Daz Studio Pro BETA - version 4.22.0.15! (*UPDATED*)

1356730

Comments

  • IceCrMn said:

    Is anyone else seeing high CPU temperatures recently while rendering or using iray preview?

    When I say high , I mean over 90C-100C.

    See this thread for more info.

    https://www.daz3d.com/forums/discussion/603371/iray-viewport-render-causing-cpu-gpu-temps-to-soar-and-crash-daz-studio-after-new-nvidia-drivers

    Disabling CPU use in Advanced Iray settings fixes it, but I'm not able to track down the cause.

    ADM Ryzen 7 3800XT

    Studio driver 517.40

    Windows 10 Info

    Edition    Windows 10 Pro
    Version    21H2
    Installed on    ‎9/‎18/‎2021
    OS build    19044.2251
    Experience    Windows Feature Experience Pack 120.2212.4180.0

    Recent OS updates

    KB5019959

    KB5020687

    ---

    Don't know if this is a Studio, GPU driver, or Windows 10 update issue.

    It could also be a hardware issue. It does not sound like a daz Studio issue per se.

  • f7eerf7eer Posts: 117

    Richard Haseltine said:

    IceCrMn said:

    Is anyone else seeing high CPU temperatures recently while rendering or using iray preview?

    When I say high , I mean over 90C-100C.

    See this thread for more info.

    https://www.daz3d.com/forums/discussion/603371/iray-viewport-render-causing-cpu-gpu-temps-to-soar-and-crash-daz-studio-after-new-nvidia-drivers

    Disabling CPU use in Advanced Iray settings fixes it, but I'm not able to track down the cause.

    ADM Ryzen 7 3800XT

    Studio driver 517.40

    Windows 10 Info

    Edition    Windows 10 Pro
    Version    21H2
    Installed on    ‎9/‎18/‎2021
    OS build    19044.2251
    Experience    Windows Feature Experience Pack 120.2212.4180.0

    Recent OS updates

    KB5019959

    KB5020687

    ---

    Don't know if this is a Studio, GPU driver, or Windows 10 update issue.

    It could also be a hardware issue. It does not sound like a daz Studio issue per se.

    This is normal behavior when rendering, and it is one reason why I don't use interactive iray. It happens to a lesser degree when you are using a GPU, but it eventually happens anyway.

  • IceCrMnIceCrMn Posts: 2,106

    I'm only seeing this issue with Studio.

    I can play fallout 4 for hours and the CPU stays under 75C.

    I can also edit video with handbrake for hours with no over heating issues. 

    So far, the only way I can trigger this is to use iray CPU assisted rendering.

    Have either of you monitored you CPU temps to see if you are experiencing the same issue?

    At least one other user in the thread I linked is.

    Would be helpful to know if anyone else is besides the two of us.

  • f7eerf7eer Posts: 117

    f7eer said:

    1. The dForce problem was introduced in nVidia driver 522.30, is known by nVidia, and a fix has not been released by nVidia yet. The workaround is to roll back to driver 517.48 or earlier.

    2. The render problem with OOT hair was introduced in DAZ Studio 4.21.1.13 (Current beta release), and fails with both nVidia drivers 522.30 and 517.40. The workaround is to roll back to DAZ Studio 4.21.0.5 (current general release).

    When I render a scene which contains OOT hair, I get the following errors:

    2022-11-08 07:19:50.362 [WARNING] :: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(377): Iray [ERROR] - IRAY:RENDER ::   1.12  IRAY   rend error: CUDA device 0 (NVIDIA GeForce RTX 2080 Super): Kernel [8] (ShadowEvalSr      ) failed after 0.009s
    2022-11-08 07:19:50.362 [WARNING] :: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(377): Iray [ERROR] - IRAY:RENDER ::   1.12  IRAY   rend error: CUDA device 0 (NVIDIA GeForce RTX 2080 Super): an illegal memory access was encountered (while launching CUDA renderer in <internal>:951)

    Is development aware of this problem? It seems pretty severe. Most (all?) OOT hair uses a custom MDL shader, and the problem may lie there, or it may be a Daz Studio problem, or it may be an nVidia issue. Is this known? Even the oldest OOT hairs I own (dating from 2017) have the problem.

