Found 260 posts.
Search results Show results as topic list.
Technical Discussion » system crash when writing out large(1gb) files
- tricecold
- 260 posts
- Online
just to be sure, if its an intense sim utilizing %100 of the cpu, it could be a PSU issue also, Usually PSU issues restart the entire systems.
Technical Discussion » Stop containers from a cluster simulation (Pyro)
- tricecold
- 260 posts
- Online
Technical Discussion » cache loading very heavy
- tricecold
- 260 posts
- Online
Joker386
Yes , delete unused attributes make a little faster reading time for cache files.
However that's obvious reading data from RAM is more faster than HDD !
I usually using SSD drive for reading and writing large cache files .
Also reading and writing data from EX4 partitions (on Linux) is faster than NTFS (on Win) , Of course in my experience !
Lol I can buy a HP workstation with the cost of massive SDDs I would need inorder to be able to work.
Technical Discussion » redhat/rhel/centos 7 stability?
- tricecold
- 260 posts
- Online
Should have mentioned that, you are right, thanks for correcting
MartybNztricecold
H13 came with a new memory allocation tool,this new memory allocation thingy would make use of all your ram after a while running your simulation. It was told that ``it was fragmenting the memory but it is not a memory leak``.
This only affected Windows. The memory allocator in Linux is jemalloc.
http://www.canonware.com/jemalloc/ [canonware.com]
Technical Discussion » cache loading very heavy
- tricecold
- 260 posts
- Online
you probaly had fields you did not need, maybe some volume attributes etc, and resulting bgeo files were probably big.
Work in Progress » NodeZArt - art with nodes
- tricecold
- 260 posts
- Online
Technical Discussion » freez wire dinamics?
- tricecold
- 260 posts
- Online
I am not sure if it is the same but when glue strength is -1 calculations are much faster, maybe solver automatically disables if strength is -1, so you can try to add some glue by proximity to the ground with the strength of -1
Technical Discussion » Xeon phi support
- tricecold
- 260 posts
- Online
I agree, I have big hopes for XEON PHI, the thing is as much as I know, XEON PHI runs on its own linux. But from the articles I read about it, it seem even an advanced TD can not do this without support.
If there was a way to use this, I would invest in one or two
If there was a way to use this, I would invest in one or two
Technical Discussion » measure different group Particle distance?
- tricecold
- 260 posts
- Online
I bump this also, I may need to affect several groups attributes based on distance in DOPS. anyone ?
Technical Discussion » Has anyone used the "Script SOP" ?
- tricecold
- 260 posts
- Online
Joker386
Yes , for example I made a Progress reporter before with this node , Just download my project file from this Topic :
http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=30787&highlight=Progress [sidefx.com]
and it worked very good, thanks again man
Technical Discussion » Request: PhysBam Multi-physics integration (for option 2.)
- tricecold
- 260 posts
- Online
I think Bullet does wonders in Houdini. recently finished a project for a movie, where we had 4 buildings to collapse, with tens of thousands of pieces, windows, bricks, wooden parts, several glue constraints ,with activation, propagation etc etc. A single pc with 2x - Xeon E-5 2620 could deliver each building in few hours.
But there definitely is a space for improvement. For test sake, I did the same setup test on my home machine an i7 at 4.5 ghz, and I outrun the Dual Cpu Xeon machine, which was the result of weak multhithreading in pretty much all solvers except flip and pyro.
RBD solvers and the new SOLID solver does not seem to be able to utilize the Cores when there are a lot of collisions happening.
But there definitely is a space for improvement. For test sake, I did the same setup test on my home machine an i7 at 4.5 ghz, and I outrun the Dual Cpu Xeon machine, which was the result of weak multhithreading in pretty much all solvers except flip and pyro.
RBD solvers and the new SOLID solver does not seem to be able to utilize the Cores when there are a lot of collisions happening.
Technical Discussion » PBR and object scale
- tricecold
- 260 posts
- Online
I also agree with Tamte, Camera is scale is a NO NO. We were scaling maya cameras 100 times smaller, it made all the Volumes render super dense, created crazy amount of noise, etc.
Technical Discussion » redhat/rhel/centos 7 stability?
- tricecold
- 260 posts
- Online
Forgot to mention also on centos 6 for 1.5 years and no problems.
Maybe freezes got something to do with all these GUI olverlays and pretty GUI stuff, I never had any of these problems on Ubuntu , but on Fedora I am on Mate, which I beleive makes Houdini run more smooth in my experience.
The only reason I can think of any OS to slow down during simming is because you must be simply running out of ram. I don`t think there is any other reason to this, I am actually pretty sure that thats the reason.
H13 came with a new memory allocation tool,this new memory allocation thingy would make use of all your ram after a while running your simulation. It was told that ``it was fragmenting the memory but it is not a memory leak``.
