I have a large h5 file with 5-dimensional numpy array in HDFS. File size is ~130Gb. I am facing memory issues while loading the file with process gets killed with OOM Error even though machine has 256Gb RAM. How can I write the file in chunks and load back in chunks? I looked around and found that h5py provides method to chunk the dataset like so but how do I load back the data in chunks? Also will it work if the file resides in HDFS?
dset = f.create_dataset("Images2", (100,480,640), 'f', chunks=True)
Idea is to load the file in batches for less I/O time as well as memory issues. Any help would be much appreciated.
Two similar (but different) h5py I/O concepts are mentioned in the answer and comments above:
HDF5 Chunking is used to enable chunked I/O for improved performance. Chunking may not help if you get an OOM error when you try to read a large dataset with insufficient memory.
NumPy style Slicing is used to read a slice of the data from the drive to memory (or write a slice of data to the drive). Slicing is the key to avoid OOM errors when reading very large files.
Also, when creating very large datasets, you generally need to make
it resizeable. You can allocate an initial size, then use the ".resize()" method to increase the size on disk.
I wrote a simple example that shows how to use both slicing and chunking. It loads 100 images at a time into a resizeable dataset. It then closes the file and reopens (read-only) to read 100 images at a time into a NumPy array.
Effective chunking requires appropriate size/shape and is based on your array shape and I/O needs. I set the chunk size/shape in my example to match the size of 100 image array I was writing/reading.
This example should get you started. You will need to modify to use a 5-d array/dataset.
import numpy as np
import h5py
with h5py.File('SO_64645940.h5','w') as h5w:
img_ds = h5w.create_dataset('Images', shape=(100,480,640), dtype='f', maxshape=(None,480,640),chunks=(10,480,640))
next_img_row = 0
arr = np.random.random(100*480*640).reshape(100,480,640)
for cnt in range(1,10):
# print(cnt,img_ds.len(),next_img_row)
if img_ds.len() == next_img_row :
img_ds.resize(100*cnt,axis=0)
print('new ds size=',img_ds.len())
h5w['Images'][next_img_row:next_img_row+100] = arr
next_img_row += 100
with h5py.File('SO_64645940.h5','r') as h5r:
for cnt in range(10):
print('get slice#',str(cnt))
img_arr = h5r['Images'][cnt*100:(cnt+1)*100]
Chunking in HDF5 means that the data is not stored contigous, but in chunks.
See information here: https://docs.h5py.org/en/stable/high/dataset.html#chunked-storage
--> So this doesn't help you with your problem.
The solution might be that you build a function yourself to load the data chunkwise.
I made it for example this way for getting the data chunked:
def get_chunked(data, chunk_size=100):
for i in give_chunk(len(data), chunk_size):
chunked_array = data[i]
yield chunked_array
def give_chunk(length, chunk_size):
it = iter(range(length))
while True:
chunk = list(itertools.islice(it, chunk_size))
if not chunk:
break
yield chunk
For writing the data to HDF5 you can create the dataset first and then write the data chunk wise with slicing, see h5py documentation: https://docs.h5py.org/en/stable/high/dataset.html#reading-writing-data
I really can recommend this book for basic knowledge about HDF5: https://www.oreilly.com/library/view/python-and-hdf5/9781491944981/
I want to build recommendation system using association rules with implemented in mlxtend library apriori algorithm. In my sales data there is information about 36 millions of transactions and 50k unique products.
I tried to use sklearn OneHotEncoder and pandas get_dummies() but both are giving OOM error as they are not able to create frame in shape of (36 mil, 50k)
MemoryError: Unable to allocate 398. GiB for an array with shape (36113798, 50087) and data type uint8
Is there any other solution?
Like you, I too had out of memory error with mlxtend at first, but the following small changes fixed the problem completely.