    When the error occurs, Daz Studio can no longer render anything, and must be restarted. Not even clearing the scene fixes it.

  • f7eer said:

    f7eer said:

    1. The dForce problem was introduced in nVidia driver 522.30, is known by nVidia, and a fix has not been released by nVidia yet. The workaround is to roll back to driver 517.48 or earlier.

    2. The render problem with OOT hair was introduced in DAZ Studio 4.21.1.13 (Current beta release), and fails with both nVidia drivers 522.30 and 517.40. The workaround is to roll back to DAZ Studio 4.21.0.5 (current general release).

    When I render a scene which contains OOT hair, I get the following errors:

    2022-11-08 07:19:50.362 [WARNING] :: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(377): Iray [ERROR] - IRAY:RENDER ::   1.12  IRAY   rend error: CUDA device 0 (NVIDIA GeForce RTX 2080 Super): Kernel [8] (ShadowEvalSr      ) failed after 0.009s
    2022-11-08 07:19:50.362 [WARNING] :: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(377): Iray [ERROR] - IRAY:RENDER ::   1.12  IRAY   rend error: CUDA device 0 (NVIDIA GeForce RTX 2080 Super): an illegal memory access was encountered (while launching CUDA renderer in <internal>:951)

    Is development aware of this problem? It seems pretty severe. Most (all?) OOT hair uses a custom MDL shader, and the problem may lie there, or it may be a Daz Studio problem, or it may be an nVidia issue. Is this known? Even the oldest OOT hairs I own (dating from 2017) have the problem.

    When the error occurs, Daz Studio can no longer render anything, and must be restarted. Not even clearing the scene fixes it.

    Yes, Daz is aware of and investigating the issue.

  • IceCrMn said:

    I'm only seeing this issue with Studio.

    I can play fallout 4 for hours and the CPU stays under 75C.

    I can also edit video with handbrake for hours with no over heating issues. 

    So far, the only way I can trigger this is to use iray CPU assisted rendering.

    Have either of you monitored you CPU temps to see if you are experiencing the same issue?

    At least one other user in the thread I linked is.

    Would be helpful to know if anyone else is besides the two of us.

    Iray uses a highly optimized multi-core CPU rendering path so, like say Linx AVX2 benchmark (based on Intel's Linpack) it can cause CPU to use a lot more power than running your average applications or games.

    Your CPU is 105W base TDP, model which is unlocked for overclocking, and comes without HSF assembly in the box which means you are using a custom HSF (cooler).

    Without knowing whether that cooler's capacity is enough for your CPU under full load (100% on all cores for extended periods of time), there's not much else I can say to help you.

    I can only tell you that when my i7-9800X (which has a 165W base TDP and cannot be compared directly with AMD TDP) starts executing AVX512 instructions, it lowers the clock by 400 MHz by default to stay within thermal and power limit. Both AVX2 and especially AVX512 instruction sets are power hogs and Iray is certainly using them in CPU rendering path. On the other hand, CPU is not doing much during gaming (most things are done using GPU, including physics sim).

    Therefore, do some stress tests with Linx and check your cooler.

  • f7eerf7eer Posts: 117

    IceCrMn said:

    I'm only seeing this issue with Studio.

    Me too. Studio is the only program I have that consistently runs the CPU so hard that it reaches 100C, at which point the thermal protection hardware kicks in and lowers the CPU speed. It only happens when I am rendering. Even though the CPU doesn't do any rendering itself, it handles the traffic to and fro with the GPU. The GPU is much faster, and is only 80% busy, and the CPU can't keep up with it. The CPU itself is only 20%, but it's running at such a high clock rate that it heats up much more than the GPU.

    I can play fallout 4 for hours and the CPU stays under 75C.

    I can also edit video with handbrake for hours with no over heating issues. 

    So far, the only way I can trigger this is to use iray CPU assisted rendering.

    Have either of you monitored you CPU temps to see if you are experiencing the same issue?

    Yes, as I reported above. It is completely normal.

