NEW: Octane X - is finally here! [MEGA FAQ THREAD]
Posted: Mon Jul 20, 2020 12:49 pmDear Octane community,
Well, here we are. Octane X is live. What a journey it's been.
You can read the full press release here.
The incredible team at OTOY did truly mind blowing work to make this happen. I am humbled to be part of it.
Maybe every GPU renderer will one day have to do what we did, and maybe it will be simpler. Being out of the gate first does seem hard, especially because so much has to be invented along the way.
I must thank the great folks at Apple who helped us with the hardest parts, some of whom we have been working with for the better part of a decade on this. There is basically no way Octane has any future on Apple's hardware going forward without us doing this work, and we needed their help to make it happen this year.
The rest of this thread will begin to answer the questions everyone must have, and there are many. I'll start with the obvious ones and we will update this post with more once we get additional feedback from users
How do I get Octane X?
If you have an Octane Studio or Enterprise sub, then Octane X standalone, and all DCC plug-ins will be available in the downloads section. If you are on studio, make sure you have a sub with all access - i.e. if you are on an old PayPal sub, make sure to get on the new one using Stripe, which includes this.
What does Octane PR1 mean? What is 'PR1' vs XB1/RC1 etc ?
Octane X went through XB/RC releases in a months long closed beta PR1 is our first public release following that closed beta. All plug-ins are ready on day 1, and everything is available on the downloads section as we normally do when we have a stable release. That does not mean PR1 is free of bugs - that would be optimistic - but we fixed all the issues that our beta testers found, and then some.
What hardware and macOS versions are supported in PR1:
For PR1, we are supporting AMD Vega and Navi GPUs on macOS 10.15.6. We need 10.15.6 to support out of core memory, there wasn't much point in releasing without it, so that was part of why we wanted to wait until it was out last week.
Intel and other GPUs will be coming in a future update for macOS 11. We will be keeping the closed beta group going to test macOS 11 builds on Intel and other GPUs.
Will there be support for pre-Vega AMD GPUs (like Radeon Pro 580X)?
Support for pre-Vega AMD GPUs (like Radeon Pro 580X) is currently blocked due to Metal drivers for these earlier generation AMD cards missing support new Metal features Octane X needs to run on macos 10.15.6. If this situation ever changes with a new driver release, we can recompile Octane X to support this hardware. Based on our current understanding, a driver update is not likely to happen soon, if ever. We therefore recommend users looking to run Octane X on AMD, plan on getting Vega or later generation AMD GPUs (at least as an eGPU if no other option is possible).
UPDATE December 1st 2020: PR7 (Big Sur 11.1) will include support for AMD Polaris GPUs, Intel SkyLake and Apple M1 (Beta)
Demo version: If I am not a subscriber, is there a demo version of Octane X I can try out now, and if so can I get it for my DCC plug-in?
Yes, there is an Octane X demo (Metal macoS) version available here - currently the demo version is just for standalone (the macOS dcc demo plug-ins are currently all CUDA), but we will be adding Octane X Metal demo versions of all available macOS plug-ins as soon as the developers get a chance to build them.
What versions of macOS dcc plug-ins work with Octane X macOS?
Note - the Octane X plug-in for C4D should be run on R21. Ahmet has noted instability in <R20 and S22, so R21 is the version to use with Octane X. Same with Maya 2020 and SketchUp 2020 - these are the versions to to use with Octane X vs. the 2019 versions of these applications.
Will Octane X PR1 work on macOS 11 beta?
The current PR1 build is only meant to run on 10.15.6, not an earlier or later version of macOS. There will be separate release down the line specifically with macOS 11.
Does Octane X work with NVIDIA cards on macOS:
Octane X is built entirely on the Apple Metal API for macOS 10.15.6 and later. At this time, there are no Metal drivers on this OS for NVIDIA, so Octane X only supports AMD. You can use macOS OctaneRender 2020 CUDA as we update it to 2020 final this year on 10.13, but this is not a long term solution - so we suggest mac users consider this when planning around future hardware investments
Does Octane X work with eGPUs?
Yes, perfectly with AMD Vega and Navi GPUs - such as the Radeon VII and RX5700XT.
Is Octane X really going to be free with macOS 11?
Yes, on new macs, and for new users who would normally sign up for Prime Tier. More details will be shared in the coming weeks. The free Octane X offer is for users who don't have Octane subs, and who do not intend to use OctaneRender CUDA (the free offer will not include CUDA licences, so if you have an existing sub, stick with it - you will get both CUDA and Metal licences - even if you can only use one at a time)
Where is OctaneBench for Metal?