`
from mlxtend.preprocessing import TransactionEncoder
import pandas as pd
te = TransactionEncoder()
#te_ary = te.fit(itemSetList).transform(itemSetList)
#df = pd.DataFrame(te_ary, columns=te.columns_)
fitted = te.fit(itemSetList)
te_ary = fitted.transform(itemSetList, sparse=True) # seemed to work good
df = pd.DataFrame.sparse.from_spmatrix(te_ary, columns=te.columns_) # seemed to work good
# now you can call mlxtend's fpgrowth() followed by association_rules()
`
You should also use fpgrowth instead of apriori on the big transaction datasets because apriori is too primitive. fpgrowth is more intelligent and modern than apriori but gives equivalent results. The mlxtend lib supports both apriori and fpgrowth.
I think a good solution would be to use embeddings instead of one-hot encoding for your problem. In addition, I recommend that you split your dataset into smaller subsets to further avoid the memory consumption problems.
You should also consult this thread : https://datascience.stackexchange.com/questions/29851/one-hot-encoding-vs-word-embeding-when-to-choose-one-or-another
I am trying make some common computations, like matrix multiplication, but without gradient computation. An example of my computation is like
import numpy as np
from scipy.special import logsumexp
var = 1e-8
a = np.random.randint(0,10,(128,20))
result = np.logsumexp(a, axis=1) / 2. + np.log(np.pi * var)
I want to use torch (gpu) to speed up the computation. Here is the code
import numpy as np
import torch
var = 1e-8
a = np.random.randint(0,10,(128,20))
a = torch.numpy_from(a).cuda()
result = torch.logsumexp(a, dim=1)/ 2. + np.log(np.pi*var)
but i have some questions:
Could the above code speed up the computation? I don't know if it works.
Do I need to convert all values into torch.tensor, like from var to torch.tensor(var).cuda() and from np.log(np.pi*var) to a torch.tensor?
Do I need to convert all tensors into gpu by myself, especially for some intermediate variable?
If the above code doesn't work, how can I speed up the computation with gpu?
You could use torch only to do the computations.
import torch
# optimization by passing device argument, tensor is created on gpu and hence move operation is saved
# convert to float to use with logsumexp
a = torch.randint(0,10, (128,20), device="cuda").float()
result = torch.logsumexp(a, dim=1)/ 2.
Answers to your some of your questions:
Could the above code speed up the computation?
It depends. If you have too many matrix multiplication, using gpu can give speed up.
Do I need to convert all values into torch.tensor, like from var to torch.tensor(var).cuda() and from np.log(np.pi*var) to a torch.tensor?
Yes
Do I need to convert all tensors into gpu by myself, especially for some intermediate variable?
Only leaf variables need to converted, intermediate variable will be placed on device on which the operations are done. For ex: if a and b are on gpu, then as a result of operation c=a+b, c will also be on gpu.
As of August 2017, Pandas DataFame.apply() is unfortunately still limited to working with a single core, meaning that a multi-core machine will waste the majority of its compute-time when you run df.apply(myfunc, axis=1).
How can you use all your cores to run apply on a dataframe in parallel?
You may use the swifter package:
pip install swifter
(Note that you may want to use this in a virtualenv to avoid version conflicts with installed dependencies.)
Swifter works as a plugin for pandas, allowing you to reuse the apply function:
import swifter
def some_function(data):
return data * 10
data['out'] = data['in'].swifter.apply(some_function)
It will automatically figure out the most efficient way to parallelize the function, no matter if it's vectorized (as in the above example) or not.
More examples and a performance comparison are available on GitHub. Note that the package is under active development, so the API may change.
Also note that this will not work automatically for string columns. When using strings, Swifter will fallback to a “simple” Pandas apply, which will not be parallel. In this case, even forcing it to use dask will not create performance improvements, and you would be better off just splitting your dataset manually and parallelizing using multiprocessing.
