If you'd like to support my channel you can use these links to shop at: Sweetwater: sweetwater.sjv.io/NkYJMb Thomann: www.thomann.de/de/index.html?offid=1&affid=2481 ZZounds: www.zzounds.com/a--3989122
Apple makes Metal and OpenCL available to developers. It's that IK is utilizing CUDA, which is proprietary to nVidia GPUs. It's becoming more of a problem in the computer world because of AI, so expect to see some changes over time. Possibly a licensing of CUDA, or a shift toward open computer platforms. Same limitation would happen on a Windows machine with an AMD GPU. Basically the distinction here is nVidia accelleration, vs. not.
I'm understanding that more and more...I need to be careful going to deep into under the hood issues...but it's the performance that ultimately matters to me! :) Thanks for the insight there...
@@lalnuntluangachhakchhuak5767 not at all. They use their own GPUs in their Apple Silicon chips, and nVidia currently doesn't license CUDA for use outside of their architecture. OpenCL is the open source alternative, and Apple also has Metal which is their proprietary alternative. Both could perform similarly to CUDA, but IK has only coded in CUDA support so far. They could choose to include OpenCL which is compatible with nVidia, AMD, intel, and Apple alike.
Thanks! I’m having fun with that. Yes. Just look for Lonely Rocker on tone.net. I have a bunch of tone models up there. Let me know what you think of them!
I believe the base machine learning libraries that drive the model learning process are optimized for NVidia GPUs. My 5 year old RTX2070S does an advanced model in about 30 minutes.
They're just giving away ram these days! I thought having 16gb in my laptop was a major upgrade (well it was) but holy moly... Kreative Devices doing some great things for creative people in need without the professional budget to do so. Cheers!
interesting insight, i can imagine the disappointment in the mac considering the price/spec but there are certainly other things that it excels at maybe that will change in the future, as software gets more optimised, but i have a feeling it's going to remain with GPU optimisations in mind it'd be interesting to see how NAM training would compare on each system
It was quite the discovery. I got the Mac Studio when I REALLY needed it. And it's been a good machine but not as good as I thought it would have been. That's a lot of money for a computer. I haven't gotten into NAM yet...I should get on that....
@@LonelyRocker i can imagine for everything else it's rock solid, unfortunately there will be discrepancies between the two platforms, I've been on PC forever, for gaming it's a no brainer i guess, and for audio you need to invest more i feel in a good interface /drivers (RME are really good on PC) than an equivalent mac setup (their audio infrastructure just seems more structuraly efficient than asio depending on drivers)
I disabled HAGS on my desktop (7900x and RTX4090) and it cut rendering time exactly in half. I've since "upgraded" to a 7800x3D for gaming, but it is still fast. Like in "normal" rendering it takes less than 3 minutes to render a capture. I honestly couldn't hear a difference between normal and advanced rendering.
Nice. I’m really enjoying the speed boosts now. The differences will be subtle but distortion is very complex so while the befits may not be immediately heard they are there. Sometimes it could be deficiencies on the gear we are listening to them in.
Update to my update. New M4 Pro Mini base with 24/512. Advanced training 78 min. That's 25 min slower than my M1. Didn't figure that one. Was there an update to ToneX training in the past month? Not adding up here. Just ran an advanced training on my Ryzen 7 PC and got 17 min, so an hour faster than my M4Pro.
It’s all about the GPU. Thats the limitation on Mac. Nvidia cards deliver the fastest results with TONEX currently. Nothing much to do with the processor as long as it is reasonably modern.
I use both Mac and Windows based machines, and the Apple people will get their feelings hurt but anytime you can take dedicated hardware to accelerate (Nvidia, AMD GPU's) it's going to absolutely crush the Apple hardware. Even if Tone-X was accelerating on Apple with their hardware GPU's (part of the M series architecture) the sheer power of NVDIA or AMD GPU vs Apple is just comical. If apple wanted to compete, they should have stuck with Intel/3rd Party GPU solutions understanding that the future was with GPU (specialized task processing) as the way forward. For day to day the M series is fantastic, but it's just a jack of all things with outstanding perf all around, but really not a master of anything outside of the box.
I’m not going to lie. I miss the modular tower Macs. The closed integrated systems take away a lot of the fun of owning a big computer. Sadly this is where Apple is now.
ToneX and NAM both train MUCH faster with a compatible NVidia GPU. My ToneX advanced captures went from >8 hours to about 30 minutes when I swapped out my AMD GPU for one of the slowest NVidia GPU cards on the HCL for ToneX.