  • Heads up, driver (both Game Ready and Studio) 526.98 is out, it fixes Daz Studio dForce issue. I tested it and I can confirm dForce works for me on 4090.

  • IceCrMnIceCrMn Posts: 2,106
    edited November 2022

    @F7eer Currently I'm running an MSI magcore 360 in a Phantec 719 case with 7 140mm case fans, and 3 120mm case fans.

    The AIO and GPU both have 3 fans each.

    Total 16 fans in the system.

    The case fans are in a positive pressure configuration.

    https://www.newegg.com/msi-liquid-cooling-system-mag-series/p/N82E16835846009?Item=N82E16835846009

    I ordered an Noctua nh-d15 chromax.black(2 140mm fans) and Corsair TM30 thermal paste, should be here by early next week.

    https://www.newegg.com/noctua-nh-d15-chromax-black/p/13C-0005-001H5?Item=9SIAADYA5G4033

    If the noctua( I have no doubt's in noctuas' performance.) doesn't work, my only option will be a different CPU and maybe active cooling.

    Here's a recent picture for reference.

    I've replaced the monitor with an Asus ProArt PA329 about a month or so ago.

    I also removed the cd/dvd drive I had sitting on the top.I no longer play the games that required a physical disk so it was no longer needed.

    edit: link to case

    https://phanteks.com/Enthoo-719.html#specs

    MyPC.jpg
    2227 x 1083 - 894K
    Post edited by IceCrMn on
  • IceCrMnIceCrMn Posts: 2,106

    I've done some more testing, I've found a solution to my over heating problem.

    The short version is,I had to switch my Windows Power plan from Ryzen High Performance to Ryzen Balanced.

  • f7eerf7eer Posts: 117

    OK, that makes a lot of sense. Sorry for not matching the level of detail on my setup as you provided for yours. I run DS on an Alienware laptop, which has an extremely efficient fan system, compared to my previous non-gaming laptop. For rendering, I used to use the Alienware "Performance" thermal power settings, which results in the behavior you describe, and which I referred to as "normal".

    When I'm not rendering I change to Alienware Balanced.

    (I found that I could eke out more performance rendering by moving to the Alienware "Full Speed" power settings, which is just all fans all the time. I found it was beneficial to let the system cool to a baseline temperature before starting the next render in a sequence.)

  • Is there a particular reason why versions of Daz after 4.16 are ridiculously unstable when gpu rendering, even when it's the exact same scene? I've tested multiple scenes at this point, and most of them produce a ton of errors in the log in 4.21, yet work perfectly fine in 4.16. This is on a 3080 as well.

     

    Screenshot 2022-11-17 184307.png
    2367 x 1149 - 386K
  • f7eerf7eer Posts: 117

    Richard Haseltine said:

    f7eer said:

    f7eer said:

    1. The dForce problem was introduced in nVidia driver 522.30, is known by nVidia, and a fix has not been released by nVidia yet. The workaround is to roll back to driver 517.48 or earlier.

    2. The render problem with OOT hair was introduced in DAZ Studio 4.21.1.13 (Current beta release), and fails with both nVidia drivers 522.30 and 517.40. The workaround is to roll back to DAZ Studio 4.21.0.5 (current general release).

    When I render a scene which contains OOT hair, I get the following errors:

    2022-11-08 07:19:50.362 [WARNING] :: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(377): Iray [ERROR] - IRAY:RENDER ::   1.12  IRAY   rend error: CUDA device 0 (NVIDIA GeForce RTX 2080 Super): Kernel [8] (ShadowEvalSr      ) failed after 0.009s
    2022-11-08 07:19:50.362 [WARNING] :: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(377): Iray [ERROR] - IRAY:RENDER ::   1.12  IRAY   rend error: CUDA device 0 (NVIDIA GeForce RTX 2080 Super): an illegal memory access was encountered (while launching CUDA renderer in <internal>:951)

    Is development aware of this problem? It seems pretty severe. Most (all?) OOT hair uses a custom MDL shader, and the problem may lie there, or it may be a Daz Studio problem, or it may be an nVidia issue. Is this known? Even the oldest OOT hairs I own (dating from 2017) have the problem.