We will release an official OB app with macOS 11, where we think we can probably get another 10-15% speed over today's release on 10.15 and also properly support Intel and other Metal GPUs.
I just want to know the OB score of all the AMD/Intel GPUs on 10.15 now
Just to be clear, these are still early numbers and have a +/- variance of 7% or so, but one thing that will likely hold is the relative performance in OB between these GPUs :
Radeon Pro Vega II Duo - 412 OB*
Radeon Pro Vega II - 206 OB
Radeon VII - 200 OB
5700 XT - 170 OB
Vega FE - ~161 OB
Vega 64 -~148 OB
Vega 56 - ~117 OB
Vega 48 - 108 OB
5600M - 108 OB
5500M - 77 OB
Vega 20 - ~50 OB
Intel iris 640 - ~17 OB
Intel iris 630 - ~15 OB
Apple A13 - ~15 OB
*dual GPUs on single graphics board (the 415 in the PR and 419 in the blog post are different runs that illustrate the variance noted above). Note that we had referred to this initially in the announcement post as a single slot card, but as some have pointed out, it’s more correct terminology to label this a single ‘board’ graphics card, so as to avoid confusion with measurement of the card’s slot ‘width’ space. We do want to be accurate on every aspect of OctaneBench scoring and categorization.
UPDATE - December 5th 2020: PR7 on macOS 11.1 Big Sur is about 8-10% faster than PR1 on Catalina:
(with more tuning to be done in the months ahead for 2021 version releases)
Radeon Pro Vega II Duo - 446 OB (dual ASIC GPU)
Radeon Pro Vega II - 223 OB
Radeon VII - 214 OB
5700 XT - 184 OB
Vega FE - ~174 OB
Vega 64 -~160 OB
Vega 56 - ~126 OB
Vega 48 - 117 OB
5600M - 117 OB
5500M - 83 OB
Vega 20 - ~55 OB
Apple M1 (MacBook Air) (TBD - still being tuned, but our Transformers scene is 7.4x slower on M1 than on a 600 OB RTX 3080 | OctaneRender 2020.2 w/ RTX on)
Apple A14 (iPhone 12/iPad Air 4) - ~20 OB
Intel iris 640 - ~17 OB
Intel iris 630 - ~15 OB
Apple A13 - ~15 OB
-----
Does rendering with Octane X on Metal give you the same results as OctaneRender on CUDA?
Yes, Octane X has pixel parity with OctaneRender 2020.1. There are some caveats with user generated OSL shaders on macOS 10.15 (more on that below), but otherwise everything is identical. You can send ORBX files from Octane X to RNDR and they will render identically on CUDA devices.
Can I mix octane X (Metal) and OctaneRender 2020 (CUDA) render nodes over net render?
You cannot mix Metal and Cuda over net render - yet - Octane X is a version ahead of CUDA, and we need to sync them together after Octane X macOS 11 and after OctaneRender 2020.2 are done. The release after those is where we would have the first opportunity to support mixed Metal/CUDA network rendering.
What is this user generated OSL code limitation on 10.15?:
Octane X on 10.15 does compile live OSL code - this functionality is actually working on Metal (thanks Apple!), and we've tested it on non-public drivers (even on iOS), but it is not ready to get into the public drivers for 10.15.
What this means is that OSL shaders that we do not bundle into the Octane X releases on 10.15 will not render correctly - this currently includes 3rd party OSL textures, Maxon OSL noises, OSL camera and OSL vectron shaders. We do plan to bundle most of these in PR2 - with a large number of OSL utility shaders already planned to be moved to core in 2020.2. For the rest, we will introduce an OLS interpreter (not the same as a compiler, it's slower) in PR3 that will provide full compliance on all scenes until we get full perf. OSL compiling working in drivers on macOS 11.
What is Octane X - why isn't it called/bundled OctaneRender like the old Mac CUDA versions?
While Octane X is feature identical to OctaneRender (CUDA), and supports all the same plug-ins and standalone as of right now, things will begin to diverge once we hit macOS 11 and put Octane X on the app store (including iOS/ipadOS). The Octane X app will be built on top of standalone, but we can't treat it as a normal desktop exe - for one thing, being on the app store means it must always be the latest version. It also can't really work as a normal DCC plug-in on iPad/iOS, and this means rethinking how we make it useful on those devices (basically bringing USD, Sculptron and EmberGenFX together is a good start).