The simplest way is to use Dask's map_partitions. You need these imports (you will need to pip install dask):
import pandas as pd
import dask.dataframe as dd
from dask.multiprocessing import get
and the syntax is
data = <your_pandas_dataframe>
ddata = dd.from_pandas(data, npartitions=30)
def myfunc(x,y,z, ...): return <whatever>
res = ddata.map_partitions(lambda df: df.apply((lambda row: myfunc(*row)), axis=1)).compute(get=get)
(I believe that 30 is a suitable number of partitions if you have 16 cores). Just for completeness, I timed the difference on my machine (16 cores):
data = pd.DataFrame()
data['col1'] = np.random.normal(size = 1500000)
data['col2'] = np.random.normal(size = 1500000)
ddata = dd.from_pandas(data, npartitions=30)
def myfunc(x,y): return y*(x**2+1)
def apply_myfunc_to_DF(df): return df.apply((lambda row: myfunc(*row)), axis=1)
def pandas_apply(): return apply_myfunc_to_DF(data)
def dask_apply(): return ddata.map_partitions(apply_myfunc_to_DF).compute(get=get)
def vectorized(): return myfunc(data['col1'], data['col2'] )
t_pds = timeit.Timer(lambda: pandas_apply())
print(t_pds.timeit(number=1))
28.16970546543598
t_dsk = timeit.Timer(lambda: dask_apply())
print(t_dsk.timeit(number=1))
2.708152851089835
t_vec = timeit.Timer(lambda: vectorized())
print(t_vec.timeit(number=1))
0.010668013244867325
Giving a factor of 10 speedup going from pandas apply to dask apply on partitions. Of course, if you have a function you can vectorize, you should - in this case the function (y*(x**2+1)) is trivially vectorized, but there are plenty of things that are impossible to vectorize.
you can try pandarallel instead: A simple and efficient tool to parallelize your pandas operations on all your CPUs (On Linux & macOS)
Parallelization has a cost (instanciating new processes, sending data via shared memory, etc ...), so parallelization is efficiant only if the amount of calculation to parallelize is high enough. For very little amount of data, using parallezation not always worth it.
Functions applied should NOT be lambda functions.
from pandarallel import pandarallel
from math import sin
pandarallel.initialize()
# FORBIDDEN
df.parallel_apply(lambda x: sin(x**2), axis=1)
# ALLOWED
def func(x):
return sin(x**2)
df.parallel_apply(func, axis=1)
see https://github.com/nalepae/pandarallel
If you want to stay in native python:
import multiprocessing as mp
with mp.Pool(mp.cpu_count()) as pool:
df['newcol'] = pool.map(f, df['col'])
will apply function f in a parallel fashion to column col of dataframe df
Just want to give an update answer for Dask
import dask.dataframe as dd
def your_func(row):
#do something
return row
ddf = dd.from_pandas(df, npartitions=30) # find your own number of partitions
ddf_update = ddf.apply(your_func, axis=1).compute()
On my 100,000 records, without Dask:
CPU times: user 6min 32s, sys: 100 ms, total: 6min 32s
Wall time: 6min 32s
With Dask:
CPU times: user 5.19 s, sys: 784 ms, total: 5.98 s
Wall time: 1min 3s
To use all (physical or logical) cores, you could try mapply as an alternative to swifter and pandarallel.
You can set the amount of cores (and the chunking behaviour) upon init:
import pandas as pd
import mapply
mapply.init(n_workers=-1)
...
df.mapply(myfunc, axis=1)
By default (n_workers=-1), the package uses all physical CPUs available on the system. If your system uses hyper-threading (usually twice the amount of physical CPUs would show up as logical cores), mapply will spawn one extra worker to prioritise the multiprocessing pool over other processes on the system.
