ericn
Temp
Balance Engineer
Posts: 16,059
|
Post by ericn on Oct 31, 2024 22:00:56 GMT -6
Nice, you all have got me curious about the upgrade now. I hadn't seen anything advertised about that dramatic of MHLink driver improvements in real DAW use (just a slight reduction in hw converter latency specs w/ phase filter and mkiv) but now I'm going to contact MH about the changes. If true, it's a much more elegant solution for my use. My sincere opinion. First, I think the MK4 upgrade is not that big a deal. Personally, I'm more of a technologist than recordist, so staying up to date is important to me, and I bought the MK4 upgrades at a significantly reduced price. Second, I'm struggling to hear a MEANINGFUL difference. Yes; I hear the differences between converter filter modes, but I don't know how to evaluate them. I think converter fetish is beyond nuts (for me). I buy, and upgrade, my MH boxes because of the tremendous functionality that it provides me, both at my day job (which is quite complex), and at my recording hobby. That's me without any clothes. Take my advice for what it's worth. Just a hunch my friend but if you were coming from say UA the upgrade would probably be more noticeable. The problem with Halo is within Halo the upgrades are minimal improvement, but you’re already starting with the conversion I would run if I didn’t have the RADAR!
|
|
|
Post by indrasnet on Oct 31, 2024 22:13:22 GMT -6
The discussion on the last few pages, however, has been specifically about NOT monitoring thru the MIOConsole, and instead about what sort of latency is achievable when monitoring thru the DAW using the MHLink protocol. That's what we're trying to determine. That's simple, the specs are Host Buffer Size S/R 32 64 128 44100 3.4 ms 4.8 ms 7.7 ms 48000 3.2 ms 4.6 ms 7.2 ms 88200 2.4 ms 3.1 ms 4.6 ms 96000 2.3 ms 3.0 ms 4.3 ms 176400 1.9 ms 2.3 ms 3.0 ms 192000 1.9 ms 2.2 ms 2.9 msThank you for this. Those numbers are better than expected. What method and software did you use to calculate this? What filter and mhlink safety buffer settings? Does this include the latency for routing through mioconsole and mh monitor controller? Thanks!
|
|
|
Post by duplemeter on Oct 31, 2024 23:11:30 GMT -6
That's simple, the specs are Host Buffer Size S/R 32 64 128 44100 3.4 ms 4.8 ms 7.7 ms 48000 3.2 ms 4.6 ms 7.2 ms 88200 2.4 ms 3.1 ms 4.6 ms 96000 2.3 ms 3.0 ms 4.3 ms 176400 1.9 ms 2.3 ms 3.0 ms 192000 1.9 ms 2.2 ms 2.9 msThank you for this. Those numbers are better than expected. What method and software did you use to calculate this? What filter and mhlink safety buffer settings? Does this include the latency for routing through mioconsole and mh monitor controller? Thanks! These numbers come from Metric Halo themselves. I don't know if they are published, but I got these directly from BJ Buchalter, the designer/founder of Metric Halo. Those are the number for monitoring directly from your DAW. You get *much* better latency numbers monitoring from MHCOnsole (0.9ms @ 96k). These are assuming the default filter & safety buffer. Changing either of those will affect these numbers. In the interest of transparency: I've been an MH user for about 20 years, and I beta test for them. If you look at the included plugins presets, you will find presets I created for them. I just want to be clear about my connection to the company.
|
|
|
Post by veggieryan on Nov 1, 2024 10:56:44 GMT -6
USB can't have deterministic latency....so a USB connection will not meet the requirements of a lot of pro users. A lot of pro users enjoy the best low latency performing drivers in existance using RME interfaces over USB so this is simply untrue. In fact the latency on RME over USB is MUCH lower with far less CPU usage than Metric Halo over any protocol currently...
|
|
|
Post by veggieryan on Nov 1, 2024 10:59:09 GMT -6
As for the posted numbers we are only interested in the actual real world usable RTL when monitoring through a DAW *including* the MHLink safety buffers as it is much higher than these numbers...
|
|
|
Post by plinker on Nov 1, 2024 16:10:16 GMT -6
My sincere opinion. First, I think the MK4 upgrade is not that big a deal. Personally, I'm more of a technologist than recordist, so staying up to date is important to me, and I bought the MK4 upgrades at a significantly reduced price. Second, I'm struggling to hear a MEANINGFUL difference. Yes; I hear the differences between converter filter modes, but I don't know how to evaluate them. I think converter fetish is beyond nuts (for me). I buy, and upgrade, my MH boxes because of the tremendous functionality that it provides me, both at my day job (which is quite complex), and at my recording hobby. That's me without any clothes. Take my advice for what it's worth. Just a hunch my friend but if you were coming from say UA the upgrade would probably be more noticeable. The problem with Halo is within Halo the upgrades are minimal improvement, but you’re already starting with the conversion I would run if I didn’t have the RADAR! Cool! Thanks, Eric. That's great to know.
|
|
|
Post by plinker on Nov 1, 2024 16:20:59 GMT -6
As for the posted numbers we are only interested in the actual real world usable RTL when monitoring through a DAW *including* the MHLink safety buffers as it is much higher than these numbers... Those numbers, for 32 @96 are inline with what I posted previously -- as reported by Logic. My RTL was slightly lower, but I had the safety buffers set to 0.6.
|
|
|
Post by veggieryan on Nov 1, 2024 18:16:09 GMT -6
Key word here being *usable*... didn't you say a few posts ago that you had to bump the buffers up to 128 in order to stop getting pops and crackles with only 8 tracks of audio in Logic?