ToneX, along with NAM and quite a few other Machine Learning-centric apps use NVidia's CUDA technology to interface with the GPU. CUDA, or Compute Unified Device Architecture, is a parallel computing platform and API that allows software to use GPUs for faster processing. NVIDIA developed CUDA to help developers speed up computing applications by using the power of GPUs. As far as I know, it's specific to NVidia GPUs so systems that don't use NVidia are S.O.L. I agree - with the rise in uses of Machine Learning and AI, Apple's gonna have to come up with an answer to this. I know their processors include what's supposed to be a ridiculously fast Neural Network for this task, I don't know to what extent they've made their API's available to developers. That said, I'm sure there's something IK could do to interface with this but it would probably need to rewrite a substantial percentage of the app's code & why would they wanna do that?
Apple silicon just wont compete with Nvidia for these specific tasks. Its a task specific thing and Nvidia is basically the world leader in this right now. Apple might be able to tweak things a little for these types of applications within software but not to the point of being competitive with Nvidia.
Apple’s closed, proprietary architecture serves their interests for control and unifying the Apple experience across validated applications but, as in this case, shows how it could fall behind. The differences in performance here make the Mac look bad. IK chose an architecture that served their best interests and that meant leaving Apple behind. It’s a shame it has to be this way sometimes.
@@LonelyRocker I dont think apple being closed is the problem. AMD and Intel are not closed but IK is not optimizing for them either they are optimizing for Nvidia CUDA tech. They really should optimize for Compute methodology and at least give other GPU's a chance to pull that time down some for people on those other platforms (apple, PC with intel or AMD GPU).
Hmmm my 4080 does it in the same time apparently. Its a shame they dont optimize the process to use more GPU resources but Im not complaining with an 18 minute advanced process
Something isn't right, my Mac Mini M1 with just 8 gigs is way faster than your Studio??? The default training is under 10 min. Cant remember exact time for Advanced. I'll be back and update, I haven't run any captures in months. Still, something is wrong with your Mac. EDIT: I'm back, not counting the capture, the Standard training is just over 8 min and the Advanced training is 53 min. I've never captured anything I'd wait 53 min for again. Just don't hear the "time value", when an IR or eq change is instant. Maybe I should try it on my gaming PC.
That's interesting. I've often felt my Studio was slower than it should be. I still have support on it. I should reach out to Apple...thanks for sharing!
IK really did a disservice to consumers when opting to only use CUDA which requires an NVIDIA GPU. Apple make their GPU and Neural processors available to software developers. OpenCL is supported on AMD, Intel, Nvidia and Apple GPUs. At some point i can see Apple only supporting the Metal API. In the 3D modeling world CUDA seems to perform better than OpenCL. IDK about Metal. Thankfully my gaming PC has an Nvidia card I can use, while I use my M3 MBP for other things.
Honestly, I can tell you from experience, anything other than a Kreative Devices computer is a waste of money... take it from a former mac user that owns 2 Kreative Devices Laptops and Displays.
I've learned that very quickly. Great company that really knows their customers and gives them what they actually need and VERY fair prices. Got to respect that 100%!
Uh no...I asked the "PC Company" to send me a laptop because I did not have a PC here to do the comparison. And I DID mention it was the GPU....I was very clear about that...but obviously you didn't actually watch the video....yet you did comment, very curious...
@@LonelyRocker Watch you own video. I sat there almost all the way to the end and all you did what shill about a budget laptop beating a mac. You kept talking about a Mac vs budget PC fight. It had nothing to do with that. There are macs with nvidia cards that dont have this issue. You should have stated it from the beginning instead of making an infomercial about that laptop. In the end you even plugged it again. I was here about what the title said.. TONEX capture solved. It got so unbearable that i stopped at the 9 minute mark where you talked about the "result" for almost 8 full minutes. I'm not mad at you brother. I just felt the format to not be direct enough for my taste...my opinion. My original comment still stands based on how I originally perceived the video. I never made any direct assumption-based comments but you did about me. I said it 'seemed". That the way your video seemed. Yet you tell me that I obviously did'nt watch the video...which I did. What is curious..I'm a bot or something? Was just constructive criticism man. Don't let the internet get to you brother.
If you'd like to support my channel you can use these links to shop at:
Sweetwater: sweetwater.sjv.io/NkYJMb
Thomann: www.thomann.de/de/index.html?offid=1&affid=2481
ZZounds: www.zzounds.com/a--3989122
Apple makes Metal and OpenCL available to developers. It's that IK is utilizing CUDA, which is proprietary to nVidia GPUs. It's becoming more of a problem in the computer world because of AI, so expect to see some changes over time. Possibly a licensing of CUDA, or a shift toward open computer platforms. Same limitation would happen on a Windows machine with an AMD GPU. Basically the distinction here is nVidia accelleration, vs. not.
I'm understanding that more and more...I need to be careful going to deep into under the hood issues...but it's the performance that ultimately matters to me! :) Thanks for the insight there...
