Modern (those made in the last 10 years or so) desktop CPUs from AMD and Intel have video logic built in to the CPU. So technically while that one main chip does both video and general computation duties, within the chip itself, it's different parts of the circuit handling those two functions. (It's not completely isolated, as typically the memory controller and system RAM are shared.) A dedicated video card is essentially taking the circuit out of the CPU and putting it on a separate PCB. In high-end gaming cards that circuit is a lot bigger. But for typical desktop duty (basically everything outside of gaming or professional CAD/CAM), the integrated GPUs within modern CPUs are beyond adequate. So you don't really "relieve" the CPU of graphics duty, you just implicitly turn off part of the CPU's circuit.
I haven't done this, but I would expect a CPU-only benchmark to be mostly unchanged when running the same CPU benchmark concurrently with a graphics benchmark. Because, again, it's different parts of the chip's circuit. (Though they certainly could impact each other if there was contention for memory bandwidth.)
If you're playing WAV files---which are literally raw PCM data---the CPU really isn't doing anything, as modern architectures all support DMA. The CPU might initiate the transfer, but basically other subsystems directly copy the PCM data from your storage device to your USB/TOSLINK/SPDIF interface (buffered by system RAM in the middle).
Even encoded formats, FLAC in particular, take truly minimal computational power to decode. It's like asking a world-class strength athlete to wiggle his finger---yes, he's technically doing work but so little that it won't even register on his vitals.
I would not argue with anything you are saying.
All I know is that is does make a difference.
You can get an old used video card, get one with the smallest onboard memory you can find and give it a try. Might cost you twenty dollars. New ones are much too power hungry, which would not make a difference if you power it separately with a riser but that is far more complicated, obviously. Yes, I know the old cards can be power hungry, too.
Are you running your CPU as "slow" as it will remain stable? I used 800 mHz without any problems.
Give it a try.
Any load you can take off the CPU seems to make a difference in my experience. This idea is not mine. The fellow who posted as "Serge" in the old cMP/PLAY thread at AA discovered this video card trick. This was on top of a minimized OS so it might not mean as much with a full on WINDOWS install. I do not know.
It also worked well with WTF the LINUX based combo OS/Player which is the best computer audio I ever heard in my system. See the posts in the PC BASED forum here at DIYaudio for more information.
All I know is that is does make a difference.
You can get an old used video card, get one with the smallest onboard memory you can find and give it a try. Might cost you twenty dollars. New ones are much too power hungry, which would not make a difference if you power it separately with a riser but that is far more complicated, obviously. Yes, I know the old cards can be power hungry, too.
Are you running your CPU as "slow" as it will remain stable? I used 800 mHz without any problems.
Give it a try.
Any load you can take off the CPU seems to make a difference in my experience. This idea is not mine. The fellow who posted as "Serge" in the old cMP/PLAY thread at AA discovered this video card trick. This was on top of a minimized OS so it might not mean as much with a full on WINDOWS install. I do not know.
It also worked well with WTF the LINUX based combo OS/Player which is the best computer audio I ever heard in my system. See the posts in the PC BASED forum here at DIYaudio for more information.
I am using moreDAMfilters NOS filter. Must admit I have not used the other filters other that what came installed on the DAM originally. The "Paul" NOS suited me and I have left it in place.
Not sure the harder leading edge of the phono is correct! So I am happy with what I am hearing for the moment. Time will tell, as it always does ...
have you any any caps above 1uf or regulators left on your lifepo4 powered sd trans or soekris?
The only caps on the SORKRIS are the ceramics between the FPGA and the shift registers. I am editing because there are caps whose value I have no idea around the FPGA that I have not touched and was not thinking about when I first wrote.
On the SDTrans I have left the two electrolytics in the FPGA rails and there is one near the display which I assume has something to do with it. I am leaving the display off since it is not of much use as long as I am only using three cards - I can keep track of three.
I will install a switch for turning it on and off once I get a sizable collection of the things and need to see what each one begins with. SDTrans remembers where you left off when you removed the disk, so that is what I mean, where it is beginning with this card insertion.
Sorry SOEKRIS for this straying but the SDTrans and the DAM make an excellent combination.
On the SDTrans I have left the two electrolytics in the FPGA rails and there is one near the display which I assume has something to do with it. I am leaving the display off since it is not of much use as long as I am only using three cards - I can keep track of three.
I will install a switch for turning it on and off once I get a sizable collection of the things and need to see what each one begins with. SDTrans remembers where you left off when you removed the disk, so that is what I mean, where it is beginning with this card insertion.
Sorry SOEKRIS for this straying but the SDTrans and the DAM make an excellent combination.
Last edited:
Allow me to join the discussion, and hope it brings the discussion more relates to dam.
The screen off thing is quite common mean for computer audiophile software/tweaking. I dont think the effect is very noticeable but it does mean something, by turning off the display task, there will be less signal being created, hence the noise floor should be lower in the computer.
Regarding this kind of myth, I just did a test myself last night. I was using a flat cat6 LAN cable to the NMR then to dam I2S input (with I2S isolator). The cable is 15 meter long from the network switch in bedroom to living room running in pipes below the floor. I did planted another cat7 LAN round type cable in the pipe before but there was no RJ45 jack at both end. Yesterday I completed the cat7 cable's endpoints and did a comparison. The conclusion is, I could hear the difference, the sound was more 3D, with a clear image. I think the possible reasons are:
1. I have bias, this is purely a psychological effect
2. The i2s isolator can't 100% "isolate" the dirt from hi-fi's perspective, as human ears still can identify the difference
3. Other reason: This is not related to i2s isolation or reclock thing but relate to something that's out of my knowledge domain
This is interesting, since your primary objective is to fight digital noise...
