RockZ Hair crashing computer
wsgentry
Posts: 572
Hi,
Does anyone besides me have a problem where RockZ hair is crashing their computer? I have isolated this to a generic G8 female, no morphs, props, clothes, anything, using perspecitve view with ONLY RockZ hair applied. It displays properly in preview, but when rendering, the computer crashes with DPC Watchdog violation about two minutes into the render. The mouse freezes for about one minute, then blue screen. All diagnostics have been run (including disk, exhaustive memory), all drivers are up to date. Bios is up to date. My comuter is this:
Alienware R3 17" laptop
32 gigs ram (ram use is 32 percent before the crash)
4TB SSD storage over three drives (one SATA the others are M2)
RTX2080ti connected via Alienware Graphics Amplifier (didn't move the cable at all during crashes and subsquent tesitng)
4k built-in display.
I have a ticket with Daz support but I figured I'd ask here, too.
Settings are tessellation 2, PR Hairs on, HD enabled (crashes no matter what these settings are), Spectral Rendering.
Note that regardless of the scene, basic or not, removing the hair allows renders to complete.
Thanks in advance!
Scott
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
If you purchased it within 30 days, you can put in a CS ticket for a refund. Some hairs are "too much" for some systems.
Thanks, Catherine. I really appreciate your suggestion.
This hair is nice, but it's a huge memory hog like most dForce hair. I've used it previously in a simple and was pretty happy with it. You need simulate it at the "Best" setting or it looks pretty limp when simulated, though. Good and Better give nearly the same result. Also, there's no explanation for the various
Tonight, in the interest of science I tried a simple render of a G8F character with a geoshell and two (2) sets of Rockz hair, one going left and one right using 4 DS lights with no HDRI, and my computer choked (32 Gb RAM - 2080ti the usual way). After around 10 minutes I had to force quit Daz Studio, which was showing a commit charge of a whopping 55 Gb for DS. It might have come around eventually, but I was seeing no activity, and it wasn't responding to trying to cancel the render.
Now my RAM usage goes way up when I render this hair. Just a single G8F, 1 RockZ, 4 lights, no HDRI had DS using 22-26 Gb while doing 3 test renders. So whatever the issue is, it's happening very early on.
Anyway, if you can't use it get a refund for it. Have you had experiences with other dForce hair (not to be confused with "hair with dForce")? If this is your first one, you might want to be careful what hair you buy in future.
PS: The product does include 4 wallpapers, so you can see what the hair could look like rendered if it doesn't crash your DS. I've not seen that "bonus" included with anything else I've bought from Daz.
DPC Watchdog?
Is this a new laptop? Have you updated Windows since you got it? This error has something to do with your SSD. It was very common with early releases of Win10 but is fixed in more recent releases. So if you have just gotten the laptop update Windows. It did cause a BSOD.
If not it may still be a good idea to check if an update is available. And no you should not trust that Dell fixed this or shipped your laptop with a recent version of Win10 installed.
For what it's worth, I got a DPC Watchdog violation (among many many others) with my computer, which I know was up to date at the time (I update it monthly). It apparently had to do with a DIMM that was going bad, and which threw off every blue screen error under the sun, for some reason. (I never knew there were so many different blue screen error codes.) All that by way of saying, since this is something that also has heavy memory usage, it might also be worth running a system hardware check on your memory -- there's a utility in Windows for that.
Failing RAM sticks will cause every wacky error under the sun but the OP said he'd run a memory check.
Ah, OK. Missed that one.
I also ran exhaustive disk diagnostics. My system is up to date, entirely. I'm pretty meticulous about that.
Last night, I found some errors in the log that were not hair or render related--more loading morphs. So, I uninstalled and reinstalled those products that were listed in the log. The hair now works, but Sevrin is right: The hair is a memory hog (and yes, I've used dForce hair). Just before the actual render starts DAZ will consume almost all available memory. Paging will happen. CPU used will peg at 100 percent, then when the render starts, CPU use will drop to about 27 percent, and memory use will gradually get down to around 50 percent with DAZ still being the largest consumer of that resource--this is due to the RTX 2080ti taking over. It's interesting to watch it happen, I don't understand why a hair has to grab as much memory as it does, but I accept it.
One factor that I thought would make a big difference is canvas size: It makes a difference, but not as much as one would think. Currently, I'm rendering a scene with this hair that is 6k x 6k with no issue. The only effect is time. Before I found the file issues, the hair would crash the system when the render was set to 1k x 1k.
To address Sevrin's "hang", I suspect that happens before your graphics card is being used -- shortly after "rendering image" appears in the render dialog? If so, that is where my system would crash before I cleaned up products. If the forum's upload allows, I'll include the snippet of the log where the "errors" are shown.
Thank you all!