Why is there a separate Octane X Render Node for standalone and plug-ins?
Standalone is actually one build ahead of the plug-ins (with a minor bug fix), and until the plug-ins catch up to this version in a few days time, we wanted to support network rendering in both.
Will EmberGenFX and Sculptron be coming to Octane X macOS/iOS?
Yes. Once we they are both out of beta and working as embedded modules. No form ETA on when that will happen.
What is the new Hydra/Solaris plug-in?
It is allows Octane to run in the Solaris viewport. It must be installed with the H18 plug-in.
What about Octane X for Unreal and Unity
Due to the unique nature of both these integrations, we never made macOS versions on CUDA. So the Metal Octane X plug-ins will be the first time these are available on Mac. We are planning to ship thee after macOS 11 is out.
Is there an RTX on mode on AMD or other Metal GPUs?
No, that doesn't exist , but if it does, we will support it fully.
Does Octane X work on the TDK?
No. Octane X does work on an iPhone 11/SE, and while that version is fully optimized for ARM64 and Apple GPUs, it needs an A13. The TDK is an A12Z. Note that Octane plug-ins on Apple Silocon will likely remain x86 so they can work with x86 apps using Rosetta2. Octane X app/standalone will be fully optimized as a Universal2 binary app for Apple Silicon.
What device handles AI Denoising on Octane X?
it is done using the AVX2 instructions on the Intel cores. This opens up the possibility that we can do AI denoising across multiple light passes. No firm ETA on that. Also you will need an AVX2 capable CPU, which means 2015+.
Is Brigade coming to Octane X?
Yes, and when we can share more about the Brigade roadmap, we will have some additional updates.
Does this mean Octane for AMD/Intel on Windows is now possible?
The work to support AMD and Intel (and others) was completed alongside our Metal work, but that hasn't really brought us much close to shipping anything on those devices. We really need Vulkan to catch up to Metal and it's not clear when that might happen. As of right now, the Vulkan drivers on AMD and Intel are not mature enough to compile (much less ship) Octane for Vulkan. The only Vulkan drivers that do run Octane for Vulkan just fine (which is how we know it works!) are NVIDIA's - and we will be using Vulkan+RTX for Brigade on NVIDIA.
....
More to be added - but this should get the ball rolling!
-J
Well, here we are. Octane X is live. What a journey it's been.
You can read the full press release here.
The incredible team at OTOY did truly mind blowing work to make this happen. I am humbled to be part of it.
Maybe every GPU renderer will one day have to do what we did, and maybe it will be simpler. Being out of the gate first does seem hard, especially because so much has to be invented along the way.
I must thank the great folks at Apple who helped us with the hardest parts, some of whom we have been working with for the better part of a decade on this. There is basically no way Octane has any future on Apple's hardware going forward without us doing this work, and we needed their help to make it happen this year.
The rest of this thread will begin to answer the questions everyone must have, and there are many. I'll start with the obvious ones and we will update this post with more once we get additional feedback from users
How do I get Octane X?
If you have an Octane Studio or Enterprise sub, then Octane X standalone, and all DCC plug-ins will be available in the downloads section. If you are on studio, make sure you have a sub with all access - i.e. if you are on an old PayPal sub, make sure to get on the new one using Stripe, which includes this.
What does Octane PR1 mean? What is 'PR1' vs XB1/RC1 etc ?
Octane X went through XB/RC releases in a months long closed beta PR1 is our first public release following that closed beta. All plug-ins are ready on day 1, and everything is available on the downloads section as we normally do when we have a stable release. That does not mean PR1 is free of bugs - that would be optimistic - but we fixed all the issues that our beta testers found, and then some.
What hardware and macOS versions are supported in PR1:
For PR1, we are supporting AMD Vega and Navi GPUs on macOS 10.15.6. We need 10.15.6 to support out of core memory, there wasn't much point in releasing without it, so that was part of why we wanted to wait until it was out last week.
Intel and other GPUs will be coming in a future update for macOS 11. We will be keeping the closed beta group going to test macOS 11 builds on Intel and other GPUs.
Will there be support for pre-Vega AMD GPUs (like Radeon Pro 580X)?
Support for pre-Vega AMD GPUs (like Radeon Pro 580X) is currently blocked due to Metal drivers for these earlier generation AMD cards missing support new Metal features Octane X needs to run on macos 10.15.6. If this situation ever changes with a new driver release, we can recompile Octane X to support this hardware. Based on our current understanding, a driver update is not likely to happen soon, if ever. We therefore recommend users looking to run Octane X on AMD, plan on getting Vega or later generation AMD GPUs (at least as an eGPU if no other option is possible).
