Audio crackling with new i7 build.

[beatz01]

Just a quick question inbetween - if it's like an external graphics card is recommended, what to do when you only have a laptop (with onboard graphics) but not a desktop ? Buy a laptop that has a dedicated card already built in ? As in a dedicated card is absolutely essential ?

Forgive my ignorance, but i'm clearly not an expert when it comes to computer hardware.

[fishmonkey]

yes, in general you are better off buying a laptop that has a discrete graphics card in addition to any integrated graphics chip. depending on your usage it may not be absolutely essential, but it's a good idea.

[Sional]

[quote="fishmonkey"]
anyone that requires glitch free, high precision graphics output knows that integrated graphics tend to be problematic. some of the later integrated graphics cards are okay for not too demanding stuff,

Fishmonkey - I agree with the above statement completely. My question to you is does a DAW that is used purely for music (i.e. not scoring to video) fall under the "not too demanding stuff" or is it in the "glitch free high precision area"?

Also, would it be fair to say that people who only ever did the "not too demanding stuff" should not find IGP graphics any more problematic than a discrete card? Note that I am talking about IGPs and not the older integrated chipset graphics.

I am not convinced that IGPs are being made solely to make things cheaper; equivalent performance VGA cards are very cheap, better performing cards are cheap and even an acceptable gaming card (e.g. Radeon HD7990) can be had for roughly the price difference between an i5 and an i7.
I suspect that different people may have a different top priority for their computer; some need blazing graphics, others value small form factor, others (say businesses) want reliability and others (like me) value silence. Therefore IGPs could suit many people's needs better than having to have an additional PCIe card in their system (which will require space, draw power, emit heat and reduce the predicted reliability of the system due to the increased parts count).

[fishmonkey]

of course an integrated graphics chip has some advantages, they just don't tend to be performance advantages.

if you are pushing your CPU hard, then a discrete graphics card will be an advantage, even if you aren't pushing your graphics hard. plus there is an increasing use of GPU power to do non-graphics calculations.

[Donnie]

Fishmonkey - I agree with the above statement completely. My question to you is does a DAW that is used purely for music (i.e. not scoring to video) fall under the "not too demanding stuff" or is it in the "glitch free high precision area"?

Also, would it be fair to say that people who only ever did the "not too demanding stuff" should not find IGP graphics any more problematic than a discrete card? Note that I am talking about IGPs and not the older integrated chipset graphics.

So you are putting DAWs into the 'not too demanding stuff' category? Really? Seriously? Every day I become more convinced that you are probably just a troll.

Lets clear this up though:

DAWs are precision high performance professional applications that require overall system stability throughout in order to ensure continuous glitch free performance.

FACT. (That's right, fact.)

[Sional]

Yes in terms of their graphics requirements I am putting DAWs into the 'not too demanding stuff' category. The graphics capabilities of current IGPs will not be stretched by a DAW.

If you dispute this please quote exact areas where IGPs fall short in meeting the DAWs graphic requirements (e.g insufficient stream processors, GPU clock speed, shaders, etc.) or point me to an acceptable reference which supports your claim.
Just because Donnie says so is not good enough and you as an engineer should be well aware of that.

By the way Ableton's stated graphics requirements merely says 1024 x 768 display. Steinberg states "Display resolution of 1280 x 800 recommended" for Cubase. Applications which do need higher performance graphics tend to quote more specific minimum VGA requirements.

I have personally tested my i5 based system both with a game quality VGA card and without. Using the forum's test as a reference (quoting one of my own projects would be meaningless as you would be unable to verify its complexity) I can achieve 96 tracks without any hint of audio breakdown with or without the graphics card.

Obviously if I wanted to play Cyris 3 whilst listening to my mix then I suspect that I would have an apparently unstable system unless I added the graphics card. I say apparently unstable because in reality I would just be exceeding the inherent performance capabilities of the system without the card. With regard to reliability, you as an engineer will also be aware that adding more components to a system decreases the inherent reliability of the system (more things to go wrong).

An engineer once said that I was a jackass polishing turds, maybe so; I just don't like myths being created due to misinformation.

Passing thought, I wonder if Microsoft Office is an imprecise, low performance, unprofessional application that does not require overall system stability. Must ask an engineer sometime.

[fishmonkey]

what is it with the engineer bullshit?

integrated graphics impact on CPU performance. the thermal load must be balanced between the CPU and the GPU. plus the shared memory architecture means that the bandwidth between the CPU and RAM is shared with the GPU.

older integrated graphics were even worse because most of the actual graphics processing was done by the CPU.

as far as this debate goes, your 96 track test doesn't tell you anything particularly useful. what would be interesting would be if you were to test both the integrated and discrete graphics to failure, in various scenarios.

a DAW places stricter realtime demands on a system than a graphics intensive game. if the system can't fill the audio buffer quickly and regularly you get glitches, whereas a game can simply drop the framerate when necessary. MS Office has no realtime constraints, that is a silly example.

this might help:

http://www.intel.com/Assets/PDF/whitepaper/323324.pdf

[Donnie]

Yes in terms of their graphics requirements I am putting DAWs into the 'not too demanding stuff' category. The graphics capabilities of current IGPs will not be stretched by a DAW.