Scott
Check your boot drive to see how much free space it has. You could be running out of swap space which could be causing the crash, just a guess but that error is definitely SSD related.
Hi Kenshaw011267,
If the pagefile was full, it would crash with a different error, something like a Pagefile full or something like that. At anyrate, the pagefile size is hard coded to larger than the physical memory in my computer. The crash I experienced can be caused by SSD's not responding, but it can also be caused by any system resource not responding. So, I've pretty much beaten all four drives to death over the last couple of days with absolutely no errors so, my best thought are the products I deleted and reinstalled as the fix--they were corrupt or not there. Note that the Warnings listed in the log fragment I posted are no longer there--and they were they when the system crashed.
Scott
Additional information: The event log indicates the CPU was in reduced power mode for over 71 seconds prior to the crash--that makes sense as the fans were running at full speed, so the firmware knocked the performance level back in order to allow the CPU to cool. Next, windbg indicates the NVIDIA driver was the current process running at the time of the crash. The DPC tick count had exceeded 500. So, it's likely the reduced CPU performace caused the NVIDIA driver to timeout--tick count exceeded.
It would have been interesting to see GPU memory use at the time just before the crash--I have CPU fallback turned off so if GPU memory became full this would be a contributing factor, in my way of thinking. As it stands now--and the render is still going--GPU memory use is at 10g out of 11.
Just some thoughts.
! Do not ever fix the size of the page file! EVER! I'm not kidding about this. Set it to system managed. And no the error will not always be pagefile full. It can result in all sorts of weird sruff.
Fixing the size of the page file is not a problem provided it's larger than physical memory with some overhead--mine is 47 gig over all drives. In fact, there are situations where a fixed paged file is actually desired. Here's a link to a microsoft doc: https://docs.microsoft.com/en-us/windows/client-management/determine-appropriate-page-file-size.
Okay, back to the hair.. I was able to get another crash by enabling HQ on the hair. The render without HQ set finished fine. Still the nvidia driver was the current active process.
Scott
Interesting.... I increased the minimum pagefile size to 48gig and the maximum to 128gig (1.5 x RAM for minimum and 4 x RAM for Maximum) and with HQ set, it's rendering.
32 percent CPU 97 percent memory.
Scott
Just let the system manage the page file. I really don't care what some "person" at MS says.
The system managing the pagefile takes resources that I don't want to dedicate to that task. Additionally, Alienware, for my configuration, recommended setting pagefile size manually. Additionally, as an MCSE, Microsoft Partner/Developer, I can see no justifyable rational reason to allow my system to manage the C: pagefile.sys. Allowing the system to manage the size of the pagefile (it does manage the size on my D: drive, however it's never used due to the size of the C: pagefile.sys) would likely work if all I did was Office Apps, but that's not the case. Now, if there was a swapfile.sys--something where entire contexts were routinely swapped out of physical memory (as they were in VAX/VMS) I'd allow the system to manage that because there would be literally no way to calculate its size--Although VAX/VMS, if memory serves, used manual allocations. Then again, in that time, having Terabytes of storage and gigabytes of memory were something way down the road, and the associate hardware was designed to work hand-in-hand with OS, in terms of virtual memory and execution contexts.
Scott
You just saw it causes unpredictable errors. But whatever your system. If you trust Dell I cannot help you.
I'm not asking for your help at this point because you haven't provided any rationale for your suggestions, though I do appreciate the time you took to make suggestions and one of those suggestions led me to increase my pagefile size, so not all is lost. I'm sorry you discount my experience with computers (over 40 years) but that's okay. I don't hold anything against you. Regarding Dell/Alienware. I've had nothing but OUTSTANDING experiences with them. When I first bought this computer, I had issues with it crashing under Daz and their recommendations removed those crashes. The support engineer I was assigned actually analyzed the crash dump file in realtime. He provided solid evidence for his suggestions and I took them. Then again, I pay to be assigned people who do not troubleshoot from scripts. Clearly, your experience differs from mine, and that's fine. I'd buy an Alienware laptop in a heartbeat. In fact, I'm considering an upgrade so I get 64gig of RAM and a much faster processor.
Take care,
Scott
^
I really don't know what else to say except I also have 40 years experience with computers, first system was an Apple II in 1978.
I'll also point out that if you do a simple search here or on pretty much any Windows support site anywhere the answer will be to always let the system manage your page file and that the reason is the errors caused are random and unpredicatble. Just exactly like the ones you had.
My first computer was an Apple II in 1978, as well. I actually "saw" and Apple 1 at the Byte Shop, didn't have the money so I had to wait. Later, I upgraded it to an Apple II+. Then an Apple IIgs. I worked for Intergraph for over 15 years. If you're interested in more about my background do the following search "AFL Scott" Allison. Make sure you include "Allison" in your search, but outside the quotes. Didn't get into Windows until Intergraph became a development partner for Windows NT.