UPDATE December 1st 2020: PR7 (Big Sur 11.1) will include support for AMD Polaris GPUs, Intel SkyLake and Apple M1 (Beta)
Demo version: If I am not a subscriber, is there a demo version of Octane X I can try out now, and if so can I get it for my DCC plug-in?
Yes, there is an Octane X demo (Metal macoS) version available here - currently the demo version is just for standalone (the macOS dcc demo plug-ins are currently all CUDA), but we will be adding Octane X Metal demo versions of all available macOS plug-ins as soon as the developers get a chance to build them.
What versions of macOS dcc plug-ins work with Octane X macOS?
Note - the Octane X plug-in for C4D should be run on R21. Ahmet has noted instability in <R20 and S22, so R21 is the version to use with Octane X. Same with Maya 2020 and SketchUp 2020 - these are the versions to to use with Octane X vs. the 2019 versions of these applications.
Will Octane X PR1 work on macOS 11 beta?
The current PR1 build is only meant to run on 10.15.6, not an earlier or later version of macOS. There will be separate release down the line specifically with macOS 11.
Does Octane X work with NVIDIA cards on macOS:
Octane X is built entirely on the Apple Metal API for macOS 10.15.6 and later. At this time, there are no Metal drivers on this OS for NVIDIA, so Octane X only supports AMD. You can use macOS OctaneRender 2020 CUDA as we update it to 2020 final this year on 10.13, but this is not a long term solution - so we suggest mac users consider this when planning around future hardware investments
Does Octane X work with eGPUs?
Yes, perfectly with AMD Vega and Navi GPUs - such as the Radeon VII and RX5700XT.
Is Octane X really going to be free with macOS 11?
Yes, on new macs, and for new users who would normally sign up for Prime Tier. More details will be shared in the coming weeks. The free Octane X offer is for users who don't have Octane subs, and who do not intend to use OctaneRender CUDA (the free offer will not include CUDA licences, so if you have an existing sub, stick with it - you will get both CUDA and Metal licences - even if you can only use one at a time)
Where is OctaneBench for Metal?
We will release an official OB app with macOS 11, where we think we can probably get another 10-15% speed over today's release on 10.15 and also properly support Intel and other Metal GPUs.
I just want to know the OB score of all the AMD/Intel GPUs on 10.15 now
Just to be clear, these are still early numbers and have a +/- variance of 7% or so, but one thing that will likely hold is the relative performance in OB between these GPUs :
Radeon Pro Vega II Duo - 412 OB*
Radeon Pro Vega II - 206 OB
Radeon VII - 200 OB
5700 XT - 170 OB
Vega FE - ~161 OB
Vega 64 -~148 OB
Vega 56 - ~117 OB
Vega 48 - 108 OB
5600M - 108 OB
5500M - 77 OB
Vega 20 - ~50 OB
Intel iris 640 - ~17 OB
Intel iris 630 - ~15 OB
Apple A13 - ~15 OB
*dual GPUs on single graphics board (the 415 in the PR and 419 in the blog post are different runs that illustrate the variance noted above). Note that we had referred to this initially in the announcement post as a single slot card, but as some have pointed out, it’s more correct terminology to label this a single ‘board’ graphics card, so as to avoid confusion with measurement of the card’s slot ‘width’ space. We do want to be accurate on every aspect of OctaneBench scoring and categorization.
UPDATE - December 5th 2020: PR7 on macOS 11.1 Big Sur is about 8-10% faster than PR1 on Catalina:
(with more tuning to be done in the months ahead for 2021 version releases)
Radeon Pro Vega II Duo - 446 OB (dual ASIC GPU)
Radeon Pro Vega II - 223 OB
Radeon VII - 214 OB
5700 XT - 184 OB
Vega FE - ~174 OB
Vega 64 -~160 OB
Vega 56 - ~126 OB
Vega 48 - 117 OB
5600M - 117 OB
5500M - 83 OB
Vega 20 - ~55 OB
Apple M1 (MacBook Air) (TBD - still being tuned, but our Transformers scene is 7.4x slower on M1 than on a 600 OB RTX 3080 | OctaneRender 2020.2 w/ RTX on)
Apple A14 (iPhone 12/iPad Air 4) - ~20 OB
Intel iris 640 - ~17 OB
Intel iris 630 - ~15 OB
Apple A13 - ~15 OB
-----
Does rendering with Octane X on Metal give you the same results as OctaneRender on CUDA?
