Page 4 of 4

Re: MOTU announces new 828 interface

Posted: Sun Mar 03, 2024 7:03 pm
by mikehalloran
miles_b wrote: Sun Mar 03, 2024 1:41 pm

The quoted "fastest" latency here isn't quite right. When my 828es was working properly, I was achieving 1.0-1.1ms RTL in Logic Pro (at 96kHz) over Thunderbolt 2. If memory serves, I was getting 1.3ms on my Apogee Element 24. The DAD AX Center is able to achieve ~0.6ms RTL over Thunderbolt 3. There are plenty of reasons for new Thunderbolt interfaces to exist, particularly in high-end applications. A key technical advantage, in terms of low-latency monitoring, is that Thunderbolt has DMA, while USB does not. Of course that requires custom drivers, but the payoff can absolutely be worth it. Custom drivers are preferable for USB applications as well, because CoreAudio will not deliver your device's fastest-possible latency. The future of pro recording isn't necessarily centered around iOS / iPadOS, and the presence of Thunderbolt doesn't preclude an interface from also supporting USB.
I went by published specs only. Nobody's marketing department is using your numbers.

Re: MOTU announces new 828 interface

Posted: Mon Mar 04, 2024 3:47 pm
by miles_b
mikehalloran wrote: Sun Mar 03, 2024 7:03 pm
miles_b wrote: Sun Mar 03, 2024 1:41 pm

The quoted "fastest" latency here isn't quite right. When my 828es was working properly, I was achieving 1.0-1.1ms RTL in Logic Pro (at 96kHz) over Thunderbolt 2. If memory serves, I was getting 1.3ms on my Apogee Element 24. The DAD AX Center is able to achieve ~0.6ms RTL over Thunderbolt 3. There are plenty of reasons for new Thunderbolt interfaces to exist, particularly in high-end applications. A key technical advantage, in terms of low-latency monitoring, is that Thunderbolt has DMA, while USB does not. Of course that requires custom drivers, but the payoff can absolutely be worth it. Custom drivers are preferable for USB applications as well, because CoreAudio will not deliver your device's fastest-possible latency. The future of pro recording isn't necessarily centered around iOS / iPadOS, and the presence of Thunderbolt doesn't preclude an interface from also supporting USB.
I went by published specs only. Nobody's marketing department is using your numbers.
DAD have published on their website 670 microseconds RTL on Thunderbolt 3 for the AX Center: (https://www.digitalaudio.dk/PRODUCTS/AX ... plication.). Yes, it's a more expensive interface, but it's a real product, and shows what's capable. MOTU was publishing 1.6ms on their Thunderbolt 2 interfaces (https://motu.com/products/proaudio/828es), but Logic Pro was actually showing even shorter latency on my system. Apogee's published spec for the Element series (Thunderbolt 2) was 1.4ms (https://knowledge.apogeedigital.com/wha ... ent-system), though again, I actually saw lower latency in real-world use.

Re: MOTU announces new 828 interface

Posted: Tue Mar 05, 2024 2:40 pm
by mikehalloran
miles_b wrote: Mon Mar 04, 2024 3:47 pm
MOTU was publishing 1.6ms on their Thunderbolt 2 interfaces
MOTU does not offer a Thunderbolt 2 interface. Though compatible with TB2 & 3 with the right cables and adapters, the 828es and other MOTU TB interfaces are Thunderbolt aka TB1.

None of this changes my statement that I do not see the need for new Thunderbolt interfaces. I saw none at NAMM six weeks ago and am not sure that any were released in 2023 either. The DAD product you referenced was introduced in 2022 and claiming to be 230 microseconds faster than Brand P introduced in 2020... I don't see either product setting the world on fire (a friend of mine owns one) or we should have seen a slew of competitors by now. Where are they?

This might change when USB 4.2/TB5 is released later this year or next but only if corresponding ADC/DAC chips are released also. For the moment, it appears to me that the industry has decided that Fast USB 2 is fast enough — at least for now.

Re: MOTU announces new 828 interface

Posted: Fri Mar 08, 2024 7:20 pm
by miles_b
mikehalloran wrote: Tue Mar 05, 2024 2:40 pm
miles_b wrote: Mon Mar 04, 2024 3:47 pm
MOTU was publishing 1.6ms on their Thunderbolt 2 interfaces
MOTU does not offer a Thunderbolt 2 interface. Though compatible with TB2 & 3 with the right cables and adapters, the 828es and other MOTU TB interfaces are Thunderbolt aka TB1.

None of this changes my statement that I do not see the need for new Thunderbolt interfaces. I saw none at NAMM six weeks ago and am not sure that any were released in 2023 either. The DAD product you referenced was introduced in 2022 and claiming to be 230 microseconds faster than Brand P introduced in 2020... I don't see either product setting the world on fire (a friend of mine owns one) or we should have seen a slew of competitors by now. Where are they?

This might change when USB 4.2/TB5 is released later this year or next but only if corresponding ADC/DAC chips are released also. For the moment, it appears to me that the industry has decided that Fast USB 2 is fast enough — at least for now.
When you look at the AVB functionality that MOTU offer[s/ed] on their interfaces with Thunderbolt, that provides enough channel throughput to more than saturate a USB 2 connection. The 828es supports 128 channels in and 128 channels out simultaneously, but only over Thunderbolt. Keep in mind that it can do 32-bit float, up to 192k, so the bandwidth needs are much higher. When you connect 828es to Logic Pro, it defaults to 32-bit float. It takes USB 3.2 to achieve the same link speed as the 828es over Thunderbolt 1, and even then, latency is higher over USB due to the lack of DMA. As far as low-latency competitors, again the use case here is high-end. High-end products tend to not sell many units. But what I can tell you, as someone who not only records, but is a musician, is that lower latencies below the really obviously audibly-perceptible level, still lead to tighter-sounding performances in the end recording.