Since changing the pagefile size as I stated earlier, I was able to render full scenes with RockZ hair with no crashes. Maybe I wasn't clear on that. The hair does consume a lot of memory so it makes sense to change the pagefile allocation. Managing it automatically will consume system resources that might be necessary for the render to start. Also, I don't take changing the pagefile size lightly. I have NOT had a crash since doing so. My issue is not with Windows support sites saying to let the pagefile be automatically adjusted, but this is a special case where a hair--based on evidence provided by monitoring system resource during a render--is consuming memory rapidly as well as causing CPU use to peg at 100 percent. At 100 percent use, there is likely to be race/timing conditions that will cause the system to page. If the page file needs to grow, it is higlhly probable that--given all that is going on at this time (granted you havent seen what I've seen), that a crash will occur while the system is attempting to grow the pagefile. All I've done is set the theorhetical minimum and the absolute maxium pagefile size for a 32gb system. If you can find me a resource that states setting minimums and maximums as I have done, is detrimental and will cause problems, I'll research that, otherwise I can see no harm in doing as I've done. Remember, I've had ZERO crashes since doing so.
Scott
What is your result when you let Windows manage the pagefile?
Sevrin,
Just did it (yeah, it's really late). It crashed with a DPC Watchdog violation with the exact indications as the earlier crashes--CPU pegged, memory used up, disk activity, etc. So, I disabled some stuff like antivirus, etc. Same crash. Reset to maximum low values and maximum high values, renabled what I disabled, and no crashes, It's easy to test because the crash happens when, I believe textures are being loaded into the GPU, so it's early on. So, that's done. Sincerely, it's fun doing this stuff. There is a race/timing issue. That it's caused by a hair product is a trip. Makes me wonder exactly when textures get decompressed.
Scott
There is something badly mangled in your system. But Dell so no surprise there.
Weird as dForce EXzela Long Hair from the same HM vendor computes really fast and works very well.
Try Simulation Settings > Collision Mode : Good Discrete (I always use that for hair, makes the computation extremely faster and result as far as I could test looks identical).
Also Surfaces > Selft Collide OFF (but it's usually that by default).
And just an advice : NEVER buy Dell or any already mounted "branded" computer. Take your time, choose the parts one by one, pay the seller to mount it. Way cheaper for a much better rig.
System bashing. Whatever.
The problem is fixed. it was fixed by manually adjusting virtual memory settings. Let the system decide--crash. Manually adjust, no crash. Simple.
Thank you for your suggestion about dForce settings.
I can build my own computers but I can't build a laptop that has the capability my Alienware laptop has. I bought it because when I travel in my motor home, I can just bring it and the graphics amplifier in and have fun with DAZ. What I needed was something portable. No other laptop I've researched is faster and no other graphics amplifier-like product is as fast as the Alienware version--that's because they extend the PCI bus and others use USB3 variants. So, there were very sound reasons why I chose as I did.
I appreciate all the responses, but system bashing isn't helpful.
This is my last post on this... this is what I told DAZ in the support ticket.
Just for the record... There are two fixes:
1. Removing and reinstalling the products listed in the attached file fixed it until I enabled HQ for the hair; and
2. Extending virtual memory to the maximum recommended sizes allows rendering with any option selected.
Conclusion: It should be noted that this hair is a memory hog. The crashes happen prior to the GPU being loaded with textures: CPU pegs at 100 and paging thrashes. Because the CPU is pegged at 100 percent for an extended period, the CPU gets hot and firmware slows it down to mitigate heat. This causes a race/timing condition in the system because Windows is trying to compensate for the lack of memory by paging, but the system is too slow to respond because the CPU has been throttled (all of this is supported by logs). As a result, the DPC watchdog timer exceeds 500 and the system suffers a bugcheck. Crash dump analysis shows the nvidia driver as the current active process at the time of the bugcheck.
There should be a warning attached to the hair product which states it consumes memory and 32 gig memory is not enough for this product to work properly in all configurations.
Hello good morning, I use 16gb of ram, if I put the tesselation with 3 or 5 it crashes me. For better results in the render it is better to use 1, the problem is that if you put 3 or 5 it creates geometry and the memory consumption is very high, those values are for high CPUs.
Hi!
I used all the hair's default settings in my experiment. Please keep in mind when making your products that many, if not most, Daz users, don't have "high CPUs".
My line tessellation was set to default as well. Generally, I'll set it to two if I want to see a viewport render of the hair, but set it back to default for full renders. What I did notice, but didn't write about is this: I did complete shutdowns and restarts of the computer before I did any render with this hair.
There's a new Studio Driver: 452.06 released the 18th. The release notes mention nothing of the crashes, but I'm going to test it this weekend with default pagefile settings. I suspect nothing will have changed.
Scott