Yes, Octane X has pixel parity with OctaneRender 2020.1. There are some caveats with user generated OSL shaders on macOS 10.15 (more on that below), but otherwise everything is identical. You can send ORBX files from Octane X to RNDR and they will render identically on CUDA devices.
Can I mix octane X (Metal) and OctaneRender 2020 (CUDA) render nodes over net render?
You cannot mix Metal and Cuda over net render - yet - Octane X is a version ahead of CUDA, and we need to sync them together after Octane X macOS 11 and after OctaneRender 2020.2 are done. The release after those is where we would have the first opportunity to support mixed Metal/CUDA network rendering.
What is this user generated OSL code limitation on 10.15?:
Octane X on 10.15 does compile live OSL code - this functionality is actually working on Metal (thanks Apple!), and we've tested it on non-public drivers (even on iOS), but it is not ready to get into the public drivers for 10.15.
What this means is that OSL shaders that we do not bundle into the Octane X releases on 10.15 will not render correctly - this currently includes 3rd party OSL textures, Maxon OSL noises, OSL camera and OSL vectron shaders. We do plan to bundle most of these in PR2 - with a large number of OSL utility shaders already planned to be moved to core in 2020.2. For the rest, we will introduce an OLS interpreter (not the same as a compiler, it's slower) in PR3 that will provide full compliance on all scenes until we get full perf. OSL compiling working in drivers on macOS 11.
What is Octane X - why isn't it called/bundled OctaneRender like the old Mac CUDA versions?
While Octane X is feature identical to OctaneRender (CUDA), and supports all the same plug-ins and standalone as of right now, things will begin to diverge once we hit macOS 11 and put Octane X on the app store (including iOS/ipadOS). The Octane X app will be built on top of standalone, but we can't treat it as a normal desktop exe - for one thing, being on the app store means it must always be the latest version. It also can't really work as a normal DCC plug-in on iPad/iOS, and this means rethinking how we make it useful on those devices (basically bringing USD, Sculptron and EmberGenFX together is a good start).
Why is there a separate Octane X Render Node for standalone and plug-ins?
Standalone is actually one build ahead of the plug-ins (with a minor bug fix), and until the plug-ins catch up to this version in a few days time, we wanted to support network rendering in both.
Will EmberGenFX and Sculptron be coming to Octane X macOS/iOS?
Yes. Once we they are both out of beta and working as embedded modules. No form ETA on when that will happen.
What is the new Hydra/Solaris plug-in?
It is allows Octane to run in the Solaris viewport. It must be installed with the H18 plug-in.
What about Octane X for Unreal and Unity
Due to the unique nature of both these integrations, we never made macOS versions on CUDA. So the Metal Octane X plug-ins will be the first time these are available on Mac. We are planning to ship thee after macOS 11 is out.
Is there an RTX on mode on AMD or other Metal GPUs?
No, that doesn't exist , but if it does, we will support it fully.
Does Octane X work on the TDK?
No. Octane X does work on an iPhone 11/SE, and while that version is fully optimized for ARM64 and Apple GPUs, it needs an A13. The TDK is an A12Z. Note that Octane plug-ins on Apple Silocon will likely remain x86 so they can work with x86 apps using Rosetta2. Octane X app/standalone will be fully optimized as a Universal2 binary app for Apple Silicon.
What device handles AI Denoising on Octane X?
it is done using the AVX2 instructions on the Intel cores. This opens up the possibility that we can do AI denoising across multiple light passes. No firm ETA on that. Also you will need an AVX2 capable CPU, which means 2015+.
Is Brigade coming to Octane X?
Yes, and when we can share more about the Brigade roadmap, we will have some additional updates.
Does this mean Octane for AMD/Intel on Windows is now possible?
The work to support AMD and Intel (and others) was completed alongside our Metal work, but that hasn't really brought us much close to shipping anything on those devices. We really need Vulkan to catch up to Metal and it's not clear when that might happen. As of right now, the Vulkan drivers on AMD and Intel are not mature enough to compile (much less ship) Octane for Vulkan. The only Vulkan drivers that do run Octane for Vulkan just fine (which is how we know it works!) are NVIDIA's - and we will be using Vulkan+RTX for Brigade on NVIDIA.
....
More to be added - but this should get the ball rolling!
-J
