8000 `torch.ldexp` upcasts 16-bit inputs to 32 bits. · Issue #133264 · pytorch/pytorch · GitHub
[go: up one dir, main page]
Skip to content

torch.ldexp upcasts 16-bit inputs to 32 bits. #133264

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
roman-openai opened this issue Aug 12, 2024 · 3 comments
Closed

torch.ldexp upcasts 16-bit inputs to 32 bits. #133264

roman-openai opened this issue Aug 12, 2024 · 3 comments
Assignees
@malfet
Labels
module: linear algebra Issues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmul triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module

Comments

@roman-openai
Copy link
roman-openai commented Aug 12, 2024 •
edited by pytorch-bot bot
Loading

🐛 Describe the bug

import torch
x = torch.tensor([1], dtype=torch.float16, device='cuda')
exp = torch.tensor([1], dtype=torch.int8, device='cuda')

torch.ldexp(x, exp).dtype

gives
torch.float32, but should preserve the input type torch.float16 and return infinity if going out of range (which is how it behaves if x is float32 - there is no upcast to float64).

Versions

Collecting environment information...
PyTorch version: 2.3.1+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.3 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: 14.0.0-1ubuntu1.1
CMake version: version 3.30.2
Libc version: glibc-2.35

Python version: 3.10.12 (main, Jul 29 2024, 16:56:48) [GCC 11.4.0] (64-bit runtime)
Python platform: Linux-6.1.85+-x86_64-with-glibc2.35
Is CUDA available: True
CUDA runtime version: 12.2.140
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration: GPU 0: Tesla T4
Nvidia driver version: 535.104.05
cuDNN version: Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.9.6
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
Address sizes:                        46 bits physical, 48 bits virtual
Byte Order:                           Little Endian
CPU(s):                               2
On-line CPU(s) list:                  0,1
Vendor ID:                            GenuineIntel
Model name:                           Intel(R) Xeon(R) CPU @ 2.00GHz
CPU family:                           6
Model:                                85
Thread(s) per core:                   2
Core(s) per socket:                   1
Socket(s):                            1
Stepping:                             3
BogoMIPS:                             4000.22
Flags:                                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid tsc_known_freq pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch invpcid_single ssbd ibrs ibpb stibp fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm mpx avx512f avx512dq rdseed adx smap clflushopt clwb avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves arat md_clear arch_capabilities
Hypervisor vendor:                    KVM
Virtualization type:                  full
L1d cache:                            32 KiB (1 instance)
L1i cache:                            32 KiB (1 instance)
L2 cache:                             1 MiB (1 instance)
L3 cache:                             38.5 MiB (1 instance)
NUMA node(s):                         1
NUMA node0 CPU(s):                    0,1
Vulnerability Gather data sampling:   Not affected
Vulnerability Itlb multihit:          Not affected
Vulnerability L1tf:                   Mitigation; PTE Inversion
Vulnerability Mds:                    Vulnerable; SMT Host state unknown
Vulnerability Meltdown:               Vulnerable
Vulnerability Mmio stale data:        Vulnerable
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed:               Vulnerable
Vulnerability Spec rstack overflow:   Not affected
Vulnerability Spec store bypass:      Vulnerable
Vulnerability Spectre v1:             Vulnerable: __user pointer sanitization and usercopy barriers only; no swapgs barriers
Vulnerability Spectre v2:             Vulnerable; IBPB: disabled; STIBP: disabled; PBRSB-eIBRS: Not affected; BHI: Vulnerable (Syscall hardening enabled)
Vulnerability Srbds:                  Not affected
Vulnerability Tsx async abort:        Vulnerable

Versions of relevant libraries:
[pip3] numpy==1.26.4
[pip3] optree==0.12.1
[pip3] torch==2.3.1+cu121
[pip3] torchaudio==2.3.1+cu121
[pip3] torchsummary==1.5.1
[pip3] torchtext==0.18.0
[pip3] torchvision==0.18.1+cu121
[pip3] triton==2.3.1
[conda] Could not collect

cc @jianyuh @nikitaved @pearu @mruberry @walterddr @xwang233 @lezcano
@Skylion007
8000 Copy link
Collaborator

Interestingly, we have a comment which notes this for the NP tests:

pytorch/torch/_numpy/_ufuncs.py

Line 147 in 7756175

# ldexp casting is special : the dtype of the result == dtype of the 1st arg

@isuruf
Copy link
Collaborator
isuruf commented Aug 13, 2024

All the ldexp issues you opened is due to the fact that ldexp(x, exp) is implemented as x * 2.0**(exp) where 2.0 is interpreted as a float32. We should just use the CUDA and libc implementations instead.

@janeyx99 janeyx99 added triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module module: linear algebra Issues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmul labels Aug 14, 2024
This was referenced Aug 14, 2024
Closed
Closed
@malfet malfet self-assigned this Aug 14, 2024
@malfet
Copy link
Contributor
malfet commented Aug 14, 2024

Closing as duplicate of #133267 because fix for 64 dtype promotion also works for preserving 16-bit dtypes

@malfet malfet closed this as not planned Won't fix, can't repro, duplicate, stale Aug 14, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@malfet malfet

Labels
module: linear algebra Issues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmul triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
5 participants
@Skylion007 @malfet @isuruf @janeyx99 @roman-openai
0