So yeah, the actual RTL there is actually much higher because anything under 128 samples is not usable... and you have to add the safety buffer twice, once in and once out.
The numbers posted leave out the safety buffers which is dubious...
MHLink ends up being significantly worse than the built in class compliant USB driver on Mac which is already too high latency for real time monitoring of FX in a DAW.
We really need to get below 5 ms total RTL for players with good timing and to avoid issues with phasing in headphones for vocalists.
Getting closer to 3 ms RTL gives us some valuable cushion for plugins that add latency like the new UAFX amp plugins: 75 sample for Dream, 145 samples for Lion...
|
|
|
Post by gravesnumber9 on Nov 1, 2024 21:45:50 GMT -6
For comparison's sake the reported numbers for S1 for MOTU 828ES Thunderbolt with a 24 AO connected via AVB are as follows
All measurements are 48khz
1.4 ms at 16 samples 2ms at 32 samples 2.4ms at 64 samples 6ms at 128 samples 11ms at 256 samples
I've measured these in real life and the real numbers are slightly lower but still close. Never had trouble tracking at any of these sample rates but I don't track with a lot of plugins typically. Still, I can run at 16 samples on a MacBook Air M2 with 24 tracks and a few light CPU EQ's and verbs without a hitch.
|
|
|
Post by the other mark williams on Nov 2, 2024 19:20:54 GMT -6
Apologies for my absence from much of the thread over the past week or so. As many of you know, I'm also a longtime MH user. I think my first MH interface was back in 2004, and I've been using them as my primary interfaces ever since. So it's safe to say I'm a fan of the company. I've quite possibly received better customer service from MH than any other company who I've ever owned anything from - not just audio, I mean in LIFE. So that does breed a lot of goodwill. Having said that, I tend to agree with the critiques of the last few pages. It has long bugged me that the RTL numbers are not lower, and I was similarly frustrated when 3D came out and it used ethernet connections instead of Thunderbolt. I wish to goodness MH had gone with Thunderbolt, but they didn't. As I think became evident from duplemeter 's posts (welcome, Steve! Great to see you over here - your posts on the MH Listserv have been very helpful to me over the years), MH had different priorities. Their boxes are used a lot in live touring rigs and installs where the ability to run ethernet cable for hundreds of feet is a huge boon, and I think the company really wanted to continue meeting the needs of that market and its established customer base there. But I continue to find it frustrating that RTL figures haven't been the priority, because that matters more TO ME in MY use cases. To be clear, as long as you are OK with using a separate application for your cue mix(es): live inputs vs. already-recorded material, I'm not sure there's a better all-in-one box on the market than the ULN-8/LIO-8. For real. The MIO Console app is fantastic, and very logically set up if you're used to a console-style workflow. The latency there is negligible/unnoticeable, just like when using an Apollo's included DSP mixer or any other company's DSP mixer layer. MH continue to invest in and develop recording functions in MIO Console, as well. It's nearly a DAW itself at this point. I've used those recording functions for live location work for years now, and it's rock solid. HOWEVER, even after many years of doing so, I am still of the opinion that using a separate application from my DAW for cue mixes is a pain in the ass. I have my setup about as streamlined as it's possible to be, including using a dedicated control surface (X-Touch in my case) that only communicates with the MIO Console, so I have physical faders right in front of me to control listening levels of my inputs vs. my 2-bus. It's pretty slick and totally solid, and yet I would still prefer to not have to use a separate app to my DAW. plinkerJim, you asked a couple pages ago about why round trip latency numbers matter so much to some of us when the +DSP mixer already exists and is so incredibly low latency. Others have answered already, so maybe my response will be redundant, but I'd like to take a crack at my own explanation. I think once you've worked with a system like Pro Tools HD (or, I gather, Luna on an Apollo), the incredible simplicity of monitoring live inputs through the same exact mixer all your recorded tracks are in is really hard to give up. Everything else feels like a compromise at best. EVERYTHING else. Even after recording for 25+ years now and being totally confident in my abilities, anything that comes between me and the music is a hindrance, and having a second app that acts as my "mixer" when tracking means there's some part of my brain paying attention to THAT instead of the song, the performance, the vibe. And the only way to achieve that outside of Pro Tools or Luna with an Apollo (or what, PARIS??) is by having low enough latency that one can simply use one's DAW as the mixer for both live inputs and already-recorded tracks. Well, you could of course use a LFAC and enough channels of conversion to pull everything up on the board - I've certainly considered that route over the years, too. I don't know if all that makes sense or not. As for the posted numbers we are only interested in the actual real world usable RTL when monitoring through a DAW *including* the MHLink safety buffers as it is much higher than these numbers... Those numbers, for 32 @96 are inline with what I posted previously -- as reported by Logic. My RTL was slightly lower, but I had the safety buffers set to 0.6. Over here at my place, I can't run consistently with the safety offset at 0.6ms. I can't even always run with it at the default of 1.0ms. Hell, when I'm editing and grading video, I can't even always run it at 60ms. I have it parked at 100ms(!) when doing video work. I'm on a Mac Mini M1 over here, same as you, it sounds like. (Now, in video work it obviously doesn't matter what the safety offset is - there are no live audio inputs.) Key word here being *usable*... didn't you say a few posts ago that you had to bump the buffers up to 128 in order to stop getting pops and crackles with only 8 tracks of audio in Logic? So yeah, the actual RTL there is actually much higher because anything under 128 samples is not usable... and you have to add the safety buffer twice, once in and once out. The numbers posted leave out the safety buffers which is dubious... MHLink ends up being significantly worse than the built in class compliant USB driver on Mac which is already too high latency for real time monitoring of FX in a DAW. We really need to get below 5 ms total RTL for players with good timing and to avoid issues with phasing in headphones for vocalists. Getting closer to 3 ms RTL gives us some valuable cushion for plugins that add latency like the new UAFX amp plugins: 75 sample for Dream, 145 samples for Lion... I'll say this: MHLink is not slower or worse than the class-compliant USB driver in my personal usage. With MHLink and the mkIV converters, I've definitely been able to achieve the lowest latency through my DAW that I've ever gotten with a Metric Halo product in nearly 20 years now. I do find it more usable when composing with things like sampled pianos and other VIs as compared to life before MHLink, and certainly when compared to life before the current mkIV converters. Some of my heavier Spitfire libraries in particular were very challenging to play--if not nearly impossible--with the USB driver in the mkIII box. And the most recent firmware, MIO Console, and driver update (from like two weeks ago) have made things even more stable on my system. Having said ALL of this, if someone is considering entering the MH ecosystem, but has "lowest possible latency" and "most-efficient-drivers-on-the-market" as primary requirements, you would need to be prepared to go with either the RME HDSPe AES PCI-e card or the Lynx AES card and interface with either of those via AES from the ULN/LIO-8. That is still the lowest latency way of getting sounds into and out of your computer with MH interfaces. If Apple had not made it so damn expensive to buy a computer with PCI slots, I would've gone that route ages ago, and may still do so with a Sonnet expansion box. It's a difficult decision when committing oneself to an audio interface for the long term. As I said earlier, Metric Halo has phenomenal customer service. They have never once been caught flat-footed with driver problems at the release of a new Mac OS. And they kept me up and running even when I lived on a near-desert island from 2009-2011 with unpredictable electricity. They're good folks. Yes, I wish the boxes were Thunderbolt and had lower RTL and world-beating lowest-CPU-overhead drivers. But you can't have everything in life. So Quint, I'm not sure what else to say, but I'm happy to answer any outstanding questions you might have. I know it's a big deal to reconfigure one's studio and commit to new interfaces that hopefully take you for at least a decade. Not the most fun spot to be in.
|
|
|
Post by duplemeter on Nov 2, 2024 20:54:08 GMT -6
USB can't have deterministic latency....so a USB connection will not meet the requirements of a lot of pro users. A lot of pro users enjoy the best low latency performing drivers in existance using RME interfaces over USB so this is simply untrue. In fact the latency on RME over USB is MUCH lower with far less CPU usage than Metric Halo over any protocol currently... I think I need to clarify. those who mix integrating external HW as HW inserts in Pro Tools require 100% deterministic RTL. There is no other way to do that & not have phase issues. Most just simply stay within the Avid HDX ecosystem because Avid is able to tie it all together very neatly. I use external HW, but find the HDX offerings don't sound as good as I'd like. So I chose MH because I get deterministic latency & the sound I want. I'm 100% happy to setup headphone mixes via MHConsole, which is near zero latency, as low as 8 samples...without any other processing). I suppose adding an analogue console is also an option, since it neutralizes the latency issue, so there's that. That said, if you wanted the lowest RTL, want to monitor through your DAW & don't require deterministic latency, yes RME is very attractive. The only option that beats it is Pro Tools with an HDX interface (like Carbon or a Matrix). Everyone favors one aspect over others & that will narrow their options. For the record, the mkIV upgrade from MH was mind blowing for me. Upgrading the analogue path to/from the converters added even more dimension & clarity to the soundfield. I certainly didn't experience it as "not a big deal". But my #1 priority is sound quality. I'll give up convenience for that.
|
|
|
Post by duplemeter on Nov 2, 2024 20:57:13 GMT -6
(welcome, Steve! Great to see you over here - your posts on the MH Listserv have been very helpful to me over the years) Thanks! Glad to be here, and very happy to hear I've been able to help you out over on the MIO list.