If you dispute this please quote exact areas where IGPs fall short in meeting the DAWs graphic requirements (e.g insufficient stream processors, GPU clock speed, shaders, etc.) or point me to an acceptable reference which supports your claim.
Just because Donnie says so is not good enough and you as an engineer should be well aware of that.

By the way Ableton's stated graphics requirements merely says 1024 x 768 display. Steinberg states "Display resolution of 1280 x 800 recommended" for Cubase. Applications which do need higher performance graphics tend to quote more specific minimum VGA requirements.

Minimum requirements are just that. They do not exemplify what is preferred for an ideal situation.

Another fundamental concept that shouldn't need to be explained to someone who claims to know what they are talking about.

More components to a system decreases the inherent reliability of the system

Irrelevant if the component being added leads to better overall stability (disagree or not). It's also an easy replacement part should something go wrong.

But yeah, I think ive concluded that any further discussion with Sional is complete waste of time. Im sure that the valid points raised by fishmonkey will fall on deaf ears as well. My conclusion: Troll. Its been fun though.

[Sional]

what is it with the engineer bullshit?

Ask Donnie

as far as this debate goes, your 96 track test doesn't tell you anything particularly useful. what would be interesting would be if you were to test both the integrated and discrete graphics to failure, in various scenarios.

The 96 track test is a test to failure; if I add another track I begin to hear slight audio degradation. Yes it would be nice to have different tests but, as far as I am aware, currently this is the only test available to forum members and therefore is the only reference available (everybody can carry out this test).

a DAW places stricter realtime demands on a system than a graphics intensive game. if the system can't fill the audio buffer quickly and regularly you get glitches, whereas a game can simply drop the framerate when necessary. MS Office has no realtime constraints, that is a silly example.

A gamer will notice video glitches just as easily as you will notice audio dropouts. Gamers can lower the resolution settings of the game to avoid glitches, which I suppose roughly equates to having less tracks/effects to avoid audio glitches. I was not aware that game software can detect glitches and automatically lower resolution settings (the only game I play on my computer is MS Chess Titans).

Yes MS Office has no real time constraints but I would suggest that it is still a precision, high performance, professional application that does require overall system reliability (just think of the number of Access applications that are used in various industries). But I assume no one would suggest that you need a discrete graphics card to run MS Office.

integrated graphics impact on CPU performance. the thermal load must be balanced between the CPU and the GPU. plus the shared memory architecture means that the bandwidth between the CPU and RAM is shared with the GPU.

this might help:

http://www.intel.com/Assets/PDF/whitepaper/323324.pdf

"Starting with Sandy Bridge, the graphics core(s) are included on the same die as the
processor cores. This enables greater efficiencies when managing power and thermal
budgets. The integration also removes the need for additional software support, since
the capabilities are now managed by hardware engines. Integration also makes it
possible for the die to actually exceed TDP for short periods of time that are not
thermally relevant."

"Upon the launch of an application, that application may be very processor intensive.
If there is additional thermal and power budget, the processor can make use of Intel®
Turbo Boost Technology to increase both frequency and voltage, to help speed up the
application. In essence, the processor is borrowing the unused graphics power and
thermal budget"

The above two paragraphs are direct lifts from your referenced document. I read the entire document and studied the nice graphs. It is full of praise of the benefits of Intel's integrated HD graphics, which is not surprising as it is an Intel document.

[fishmonkey]

I read the entire document and studied the nice graphs. It is full of praise of the benefits of Intel's integrated HD graphics, which is not surprising as it is an Intel document.

the basic gist of the graphs is that integrated graphics are inextricably linked with and impact on CPU performance (and vice-versa).

[Sional]

the basic gist of the graphs is that integrated graphics are inextricably linked with and impact on CPU performance (and vice-versa).

And the basic gist of the accompanying words are that this is a benefit. Not unexpected as this document was written by Intel to describe an Intel product ( I doubt they would knock their own product).

[fishmonkey]

the basic gist of the graphs is that integrated graphics are inextricably linked with and impact on CPU performance (and vice-versa).

And the basic gist of the accompanying words are that this is a benefit. Not unexpected as this document was written by Intel to describe an Intel product ( I doubt they would knock their own product).

what? the doc is talking about the various hoops they are jumping through to manage the resource conflicts caused by having both the GPU and CPU on the same chip.

the clever thermal management involves continuously ramping up and down clock frequencies and voltages. it's complicated and it is more prone to causing problems with strict realtime constraints than using a discrete graphics card. contrary to what you are saying, an integrated graphics system is not simpler from an engineering viewpoint at all.