Something to do with Apple and their Egos. They had a long lasting fued with Nvidia, which is the biggest player in AI and Graphic processing unit
@@lalnuntluangachhakchhuak5767 not at all. They use their own GPUs in their Apple Silicon chips, and nVidia currently doesn't license CUDA for use outside of their architecture. OpenCL is the open source alternative, and Apple also has Metal which is their proprietary alternative. Both could perform similarly to CUDA, but IK has only coded in CUDA support so far. They could choose to include OpenCL which is compatible with nVidia, AMD, intel, and Apple alike.
That’s a cool setup 😊, I use tonex myself , I don’t capture myself sadly but love downloading other peoples.
Do you upload your captures ?
Thanks! I’m having fun with that. Yes. Just look for Lonely Rocker on tone.net. I have a bunch of tone models up there. Let me know what you think of them!
I believe the base machine learning libraries that drive the model learning process are optimized for NVidia GPUs. My 5 year old RTX2070S does an advanced model in about 30 minutes.
It really seems so...what a difference!!! I can't believe how many hours I have wasted!!!
Good video Dan!!
Thank you kindest of strangers…..😂
They're just giving away ram these days! I thought having 16gb in my laptop was a major upgrade (well it was) but holy moly... Kreative Devices doing some great things for creative people in need without the professional budget to do so. Cheers!
16GB? That's so 2004!!!! I recommend at least 48GB for a machine where you are recording and mixing....
You need every GB of it, devs are lazy af these days
interesting insight, i can imagine the disappointment in the mac considering the price/spec but there are certainly other things that it excels at
maybe that will change in the future, as software gets more optimised, but i have a feeling it's going to remain with GPU optimisations in mind
it'd be interesting to see how NAM training would compare on each system
It was quite the discovery. I got the Mac Studio when I REALLY needed it. And it's been a good machine but not as good as I thought it would have been. That's a lot of money for a computer. I haven't gotten into NAM yet...I should get on that....
@@LonelyRocker i can imagine for everything else it's rock solid, unfortunately there will be discrepancies between the two platforms, I've been on PC forever, for gaming it's a no brainer i guess, and for audio you need to invest more i feel in a good interface /drivers (RME are really good on PC) than an equivalent mac setup (their audio infrastructure just seems more structuraly efficient than asio depending on drivers)
I disabled HAGS on my desktop (7900x and RTX4090) and it cut rendering time exactly in half. I've since "upgraded" to a 7800x3D for gaming, but it is still fast. Like in "normal" rendering it takes less than 3 minutes to render a capture. I honestly couldn't hear a difference between normal and advanced rendering.
Nice. I’m really enjoying the speed boosts now. The differences will be subtle but distortion is very complex so while the befits may not be immediately heard they are there. Sometimes it could be deficiencies on the gear we are listening to them in.
@@LonelyRockerI feel it in the bounce of the string feel idk
Update to my update. New M4 Pro Mini base with 24/512. Advanced training 78 min. That's 25 min slower than my M1. Didn't figure that one. Was there an update to ToneX training in the past month? Not adding up here. Just ran an advanced training on my Ryzen 7 PC and got 17 min, so an hour faster than my M4Pro.
It’s all about the GPU. Thats the limitation on Mac. Nvidia cards deliver the fastest results with TONEX currently. Nothing much to do with the processor as long as it is reasonably modern.
I use both Mac and Windows based machines, and the Apple people will get their feelings hurt but anytime you can take dedicated hardware to accelerate (Nvidia, AMD GPU's) it's going to absolutely crush the Apple hardware. Even if Tone-X was accelerating on Apple with their hardware GPU's (part of the M series architecture) the sheer power of NVDIA or AMD GPU vs Apple is just comical. If apple wanted to compete, they should have stuck with Intel/3rd Party GPU solutions understanding that the future was with GPU (specialized task processing) as the way forward. For day to day the M series is fantastic, but it's just a jack of all things with outstanding perf all around, but really not a master of anything outside of the box.
I’m not going to lie. I miss the modular tower Macs. The closed integrated systems take away a lot of the fun of owning a big computer. Sadly this is where Apple is now.
ToneX and NAM both train MUCH faster with a compatible NVidia GPU. My ToneX advanced captures went from >8 hours to about 30 minutes when I swapped out my AMD GPU for one of the slowest NVidia GPU cards on the HCL for ToneX.
Nvidia seems like the trick....but damn, that Mac needs to be better than that!! We;;, glad I have a great laptop now to handle the job!
