The Wiert Corner – irregular stream of stuff

Jeroen W. Pluimers on .NET, C#, Delphi, databases, and personal interests

  • My badges

  • Twitter Updates

  • My Flickr Stream

  • Pages

  • All categories

  • Enter your email address to subscribe to this blog and receive notifications of new posts by email.

    Join 2,115 other followers

Archive for the ‘FastMM’ Category

Rewritten version (free for non-commercial; small price for commercial use) GitHub – pleriche/FastMM5: FastMM is a fast replacement memory manager for Embarcadero Delphi applications that scales well across multiple threads and CPU cores, is not prone to memory fragmentation, and supports shared memory without the use of external .DLL files.

Posted by jpluimers on 2020/05/05

It has been mentioned a few times already, but for my link archive: [WayBack] GitHub – pleriche/FastMM5: FastMM is a fast replacement memory manager for Embarcadero Delphi applications that scales well across multiple threads and CPU cores, is not prone to memory fragmentation, and supports shared memory without the use of external .DLL files.

From the [WayBack] README.md:

Version 5 is a complete rewrite of FastMM. It is designed from the ground up to simultaneously keep the strengths and address the shortcomings of version 4.992:

  • Multithreaded scaling across multiple CPU cores is massively improved, without memory usage blowout. It can be configured to scale close to linearly for any number of CPU cores.
  • In the Fastcode memory manager benchmark tool FastMM 5 scores 15% higher than FastMM 4.992 on the single threaded benchmarks, and 30% higher on the multithreaded benchmarks. (I7-8700K CPU, EnableMMX and AssumeMultithreaded options enabled.)
  • It is fully configurable runtime. There is no need to change conditional defines and recompile to change options. (It is however backward compatible with many of the version 4 conditional defines.)
  • Debug mode uses the same debug support library as version 4 (FastMM_FullDebugMode.dll) by default, but custom stack trace routines are also supported. Call FastMM_EnterDebugMode to switch to debug mode (“FullDebugMode”) and call FastMM_ExitDebugMode to return to performance mode. Calls may be nested, in which case debug mode will be exited after the last FastMM_ExitDebugMode call.
  • Supports 8, 16, 32 or 64 byte alignment of all blocks. Call FastMM_EnterMinimumAddressAlignment to request a minimum block alignment, and FastMM_ExitMinimumAddressAlignment to rescind a prior request. Calls may be nested, in which case the coarsest alignment request will be in effect.
  • All event notifications (errors, memory leak messages, etc.) may be routed to the debugger (via OutputDebugString), a log file, the screen or any combination of the three. Messages are built using templates containing mail-merge tokens. Templates may be changed runtime to facilitate different layouts and/or translation into any language. Templates fully support Unicode, and the log file may be configured to be written in UTF-8 or UTF-16 format, with or without a BOM.
  • It may be configured runtime to favour speed, memory usage efficiency or a blend of the two via the FastMM_SetOptimizationStrategy call.

Read the rest of this entry »

Posted in Delphi, Development, FastMM, Software Development | 6 Comments »

Delphi memory allocators and configuration notes in multi-threaded environments with high allocation/deallocation rates

Posted by jpluimers on 2020/02/20

From an interesting discussion at [Archive.is/WayBack] FaceBook: Delphi developer thread by Jarto Tarpio with some measurements by Jarto Tarpio and André Mussche.

  • Manipulation of strings and lists in Delphi have high memory allocation/deallocation rates, so HTTP related services with high call rates are affected more than regular services
  •  FastMM:
    • conditional defines that can help are NeverSleepOnThreadContention, UseSwitchToThread, and UseReleaseStack then measure.
    • has one huge advantage: It’s very, very good at keeping memory fragmentation at bay
    • default settings are for applications that use lots of CPU, but have no really high memory allocation/deallocation rates
    • has very good debugging facilities
    • Under FullDebugMode address space is never released back to the operating system so once the address space has been exhausted there is very little room to manoeuvre.
  • TCMalloc:
    • is very good at multi-threaded memory management with high allocation/deallocation rates
    • needs to be persuaded to releases memory to the OS:
      it only releases to the system under two occasions: Freeing another part of the memory, or asking it to release all parts marked as freed.

    • has no debugging facilities

The differences make it a challenge to integrate in your development and deployment process: because of the debugging facilities, you’d like FastMM in all your environments, but TCMalloc in multi-threaded environments with high allocation/deallocation rates.