|
|
ericn
Temp
Balance Engineer
Posts: 16,059
|
Post by ericn on Nov 2, 2024 21:19:34 GMT -6
Apologies for my absence from much of the thread over the past week or so. As many of you know, I'm also a longtime MH user. I think my first MH interface was back in 2004, and I've been using them as my primary interfaces ever since. So it's safe to say I'm a fan of the company. I've quite possibly received better customer service from MH than any other company who I've ever owned anything from - not just audio, I mean in LIFE. So that does breed a lot of goodwill. Having said that, I tend to agree with the critiques of the last few pages. It has long bugged me that the RTL numbers are not lower, and I was similarly frustrated when 3D came out and it used ethernet connections instead of Thunderbolt. I wish to goodness MH had gone with Thunderbolt, but they didn't. As I think became evident from duplemeter 's posts (welcome, Steve! Great to see you over here - your posts on the MH Listserv have been very helpful to me over the years), MH had different priorities. Their boxes are used a lot in live touring rigs and installs where the ability to run ethernet cable for hundreds of feet is a huge boon, and I think the company really wanted to continue meeting the needs of that market and its established customer base there. But I continue to find it frustrating that RTL figures haven't been the priority, because that matters more TO ME in MY use cases. To be clear, as long as you are OK with using a separate application for your cue mix(es): live inputs vs. already-recorded material, I'm not sure there's a better all-in-one box on the market than the ULN-8/LIO-8. For real. The MIO Console app is fantastic, and very logically set up if you're used to a console-style workflow. The latency there is negligible/unnoticeable, just like when using an Apollo's included DSP mixer or any other company's DSP mixer layer. MH continue to invest in and develop recording functions in MIO Console, as well. It's nearly a DAW itself at this point. I've used those recording functions for live location work for years now, and it's rock solid. HOWEVER, even after many years of doing so, I am still of the opinion that using a separate application from my DAW for cue mixes is a pain in the ass. I have my setup about as streamlined as it's possible to be, including using a dedicated control surface (X-Touch in my case) that only communicates with the MIO Console, so I have physical faders right in front of me to control listening levels of my inputs vs. my 2-bus. It's pretty slick and totally solid, and yet I would still prefer to not have to use a separate app to my DAW. plinkerJim, you asked a couple pages ago about why round trip latency numbers matter so much to some of us when the +DSP mixer already exists and is so incredibly low latency. Others have answered already, so maybe my response will be redundant, but I'd like to take a crack at my own explanation. I think once you've worked with a system like Pro Tools HD (or, I gather, Luna on an Apollo), the incredible simplicity of monitoring live inputs through the same exact mixer all your recorded tracks are in is really hard to give up. Everything else feels like a compromise at best. EVERYTHING else. Even after recording for 25+ years now and being totally confident in my abilities, anything that comes between me and the music is a hindrance, and having a second app that acts as my "mixer" when tracking means there's some part of my brain paying attention to THAT instead of the song, the performance, the vibe. And the only way to achieve that outside of Pro Tools or Luna with an Apollo (or what, PARIS??) is by having low enough latency that one can simply use one's DAW as the mixer for both live inputs and already-recorded tracks. Well, you could of course use a LFAC and enough channels of conversion to pull everything up on the board - I've certainly considered that route over the years, too. I don't know if all that makes sense or not. Those numbers, for 32 @96 are inline with what I posted previously -- as reported by Logic. My RTL was slightly lower, but I had the safety buffers set to 0.6. Over here at my place, I can't run consistently with the safety offset at 0.6ms. I can't even always run with it at the default of 1.0ms. Hell, when I'm editing and grading video, I can't even always run it at 60ms. I have it parked at 100ms(!) when doing video work. I'm on a Mac Mini M1 over here, same as you, it sounds like. (Now, in video work it obviously doesn't matter what the safety offset is - there are no live audio inputs.) Key word here being *usable*... didn't you say a few posts ago that you had to bump the buffers up to 128 in order to stop getting pops and crackles with only 8 tracks of audio in Logic? So yeah, the actual RTL there is actually much higher because anything under 128 samples is not usable... and you have to add the safety buffer twice, once in and once out. The numbers posted leave out the safety buffers which is dubious... MHLink ends up being significantly worse than the built in class compliant USB driver on Mac which is already too high latency for real time monitoring of FX in a DAW. We really need to get below 5 ms total RTL for players with good timing and to avoid issues with phasing in headphones for vocalists. Getting closer to 3 ms RTL gives us some valuable cushion for plugins that add latency like the new UAFX amp plugins: 75 sample for Dream, 145 samples for Lion... I'll say this: MHLink is not slower or worse than the class-compliant USB driver in my personal usage. With MHLink and the mkIV converters, I've definitely been able to achieve the lowest latency through my DAW that I've ever gotten with a Metric Halo product in nearly 20 years now. I do find it more usable when composing with things like sampled pianos and other VIs as compared to life before MHLink, and certainly when compared to life before the current mkIV converters. Some of my heavier Spitfire libraries in particular were very challenging to play--if not nearly impossible--with the USB driver in the mkIII box. And the most recent firmware, MIO Console, and driver update (from like two weeks ago) have made things even more stable on my system. Having said ALL of this, if someone is considering entering the MH ecosystem, but has "lowest possible latency" and "most-efficient-drivers-on-the-market" as primary requirements, you would need to be prepared to go with either the RME HDSPe AES PCI-e card or the Lynx AES card and interface with either of those via AES from the ULN/LIO-8. That is still the lowest latency way of getting sounds into and out of your computer with MH interfaces. If Apple had not made it so damn expensive to buy a computer with PCI slots, I would've gone that route ages ago, and may still do so with a Sonnet expansion box. It's a difficult decision when committing oneself to an audio interface for the long term. As I said earlier, Metric Halo has phenomenal customer service. They have never once been caught flat-footed with driver problems at the release of a new Mac OS. And they kept me up and running even when I lived on a near-desert island from 2009-2011 with unpredictable electricity. They're good folks. Yes, I wish the boxes were Thunderbolt and had lower RTL and world-beating lowest-CPU-overhead drivers. But you can't have everything in life. So Quint, I'm not sure what else to say, but I'm happy to answer any outstanding questions you might have. I know it's a big deal to reconfigure one's studio and commit to new interfaces that hopefully take you for at least a decade. Not the most fun spot to be in. Mark, as far as the need to use a separate App other than the DAW for Low latency monitoring especially with DSP it really comes down to one simple fact of life in the pro audio world. Nobody wants to work with anyone else. As an industry it would not be all that hard for everyone who uses DSP to develop a standard control protocol but nobody wants to. Remember if we were to go back in time to about 96-99 you would be in a world where DP, Logic, and one or 2 others were all capable of running on a PT TDM rig! As customers we all need to be reminding everyone we are not as stupid as their marketing departments! We want standards and simplicity!