Depending on your definition of all your cores, you could also use all logical cores instead (beware that like this the CPU-bound processes will be fighting for physical CPUs, which might slow down your operation):
import multiprocessing
n_workers = multiprocessing.cpu_count()
# or more explicit
import psutil
n_workers = psutil.cpu_count(logical=True)
Here is an example of sklearn base transformer, in which pandas apply is parallelized
import multiprocessing as mp
from sklearn.base import TransformerMixin, BaseEstimator
class ParllelTransformer(BaseEstimator, TransformerMixin):
def __init__(self,
n_jobs=1):
"""
n_jobs - parallel jobs to run
"""
self.variety = variety
self.user_abbrevs = user_abbrevs
self.n_jobs = n_jobs
def fit(self, X, y=None):
return self
def transform(self, X, *_):
X_copy = X.copy()
cores = mp.cpu_count()
partitions = 1
if self.n_jobs <= -1:
partitions = cores
elif self.n_jobs <= 0:
partitions = 1
else:
partitions = min(self.n_jobs, cores)
if partitions == 1:
# transform sequentially
return X_copy.apply(self._transform_one)
# splitting data into batches
data_split = np.array_split(X_copy, partitions)
pool = mp.Pool(cores)
# Here reduce function - concationation of transformed batches
data = pd.concat(
pool.map(self._preprocess_part, data_split)
)
pool.close()
pool.join()
return data
def _transform_part(self, df_part):
return df_part.apply(self._transform_one)
def _transform_one(self, line):
# some kind of transformations here
return line
for more info see https://towardsdatascience.com/4-easy-steps-to-improve-your-machine-learning-code-performance-88a0b0eeffa8
The native Python solution (with numpy) that can be applied on the whole DataFrame as the original question asks (not only on a single column)
import numpy as np
import multiprocessing as mp
dfs = np.array_split(df, 8000) # divide the dataframe as desired
def f_app(df):
return df.apply(myfunc, axis=1)
with mp.Pool(mp.cpu_count()) as pool:
res = pd.concat(pool.map(f_app, dfs))
Here another one using Joblib and some helper code from scikit-learn. Lightweight (if you already have scikit-learn), good if you prefer more control over what it is doing since joblib is easily hackable.
from joblib import parallel_backend, Parallel, delayed, effective_n_jobs
from sklearn.utils import gen_even_slices
from sklearn.utils.validation import _num_samples
def parallel_apply(df, func, n_jobs= -1, **kwargs):
""" Pandas apply in parallel using joblib.
Uses sklearn.utils to partition input evenly.
Args:
df: Pandas DataFrame, Series, or any other object that supports slicing and apply.
func: Callable to apply
n_jobs: Desired number of workers. Default value -1 means use all available cores.
**kwargs: Any additional parameters will be supplied to the apply function
Returns:
Same as for normal Pandas DataFrame.apply()
"""
if effective_n_jobs(n_jobs) == 1:
return df.apply(func, **kwargs)
else:
ret = Parallel(n_jobs=n_jobs)(
delayed(type(df).apply)(df[s], func, **kwargs)
for s in gen_even_slices(_num_samples(df), effective_n_jobs(n_jobs)))
return pd.concat(ret)
Usage: result = parallel_apply(my_dataframe, my_func)
Instead of
df["new"] = df["old"].map(fun)
do
from joblib import Parallel, delayed
df["new"] = Parallel(n_jobs=-1, verbose=10)(delayed(fun)(i) for i in df["old"])
To me this is a slight improvement over
import multiprocessing as mp
with mp.Pool(mp.cpu_count()) as pool:
df["new"] = pool.map(fun, df["old"])
as you get a progress indication and automatic batching if the jobs are very small.
Since the question was "How can you use all your cores to run apply on a dataframe in parallel?", the answer can also be with modin. You can run all cores in parallel, though the real time is worse.
See https://github.com/modin-project/modin . It runs of top of dask or ray. They say "Modin is a DataFrame designed for datasets from 1MB to 1TB+." I tried: pip3 install "modin"[ray]". Modin vs pandas was - 12 sec on six cores vs. 6 sec.