ToneX, along with NAM and quite a few other Machine Learning-centric apps use NVidia's CUDA technology to interface with the GPU. CUDA, or Compute Unified Device Architecture, is a parallel computing platform and API that allows software to use GPUs for faster processing. NVIDIA developed CUDA to help developers speed up computing applications by using the power of GPUs. As far as I know, it's specific to NVidia GPUs so systems that don't use NVidia are S.O.L.
I agree - with the rise in uses of Machine Learning and AI, Apple's gonna have to come up with an answer to this. I know their processors include what's supposed to be a ridiculously fast Neural Network for this task, I don't know to what extent they've made their API's available to developers. That said, I'm sure there's something IK could do to interface with this but it would probably need to rewrite a substantial percentage of the app's code & why would they wanna do that?
Apple silicon just wont compete with Nvidia for these specific tasks. Its a task specific thing and Nvidia is basically the world leader in this right now. Apple might be able to tweak things a little for these types of applications within software but not to the point of being competitive with Nvidia.
Apple’s closed, proprietary architecture serves their interests for control and unifying the Apple experience across validated applications but, as in this case, shows how it could fall behind. The differences in performance here make the Mac look bad. IK chose an architecture that served their best interests and that meant leaving Apple behind. It’s a shame it has to be this way sometimes.
@@LonelyRocker I dont think apple being closed is the problem. AMD and Intel are not closed but IK is not optimizing for them either they are optimizing for Nvidia CUDA tech. They really should optimize for Compute methodology and at least give other GPU's a chance to pull that time down some for people on those other platforms (apple, PC with intel or AMD GPU).
Hmmm my 4080 does it in the same time apparently. Its a shame they dont optimize the process to use more GPU resources but Im not complaining with an 18 minute advanced process
Glad that's working for you....
Something isn't right, my Mac Mini M1 with just 8 gigs is way faster than your Studio??? The default training is under 10 min. Cant remember exact time for Advanced. I'll be back and update, I haven't run any captures in months. Still, something is wrong with your Mac. EDIT: I'm back, not counting the capture, the Standard training is just over 8 min and the Advanced training is 53 min. I've never captured anything I'd wait 53 min for again. Just don't hear the "time value", when an IR or eq change is instant. Maybe I should try it on my gaming PC.
That's interesting. I've often felt my Studio was slower than it should be. I still have support on it. I should reach out to Apple...thanks for sharing!
@@LonelyRocker Id love to hear what you find out on this hopefully you do a video.
@@StoneXue Because of this comment I called Apple Support. I have an appointment on Friday for them to look at it. Let’s see what happens!
Mac Studio and battery?????
Huh?
IK really did a disservice to consumers when opting to only use CUDA which requires an NVIDIA GPU. Apple make their GPU and Neural processors available to software developers. OpenCL is supported on AMD, Intel, Nvidia and Apple GPUs. At some point i can see Apple only supporting the Metal API. In the 3D modeling world CUDA seems to perform better than OpenCL. IDK about Metal. Thankfully my gaming PC has an Nvidia card I can use, while I use my M3 MBP for other things.
I'm sure there are reasons why they choose the dev paths they do....
Love Kreative Devices! Windows FTW! ;)
Great machine!! I have seen the light! ;-)
Straight 🔥
I have a Nvidia 3070 gets it done in about 10 min
Oh nice!
Honestly, I can tell you from experience, anything other than a Kreative Devices computer is a waste of money... take it from a former mac user that owns 2 Kreative Devices Laptops and Displays.
I've learned that very quickly. Great company that really knows their customers and gives them what they actually need and VERY fair prices. Got to respect that 100%!
It's about the GPU being used, not the computer. It seems like you just wanna plug for the PC company that sent you the laptop.
Uh no...I asked the "PC Company" to send me a laptop because I did not have a PC here to do the comparison. And I DID mention it was the GPU....I was very clear about that...but obviously you didn't actually watch the video....yet you did comment, very curious...
@@LonelyRocker Watch you own video. I sat there almost all the way to the end and all you did what shill about a budget laptop beating a mac. You kept talking about a Mac vs budget PC fight. It had nothing to do with that. There are macs with nvidia cards that dont have this issue. You should have stated it from the beginning instead of making an infomercial about that laptop. In the end you even plugged it again. I was here about what the title said.. TONEX capture solved. It got so unbearable that i stopped at the 9 minute mark where you talked about the "result" for almost 8 full minutes. I'm not mad at you brother. I just felt the format to not be direct enough for my taste...my opinion. My original comment still stands based on how I originally perceived the video. I never made any direct assumption-based comments but you did about me. I said it 'seemed". That the way your video seemed. Yet you tell me that I obviously did'nt watch the video...which I did. What is curious..I'm a bot or something? Was just constructive criticism man. Don't let the internet get to you brother.
I'll never go back to mac
That would be a day hell freezes over ;-)