|
|
|
Post by the other mark williams on Nov 2, 2024 21:30:08 GMT -6
[whole bunch of stuff in original post...] Mark, as far as the need to use a separate App other than the DAW for Low latency monitoring especially with DSP it really comes down to one simple fact of life in the pro audio world. Nobody wants to work with anyone else. As an industry it would not be all that hard for everyone who uses DSP to develop a standard control protocol but nobody wants to. Remember if we were to go back in time to about 96-99 you would be in a world where DP, Logic, and one or 2 others were all capable of running on a PT TDM rig! As customers we all need to be reminding everyone we are not as stupid as their marketing departments! We want standards and simplicity! Absolutely, Eric, and great point. I was still going between tape, ADATs, and PT TDM in that 96-99 era, and didn't move over to Logic until like 2002? I had Logic 6 which could still utilize TDM hardware, but I never actually tried it out before that feature got dropped. I really wish we had more standards in our industry...
|
|
|
Post by plinker on Nov 2, 2024 23:36:29 GMT -6
Plinker, you asked a couple pages ago about why round trip latency numbers matter so much to some of us when the +DSP mixer already exists and is so incredibly low latency. Others have answered already, so maybe my response will be redundant, but I'd like to take a crack at my own explanation. I think once you've worked with a system like Pro Tools HD (or, I gather, Luna on an Apollo), the incredible simplicity of monitoring live inputs through the same exact mixer all your recorded tracks are in is really hard to give up. Everything else feels like a compromise at best. EVERYTHING else. Even after recording for 25+ years now and being totally confident in my abilities, anything that comes between me and the music is a hindrance, and having a second app that acts as my "mixer" when tracking means there's some part of my brain paying attention to THAT instead of the song, the performance, the vibe. And the only way to achieve that outside of Pro Tools or Luna with an Apollo (or what, PARIS??) is by having low enough latency that one can simply use one's DAW as the mixer for both live inputs and already-recorded tracks. Well, you could of course use a LFAC and enough channels of conversion to pull everything up on the board - I've certainly considered that route over the years, too. I don't know if all that makes sense or not. Thank, Mark. Another person responded about using virtual instruments and such -- I appreciate your perspective as well. Those numbers, for 32 @96 are inline with what I posted previously -- as reported by Logic. My RTL was slightly lower, but I had the safety buffers set to 0.6. Yeah. Like I said earlier, I was just posting RTL under ideal (i.e., not realistic for my setup) circumstances as a reference. I don't run with those parameters either.
|
|
|
Post by veggieryan on Nov 3, 2024 11:55:24 GMT -6
A lot of pro users enjoy the best low latency performing drivers in existance using RME interfaces over USB so this is simply untrue. In fact the latency on RME over USB is MUCH lower with far less CPU usage than Metric Halo over any protocol currently... I think I need to clarify. those who mix integrating external HW as HW inserts in Pro Tools require 100% deterministic RTL. There is no other way to do that & not have phase issues. Most just simply stay within the Avid HDX ecosystem because Avid is able to tie it all together very neatly. I use external HW, but find the HDX offerings don't sound as good as I'd like. So I chose MH because I get deterministic latency & the sound I want. I'm 100% happy to setup headphone mixes via MHConsole, which is near zero latency, as low as 8 samples...without any other processing). I suppose adding an analogue console is also an option, since it neutralizes the latency issue, so there's that. That said, if you wanted the lowest RTL, want to monitor through your DAW & don't require deterministic latency, yes RME is very attractive. The only option that beats it is Pro Tools with an HDX interface (like Carbon or a Matrix). Everyone favors one aspect over others & that will narrow their options. For the record, the mkIV upgrade from MH was mind blowing for me. Upgrading the analogue path to/from the converters added even more dimension & clarity to the soundfield. I certainly didn't experience it as "not a big deal". But my #1 priority is sound quality. I'll give up convenience for that. Actually, I think you first need to verify rather than clarify. First you are using the phrase "deterministic latency" when you really mean repeatable deterministic latency specifically. This means the RTL remains consistent over time AND after reboots for a given sample rate and buffer size. This way the ping and latency compensation of hardware inserts in DAW's remains consistent. The claim that RME USB interfaces don't provide repeatable deterministic latency is 100% false. Ironically, hardware inserts via the I/O plugin in the Metric Halo MIOConsole mixer are still not latency compensated unlike most DAWs. This might explain why the mkiv upgrade was such a big improvement for you as it would reduce the phase issues introduced by external HW in the MIOConsole due to the reduced converter latency. Unfortunately Metric Halo still have not addressed this issue after many years so I assume that it is either too difficult for them or not a priority. As of now every hardware insert in the MIOConsole still causes audio to be out of phase which is one of the main reasons I sold my ULN-8 and can't consider it a viable option along with the terrible performance and RTL of the MHLink driver. At this point it's been years since 3D was released. I have to assume they have not solved these problems because they can't and never will.
