Houdini Crashes With No Error Message: Reading Exit Codes and Logs
When Houdini closes with no dialog, the operating system usually ended it before Houdini could report anything. The clues are in the terminal, the memory graph, the system log and the exit code.
- First step: launch Houdini from a terminal so the last log lines survive the crash
- Memory: a RAM graph that hits the top just before the crash points to running out of memory
- Exit codes: 137 = killed (often the Linux OOM Killer), 139 = segmentation fault, 0xC0000005 = Windows access violation
- Fixes: cap the DOP cache, cache to disk, use checkpoints; broken data needs a clean file, not more hardware
- On iRender: 256 GB ECC RAM servers; you install Houdini with your own SideFX licence
Few things in Houdini are as frustrating as the application disappearing in the middle of a simulation or render, with no crash dialog and no message. It feels random, but it rarely is. In most cases the operating system closed Houdini, often because memory ran out, before Houdini had a chance to explain itself.
This guide shows where to look for the evidence, how to read the exit code, how to tell a memory problem from a broken scene, and which Houdini settings keep heavy simulations from taking your session down.
Checked against SideFX Houdini documentation, GNU Bash and Microsoft documentation, November 2026.
Where to look after a silent crash
Check four places, in this order:
- The terminal. If Houdini was started from Command Prompt or Terminal, the window stays open and keeps the last log lines.
- The memory monitor. Task Manager, Activity Monitor or a system monitor shows the RAM and VRAM history right before the crash.
- The system log. Event Viewer on Windows, or dmesg and the system journal on Linux, records driver resets and processes killed for lack of memory.
- The exit code. The number the process returned tells you how it ended.
How to read the figure. The four numbered panels are the four places to check, in order. Panel 1 is the terminal: lines such as Out of memory or Segmentation fault right before the prompt returns are the strongest clue. Panel 2 is the memory graph: a climb to 100% followed by the crash means the machine ran out of RAM. Panel 3 is the system log, where a display driver reset or an oom-kill line pins the cause on the GPU driver or on memory. Panel 4 is the exit code, which tells you how Houdini ended and is only useful next to the other three.
Keep the log: launch Houdini from a terminal
Started from a shortcut, Houdini takes its console with it when it crashes. Start it from a terminal instead:
- Open Command Prompt (Windows) or Terminal (macOS, Linux).
- Run the Houdini executable from its
binfolder, for example"C:\Program Files\Side Effects Software\Houdini 22.0.xxx\bin\houdini.exe"on Windows. - Work as usual. After a crash, scroll up through the last lines for out-of-memory, segmentation fault or driver messages.
While you work, keep a memory monitor visible. On Windows, Task Manager > Performance > Memory shows both RAM in use and committed memory; if the pagefile is full, Windows can end a process even before physical RAM reaches 100%. Inside Houdini, the Performance Monitor can sample Memory Stats and DOP Solves to show which node is using memory, though SideFX notes that memory stats may be inaccurate with multi-threaded cooking.
What the exit code tells you
On Linux and macOS, run echo $? right after the crash; on Windows, echo %errorlevel%. Shells report a process ended by a signal as 128 plus the signal number, which makes the common codes easy to decode.
How to read the figure. Each row is one exit code, with the system it appears on, what it means and what usually causes it. Orange is the memory row: 137 means the process was killed with signal 9, which on Linux is what the out-of-memory killer sends, so check dmesg for an out-of-memory line at the same time. Blue rows are memory-access faults, 139 on Linux and 0xC0000005 on Windows, which can come from memory pressure but also from a broken HDA, a plugin or a driver. The grey row is the warning: a generic code tells you that Houdini failed, not why, so always confirm with the log and the memory graph.
Software problem or hardware limit?
Before you change anything, decide which kind of crash you have. The fix for one is useless for the other.
| Symptom | Likely cause | Where to look | Fix |
|---|---|---|---|
| Houdini vanishes, RAM at 100% | Out of memory | Memory monitor, dmesg | Cache to disk, lower resolution, more RAM |
| Screen flickers, “display driver stopped responding and has recovered” | GPU driver timeout (TDR) | Event Viewer | Update the driver, reduce GPU load |
| Viewport frozen, CPU busy | Heavy cook, not a crash | Task Manager | Wait or press Esc |
| Viewport frozen, CPU and GPU at 0% | Hang or deadlock | Terminal, Event Viewer | Force close, check custom HDAs |
How to read the figure. The figure splits crashes into two columns. On the left, a software problem, such as a corrupt HDA, a broken node or a scene bug, crashes on your machine and on a bigger machine alike, while memory stays well below the limit; the fix is to clean the scene and isolate the broken element in a new file. On the right, a hardware limit shows the RAM graph hitting the top just before the crash, and the same scene finishes on a machine with more RAM or VRAM. Decide which column you are in before you spend money on hardware.
