I use a custom heap implementation in one of my projects. It consists of two major parts:
Fixed size-block heap. I.e. a heap that allocates blocks of a specific size only. It allocates larger memory blocks (either virtual memory pages or from another heap), and then divides them into atomic allocation units.
It performs allocation/freeing fast (in O(1)) and there's no memory usage overhead, not taking into account things imposed by the external heap.
Global general-purpose heap. It consists of buckets of the above (fixed-size) heaps. WRT the requested allocation size it chooses the appropriate bucket, and performs the allocation via it.
Since the whole application is (heavily) multi-threaded - the global heap locks the appropriate bucket during its operation.
Note: in contrast to the traditional heaps, this heap requires the allocation size not only for the allocation, but also for freeing. This allows to identify the appropriate bucket without searches or extra memory overhead (such as saving the block size preceding the allocated block). Though somewhat less convenient, this is ok in my case. Moreover, since the "bucket configuration" is known at compile-time (implemented via C++ template voodoo) - the appropriate bucket is determined at compile time.
So far everything looks (and works) good.
Recently I worked on an algorithm that performs heap operations heavily, and naturally affected significantly by the heap performance. Profiling revealed that its performance is considerably impacted by the locking. That is, the heap itself works very fast (typical allocation involves just a few memory dereferencing instructions), but since the whole application is multi-threaded - the appropriate bucket is protected by the critical section, which relies on interlocked instructions, which are much heavier.
I've fixed this meanwhile by giving this algorithm its own dedicated heap, which is not protected by a critical section. But this imposes several problems/restrictions at the code level. Such as the need to pass the context information deep within the stack wherever the heap may be necessary. One may also use TLS to avoid this, but this may cause some problems with re-entrance in my specific case.
This makes me wonder: Is there a known technique to optimize the heap for (but not limit to) single-threaded usage?
EDIT:
Special thanks to #Voo for suggesting checking out the google's tcmalloc.
It seems to work similar to what I did more-or-less (at least for small objects). But in addition they solve the exact issue I have, by maintaining per-thread caching.
I too thought in this direction, but I thought about maintaining per-thread heaps. Then freeing a memory block allocated from the heap belonging to another thread is somewhat tricky: one should insert it in a sort of a locked queue, and that other thread should be notified, and free the pending allocations asynchronously. Asynchronous deallocation may cause problems: if that thread is busy for some reason (for instance performs an aggressive calculations) - no memory deallocation actually occurs. Plus in multi-threaded scenario the cost of deallocation is significantly higher.
OTOH the idea with caching seems much simpler, and more efficient. I'll try to work it out.
Thanks a lot.
P.S.:
Indeed google's tcmalloc is great. I believe it's implemented pretty much similar to what I did (at least fixed-size part).
But, to be pedantic, there's one matter where my heap is superior. According to docs, tcmalloc imposes an overhead roughly 1% (asymptotically), whereas my overhead is 0.0061%. It's 4/64K to be exact.
:)
One thought is to maintain a memory allocator per-thread. Pre-assign fairly chunky blocks of memory to each allocator from a global memory pool. Design your algorithm to assign the chunky blocks from adjacent memory addresses (more on that later).
When the allocator for a given thread is low on memory, it requests more memory from the global memory pool. This operation requires a lock, but should occur far less frequently than in your current case. When the allocator for a given thread frees it's last byte, return all memory for that allocator to the global memory pool (assume thread is terminated).
This approach will tend to exhaust memory earlier than your current approach (memory can be reserved for one thread that never needs it). The extent to which that is an issue depends on the thread creation / lifetime / destruction profile of your app(s). You can mitigate that at the expense of additional complexity, e.g. by introducing a signal that a memory allocator for given thread is out of memory, and the global pool is exhaused, that other memory allocators can respond to by freeing some memory.
An advantage of this scheme is that it will tend to eliminate false sharing, as memory for a given thread will tend to be allocated in contiguous address spaces.
