# I keep failing to install Flash Attention 3 in the LTX-2 UV environment
**URL:** https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560
**Category:** DGX Spark / GB10
**Created:** [January 14, 2026, 12:06pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560 "2026-01-14T12:06:40Z")
**Posts on this page:** 11
**Page:** 1
### Author:  [@hongde](https://forums.developer.nvidia.com/u/hongde)
#### Post date: [January 14, 2026, 12:06pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/1 "2026-01-14T12:06:40Z")
copying /tmp/tmpk4223xzz.build-lib/flash\_attn\_3/\_C.abi3.so →
flash\_attn\_3
```auto
[stderr]
/home/xiaohong/LTX-2/.venv/lib/python3.12/site-packages/setuptools/dist.py:332:
InformationOnly: Normalizing '3.0.0.b1' to '3.0.0b1'
self.metadata.version = self._normalize_version(self.metadata.version)
/home/xiaohong/LTX-2/.venv/lib/python3.12/site-packages/setuptools/dist.py:759:
SetuptoolsDeprecationWarning: License classifiers are deprecated.
!!
********************************************************************************
Please consider removing the following classifiers in favor of
a SPDX license expression:
License :: OSI Approved :: Apache Software License
See
https://packaging.python.org/en/latest/guides/writing-pyproject-toml/#license
for details.
********************************************************************************
!!
self._finalize_license_expression()
W0114 07:08:01.022000 1591534 torch/utils/cpp_extension.py:531] There
are no /usr/bin/g++-11 version bounds defined for CUDA version 13.0
error: could not create 'flash_attn_3/_C.abi3.so': No such file or
directory
hint: This usually indicates a problem with the package or the build
environment.
```
DEBUG Released lock at `/home/xiaohong/LTX-2/.venv/.lock`
DEBUG Released lock at `/home/xiaohong/.cache/uv/.lock`
---
### Author:  [@aniculescu](https://forums.developer.nvidia.com/u/aniculescu)
#### Post date: [January 14, 2026, 7:11pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/3 "2026-01-14T19:11:04Z")
Can you provide more details so we can attempt to reproduce this issue? What scripts or commands did you run?
---
### Author:  [@hongde](https://forums.developer.nvidia.com/u/hongde)
#### Post date: [January 15, 2026, 9:33am UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/6 "2026-01-15T09:33:16Z")
export CUDA\_HOME=/usr/local/cuda
export PATH=$CUDA\_HOME/bin:$PATH
export LD\_LIBRARY\_PATH=/usr/local/lib:$CUDA\_HOME/lib:$CUDA\_HOME/lib64:$LD\_LIBRARY\_PATH
export CC=/usr/bin/gcc-11
export CXX=/usr/bin/g+±11
export TORCH\_CUDA\_ARCH\_LIST=“12.1”
MAX\_JOBS=12 uv pip install -e . --verbose --no-build-isolation
---
### Author:  [@cyuen1](https://forums.developer.nvidia.com/u/cyuen1)
#### Post date: [January 16, 2026, 9:38pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/8 "2026-01-16T21:38:58Z")
Hello Hongde,
What version of PyTorch are you using/are you using the PyTorch NGC container?
---
### Author:  [@cyuen1](https://forums.developer.nvidia.com/u/cyuen1)
#### Post date: [January 16, 2026, 11:42pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/9 "2026-01-16T23:42:25Z")
Hello again,
I also was able to reproduce your error. Before running your commands, I created the directory first.
For me, I was working out of a NGC PyTorch container where my working directory was `/workspace/flash-attention/hopper` (via the flash-attention repo). I did `mkdir -p flash_attn_3`and it worked for me!
Please try that fix and let me know if you have any other questions or concerns!
---
### Author:  [@hongde](https://forums.developer.nvidia.com/u/hongde)
#### Post date: [January 18, 2026, 8:51am UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/10 "2026-01-18T08:51:55Z")
thank you
(LTX-2) root@c0e2a06cb538:/workspace/flash-attention/hopper# pip show torch
Name: torch
Version: 2.10.0a0+b4e4ee81d3.nv25.12
Summary: Tensors and Dynamic neural networks in Python with strong GPU acceleration
Home-page: [https://pytorch.org](https://pytorch.org)
Author:
Author-email: PyTorch Team [packages@pytorch.org](mailto:packages@pytorch.org)
License: BSD-3-Clause
Location: /usr/local/lib/python3.12/dist-packages
Requires: filelock, fsspec, jinja2, networkx, setuptools, sympy, typing-extensions
Required-by: flash\_attn, flash\_attn\_3, lightning-thunder, nvidia-modelopt, nvidia-resiliency-ext, torchdata, torchprofile, torchvision, transformer\_engine
---
### Author:  [@cyuen1](https://forums.developer.nvidia.com/u/cyuen1)
#### Post date: [January 20, 2026, 5:52pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/11 "2026-01-20T17:52:38Z")
Great. That PyTorch version should be sufficient. Were you able to get flash attention 3 installed properly?