[Nokatus]

integrated graphics impact on CPU performance. the thermal load must be balanced between the CPU and the GPU.

a DAW places stricter realtime demands on a system than a graphics intensive game.

All of these "GPU load impacts CPU performance" and "the thermal load must be balanced" concerns are only factors when the GPU is actually under a load. In other words, if it's doing actual graphics calculations. When your workstation is 100% for 2D use, and without any intensive GPU compositing going on at that, the GPU pretty much just sits there -- that is, when compared to its workload when rendering realtime 3D graphics, for example. If you have actual data on the straining calculations the GPU is doing during typical DAW use, show it.

DAW use has placed the strictest realtime demands on any system I have personally needed in any computerized application, yes. The catch is, while these demands are pushing the CPU and audio throughput, they definitely aren't pushing the GPU. I would be VERY surprised to see other than negligible differences in realtime DAW performance (number of realtime plugin instances, length of buffer, etc.) between, for example, a fully functioning problem-free i7 system using the Intel HD Graphics solution versus that same system outfitted with any external display adapter.

On the anecdotal side: companies like Reyniers Audio are listing i7 integrated graphics as a valid choice for systems used strictly for DAW applications. "Purpose: strictly DAW" --> "Recommended choices: Sandy Bridge Onboard, nVidia Geforce 210." http://www.reyniersaudio.com/daw/19-Vir ... Pro-Studio

I am quite sure reputable DAW system vendors would warn their customers of the dangers of Intel HD Graphics if it was an actual factor in a system with tried and tested components that are chosen to be a rock solid combination from the get go. The thing is, when going through their recommendations and especially when asking in person, they don't. If this was a problem, wouldn't this behavior be weird, as this could be so inexpensively remedied and it would also mean that small amount of added sales to them?

Another anecdote: I've been using an i7 system with carefully chosen parts (not a machine from Reyniers or any other such place, mind you, but I digress), including an RME HDSPe AIO interface, and it has been absolutely trouble free using Intel HD Graphics. I've been pushing it with large commercial projects, and the amount of power and stability when using even minuscule buffer sizes is just silly. Again, no CPU load anomalies, no dropouts, no shit -- and I would be very surprised if I could catch an actually significant difference trying an external graphics adapter.

Just a perspective to take into account. Not interested in a huge dick-waving debate on this, however, so carry on

[fishmonkey]

All of these "GPU load impacts CPU performance" and "the thermal load must be balanced" concerns are only factors when the GPU is actually under a load. In other words, if it's doing actual graphics calculations. When your workstation is 100% for 2D use, and without any intensive GPU compositing going on at that, the GPU pretty much just sits there -- that is, when compared to its workload when rendering realtime 3D graphics, for example. If you have actual data on the straining calculations the GPU is doing during typical DAW use, show it.

DAW use has placed the strictest realtime demands on any system I have personally needed in any computerized application, yes. The catch is, while these demands are pushing the CPU and audio throughput, they definitely aren't pushing the GPU. I would be VERY surprised to see other than negligible differences in realtime DAW performance (number of realtime plugin instances, length of buffer, etc.) between, for example, a fully functioning problem-free i7 system using the Intel HD Graphics solution versus that same system outfitted with any external display adapter.

to be fair, the thermal load issues are more pertinent when talking about laptops, where heat is more of an issue. how many external monitors you use is another factor.

i agree that for pure DAW use, the latest integrated graphics systems are more than powerful enough, and actually i don't doubt that the OP's problems could have been resolved without resorting to a discrete card.

there are no available benchmarks for this, which makes speculating more fun. it would be good if there were some comprehensive tests. i would guess that one case where the shared memory bandwidth might have a measurable impact is when you are doing production that is heavily sample library based.

and looking into it further, there are actually some performance advantages built in to the latest Intel integrated graphics systems, like Quick Sync which is supposed to be really good for video transcoding...

[Sional]

All of these "GPU load impacts CPU performance" and "the thermal load must be balanced" concerns are only factors when the GPU is actually under a load. In other words, if it's doing actual graphics calculations. When your workstation is 100% for 2D use, and without any intensive GPU compositing going on at that, the GPU pretty much just sits there -- that is, when compared to its workload when rendering realtime 3D graphics, for example. If you have actual data on the straining calculations the GPU is doing during typical DAW use, show it.

I have not found actual data pertaining to any DAW. They are obviously not 100% 2D, like streaming video, as user interaction does affect the displayed frame. There must be some degree of 3D in terms of the modelling to create the displayed frame but nothing that can't be handled by the IGP.