|
|
|
Post by Quint on Nov 3, 2024 12:22:24 GMT -6
Apologies for my absence from much of the thread over the past week or so. As many of you know, I'm also a longtime MH user. I think my first MH interface was back in 2004, and I've been using them as my primary interfaces ever since. So it's safe to say I'm a fan of the company. I've quite possibly received better customer service from MH than any other company who I've ever owned anything from - not just audio, I mean in LIFE. So that does breed a lot of goodwill. Having said that, I tend to agree with the critiques of the last few pages. It has long bugged me that the RTL numbers are not lower, and I was similarly frustrated when 3D came out and it used ethernet connections instead of Thunderbolt. I wish to goodness MH had gone with Thunderbolt, but they didn't. As I think became evident from duplemeter 's posts (welcome, Steve! Great to see you over here - your posts on the MH Listserv have been very helpful to me over the years), MH had different priorities. Their boxes are used a lot in live touring rigs and installs where the ability to run ethernet cable for hundreds of feet is a huge boon, and I think the company really wanted to continue meeting the needs of that market and its established customer base there. But I continue to find it frustrating that RTL figures haven't been the priority, because that matters more TO ME in MY use cases. To be clear, as long as you are OK with using a separate application for your cue mix(es): live inputs vs. already-recorded material, I'm not sure there's a better all-in-one box on the market than the ULN-8/LIO-8. For real. The MIO Console app is fantastic, and very logically set up if you're used to a console-style workflow. The latency there is negligible/unnoticeable, just like when using an Apollo's included DSP mixer or any other company's DSP mixer layer. MH continue to invest in and develop recording functions in MIO Console, as well. It's nearly a DAW itself at this point. I've used those recording functions for live location work for years now, and it's rock solid. HOWEVER, even after many years of doing so, I am still of the opinion that using a separate application from my DAW for cue mixes is a pain in the ass. I have my setup about as streamlined as it's possible to be, including using a dedicated control surface (X-Touch in my case) that only communicates with the MIO Console, so I have physical faders right in front of me to control listening levels of my inputs vs. my 2-bus. It's pretty slick and totally solid, and yet I would still prefer to not have to use a separate app to my DAW. plinker Jim, you asked a couple pages ago about why round trip latency numbers matter so much to some of us when the +DSP mixer already exists and is so incredibly low latency. Others have answered already, so maybe my response will be redundant, but I'd like to take a crack at my own explanation. I think once you've worked with a system like Pro Tools HD (or, I gather, Luna on an Apollo), the incredible simplicity of monitoring live inputs through the same exact mixer all your recorded tracks are in is really hard to give up. Everything else feels like a compromise at best. EVERYTHING else. Even after recording for 25+ years now and being totally confident in my abilities, anything that comes between me and the music is a hindrance, and having a second app that acts as my "mixer" when tracking means there's some part of my brain paying attention to THAT instead of the song, the performance, the vibe. And the only way to achieve that outside of Pro Tools or Luna with an Apollo (or what, PARIS??) is by having low enough latency that one can simply use one's DAW as the mixer for both live inputs and already-recorded tracks. Well, you could of course use a LFAC and enough channels of conversion to pull everything up on the board - I've certainly considered that route over the years, too. I don't know if all that makes sense or not. Those numbers, for 32 @96 are inline with what I posted previously -- as reported by Logic. My RTL was slightly lower, but I had the safety buffers set to 0.6. Over here at my place, I can't run consistently with the safety offset at 0.6ms. I can't even always run with it at the default of 1.0ms. Hell, when I'm editing and grading video, I can't even always run it at 60ms. I have it parked at 100ms(!) when doing video work. I'm on a Mac Mini M1 over here, same as you, it sounds like. (Now, in video work it obviously doesn't matter what the safety offset is - there are no live audio inputs.) Key word here being *usable*... didn't you say a few posts ago that you had to bump the buffers up to 128 in order to stop getting pops and crackles with only 8 tracks of audio in Logic? So yeah, the actual RTL there is actually much higher because anything under 128 samples is not usable... and you have to add the safety buffer twice, once in and once out. The numbers posted leave out the safety buffers which is dubious... MHLink ends up being significantly worse than the built in class compliant USB driver on Mac which is already too high latency for real time monitoring of FX in a DAW. We really need to get below 5 ms total RTL for players with good timing and to avoid issues with phasing in headphones for vocalists. Getting closer to 3 ms RTL gives us some valuable cushion for plugins that add latency like the new UAFX amp plugins: 75 sample for Dream, 145 samples for Lion... I'll say this: MHLink is not slower or worse than the class-compliant USB driver in my personal usage. With MHLink and the mkIV converters, I've definitely been able to achieve the lowest latency through my DAW that I've ever gotten with a Metric Halo product in nearly 20 years now. I do find it more usable when composing with things like sampled pianos and other VIs as compared to life before MHLink, and certainly when compared to life before the current mkIV converters. Some of my heavier Spitfire libraries in particular were very challenging to play--if not nearly impossible--with the USB driver in the mkIII box. And the most recent firmware, MIO Console, and driver update (from like two weeks ago) have made things even more stable on my system. Having said ALL of this, if someone is considering entering the MH ecosystem, but has "lowest possible latency" and "most-efficient-drivers-on-the-market" as primary requirements, you would need to be prepared to go with either the RME HDSPe AES PCI-e card or the Lynx AES card and interface with either of those via AES from the ULN/LIO-8. That is still the lowest latency way of getting sounds into and out of your computer with MH interfaces. If Apple had not made it so damn expensive to buy a computer with PCI slots, I would've gone that route ages ago, and may still do so with a Sonnet expansion box. It's a difficult decision when committing oneself to an audio interface for the long term. As I said earlier, Metric Halo has phenomenal customer service. They have never once been caught flat-footed with driver problems at the release of a new Mac OS. And they kept me up and running even when I lived on a near-desert island from 2009-2011 with unpredictable electricity. They're good folks. Yes, I wish the boxes were Thunderbolt and had lower RTL and world-beating lowest-CPU-overhead drivers. But you can't have everything in life. So Quint , I'm not sure what else to say, but I'm happy to answer any outstanding questions you might have. I know it's a big deal to reconfigure one's studio and commit to new interfaces that hopefully take you for at least a decade. Not the most fun spot to be in. Thanks for the thorough response. The customer service part is definitely something that is attractive to me, considering what has happened with customer service at UA. However, the latency thing appears to be a deal breaker for my use case, unfortunately. But there IS a lot to like with MH. I just wish they would go thunderbolt. Maybe one day.
|
|
|
Post by duplemeter on Nov 3, 2024 23:51:51 GMT -6
I think I need to clarify. those who mix integrating external HW as HW inserts in Pro Tools require 100% deterministic RTL. There is no other way to do that & not have phase issues. Most just simply stay within the Avid HDX ecosystem because Avid is able to tie it all together very neatly. I use external HW, but find the HDX offerings don't sound as good as I'd like. So I chose MH because I get deterministic latency & the sound I want. I'm 100% happy to setup headphone mixes via MHConsole, which is near zero latency, as low as 8 samples...without any other processing). I suppose adding an analogue console is also an option, since it neutralizes the latency issue, so there's that. That said, if you wanted the lowest RTL, want to monitor through your DAW & don't require deterministic latency, yes RME is very attractive. The only option that beats it is Pro Tools with an HDX interface (like Carbon or a Matrix). Everyone favors one aspect over others & that will narrow their options. For the record, the mkIV upgrade from MH was mind blowing for me. Upgrading the analogue path to/from the converters added even more dimension & clarity to the soundfield. I certainly didn't experience it as "not a big deal". But my #1 priority is sound quality. I'll give up convenience for that. Actually, I think you first need to verify rather than clarify. First you are using the phrase "deterministic latency" when you really mean repeatable deterministic latency specifically. This means the RTL remains consistent over time AND after reboots for a given sample rate and buffer size. This way the ping and latency compensation of hardware inserts in DAW's remains consistent. The claim that RME USB interfaces don't provide repeatable deterministic latency is 100% false. Ironically, hardware inserts via the I/O plugin in the Metric Halo MIOConsole mixer are still not latency compensated unlike most DAWs. This might explain why the mkiv upgrade was such a big improvement for you as it would reduce the phase issues introduced by external HW in the MIOConsole due to the reduced converter latency. Unfortunately Metric Halo still have not addressed this issue after many years so I assume that it is either too difficult for them or not a priority. As of now every hardware insert in the MIOConsole still causes audio to be out of phase which is one of the main reasons I sold my ULN-8 and can't consider it a viable option along with the terrible performance and RTL of the MHLink driver. At this point it's been years since 3D was released. I have to assume they have not solved these problems because they can't and never will. Sure...we can call it "repeatable deterministic latency". I also think you've missed a couple updates on MHConsole, which now includes latency compensation (except for a few edge cases that bypass bus assignments). For the record, I've never used the I/O plugin for HW inserts. I setup routing via input strips, which have included latency compensation since late 3d/early mkIV (which was a necessity when they turned the record panel into the Session window). Before that I had moved ITB after being hybrid with a console, which negates the whole latency thing, anyway. I don't know what you encountered, but my ULN-8 is one of the best performing & best sounding interfaces I've ever used. We even did a converter shootout at Production Expert & in a blind test, all involved picked the Metric Halo over other options (including 2 DSD options, a UA converter & RADAR Nyquist). If anyone's interested, you can catch the podcast here: podcasts.apple.com/us/podcast/production-expert-podcast/id501987225?i=1000562215060I use these things daily for my livelihood & in multiple places. I have 3 boxes in my home Atmos mix room, we have 4 in the studio & we have 5 in the remote truck. I'm not saying you can't dislike them, I'm just saying I have a lot of experience with them in professional settings & am really happy with everything about them. You are free to have your own opinions, but you should also make sure you are correct when you make statements like MHConsole not being latency compensated.
|
|
|
Post by veggieryan on Nov 4, 2024 0:40:46 GMT -6
I'm not sure why you didn't verify this yourself but it's confirmed multiple times in the current MIOConsole documentation:
Page 421: "Of course, any signal path which leaves the mixer and returns via an Input strip (shown in red) can
not be automatically latency compensated. This includes the I/O routing inserts, and any other route
that loops out to external devices via analog, digital, MHLink or USB."
Page 426: "Reminder: Since ‘I/O’ inserts route to physical ports outside the 3d mixer, they can not be automagically latency-compensated like plug-in processor Inserts. All physical routes outside and back into the Mixer will incur some form of latency - whether that bit of delay is acceptable will depend on the individual use case."