One possibility is to have your CI environment deliver both in all stages, run all tests on both, then choose the final one depending on your run-time configuration.

That gives a burden on configuring your Continuous Integration, but the gain might outweigh this cost.

Relevant links from the Facebook thread:

–jeroen

Posted in Delphi, Development, FastMM, Software Development | 5 Comments »

Useful settings for FastMM4 to tracking bottlenecks in multi-threaded memory allocation

Posted by jpluimers on 2019/05/23

Interesting thread that reminded me of the relatively new FastMM feature to track down lock contention and cope with it using release stacks:

[WayBack] I try use FastMM4 to tracking bottleneck in allocating memory. I was inspired by this Primož movie: https://www.youtube.com/watch?v=p-5mJyXvmrc When I… – Jacek Laskowski – Google+

The changes by Primož have added these two new conditional defines to the FastMM4 codebase:

These options are mutually exclusive.

This was the original post it got introduced in [WayBack] The Delphi Geek: Finding Memory Allocation Bottlenecks with FastMM4 and merged into the main repository.

FastMM introduction: [WayBack] The New Memory Manager In BDS 2006 – by Pierre le Riche.

It also taught me about [WayBack] RAMDisk – Software that Accelerates, Protects, Optimizes – Server Memory Products & Services – Dataram.

Primož stores his DCU files there: much faster than SSD, and far less wear on your SSD; see https://youtu.be/p-5mJyXvmrc?t=2675

–jeroen

Read the rest of this entry »

Posted in Conference Topics, Conferences, Delphi, Development, Event, FastMM, Software Development | Leave a Comment »

Have NoMessageBoxes depend on a global boolean · Issue #58 · pleriche/FastMM4 · GitHub

Posted by jpluimers on 2018/08/16

Reminder to self: [WayBackHave NoMessageBoxes depend on a global boolean · Issue #58 · pleriche/FastMM4 · GitHub

–jeroen

Posted in Delphi, Development, FastMM, Software Development | Leave a Comment »

A success story – kudos to ScaleMM, +André Mussche… – Ondrej Kelle – Google+

Posted by jpluimers on 2018/08/15

Though no maintenance for over 2 years, it ScaleMM seems among the fastest of Delphi memory managers: [WayBack] GitHub – andremussche/scalemm: Fast scaling memory manager for Delphi

Full story via [WayBackA success story – kudos to ScaleMM, +André Mussche… – Ondrej Kelle – Google+ at [WayBack] just tried Scaled MM – cut the time for 100k down from 51 minutes to 40 minutes. That’s with 18 (9+9) cores allocated to the app and 2 to everything else. Wow, allocating 8 + 8 and the time drops even more… – Russell Weetch

Two lessons on multi-threading here:

  • use a memory manager that copes well with threads
  • do not allocate more busy threads than available (hyper-threaded) cores

Some history (as ScaleMM, TopMM and FastMM seem to be related):

  • A bit of history here: at the time we had the mm contest. Not many people had access to 8 cores+, we had…. Andre and Ivo were both colleagues of mine at that point, although working for different companies. We needed (stock trading, Ivo and me) something to scale and Ivo wrote topmm, which still performs better on multi-cores than fastmm and has less assembler code. Andre improved on Ivo’s concepts and yes, it really outperforms fastmm on multi-cores today.
    It also out-performs most C family provided mm’s. Note I was not massively involved, but both Ivo and Andre were. And both did a proper job. So KUDOS to them.
    Note these (fastmm and topmm) were written with multi-core in mind. The practical results at the time were often under-estimated, because few people had access to the real hardware. Most of us running two cores at most. Nice to see that the concept those two programmers pursued still pays dividend in 2018!
  • Read: ScaleMM and TopMM and FastMM… Keyboard left me..
    (I wrote COMMM as a joke..!).
  • [WayBack] TopMemory v3.55. High Performance. Fully Scalable. Free. Memory Manager. for. Delphi – PDF
  • [WayBack] FPC Anagrams

There are also some IntelTBB memory manager references at https://plus.google.com/+RussellWeetch/posts/W4EQgLme5ud [WayBack]

–jeroen

Posted in Delphi, Development, FastMM, Software Development | Leave a Comment »

 
%d bloggers like this: