Investigating the performance of np.einsum
A reader of my last blog post pointed out to me that for slicewise matmul-like operations, np.einsum is considerably slower than np.matmul unless you turn on the optimize flag in the parameters list: np.einsum(..., optimize = True).
Being somewhat skeptical, I fired up a Jupyter notebook and did some preliminary tests. And I'll be damned, it's completely true - even for two-operand cases where optimize shouldn't make any difference at all!
Test 1 is pretty simple - matrix multiplication of two C-order (aka row major order) matrices of varying dimensions. np.matmul is consistently about twenty times faster.
M1 | M2 | np.einsum | np.matmul | np.einsum / np.matmul |
---|---|---|---|---|
(100, 500) | (500, 100) | 0.765 | 0.045 | 17.055 |
(100, 1000) | (1000, 100) | 1.495 | 0.073 | 20.554 |
(100, 10000) | (10000, 100) | 15.148 | 0.896 | 16.899 |
For Test 2, with optimize=True, the results are drastically different. np.einsum is still slower, but only about 1.5 times slower at worst!
M1 | M2 | np.einsum | np.matmul | np.einsum / np.matmul |
---|---|---|---|---|
(100, 500) | (500, 100) | 0.063 | 0.043 | 1.474 |
(100, 1000) | (1000, 100) | 0.086 | 0.067 | 1.284 |
(100, 10000) | (10000, 100) | 1.000 | 0.936 | 1.068 |
Why?
My understanding of the optimize flag is that it determines an optimal contraction order when there are three or more operands. Here, we have only two operands. So optimize shouldn't make a difference, right?
But maybe optimize is doing something more than just picking a contraction order? Maybe optimize is aware of memory layout, and this has something to do with row-major vs. column-major layout?
In the grade-school method of matrix multiplication, to calculate a single entry, you iterate over a row in op1 while you iterate over a column in op2, so putting the second argument in column-major order might result in a speedup for np.einsum (assuming that np.einsum is sort of like a generalized version of the grade-school method of matrix multiplication under the hood, which I suspect is true).
So, for Test 3, I passed in a column-major matrix for the second operand to see if this sped up np.einsum when optimize=False.
Here are the results. Surprisingly, np.einsum is still considerably worse. Clearly, something is going on that I don't understand - perhaps np.einsum uses an entirely different code-path when optimize is True? Time to start digging.
M1 | M2 | np.einsum | np.matmul | np.einsum / np.matmul |
---|---|---|---|---|
(100, 500) | (500, 100) | 1.486 | 0.056 | 26.541 |
(100, 1000) | (1000, 100) | 3.885 | 0.125 | 31.070 |
(100, 10000) | (10000, 100) | 49.669 | 1.047 | 47.444 |
Going Deeper
The release notes of Numpy 1.12.0 mention the introduction of the optimize flag. However, the purpose of optimize seems to be to determine the order that arguments in a chain of operands are combined (i.e.; the associativity) - so optimize shouldn't make a difference for just two operands, right? Here's the release notes:
np.einsum now supports the optimize argument which will optimize the order of contraction. For example, np.einsum would complete the chain dot example np.einsum(‘ij,jk,kl->il’, a, b, c) in a single pass which would scale like N^4; however, when optimize=True np.einsum will create an intermediate array to reduce this scaling to N^3 or effectively np.dot(a, b).dot(c). Usage of intermediate tensors to reduce scaling has been applied to the general einsum summation notation. See np.einsum_path for more details.
To compound the mystery, some later release notes indicate that np.einsum was upgraded to use tensordot (which itself uses BLAS where appropriate). Now, that seems promising.
But then, why are we only seeing the speedup when optimize is True? What's going on?
If we read def einsum(*operands, out=None, optimize=False, **kwargs) in numpy/numpy/_core/einsumfunc.py, we'll come to this early-out logic almost immediately:
# If no optimization, run pure einsum if optimize is False: if specified_out: kwargs['out'] = out return c_einsum(*operands, **kwargs)
Does c_einsum utilize tensordot? I doubt it. Later on in the code, we see the tensordot call that the 1.14 notes seem to be referencing:
# Start contraction loop for num, contraction in enumerate(contraction_list): ... # Call tensordot if still possible if blas: ... # Contract! new_view = tensordot( *tmp_operands, axes=(tuple(left_pos), tuple(right_pos)) )
So, here's what's happening:
- The contraction_list loop is executed if optimize is True - even in the trivial two operand case.
- tensordot is only invoked in the contraction_list loop.
- Therefore, we invoke tensordot (and therefore BLAS) when optimize is True.
To me, this seems like a bug. IMHO, the "early-out" at the beginning of np.einsum should still detect if the operands are tensordot-compatible and call out to tensordot if possible. Then, we would get the obvious BLAS speedups even when optimize is False. After all, the semantics of optimize pertain to contraction order, not to usage of BLAS, which I think should be a given.
The boon here is that persons invoking np.einsum for operations that are equivalent to a tensordot invocation would get the appropriate speedups, which makes np.einsum a bit less dangerous from a performance point of view.
How does c_einsum actually work?
I spelunked some C code to check it out. The heart of the implementation is here.
After a great deal of argument parsing and parameter preparation, the axis iteration order is determined and a special-purpose iterator is prepared. Each yield from the iterator represents a different way to stride over all operands simultaneously.
# If no optimization, run pure einsum if optimize is False: if specified_out: kwargs['out'] = out return c_einsum(*operands, **kwargs)
Assuming certain special-case optimizations don't apply, an appropriate sum-of-products (sop) function is determined based on the datatypes involved:
# Start contraction loop for num, contraction in enumerate(contraction_list): ... # Call tensordot if still possible if blas: ... # Contract! new_view = tensordot( *tmp_operands, axes=(tuple(left_pos), tuple(right_pos)) )
And then, this sum-of-products (sop) operation is invoked on each multi-operand stride returned from the iterator, as seen here:
/* Allocate the iterator */ iter = NpyIter_AdvancedNew(nop+1, op, iter_flags, order, casting, op_flags, op_dtypes, ndim_iter, op_axes, NULL, 0);
That's my understanding of how einsum works, which is admittedly still a little thin - it really deserves more than the hour I've given it.
But it does confirm my suspicions, however, that it acts like a generalized, gigabrain version of the grade-school method of matrix multiplication. Ultimately, it delegates out to a series of "sum of product" operations which rely on "striders" moving through the operands - not too different from what you do with your fingers when you learn matrix multiplication.
Summary
So why is np.einsum generally faster when you call it with optimize=True? There are two reasons.
The first (and original) reason is it tries to find an optimal contraction path. However, as I pointed out, that shouldn't matter when we have just two operands, as we do in our performance tests.
The second (and newer) reason is that when optimize=True, even in the two operand case it activates a codepath that calls tensordot where possible, which in turn tries to use BLAS. And BLAS is as optimized as matrix multiplication gets!
So, the two-operands speedup mystery is solved! However, we haven't really covered the characteristics of the speedup due to contraction order. That will have to wait for a future post! Stay tuned!
The above is the detailed content of Investigating the performance of np.einsum. For more information, please follow other related articles on the PHP Chinese website!

Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

Video Face Swap
Swap faces in any video effortlessly with our completely free AI face swap tool!

Hot Article

Hot Tools

Notepad++7.3.1
Easy-to-use and free code editor

SublimeText3 Chinese version
Chinese version, very easy to use

Zend Studio 13.0.1
Powerful PHP integrated development environment

Dreamweaver CS6
Visual web development tools

SublimeText3 Mac version
God-level code editing software (SublimeText3)

Hot Topics











Python is easier to learn and use, while C is more powerful but complex. 1. Python syntax is concise and suitable for beginners. Dynamic typing and automatic memory management make it easy to use, but may cause runtime errors. 2.C provides low-level control and advanced features, suitable for high-performance applications, but has a high learning threshold and requires manual memory and type safety management.

Is it enough to learn Python for two hours a day? It depends on your goals and learning methods. 1) Develop a clear learning plan, 2) Select appropriate learning resources and methods, 3) Practice and review and consolidate hands-on practice and review and consolidate, and you can gradually master the basic knowledge and advanced functions of Python during this period.

Python is better than C in development efficiency, but C is higher in execution performance. 1. Python's concise syntax and rich libraries improve development efficiency. 2.C's compilation-type characteristics and hardware control improve execution performance. When making a choice, you need to weigh the development speed and execution efficiency based on project needs.

Python and C each have their own advantages, and the choice should be based on project requirements. 1) Python is suitable for rapid development and data processing due to its concise syntax and dynamic typing. 2)C is suitable for high performance and system programming due to its static typing and manual memory management.

Pythonlistsarepartofthestandardlibrary,whilearraysarenot.Listsarebuilt-in,versatile,andusedforstoringcollections,whereasarraysareprovidedbythearraymoduleandlesscommonlyusedduetolimitedfunctionality.

Python excels in automation, scripting, and task management. 1) Automation: File backup is realized through standard libraries such as os and shutil. 2) Script writing: Use the psutil library to monitor system resources. 3) Task management: Use the schedule library to schedule tasks. Python's ease of use and rich library support makes it the preferred tool in these areas.

Python's applications in scientific computing include data analysis, machine learning, numerical simulation and visualization. 1.Numpy provides efficient multi-dimensional arrays and mathematical functions. 2. SciPy extends Numpy functionality and provides optimization and linear algebra tools. 3. Pandas is used for data processing and analysis. 4.Matplotlib is used to generate various graphs and visual results.

Key applications of Python in web development include the use of Django and Flask frameworks, API development, data analysis and visualization, machine learning and AI, and performance optimization. 1. Django and Flask framework: Django is suitable for rapid development of complex applications, and Flask is suitable for small or highly customized projects. 2. API development: Use Flask or DjangoRESTFramework to build RESTfulAPI. 3. Data analysis and visualization: Use Python to process data and display it through the web interface. 4. Machine Learning and AI: Python is used to build intelligent web applications. 5. Performance optimization: optimized through asynchronous programming, caching and code