Houdini settings that prevent memory crashes
For simulations, most silent crashes come from frames piling up in RAM. These SideFX settings keep memory under control:
How to read the figure. Badge 1 lowers Cache Memory (MB) on the DOP Network so the simulation cache cannot grow past the RAM you have free. Badge 2 allows that cache to spill to disk, and badge 3 saves checkpoints so a long simulation can resume from the last checkpoint after a crash. Badge 4 writes the simulation out with a File Cache SOP, the default .bgeo.sc in $HIP/geo, so each frame leaves memory once it is on disk. Badge 5 is the diagnostic: sample Memory Stats in the Performance Monitor to find the node responsible.
Work through these before you start a long cook or render.
SaaS or IaaS for heavy Houdini scenes
When a scene really needs more memory than your machine has, a render farm can help, but the two farm models handle Houdini differently.
Running memory-hungry Houdini jobs elsewhere
A neutral comparison for Houdini artists.
| Your situation | SaaS render farm | IaaS render farm (iRender) |
|---|---|---|
| You need to watch memory and logs yourself | Limited; you see the farm's job log | Full desktop: run Houdini from a terminal and watch RAM as at home |
| You use custom HDAs and packages | Must match what the farm supports | Install your own Houdini build and packages; the server image is kept |
| You need to resume a crashed sim | Resubmit the job | Resume from your checkpoints on the same server |
| Licensing | Usually provided by the farm for supported renderers | You bring your own SideFX licences |
Running heavy Houdini scenes on an iRender server
How to read the figure. The top row is the setup: your own PC or Mac connects by remote desktop to a whole bare-metal server with 1, 2, 4 or 8 GPUs, so you open your Houdini project there exactly as at home. The numbered row is the order of a session: register and top up, upload to iRender Drive before the server runs, start and connect, render a test frame and the job, then shut down so billing stops.
- Register an iRender account and top up; the first top-up of $50 or more gets a 100% bonus.
- Upload your project and caches to iRender Drive before you start the server.
- Start a server and connect by remote desktop. Every server has 256 GB ECC RAM; choose 1, 2, 4 or 8 RTX 4090 or RTX 5090 GPUs.
- Install Houdini and your packages once and license them with your own SideFX licence; the server image is kept between sessions.
- Run Houdini from a terminal, cache simulations to disk, then render.
- Download the output and shut the server down so billing stops.
Frequently asked questions
1. Why does Houdini close without an error message?
Usually because the operating system ended it, most often for running out of memory or after a driver failure, before Houdini could show a dialog. The terminal output, memory graph, system log and exit code show which.
2. What does exit code 139 mean in Houdini on Linux?
139 is 128 plus signal 11, a segmentation fault: Houdini accessed memory it was not allowed to. It can come from memory pressure, but also from a broken HDA, plugin or driver.
3. What does exit code 137 mean?
137 is 128 plus signal 9: the process was killed. On Linux this is what the out-of-memory killer does, so check dmesg for an out-of-memory line at the same time.
4. What is 0xC0000005 on Windows?
It is an access violation, reported as -1073741819 by %errorlevel%. Like 139 on Linux, it says how Houdini ended, not why; check memory and the Event Viewer.
5. How do I stop a pyro or FLIP sim from crashing Houdini?
Cap Cache Memory on the DOP Network, allow caching to disk, save checkpoints and write the sim out with a File Cache SOP. If RAM still runs out, lower the resolution or use a machine with more memory.
6. Is Houdini installed on iRender servers?
No. You install the Houdini build you use and license it with your own SideFX licence. The server image is kept, so you set it up once.
Run your Houdini scenes on iRender
iRender is an IaaS GPU render farm: you rent a bare-metal server with 4x NVIDIA RTX 5090 (32 GB) GPUs, an AMD Ryzen™ Threadripper™ PRO CPU, 256 GB ECC RAM and a 2 TB NVMe SSD, connect by remote desktop and work on it like your own workstation, with full administrator access. You install Houdini and use your own SideFX license; your server image is saved between sessions, so Houdini, your packages and caches stay installed. For Karma XPU and GPU-accelerated solvers we recommend the 4-GPU servers.
Multi-GPU servers are billed per minute (1-GPU servers have a one-hour minimum), and sessions of 3 hours or more get 10% off. Our 4-GPU servers are liquid-cooled; in our own measurements the GPUs hold 60–70°C under sustained render load. Elite support is available 7 days a week by hotline, live chat, Skype, WhatsApp and email.
This video walks through rendering a Houdini Karma XPU scene on an iRender server in five steps:
New users get a 100% bonus on the first top-up (minimum $50). See the Karma XPU render farm page for server options, or download the iRender GPU app.
If Karma XPU renders are slow on your machine, iRender’s Karma XPU render farm gives you a dedicated remote server where you install Houdini, bring your own license and render on multiple RTX 5090 GPUs (for Karma XPU, 4 GPUs is usually the most efficient setup).