Therefore any attempt to patch hardware via the I/O plugin or routing an output back to an input channel is not latency compensated in the MIOConsole. I also personally verified this problem with BJ and he gave no indication a solution was forthcoming. This along with terrible RTL performance on MHLink and the fact that the latency reported by MHLink to the DAW incorrectly ommits the safety buffer and mixer latency which leads to issues with the time alignment of overdubs in DAWs, etc... all led me to move on from my ULN-8. I had hoped at least some of those issues were solved by now after a few years but sadly that is not the case.
|
|
|
Post by Dan on Nov 4, 2024 10:14:17 GMT -6
I think I need to clarify. those who mix integrating external HW as HW inserts in Pro Tools require 100% deterministic RTL. There is no other way to do that & not have phase issues. Most just simply stay within the Avid HDX ecosystem because Avid is able to tie it all together very neatly. I use external HW, but find the HDX offerings don't sound as good as I'd like. So I chose MH because I get deterministic latency & the sound I want. I'm 100% happy to setup headphone mixes via MHConsole, which is near zero latency, as low as 8 samples...without any other processing). I suppose adding an analogue console is also an option, since it neutralizes the latency issue, so there's that. That said, if you wanted the lowest RTL, want to monitor through your DAW & don't require deterministic latency, yes RME is very attractive. The only option that beats it is Pro Tools with an HDX interface (like Carbon or a Matrix). Everyone favors one aspect over others & that will narrow their options. For the record, the mkIV upgrade from MH was mind blowing for me. Upgrading the analogue path to/from the converters added even more dimension & clarity to the soundfield. I certainly didn't experience it as "not a big deal". But my #1 priority is sound quality. I'll give up convenience for that. Actually, I think you first need to verify rather than clarify. First you are using the phrase "deterministic latency" when you really mean repeatable deterministic latency specifically. This means the RTL remains consistent over time AND after reboots for a given sample rate and buffer size. This way the ping and latency compensation of hardware inserts in DAW's remains consistent. The claim that RME USB interfaces don't provide repeatable deterministic latency is 100% false. Ironically, hardware inserts via the I/O plugin in the Metric Halo MIOConsole mixer are still not latency compensated unlike most DAWs. This might explain why the mkiv upgrade was such a big improvement for you as it would reduce the phase issues introduced by external HW in the MIOConsole due to the reduced converter latency. Unfortunately Metric Halo still have not addressed this issue after many years so I assume that it is either too difficult for them or not a priority. As of now every hardware insert in the MIOConsole still causes audio to be out of phase which is one of the main reasons I sold my ULN-8 and can't consider it a viable option along with the terrible performance and RTL of the MHLink driver. At this point it's been years since 3D was released. I have to assume they have not solved these problems because they can't and never will. RME usb is only sample accurate. Other interfaces are +/- 1 sample. PCI-E and most thunderbolt interfaces are subsample accurate. This includes digilink. I’ve never tested Dante. MOTU AVB is a mess. Keep all of your mics on one source on the same interface. Zmix made a big hoopla about this years ago, people tried to shoot him down and discredit him, but he was correct that his cheap motu pci-e interface had deterministic latency where no usb interface did.
|
|
|
Post by veggieryan on Nov 4, 2024 11:53:08 GMT -6
RME usb is only sample accurate. Other interfaces are +/- 1 sample. PCI-E and most thunderbolt interfaces are subsample accurate. This includes digilink. I’ve never tested Dante. MOTU AVB is a mess. Keep all of your mics on one source on the same interface. Zmix made a big hoopla about this years ago, people tried to shoot him down and discredit him, but he was correct that his cheap motu pci-e interface had deterministic latency where no usb interface did. Yes I am familiar with the Zmix controversy on the low latency thread on the purple site. He made the same erroneous claim being made here that *all* USB interfaces don't provide repeatable deterministic latency. Yes there are many interfaces on various protocols that amazingly do not provide repeatable deterministic latency, not just USB. This is more due to the complexity and difficulty of writing a good driver for an audio interface. Zmix never revealed which USB device(s) he tested and was universally rebuked for making bold claims and blanket statements with no supporting evidence. He also stated the he "purchased an entire 48 channel "ethernet" based system and discovered that it could not provide deterministic, sample accurate latency, and 3 years later the manufacturer still has not solved it." ... Yes because it's a driver problem, not a USB problem although ethernet and USB are more difficult to write a good driver for than PCIe/Thunderbolt. I wonder what ethernet based system he was referring to? At least we can now agree that RME USB interfaces do in fact provide sample accurate repeatable deterministic latency unlike a lot of interfaces. Dan: Where can I learn more about the claim that RME USB is not subsample accurate specifically?
|
|
|
Post by gravesnumber9 on Nov 4, 2024 16:52:51 GMT -6
Massive sale right now on MH. Taking $1k plus off.
|
|
|
Post by Quint on Nov 4, 2024 17:13:36 GMT -6
Massive sale right now on MH. Taking $1k plus off. Man, you know, if the MHLink just had lower latency, I would have considered switching to that from my Thunderbolt setup, and then gotten a MADI Edge Card to bring in MADI that way. That would have been a great solution for my needs. But the latency.... Great sale price though. Damnit!
|
|
|
Post by plinker on Nov 4, 2024 17:21:36 GMT -6
Massive sale right now on MH. Taking $1k plus off. Wow! That's serious!! I guess pro audio sales really are down, huh?! I better get my sinkers ready to do some bottom-feeding on the used market!
|
|
|
Post by veggieryan on Nov 4, 2024 17:26:46 GMT -6
Massive sale right now on MH. Taking $1k plus off. Man, you know, if the MHLink just had lower latency, I would have considered switching to that from my Thunderbolt setup, and then gotten a MADI Edge Card to bring in MADI that way. That would have been a great solution for my needs. But the latency.... Great sale price though. Damnit! At this price its easier to justify an RME MADI PCIe as the interface to/from a MADI Edge card... Then no need for the MHLink headache...
|
|