Meaning task manager may tell you that you are using all of your ram after a while you ran a flip simulation, which should not fill up your ram. But the computer would keep responding unlike when the OS uses swap due to not enough memory, which makes a pc unusable.
This issue have been fixed long time ago as I don`t experience it at my workstation anymore.
I hope I managed to answer your question
Maybe freezes got something to do with all these GUI olverlays and pretty GUI stuff, I never had any of these problems on Ubuntu , but on Fedora I am on Mate, which I beleive makes Houdini run more smooth in my experience.
The only reason I can think of any OS to slow down during simming is because you must be simply running out of ram. I don`t think there is any other reason to this, I am actually pretty sure that thats the reason.
H13 came with a new memory allocation tool,this new memory allocation thingy would make use of all your ram after a while running your simulation. It was told that ``it was fragmenting the memory but it is not a memory leak``.
Meaning task manager may tell you that you are using all of your ram after a while you ran a flip simulation, which should not fill up your ram. But the computer would keep responding unlike when the OS uses swap due to not enough memory, which makes a pc unusable.
This issue have been fixed long time ago as I don`t experience it at my workstation anymore.
I hope I managed to answer your question
Technical Discussion » redhat/rhel/centos 7 stability?
- tricecold
- 260 posts
- Online
Houdini Lounge » The average income for a HoudiniFX artist
- tricecold
- 260 posts
- Online
Here Is what I know, Daily rates at studios go from 100 Euros to 300+ from Junior to Senior FX artists. Not Software related, can be Maya can be Houdini etc.
You can also easily check these numbers at glassdoor website, just google it.
These numbers greatly differ from country to country depending on the GDP scale.
You can also easily check these numbers at glassdoor website, just google it.
These numbers greatly differ from country to country depending on the GDP scale.
Technical Discussion » system crash when writing out large(1gb) files
- tricecold
- 260 posts
- Online
You can add whatever extra fields you want to import to the BGEO froim the DOP import fields ,
But I also don't think the sim files are crashing your Houdini, because even if you export or not every field that would be in the sim file is calculated every frame, regardless if you are writing the sim files to HDD or not.
If you have a lot of inputs for the simulation, trying cashing every source to HDD first,
This way you will know your source is not the problem,
So you can focus on the DOP net
Better share a hip file if you can, That would help a lot.
But I also don't think the sim files are crashing your Houdini, because even if you export or not every field that would be in the sim file is calculated every frame, regardless if you are writing the sim files to HDD or not.
If you have a lot of inputs for the simulation, trying cashing every source to HDD first,
This way you will know your source is not the problem,
So you can focus on the DOP net
Better share a hip file if you can, That would help a lot.
Technical Discussion » system crash when writing out large(1gb) files
- tricecold
- 260 posts
- Online
No they are bgeos, with volumes with some extra fields, gradient,velocity rest etc. Oh I can actually show now since the trailer is out.
https://www.youtube.com/watch?v=sTte6BQndTQ [youtube.com]
1:57 altough thats pretty much the beginning of cache, but the sim was used on plenty of shots including close ups, which was pretty impressive we could put the camera showing %30 of the sim yet still had enough detail.
But the image quality there is pretty bad.
https://www.youtube.com/watch?v=sTte6BQndTQ [youtube.com]
1:57 altough thats pretty much the beginning of cache, but the sim was used on plenty of shots including close ups, which was pretty impressive we could put the camera showing %30 of the sim yet still had enough detail.
But the image quality there is pretty bad.
Technical Discussion » How to send commands to an open Houdini from Terminal ?
- tricecold
- 260 posts
- Online
PradeepBaruamost of the time there is more than one houdini open, and if one becomes unresponsive, I might have to kill it, but i cannot see the file names, in htop , is there a way to figure out which Houdini to kill based on scene name
PID tells you which job created first. Let's say you have 3 instances of Houdini open on your machine, now which Houdini you launch first, would have lower PID number.
You have to keep remember order of opening Houdini session or scene file.
thnx man you have been greatly helpful
Technical Discussion » How to send commands to an open Houdini from Terminal ?
- tricecold
- 260 posts
- Online
PradeepBarua
You can force Houdini to crash by
kill -11 PID
it will save your file in temp directory.
You can get PID by typing:
top | grep Houdini
thnx a lot, while we are at it, lemme ask another one, most of the time there is more than one houdini open, and if one becomes unresponsive, I might have to kill it, but i cannot see the file names, in htop , is there a way to figure out which Houdini to kill based on scene name
Technical Discussion » How to send commands to an open Houdini from Terminal ?
- tricecold
- 260 posts
- Online
Ok here was the issue, When I was working a few hours ago, Xserver crashed but the linux was keep running, so I was able to SSH to my machine from a different machine and H was running, and I really wanted to save the scene. So I tried to figure out how to send a command to Houdini to save but failed to do so, any ideas?
-
- Quick Links