    When the error occurs, Daz Studio can no longer render anything, and must be restarted. Not even clearing the scene fixes it.

    Yes, Daz is aware of and investigating the issue.

    Does the following update in DS 4.21.1.18 fix the problem with OOT hair in current beta DS 4.21.1.13?
     Update to NVIDIA Iray 2022.1.0 beta (363600.482)

        Minimum driver is 522.06 (R520) on Windows in order to support the new CUDA and OptiX features
        Minimum driver is 526.98 (R525) on Windows in order to retain OpenCL/dForce compatibility

    (From the change log http://docs.daz3d.com/doku.php/public/software/dazstudio/4/change_log)

  • f7eer said:

    Richard Haseltine said:

    f7eer said:

    f7eer said:

    1. The dForce problem was introduced in nVidia driver 522.30, is known by nVidia, and a fix has not been released by nVidia yet. The workaround is to roll back to driver 517.48 or earlier.

    2. The render problem with OOT hair was introduced in DAZ Studio 4.21.1.13 (Current beta release), and fails with both nVidia drivers 522.30 and 517.40. The workaround is to roll back to DAZ Studio 4.21.0.5 (current general release).

    When I render a scene which contains OOT hair, I get the following errors:

    2022-11-08 07:19:50.362 [WARNING] :: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(377): Iray [ERROR] - IRAY:RENDER ::   1.12  IRAY   rend error: CUDA device 0 (NVIDIA GeForce RTX 2080 Super): Kernel [8] (ShadowEvalSr      ) failed after 0.009s
    2022-11-08 07:19:50.362 [WARNING] :: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(377): Iray [ERROR] - IRAY:RENDER ::   1.12  IRAY   rend error: CUDA device 0 (NVIDIA GeForce RTX 2080 Super): an illegal memory access was encountered (while launching CUDA renderer in <internal>:951)

    Is development aware of this problem? It seems pretty severe. Most (all?) OOT hair uses a custom MDL shader, and the problem may lie there, or it may be a Daz Studio problem, or it may be an nVidia issue. Is this known? Even the oldest OOT hairs I own (dating from 2017) have the problem.

    When the error occurs, Daz Studio can no longer render anything, and must be restarted. Not even clearing the scene fixes it.

    Yes, Daz is aware of and investigating the issue.

    Does the following update in DS 4.21.1.18 fix the problem with OOT hair in current beta DS 4.21.1.13?
     Update to NVIDIA Iray 2022.1.0 beta (363600.482)

        Minimum driver is 522.06 (R520) on Windows in order to support the new CUDA and OptiX features
        Minimum driver is 526.98 (R525) on Windows in order to retain OpenCL/dForce compatibility

    (From the change log http://docs.daz3d.com/doku.php/public/software/dazstudio/4/change_log)

    I am not aware of any relation between the two.

  • nonesuch00nonesuch00 Posts: 17,922

    The thread here is pretty long but I'm having serious bugs with 4.21.1.13 Public Beta:

    1) I load Victoria 9 HD & Genesis 9. 

    2) I apply various eye colors to both eyes and they only apply to the Victoria 9 HD eye (ie. the 1st eyes loaded named Genesis 9 Eyes and not the Genesis 9 Eyes (2)

    3) I apply color to the hairs but somehow, eventually the viewport will draw pitch black hair no matter what color the hair is actually is changed to. I also lost the ability to change color.

    4) I apply poses from the Genesis Starter Essentials and only the base character mesh moves and not the mouth, eyes, tears, and so on.

    5) Exiting, starting, and reopening the scene restored functionality for a bit but the more I applied material presets the more problems start showing up again.

  • Everything goes fine until I hit the render button, then the app crashes and closes. Anybody else?

  • Either it crashes or just doesn't render for me as long as I load most of the hair I have.

  • PerttiAPerttiA Posts: 9,376

    Running out of RAM has been a reason for crashing, and running out of VRAM with not allowing fallback to CPU has caused the scene not to render.

    Difficult to pinpoint the culprit, if not given any information, but if one has less than 32GB's of RAM and 8GB's or less VRAM (especially with a non-RTX GPU), these would be the first to look at.

