, I have not taken CPU temp. It just run smooth, the heatsink of Hdplex H3 is warm only not hot.

 

Thanks.

 

If case is not as warm as expected, then this can sometimes be a bad sign (for long term CPU health). 

 

If you get a chance to monitor CPU temp in linux, would be interesting to see after 30mins then after 1 hour.

 

 

Link to comment
10 hours ago, asdf1000 said:

 

Thanks.

 

If case is not as warm as expected, then this can sometimes be a bad sign (for long term CPU health). 

 

If you get a chance to monitor CPU temp in linux, would be interesting to see after 30mins then after 1 hour.

 

 

 

You are right, I install Win10 on it and use HWMonitor, when hqp desktop4 started to play, it ramp up to > 90degC, sometimes reach 100degC. But I kept this to play songs for 2 hours, it just played on, no single dropout, very smooth, sound is very good too, no sign of any overheat or shutdown. 

I think I will keep this on and let it be. Maybe some days it died due to long term high temp, hope it may be the time there are cooler CPU to take up the job.

Link to comment
3 hours ago, salaryman said:

 

You are right, I install Win10 on it and use HWMonitor, when hqp desktop4 started to play, it ramp up to > 90degC, sometimes reach 100degC. But I kept this to play songs for 2 hours, it just played on, no single dropout, very smooth, sound is very good too, no sign of any overheat or shutdown. 

I think I will keep this on and let it be. Maybe some days it died due to long term high temp, hope it may be the time there are cooler CPU to take up the job.

 

Sounds like very hot, if the cooling is supposed to handle the specified TDP, I would look into heat joint installations. That there is enough, but not too much thermal paste and that there's a good contact between cooling parts.

 

But overall, AMD Ryzen CPUs with equivalent TDP run hotter, because they don't have built-in GPU which would be responsible for part of the TDP. Since on Intel CPUs with built-in GPU the GPU is usually not much used, it is off the total TDP budget.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
8 hours ago, salaryman said:

 

I think I will keep this on and let it be. Maybe some days it died due to long term high temp, hope it may be the time there are cooler CPU to take up the job.

 

It sounds like you are a risk taker! I respect that.

 

It's never a good sign if a fanless case never gets warm when powerful CPU is doing a lot of work.

 

So even if you willing to risk this CPU in this case (i don't think any HDPlex cases support 'K' model Intel Core CPU) you can try to cool it better doing what Jussi mentioned. 

 

At least if the case gets warmer then CPU should last longer than now. Still might die but maybe not so soon

 

Actually CPU damage is not only risk in this fanless case. The inside case temp will increase and affect other components also. 

 

You can alway try to save this CPU and move to a big quiet case with big Noctua fan.

Link to comment
8 hours ago, Miska said:

would look into heat joint installations. That there is enough, but not too much thermal paste and that there's a good contact between cooling parts.

I was thinking exactly the same

 

I have your build@salaryman

And I can run EC5 but not EC7 with certain filters and content, under Linux the kernel and AL won't report the Temps, and the intracore load distribution has issues too, I switched to Ubuntu and it works a little better, but yeah Intel is better, but the Temps in my case (measuring with laser) seems to be OK.

 

 

Link to comment
39 minutes ago, luisma said:

under Linux the kernel and AL won't report the Temps

 

Are you sure about that? I am running Ubuntu 20.04 LTS and I have a temp readout of my CPU in my system bar at the top of the screen, and if I click it I see every single sensor the computer has in it with its temps. I installed psensor.

No electron left behind.

Link to comment
55 minutes ago, luisma said:

I was thinking exactly the same

 

I have your build@salaryman

And I can run EC5 but not EC7 with certain filters and content, under Linux the kernel and AL won't report the Temps, and the intracore load distribution has issues too, I switched to Ubuntu and it works a little better, but yeah Intel is better, but the Temps in my case (measuring with laser) seems to be OK.

 

 

 

I think I will re-install all heatsink and heatpipe this weekend.

Last time when I install this 9700K system for trial, it was after uninstall the 5600X system without reapply new paste.

Will update here if there is any improvement in temperature.

Link to comment
57 minutes ago, StreamFidelity said:

 

HDPLEX 2nd Gen H5 Fanless Chassis can cool a 95W TDP CPU such as an i9-9900K CPU very efficiently. With the lid closed, at 4GHz x 16 (Hyperthreading) and DSD 256 with ASDM7EC I hardly get above 60°C. It's also a matter of undervolting in the BIOS. With the lid open, I'm around 50°C and idling at around 35°C. The open chassis can also be an eyecacher. 😉

 

 

39892039dm.jpeg

 

 

Nice! you also have tailor-made cables, those make it tidy and clean.

So you manually fixed the CPU freq to 4G? I thought mine need 4.7G or above to support 256EC7.

BTW, have you applied paste between heatpipe and side panel heatsink?

I don't see this is required in the installation manual, but I have applied to it.

Link to comment
6 minutes ago, Miska said:

 

Streacom or HDplex, there should be thermal paste at both ends of heatpipes. Between CPU and the pipe block, between pipes and the pipe block, between pipes and the mount/heatsink on the chassis.

 

I also do what you say.

But in hdplex's video, really no thermal paste applied between heatpipe and side panel, as shown in the youtube (start: 8:30):

Building a Fully Silent Fanless PC - HDPlex H3 - YouTube

 

Link to comment
3 hours ago, salaryman said:

So you manually fixed the CPU freq to 4G?

 

Yes and it's enough. 👍

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...