Quite the contrary, the NOS filter injects a large amount of signal-correlated noise, which is impossible to suppress further upstream.
Some comparisons between Paul's "Party Pak" filters would be appreciated.
If we accept that there was indeed a change in sound the technical explanation can only be found in the analog domain i.e. that the noise picked up by the cable is polluting the DAC in general i.e. it is not an impact on the digital information per say that causes this but rather, it is the clock and analog parts on the unit that is. So, the electrical galvanic connection is the problem and thus, it is solved by using a carrier that do not pick up noise - opto is such a carrier.
//
I did not say anything about turning the screen off.
I was talking about using a video card (pci-e only) to take the video load off of the CPU. Turning off the screen does not turn off video processing. It may lessen it, at best, but the video processing is still chugging away.
We found, in the cMP days, that there was a slight improvement but I would tend to agree with you, it seemed vestigial at best.
Using the video card is another thing entirely. It would seem that turning off the screen with the video card would do nothing and I hear nothing different with the screen on or off.
I would, when I was using a computer, turn off the screen just to keep it from lighting up the room not because I thought it was doing something for the sound
Unless you are confusing my comment about the LED screen on the SDTrans and turning that off does make a mild difference. And in the world of audio, there is the HOPE of cumulative effects from doing all of these little things. Like lightening a race car - you must lighten every part.
Many have found that eliminating LEDs can make a difference with any component. I tend to agree that the LED does inject something into the signal.
______________________________________________________________
Regarding digital noise - don't kid yourself - digital noise is everywhere in the spectrum - not just in the high frequencies. It is the noise (and that is using "noise" in the broadest sense) that is within every silence and note. It is not an overlay on the signal, it is intrinsic if allowed to fester.
Consider that the best clocks are selected for low phase noise at subsonic frequencies. Simple filtering of the ultra high frequencies will allay noise THERE but there is plenty more where that came from.
I was talking about using a video card (pci-e only) to take the video load off of the CPU. Turning off the screen does not turn off video processing. It may lessen it, at best, but the video processing is still chugging away.
We found, in the cMP days, that there was a slight improvement but I would tend to agree with you, it seemed vestigial at best.
Using the video card is another thing entirely. It would seem that turning off the screen with the video card would do nothing and I hear nothing different with the screen on or off.
I would, when I was using a computer, turn off the screen just to keep it from lighting up the room not because I thought it was doing something for the sound
Unless you are confusing my comment about the LED screen on the SDTrans and turning that off does make a mild difference. And in the world of audio, there is the HOPE of cumulative effects from doing all of these little things. Like lightening a race car - you must lighten every part.
Many have found that eliminating LEDs can make a difference with any component. I tend to agree that the LED does inject something into the signal.
______________________________________________________________
Regarding digital noise - don't kid yourself - digital noise is everywhere in the spectrum - not just in the high frequencies. It is the noise (and that is using "noise" in the broadest sense) that is within every silence and note. It is not an overlay on the signal, it is intrinsic if allowed to fester.
Consider that the best clocks are selected for low phase noise at subsonic frequencies. Simple filtering of the ultra high frequencies will allay noise THERE but there is plenty more where that came from.
Last edited:
Very cool! nice work!
could you post some more detailed photos of the inside? it's hard to see what you've done. also curious to see the psu side of things in the lower chassis
People keep asking for how to do AES/EBU interface....
My new recommendation is to use a RS-422 receiver, both for AES/EBU and for Coax. First a transformer, like the Murata DA101, with correct termination resistor, 75R or 110R, then to a Maxim MAX3280E or Intersil ISL3280E, power 3.3V with decoupling, then into the dam1021's ttl level spdif input port.
Pretty simple, if you need multiple ports, add more receivers and a switch.
My new recommendation is to use a RS-422 receiver, both for AES/EBU and for Coax. First a transformer, like the Murata DA101, with correct termination resistor, 75R or 110R, then to a Maxim MAX3280E or Intersil ISL3280E, power 3.3V with decoupling, then into the dam1021's ttl level spdif input port.
Pretty simple, if you need multiple ports, add more receivers and a switch.
I dint recall that we have seen anyone doing that... yet.
You don't really upload explicit filter setting - you construct a filter (with some, your choice, tool) file (file coding with special tool by soerkris) which contain filter parameters in the way you want and then upload that (one per fs and according to the internal file system structure). In order to do that one need to understand the filter file structure of the DAM. The whole procedure is really advanced. The tools involved and knowledge require is far beyond you typical "miniDSP". This is however described.... somewhere.
//
You don't really upload explicit filter setting - you construct a filter (with some, your choice, tool) file (file coding with special tool by soerkris) which contain filter parameters in the way you want and then upload that (one per fs and according to the internal file system structure). In order to do that one need to understand the filter file structure of the DAM. The whole procedure is really advanced. The tools involved and knowledge require is far beyond you typical "miniDSP". This is however described.... somewhere.
//
Last edited:
