DAC Input latency tierlist for gaming? Hasn't been done before.

[bodygunger]

I saw the battle(non)sense video on audio latency, he also says the audio codec inherently has bad latency (70ms and higher, which is about 17-20 frames on a 240hz monitor). Could be a good idea.

 

[Verig]

I have never considered gaming audio latency but for production software it's more about drivers than gear. My low tier usb adc/dac unit for home recording has very minor 1-5ms latency but with windows default drivers it can be like 100ms. I suspect that it's the same with games, simply a software thing and not dac related. But this is just a guess so please carry on.

 

[bodygunger]

I have never considered gaming audio latency but for production software it's more about drivers than gear. My low tier usb adc/dac unit for home recording has very minor 1-5ms latency but with windows default drivers it can be like 100ms. I suspect that it's the same with games, simply a software thing and not dac related. But this is just a guess so please carry on.

Windows drivers for external dacs? Because battle(non)sense just plugged a dac in and got a lot more delay. Maybe your dac would be on the high end and the ones he used would be on the low end of the tierlist.

 

[DVDdoug]

There are buffers, I think in the drivers. Buffers are also delays. You NEED buffers for "smooth" audio and video because the multitasking operating system is ALWAYS interrupting and multitasking even you're only running one application. (If you have a GPU that probably helps with the graphics.)

Some hardware has ASIO drivers for audio production, which are designed for low latency and you can adjust the buffer size. But the application has to support ASIO and I don't think any games do.

 

[Verig]

Windows drivers for external dacs? Because battle(non)sense just plugged a dac in and got a lot more delay. Maybe your dac would be on the high end and the ones he used would be on the low end of the tierlist.

"Just plug in" means that default drivers are used. Devices don't just work on their own.
I had 10ms latency even 25 years ago so it's not about high end, it's about application.

I think I wrote very vaguely and left out a part...

This is 100% a driver/software thing. Dacs themselves don't have any meaningful delay.
I was thinking about gaming soundcards / outboard dacs as whole systems (which includes the drivers for that certain device and task and possible extra chips for processing which are not dac related per se) and if there are "accelerated" versions available or any difference between them. I don't much pay attention to anything "for gaming" branded but this would be an actual case.
Are there any brands that advertise optimized latency for gaming? That would be a good starting point.

 

[Nutul]

Windows' sound stack has always been affected by latency. ASIO drivers provide good improvements on this matter.

 

[bodygunger]

Windows' sound stack has always been affected by latency. ASIO drivers provide good improvements on this matter.

not compatible with games

 

[KSTR]

not compatible with games

PC Games that "need" low-latency audio and don't support ASIO are a design failure, IMHO.

 

[kemmler3D]

not compatible with games

Which games? I'm using a DX5 lite which apparently has ASIO drivers and you can set the latency as low as 5ms, and it works fine with games. (mostly just BG3 and Cyberpunk lately tho)

 

[Tell]

This would be interesting to see how much audio latency will affect a players performance in competitive games. I don't think it's nearly as important as high framerates where there have been shown that going from 120hz to 240 or 360 can give an advantage for your aim and therefor also getting a kill or not, while audio won't matter that much in this regard. It's more for hearing where your enemies are, so it matters for getting killed or not I guess, but hearing that with a 4, 16 or 70ms delay probably won't matter that much in most situations.
But would be interesting to see the actual performance differences in some way. This might be something to ask LTT to do a video about?

 

[Verig]

True. LTT would be a right place.

 

[bodygunger]

PC Games that "need" low-latency audio and don't support ASIO are a design failure, IMHO.

yes but saying that isn't a study of what works best and tangible numbers and stuff

 

[wwenze]

This would be interesting to see how much audio latency will affect a players performance in competitive games. I don't think it's nearly as important as high framerates where there have been shown that going from 120hz to 240 or 360 can give an advantage for your aim and therefor also getting a kill or not, while audio won't matter that much in this regard. It's more for hearing where your enemies are, so it matters for getting killed or not I guess, but hearing that with a 4, 16 or 70ms delay probably won't matter that much in most situations.
But would be interesting to see the actual performance differences in some way. This might be something to ask LTT to do a video about?

I'd say it is worth experimenting. Gamers often twitch-react to sound cues, a 20-50ms reaction timing difference easily makes a difference in server ping so no reason to believe if won't for sound.

 

[KSTR]

yes but saying that isn't a study of what works best and tangible numbers and stuff

Well, for a competent DAC like any RME model I do know that latency is very likely below 2ms, for 48kHz sample rate and 32 samples ASIO buffer size, USB transfer.

Total latency of a DAC is very hard to measure directly as you need some kind of trigger pin that you set when you start outputting (the moment you pass your sample buffer to the driver) a dirac pulse signal so that you can see and measure the time difference between the trigger and the analog output with an oscilloscope. But today's PCs usually don't have any accessible hardware pins that can be accessed from software (back in the day you could use pins of the serial or parallel ports for that). Also, you have write you own code for this test even when you have access to some kind of trigger pin (like a GPIO pin on a Raspberry Pi).

In my case the workaround was that I was using a combined ADC+DAC device (RME ADI-2 Pro) and measured total round-trip latency, which was 3.5ms in this case. From that, it is very likely that the DAC latency alone is on the order of 2ms max. ADC and DAC filter settings also play a role here, I've been using the worst-case choice, one of the linear-phase filters. With minimum-phase filters, the delay is even shorter.

 

[dasdoing]

Normally the internal latency can be ignored since the OS buffer latency is much higher. the smaller the buffer, the lower the latency

 

[wwenze]

And the way Windows works, especially with USB, I don't think I can stomach a buffer below 100ms

So, still OS's fault in the end.

 

[dasdoing]

the minimum possible buffer without dropouts will depend on the chip but also on the used resource. the latter can be checked with something like this: https://www.resplendence.com/latencymon
unfortunately, often graphic card drivers will actually limit stuff.

also, you wanna make sure the DAC has its own IRQ or uses MSIs.

 

[AnalogSteph]

also, you wanna make sure the DAC has its own IRQ or uses MSIs.

I would actually say that using MSIs is most important for devices that like to become notorious interrupt hogs otherwise, like GPUs and mass storage controllers. That's the kind of stuff that likes to mess up the sound.

 

[dasdoing]

I would actually say that using MSIs is most important for devices that like to become notorious interrupt hogs otherwise, like GPUs and mass storage controllers. That's the kind of stuff that likes to mess up the sound.

I think you want to set everything to MSI that supports it. my graphic card wasn't even it does support. had to use a special software that I now forgot the name of

EDIT: here is info plus the tool https://forums.guru3d.com/threads/w...ge-signaled-based-interrupts-msi-tool.378044/

 

[nagster]

Chord Hugo 2 Review (Portable DAC & HP Amp)

For curiosity: do you know the audio latency of a modern mobo with integrated sound chip? I would not, but I reckon it varies, from card to card. The more processing there is the more delay there is. And as I mentioned before, there are hardware and software buffers deliberately implemented for...

www.audiosciencereview.com

According to previous measurements, the onboard analog audio output appears to have lower latency. However, it is not stable in any case.
It may be different if use UAC1.0 or 2.0 (CC) instead of ASIO, but I don't know how to measure it with AP. I think it is necessary to use ASIO in order for APx software to input and output audio via a computer.
Also, it seems possible that a native PCIe card with analog audio output such as RME may be able to achieve both low latency and high sound quality. If use IEMs or headphones, I think can respond about 1ms faster than speakers.
Mr. @MC_RME, if you don't mind, please tell me how to achieve low latency (preferably high sound quality) for game sound.