In case you need to do something based on the column name inside the function beware that .apply function may give you some trouble. In my case I needed to change the column type using astype() function based on the column name. This is probably not the most efficient way of doing it but suffices the purpose and keeps the column names as the original one.
import multiprocessing as mp
def f(df):
""" the function that you want to apply to each column """
column_name = df.columns[0] # this is the same as the original column name
# do something what you need to do to that column
return df
# Here I just make a list of all the columns. If you don't use .to_frame()
# it will pass series type instead of a dataframe
dfs = [df[column].to_frame() for column in df.columns]
with mp.Pool(mp.cpu_num) as pool:
processed_df = pd.concat(pool.map(f, dfs), axis=1)
I have access through ssh to a cluster of n GPUs. Tensorflow automatically gave them names gpu:0,...,gpu:(n-1).
Others have access too and sometimes they take random gpus.
I did not place any tf.device() explicitely because that is cumbersome and even if I selected gpu number j and that someone is already on gpu number j that would be problematic.
I would like to go throuh the gpus usage and find the first that is unused and use only this one.
I guess someone could parse the output of nvidia-smi with bash and get a variable i and feed that variable i to the tensorflow script as the number of the gpu to use.
I have never seen any example of this. I imagine it is a pretty common problem. What would be the simplest way to do that ? Is a pure tensorflow one available ?
I'm not aware of pure-TensorFlow solution. The problem is that existing place for TensorFlow configurations is a Session config. However, for GPU memory, a GPU memory pool is shared for all TensorFlow sessions within a process, so Session config would be the wrong place to add it, and there's no mechanism for process-global config (but there should be, to also be able to configure process-global Eigen threadpool). So you need to do on on a process level by using CUDA_VISIBLE_DEVICES environment variable.
Something like this:
import subprocess, re
# Nvidia-smi GPU memory parsing.
# Tested on nvidia-smi 370.23
def run_command(cmd):
"""Run command, return output as string."""
output = subprocess.Popen(cmd, stdout=subprocess.PIPE, shell=True).communicate()[0]
return output.decode("ascii")
def list_available_gpus():
"""Returns list of available GPU ids."""
output = run_command("nvidia-smi -L")
# lines of the form GPU 0: TITAN X
gpu_regex = re.compile(r"GPU (?P<gpu_id>\d+):")
result = []
for line in output.strip().split("\n"):
m = gpu_regex.match(line)
assert m, "Couldnt parse "+line
result.append(int(m.group("gpu_id")))
return result
def gpu_memory_map():
"""Returns map of GPU id to memory allocated on that GPU."""
output = run_command("nvidia-smi")
gpu_output = output[output.find("GPU Memory"):]
# lines of the form
# | 0 8734 C python 11705MiB |
memory_regex = re.compile(r"[|]\s+?(?P<gpu_id>\d+)\D+?(?P<pid>\d+).+[ ](?P<gpu_memory>\d+)MiB")
rows = gpu_output.split("\n")
result = {gpu_id: 0 for gpu_id in list_available_gpus()}
for row in gpu_output.split("\n"):
m = memory_regex.search(row)
if not m:
continue
gpu_id = int(m.group("gpu_id"))
gpu_memory = int(m.group("gpu_memory"))
result[gpu_id] += gpu_memory
return result
def pick_gpu_lowest_memory():
"""Returns GPU with the least allocated memory"""
memory_gpu_map = [(memory, gpu_id) for (gpu_id, memory) in gpu_memory_map().items()]
best_memory, best_gpu = sorted(memory_gpu_map)[0]
return best_gpu
You can then put it in utils.py and set GPU in your TensorFlow script before first tensorflow import. IE
import utils
import os
os.environ["CUDA_VISIBLE_DEVICES"] = str(utils.pick_gpu_lowest_memory())
import tensorflow
An implementation along the lines of Yaroslav Bulatov's solution is available on https://github.com/bamos/setGPU.