Function's created via exec() raise recompilation warning and cache_size_limit error

[fferflo]

🐛 Describe the bug

I am creating lots of different functions via Python's exec() and compiling them using torch.compile as in the example below. Even though I am calling every compiled function only once, it causes a recompilation warning and the error torch._dynamo.exc.Unsupported: cache_size_limit reached. This code reproduces the error:

import torch

code = """
import torch
def op(x):
    return torch.mean(x)
"""

for _ in range(20):
    scope = {}
    exec(code, scope, scope)
    op = scope["op"]

    op = torch.compile(op, fullgraph=True)

    op(torch.zeros((3,)))

It's possible the error might stem from this line

pytorch/torch/_dynamo/eval_frame.py

Line 353 in 72271fb

fn = external_utils.wrap_inline(fn)

which wraps the function in a custom wrapper that is mentioned in the recompilation warning. The comments say that the wrapper handles builtins by adding a frame around them, but due to how the check is written it also applies to functions created in exec(). Without that line (commenting it out, or renaming the function to _call_impl as a workaround which also skips the line) the error does not appear.

Error logs

[2024-04-16 22:28:39,856] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (8)
[2024-04-16 22:28:39,856] torch._dynamo.convert_frame: [WARNING]    function: 'inner' (.../python3.10/site-packages/torch/_dynamo/external_utils.py:15)
[2024-04-16 22:28:39,856] torch._dynamo.convert_frame: [WARNING]    last reason: ___check_obj_id(getattr(L['fn'], '__code__', None), 140699596231360)
[2024-04-16 22:28:39,856] torch._dynamo.convert_frame: [WARNING] To log all recompilation reasons, use TORCH_LOGS="recompiles".
[2024-04-16 22:28:39,856] torch._dynamo.convert_frame: [WARNING] To diagnose recompilation issues, see https://pytorch.org/docs/master/compile/troubleshooting.html.
...
torch._dynamo.exc.Unsupported: cache_size_limit reached

Minified repro

No response

Versions

Collecting environment information...
PyTorch version: 2.2.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 20.04.6 LTS (x86_64)
GCC version: (GCC) 11.3.1 20230104
Clang version: Could not collect
CMake version: version 3.25.20230104-g4fec7f9
Libc version: glibc-2.31

Python version: 3.10.9 (tags/v3.10.9-dirty:1dd9be6584, Dec 12 2023, 13:41:49) [GCC 11.3.1 20230104] (64-bit runtime)
Python platform: Linux-5.15.0-91-generic-x86_64-with-glibc2.31
Is CUDA available: True
CUDA runtime version: 12.2.140
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration:
GPU 0: Quadro RTX 6000
GPU 1: Quadro RTX 6000

Nvidia driver version: 535.129.03
cuDNN version: /usr/local/cuda-10.2/targets/x86_64-linux/lib/libcudnn.so.7.6.5
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 46 bits physical, 48 bits virtual
CPU(s): 36
On-line CPU(s) list: 0-35
Thread(s) per core: 2
Core(s) per socket: 18
Socket(s): 1
NUMA node(s): 1
Vendor ID: GenuineIntel
CPU family: 6
Model: 85
Model name: Intel(R) Core(TM) i9-9980XE CPU @ 3.00GHz
Stepping: 4
CPU MHz: 3000.000
CPU max MHz: 4500,0000
CPU min MHz: 1200,0000
BogoMIPS: 6000.00
Virtualization: VT-x
L1d cache: 576 KiB
L1i cache: 576 KiB
L2 cache: 18 MiB
L3 cache: 24,8 MiB
NUMA node0 CPU(s): 0-35
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit: KVM: Mitigation: VMX disabled
Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cache flushes, SMT vulnerable
Vulnerability Mds: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Meltdown: Mitigation; PTI
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Retbleed: Mitigation; IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; IBRS, IBPB conditional, STIBP conditional, RSB filling, PBRSB-eIBRS Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Mitigation; Clear CPU buffers; SMT vulnerable
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single pti ssbd mba ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req md_clear flush_l1d arch_capabilities

Versions of relevant libraries:
[pip3] mypy==1.8.0
[pip3] mypy-extensions==1.0.0
[pip3] numpy==1.26.2
[pip3] pytorch-benchmark==0.3.6
[pip3] torch==2.2.0
[pip3] torchvision==0.17.0
[pip3] triton==2.2.0
[conda] blas 1.0 mkl
[conda] mkl 2020.2 256
[conda] mkl-service 2.3.0 py38he904b0f_0
[conda] mkl_fft 1.2.0 py38h23d657b_0
[conda] mkl_random 1.1.1 py38h0573a6f_0
[conda] numpy 1.19.2 py38h54aff64_0
[conda] numpy-base 1.19.2 py38hfa32c7d_0
[conda] numpydoc 1.1.0 pyhd3eb1b0_1

cc @ezyang @msaroufim @bdhirsh @anijain2305 @chauhang @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @chenyang78 @kadeng

[ezyang]

Nice catch. Send us a PR?}