---
### Author:  [@nmaine](https://forums.developer.nvidia.com/u/nmaine)
#### Post date: [April 21, 2026, 5:57pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/12 "2026-04-21T17:57:57Z")
Adding a data point in case it saves someone else the 50-minute build cycle.
Following the workaround described upstream (the `mkdir -p flash_attn_3` fix combined with setting `TORCH_CUDA_ARCH_LIST`) the FA3 build completes successfully inside `nvcr.io/nvidia/pytorch:26.03-py3` on a DGX Spark (GB10, compute capability 12.1, aarch64, CUDA 13.2 forward-compat driver). For the record I used `TORCH_CUDA_ARCH_LIST="12.0"` per the `sm_120` binary-compatibility approach documented in a recent Medium writeup on FA2 builds for this hardware. `pip install -e . --verbose --no-build-isolation` finishes clean after about 50 minutes, the editable wheel registers, and `import flash_attn_interface` succeeds.
At runtime, however, a minimal `flash_attn_func(q, k, v, causal=True)` on bf16 tensors fails with:
```auto
CUDA error (flash-attention-src/hopper/flash_fwd_launch_template.h:193):
no kernel image is available for execution on the device
```
Looking at the build log, every kernel instantiation object file names its target as `sm80` or `sm90` (for example `flash_fwd_hdim128_bf16_sm90.o`, `flash_bwd_hdim128_bf16_sm80.o`). I do not see any `sm_120` or `sm_121` compilations, which is consistent with FA3’s upstream Hopper kernel set not currently including Blackwell instantiations, and appears to be the tracking issue in [Dao-AILab/flash-attention#1969](https://github.com/Dao-AILab/flash-attention/issues/1969). It also lines up with the choice made in AI2’s `open-instruct` DGX Spark branch to mark `flash-attn` as unavailable on aarch64 and fall back to SDPA.
A few questions I would genuinely appreciate correction on:
1. Is my read of the build log correct, i.e., the current FA3 code path produces only SM80 and SM90 kernels regardless of `TORCH_CUDA_ARCH_LIST`, so the “successful build” on GB10 is effectively producing Hopper and Ampere binaries with no Blackwell match? Or is there a build flag or source path I missed?
2. Has anyone produced SM120 or SM121 compatible FA3 kernels using `TORCH_CUDA_ARCH_LIST="12.1"` or the accelerated `"12.1a"` variant? Medium reports that FA2.8.3 builds usefully at `"12.0"`, but that is a different codebase.
3. For DGX Spark users, is the recommended path for now essentially “use FA2 or SDPA,” since `flash_attn` 2.7.4 is pre-installed in the NGC container and in my tests produces outputs bit-exact with `torch.nn.functional.scaled_dot_product_attention`? If so, that would be worth adding to this thread for the next person who reaches a successful FA3 build and is then surprised by a runtime kernel-image failure.
Happy to share the exact reproduction (container tag, build script, verification harness) if useful. I may still be missing something obvious here, and am genuinely interested in being corrected.
---
### Author:  [@paulsc.liu](https://forums.developer.nvidia.com/u/paulsc.liu)
#### Post date: [April 22, 2026, 2:08am UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/13 "2026-04-22T02:08:20Z")
> [@nmaine](#):
>
> use FA2 or SDPA
I think so. I enountered this problem when I try to compile SGLang-Diffusion. I end up comment out FA.
FA-4 is also a problem for “consumer” Blackwell. It require tcgen05 which only “datacenter” Blackwell B200 has it.
---
### Author:  [@AoE](https://forums.developer.nvidia.com/u/AoE)
#### Post date: [April 26, 2026, 8:56pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/14 "2026-04-26T20:56:22Z")
Thank you for the investigation. It looks like we’ll have to wait for V4 to get this functionality and in the mean time, we can rely on Sage Attention.
---
### Author:  [@system](https://forums.developer.nvidia.com/u/system)
#### Post date: [May 10, 2026, 8:56pm UTC](https://forums.developer.nvidia.com/t/i-keep-failing-to-install-flash-attention-3-in-the-ltx-2-uv-environment/357560/15 "2026-05-10T20:56:27Z")
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.