  • second_technician_rimmer_9571136c47 said:

    Either it crashes or just doesn't render for me as long as I load most of the hair I have.

    Some hairs by OutOfTouch are known to do this, the cause is being investigated.

  • nonesuch00nonesuch00 Posts: 17,922

    PerttiA said:

    Running out of RAM has been a reason for crashing, and running out of VRAM with not allowing fallback to CPU has caused the scene not to render.

    Difficult to pinpoint the culprit, if not given any information, but if one has less than 32GB's of RAM and 8GB's or less VRAM (especially with a non-RTX GPU), these would be the first to look at.

    I have CPU fallback disabled to avoid the OOT hair out of RAM crash. The render instead stop without ever having drawn to the canvas once the VRAM is full. I replicated this error myself by having too many mesh layers with opacity causing too many intermediate data structures to be create to calculate all the opaque mesh layers. I was rendering flames on "The Human Torch" though and not hair by OOT. Some of OOT's hair does the same though. 

  • If CPU temps are higher in beta vs release, it could be a result of new multi-threading enhancements. I notice renders are faster with CPU enabled. This is a good thing, but requires good themal management.

  • IceCrMnIceCrMn Posts: 2,106
    edited December 2022

    I'm not aware of any new multi-threading changes from release version and the beta.

    My AMD CPU is cooled with an MSI magcore 360mm AIO.

    I posted a pic of the system further up the thread. Lot's of case fans also.

    I've been using CPU under-volting and frequency limiting to keep the temps down in the beta.

    I have no such issues in the release version.

    Seeing as only one other user has reported this problem; I can only assume it isn't effecting many others.

    If there is a problem and it is released to the general release version then more people will be "testing" it.

    Which should help narrow it down.

     

    edit: I'm also restricting Studios affinity to 14 out of 16 threads. So it's not gettig to use 100% of my CPU.

    Post edited by IceCrMn on
  • kyoto kidkyoto kid Posts: 40,555
    edited December 2022

    ...hmmmm, because of the new driver requirement, Iray GPU rendering is no longer supported for W7.   

    That sure was quick.  

    So apparently the 4.21.05 General Release is the "last stop" on this 15 year run.. 

    Ah well, nothing lasts forever I guess.

    Post edited by kyoto kid on
  • faibolx said:

    Everything goes fine until I hit the render button, then the app crashes and closes. Anybody else?

    Yeap, newest Studio drivers and previewing in IRAY just has the yellow bar going up and up, but nothing.

    If I try to render, I get a black screen or it crashes to desktop. These are scenes with four G8 models and one room, so not intensive at all.

    I have two GPUs (3090 ti and a 3080 ti) and Ryzen 9 5900X and 128 GB, so I don't want people to tell me it's my rig.

    This software needs to be fixed well before introducing a new Gen. So frustrating when everything was working fine a month or two ago.

  • Has anyone else noticed that enabling Guided Sampling results in considerably higher brightness?

    Scene looks somewhere approx. two stops brighter to me.

    Also, mirror effect (reflection from the mirror on adjacent wall for example) is much more pronounced.

    Scene only lights, spectral on with rec709 and 1964 if it is of any importance.

  • Almost been one and a half month and still no fix for the crashes? Is there any ETA in the slightest? 

  • I'm finding that pretty much most of my trasmapped hairs crash iray.

  • IceCrMnIceCrMn Posts: 2,106

    New beta is here.

    • Added support for the “Iray Ghost Light Factor” property on standard light types when NVIDIA Iray is the active renderer

      • Experimental

    • If a DzFloatProperty (user) property named “Iray Ghost Light Factor”, with the property group path of “Display/Rendering/Iray” and a default value of 1.0 is added to a pointlight/spotlight with “Light Geometry” set to a value other than “Point”, the value of this property is used to control visibility in glossy interactions
    • See official documentation for more information

    Anyone try it out yet? 

  • whitedudeabove said:

    Almost been one and a half month and still no fix for the crashes? Is there any ETA in the slightest? 

    Which crashes are you referring to?

  • nickalaman said:

    I'm finding that pretty much most of my trasmapped hairs crash iray.

    Which hairs are you referring to?

Sign In or Register to comment.