On a side note, if you have not already read it, I suggest IBM's Inside Memory Management article for anyone implementing their own memory management.
UPDATE
If the goal is to have very fast memory allocation optimized for a multi-threaded environment (as opposed to learning how to do it yourself), have a look at alternate memory allocators. If the goal is learning, perhaps check out their source code.
Hoarde
tcmalloc (thanks Voo)
It might be a good idea to read Jeff Bonwicks classic papers on the slab allocator and vmem. The original slab allocator sounds somewhat what you're doing. Although not very multithread friendly it might give you some ideas.
The Slab Allocator: An Object-Caching Kernel Memory Allocator
Then he extended the concept with VMEM, which will definitely give you some ideas since it had very nice behavior in a multi cpu environment.
Magazines and Vmem: Extending the Slab Allocator to Many CPUs and Arbitrary Resources
Related
My program is a daemon and runs for a long time. Only some of the time it will needs a lot of memory resource.
I want to increase my program performance by increasing memory locality.
And PMR seems like a good tool for this purpose.
However, it seems that the memory resources provided by the standard does not return the memory to upstream when there are lots of memory currently not used.
(i.e. synchronized_pool_resource, unsynchronized_pool_resource, monotonic_buffer_resource)
I want that my program can use less memory when the load is not high. (kinds of like calling malloc_trim when needed)
Is there a memory resource that will only cache small amount of currenly un-used memory, and return the rest to upstream.
A memory resource can be written to do whatever you want. However, since what you've described (returning memory that is unused) is what the default allocator does (and is one of the main reasons to use it), there wasn't much point in adding more standard library memory resources that do this.
Most of the defined memory resources are all about not returning unused memory, because returning and reallocating memory is expensive. They provide different strategies for keeping that memory accessible, so that later allocation calls are as fast as possible. That is, their whole point is to avoid the cost of allocating memory from the heap.
So you'll have to write a resource with the functionality you're looking for.
I'm implementing the FineList and the LazyList classes in C++. Both the above concurrent linked lists have been implemented in Java in the book "The Art of Multiprocessor Programming". I'd like to measure the amount of memory consumed by each algorithm during it's entire course of execution. I'm not sure how do I do that. I could just keep track of the number of times "new" is called by each thread, in a counter, in both algorithms and use that as a measuring criteria. Similarly, whenever a thread calls "delete", I'd decrement the counter. Is that a fair criteria to measure memory consumption? The problem is FineList algorithm allows me to immediately "delete" a node once it's removed from the linked-list, due to it's lock-based nature. But that's not the case in LazyList algorithm, since it has lock-free methods. Is there any other way to measure memory consumption or is the above method fair to both algorithms?
If you keep an ordered log of the new and delete calls (including the size requested) and you are sure the code you are interested in only uses new and delete and not other allocation routines, you can determine more or less the theoretical memory consumption at any point in time by keeping a running tally of the excess of memory allocated over memory freed. You could perhaps generate such a log by overloading global operator new(size_t) as well as delete.
The number will be only theoretical due to several factors:
Allocators add a certain amount of overhead, so the actual memory allocated will generally be larger than the sum of the sizes of the objects allocated. This overhead includes fragmentation, since some unallocated memory could be technically-free-but-not-really-available.
Allocators may not return any memory to the OS, or they may return it, but in an unpredictable way. So if you are measuring "memory allocated" as far as the OS is concerned (versus as far as the runtime allocator is concerned), you have a harder problem and this is highly allocator dependent.
Especially in a multi-threaded scenario, not all freed memory is necessarily used by future allocations. A specific case of this for thread-aware allocators is the use of thread-local allocation buffers: memory freed on one thread might not be immediately usable on another thread, until some threshold is reached whereby allocations can move across threads. This might be relevant for your scenario if there is a disparity of nodes allocated and freed among threads.
There is a whole layer of complexity when it comes to determining how much memory has been allocated - even to define that term. For example, for a large allocation, the OS may return you a chunk of memory which doesn't actually exist in RAM, and only page it is lazily as you access it. So if you don't access everything you allocate, the numbers reported by the allocator could actually be an over-estimate.
That's just scratching the surface.
C++ allows you to provide your own operator new and matching operator delete. This is useful for such measurement tasks. You can even use this to figure out the memory used as a function of the allocation strategy, e.g. see how much more memory your algorithm needs when rounding allocation up to a multiple of 16 bytes.
The large benefit of this approach is that you don't need to touch the code of your algorithm itself. You can use the bookkeeping of your own allocator.
Mind you, the idea of lock-free programming can be overly optimistic if you actually need new. There's no guarantee in C++ whatsoever that new is lock-free. And since C++ allows you to new memory in one thread and delete it in another, some cross-thread synchronization is needed.
To my understanding, a memory pool is a block, or multiple blocks of memory allocate on the stack before runtime.
By contrast, to my understanding, dynamic memory is requested from the operating system and then allocated on the heap during run time.
// EDIT //
Memory pools are evidently not necessarily allocated on the stack ie. a memory pool can be used with dynamic memory.
Non dynamic memory is evidently also not necessarily allocated on the stack, as per the answer to this question.
The topics of 'dynamic vs. static memory' and 'memory pools' are thus not really related although the answer is still relevant.
From what I can tell, the purpose of a memory pool is to provide manual management of RAM, where the memory must be tracked and reused by the programmer.
This is theoretically advantageous for performance for a number of reasons:
Dynamic memory becomes fragmented over time
The CPU can parse static blocks of memory faster than dynamic blocks
When the programmer has control over memory, they can choose to free and rebuild data when it is best to do so, according the the specific program.
4. When multithreading, separate pools allow separate threads to operate independently without waiting for the shared heap (Davislor)
Is my understanding of memory pools correct? If so, why does it seem like memory pools are not used very often?
It seems this question is thwart with XY problem and premature optimisation.
You should focus on writing legible code, then using a profiler to perform optimisations if necessary.
Is my understanding of memory pools correct?
Not quite.
... on the stack ...
... on the heap ...
Storage duration is orthogonal to the concept of pools; pools can be allocated to have any of the four storage durations (they are: static, thread, automatic and dynamic storage duration).
The C++ standard doesn't require that any of these go into a stack or a heap; it might be useful to think of all of them as though they go into the same place... after all, they all (commonly) go onto silicon chips!
... allocate ... before runtime ...
What matters is that the allocation of multiple objects occurs before (or at least less often than) those objects are first used; this saves having to allocate each object separately. I assume this is what you meant by "before runtime". When choosing the size of the allocation, the closer you get to the total number of objects required at any given time the less waste from excessive allocation and the less waste from excessive resizing.
If your OS isn't prehistoric, however, the advantages of pools will quickly diminish. You'd probably see this if you used a profiler before and after conducting your optimisation!
Dynamic memory becomes fragmented over time
This may be true for a naive operating system such as Windows 1.0. However, in this day and age objects with allocated storage duration are commonly stored in virtual memory, which periodically gets written to, and read back from disk (this is called paging). As a consequence, fragmented memory can be defragmented and objects, functions and methods that are more commonly used might even end up being united into common pages.
That is, paging forms an implicit pool (and cache prediction) for you!
The CPU can parse static blocks of memory faster than dynamic blocks
While objects allocated with static storage duration commonly are located on the stack, that's not mandated by the C++ standard. It's entirely possible that a C++ implementation may exist where-by static blocks of memory are allocated on the heap, instead.
A cache hit on a dynamic object will be just as fast as a cache hit on a static object. It just so happens that the stack is commonly kept in cache; you should try programming without the stack some time, and you might find that the cache has more room for the heap!
BEFORE you optimise you should ALWAYS use a profiler to measure the most significant bottleneck! Then you should perform the optimisation, and then run the profiler again to make sure the optimisation was a success!
This is not a machine-independent process! You need to optimise per-implementation! An optimisation for one implementation is likely a pessimisation for another.
If so, why does it seem like memory pools are not used very often?
The virtual memory abstraction described above, in conjunction with eliminating guess-work using cache profilers virtually eliminates the usefulness of pools in all but the least-informed (i.e. use a profiler) scenarios.
A customized allocator can help performance since the default allocator is optimized for a specific use case, which is infrequently allocating large chunks of memory.
But let's say for example in a simulator or game, you may have a lot of stuff happening in one frame, allocating and freeing memory very frequently. In this case the default allocator is not as good.
A simple solution can be allocating a block of memory for all the throwaway stuff happening during a frame. This block of memory can be overwritten over and over again, and the deletion can be deferred to a later time. e.g: end of a game level or whatever.
Memory pools are used to implement custom allocators.
One commonly used is a linear allocator. It only keeps a pointer seperating allocated/free memory. Allocating with it is just a matter of incrementing the pointer by the N bytes requested, and returning it's previous value. And deallocation is done by resetting the pointer to the start of the pool.
I have seen that lot of code bases specially server codes have basic (sometimes advanced) memory managers. Is the real purpose of memory manager is to reduce number of malloc calls or mainly for the purpose of memory analysis, corruption check or may be other application centric purposes.
Is the argument of saving malloc calls reasonable enough as malloc in itself is a memory manager. The only performance gain I can reason is when we know that system always ask for same size memory.
Or the reason for having memory manager is that free does not return memory to OS but saves in the list. So over the lifetime of the process, the heap usage of the process may increase if we keep on doing malloc/free because of fragmentation.
mallocis a general purpose allocator - "not slow" is more important than "always fast".
Consider a feature that would be a 10% improvement in many common cases, but might cause significant performance degradation in a few rare cases. An application specific allocator can avoid the rare case and reap the benefits. A general purpose allocator should not.
Besides number of calls to malloc, there are other relevant attributes:
locality of allocations
On current hardware, this easily the most important factor for performance. An application has more knowledge of the access patterns and can optimize the allocations accordingly.
multithreading
A general purpose allocator must allow calls to malloc and free from different threads. This usually requires a lock or similar concurrency handling. If the heap is very busy, this leads to massive contention.
An application that knows that some high-frequency alloc/frees come only from one thread can use its own thread-specific heap, which not only avoids contention for these allocations, but also increases their locality and takes load off the default allocator.
fragmentation
This is still a problem for long running applications on systems with limited physical memory or address space. Fragmentation may require more and more memory or address space from the OS, even without the actual working set increasing. This is a significant problem for applications that need to run uninterrupted.
Last time I looked deeper into allocators (which is probably half a decade past), the consensus was that naive attempts to reduce fragmentation often conflict with the never slow rule.
Again, an application that knows (some of its) allocation patterns can take a lot of load from the default allocator. One very common use case is building a syntax tree or something similar: there are gazillions of small allocations which are never freed individually, only as a whole. Such a pattern can be served efficiently with a very trivial allocator.
resilence and diagnostics
Last not least the diagnostic and self-protection capabilities of the default allocator may not be sufficient for many applications.
Why do we have custom memory managers rather than the built-in ones?
Number one reason is probably that the codebase was originaly written 20-30years ago when the provided one wasn't any good and nobody dares change it.
But otherwise, as you say because the application needs to manage fragmentation, grab memory at startup to ensure that memory will always be available, for security or a bunch of other reasons - most of which could be acheived by correct use of the built-in manager.
C and C++ are designed to be stripped down. They don't do much that is not explicitly asked for, so when a program asks for memory, it gets the minimum possible effort required to deliver that memory.
In other words, if you don't need it, you don't pay for it.
If finer-grained control of the memory is required, that's the domain of the programmer. If the programmer wishes to trade bare metal speed for a system that will provide higher performance on the target hardware in conjunction with the program's often unique goals, better debugging support, or simply likes the look and feel and warm fuzzies that come from using a manager, that is up to them. The programmer either writes something smarter or finds a third party library to do what they want.
You briefly touched on a lot of the different reasons why you would use a memory manager in your question.
Is the real purpose of a memory manager to reduce the number of malloc calls or mainly for the purpose of memory analysis, corruption check or other application centric purposes?
This is the big question. A memory manager in any application can be generic (like malloc) or it can be more specific. The more specialized the memory manager becomes it is likely to be more efficient at the specific task it is supposed to accomplish.
Take this overly-simplified example:
#define MAX_OBJECTS 1000
Foo globalObjects[MAX_OBJECTS];
int main(int argc, char ** argv)
{
void * mallocObjects[MAX_OBJECTS] = {0};
void * customObjects[MAX_OBJECTS] = {0};
for(int i = 0; i < 1000; ++i)
{
mallocObjects[i] = malloc(sizeof(Foo));
customObjects[i] = &globalObjects[i];
}
}
In the above I am pretending that this global object list is our "custom memory allocator." This is just to simplify what I am explaining.
When you allocate with malloc there is no guarantee it is right next to the previous allocation. Malloc is a general purpose allocator and does a good job at that but doesn't necessarily make the most efficient choice for every application.
With a custom allocator you might be able to up front allocate room for 1000 custom objects and since they are a fixed size return the exact amount of memory you need to prevent fragmentation and to efficiently allocate that block.
There is also the difference between memory abstraction and custom memory allocators. STL allocators are arguably an abstraction model and not a custom memory allocator.
Take a look at this link for some more information on custom allocators and why they are useful: gamedev.net link
There are many reasons why we would want to do this and it really depends on the application itself. In fact all the reasons you mentioned are valid.
I once built a very simple memory manager that kept track of shared_ptr allocations in order for me to see what was not being released properly on application end.
I would say stick to your runtime unless you need something that it does not provide.
Memory managers are used basically to manage efficiently your memory reservation. Normally processes have access to a limited amount of memory (4GB in 32bits systems), from this you have to subtract the virtual memory space reserved for the kernel (1GB or 2GB depending on your OS configuration). Thus, virtually the process has access let's say to 3GB of memory that will be used to hold all of its segments (code, data, bss, heap and stack).
Memory managers (malloc for example) try to fulfill the different memory reservation requests issued by the process by requesting new memory pages to the OS (using sbrk or mmap system calls). Every time this happens it implies an extra cost on the program execution since the OS has to look for a suitable memory page to be assigned to the process (Physical memory is limited and all the running processes want to use it), update the process tables (TMP, etc). These operations are time consuming and hit the process execution and performance. Thus, the memory manager normally try to request the needed pages to fulfill the process reservations cleverly. For example it could ask for some more pages to avoid calling more mmap calls in the near future. Additionally, it tries to deal with issues like fragmentation, memory alignment, etc. This basically unloads the process from this responsibility, otherwise everybody writing some program that needs dynamic memory allocation has to perform this manually!
Actually, there are cases where one could be interested in doing the memory management manually. This is the case for embedded or high availability systems which have to run for 24/365. In these cases even if the memory fragmentation is low it could become a problem after very long period of running (1 year for example). So, one of the solutions that are used in this case is to use a memory pool to allocate before hand the memory for the application objects. After-wards each time you need memory for some object you just use the already reserved memory.
For server based or any application that needs to run for long periods of time or indefinitely, the main issue is paged memory fragmentation. After a long series of mallocs / new and free / delete, paged memory can end up with gaps in the pages that waste space and could eventually run out of virtual address space. Microsoft deals with this with it's .NET framework, by occasionally pausing a process to repack paged memory for a process.
To avoid slowdown during repacking of memory in a process, a server type application can use multiple processes for the application, so that during repacking of one process, the other process(es) take more of the load.
Suppose I have a memory pool object with a constructor that takes a pointer to a large chunk of memory ptr and size N. If I do many random allocations and deallocations of various sizes I can get the memory in such a state that I cannot allocate an M byte object contiguously in memory even though there may be a lot free! At the same time, I can't compact the memory because that would cause a dangling pointer on the consumers. How does one resolve fragmentation in this case?
I wanted to add my 2 cents only because no one else pointed out that from your description it sounds like you are implementing a standard heap allocator (i.e what all of us already use every time when we call malloc() or operator new).
A heap is exactly such an object, that goes to virtual memory manager and asks for large chunk of memory (what you call "a pool"). Then it has all kinds of different algorithms for dealing with most efficient way of allocating various size chunks and freeing them. Furthermore, many people have modified and optimized these algorithms over the years. For long time Windows came with an option called low-fragmentation heap (LFH) which you used to have to enable manually. Starting with Vista LFH is used for all heaps by default.
Heaps are not perfect and they can definitely bog down performance when not used properly. Since OS vendors can't possibly anticipate every scenario in which you will use a heap, their heap managers have to be optimized for the "average" use. But if you have a requirement which is similar to the requirements for a regular heap (i.e. many objects, different size....) you should consider just using a heap and not reinventing it because chances are your implementation will be inferior to what OS already provides for you.
With memory allocation, the only time you can gain performance by not simply using the heap is by giving up some other aspect (allocation overhead, allocation lifetime....) which is not important to your specific application.
For example, in our application we had a requirement for many allocations of less than 1KB but these allocations were used only for very short periods of time (milliseconds). To optimize the app, I used Boost Pool library but extended it so that my "allocator" actually contained a collection of boost pool objects, each responsible for allocating one specific size from 16 bytes up to 1024 (in steps of 4). This provided almost free (O(1) complexity) allocation/free of these objects but the catch is that a) memory usage is always large and never goes down even if we don't have a single object allocated, b) Boost Pool never frees the memory it uses (at least in the mode we are using it in) so we only use this for objects which don't stick around very long.
So which aspect(s) of normal memory allocation are you willing to give up in your app?
Depending on the system there are a couple of ways to do it.
Try to avoid fragmentation in the first place, if you allocate blocks in powers of 2 you have less a chance of causing this kind of fragmentation. There are a couple of other ways around it but if you ever reach this state then you just OOM at that point because there are no delicate ways of handling it other than killing the process that asked for memory, blocking until you can allocate memory, or returning NULL as your allocation area.
Another way is to pass pointers to pointers of your data(ex: int **). Then you can rearrange memory beneath the program (thread safe I hope) and compact the allocations so that you can allocate new blocks and still keep the data from old blocks (once the system gets to this state though that becomes a heavy overhead but should seldom be done).
There are also ways of "binning" memory so that you have contiguous pages for instance dedicate 1 page only to allocations of 512 and less, another for 1024 and less, etc... This makes it easier to make decisions about which bin to use and in the worst case you split from the next highest bin or merge from a lower bin which reduces the chance of fragmenting across multiple pages.
Implementing object pools for the objects that you frequently allocate will drive fragmentation down considerably without the need to change your memory allocator.
It would be helpful to know more exactly what you are actually trying to do, because there are many ways to deal with this.
But, the first question is: is this actually happening, or is it a theoretical concern?
One thing to keep in mind is you normally have a lot more virtual memory address space available than physical memory, so even when physical memory is fragmented, there is still plenty of contiguous virtual memory. (Of course, the physical memory is discontiguous underneath but your code doesn't see that.)
I think there is sometimes unwarranted fear of memory fragmentation, and as a result people write a custom memory allocator (or worse, they concoct a scheme with handles and moveable memory and compaction). I think these are rarely needed in practice, and it can sometimes improve performance to throw this out and go back to using malloc.
write the pool to operate as a list of allocations, you can then extended and destroyed as needed. this can reduce fragmentation.
and/or implement allocation transfer (or move) support so you can compact active allocations. the object/holder may need to assist you, since the pool may not necessarily know how to transfer types itself. if the pool is used with a collection type, then it is far easier to accomplish compacting/transfers.