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 2,949 other followers

Archive for the ‘x86’ Category

When floating point code suddenly becomes orders magnitudes slower (via C++ – Why does changing 0.1f to 0 slow down performance by 10x? – Stack Overflow)

Posted by jpluimers on 2022/01/26

When working with converging algorithms, sometimes floating code can become very slow. That is: orders of magnitude slower than you would expect.

A very interesting answer to [Wayback] c++ – Why does changing 0.1f to 0 slow down performance by 10x? – Stack Overflow.

I’ve only quoted a few bits, read the full question and answer for more background information.

Welcome to the world of denormalized floating-point! They can wreak havoc on performance!!!

Denormal (or subnormal) numbers are kind of a hack to get some extra values very close to zero out of the floating point representation. Operations on denormalized floating-point can be tens to hundreds of times slower than on normalized floating-point. This is because many processors can’t handle them directly and must trap and resolve them using microcode.

If you print out the numbers after 10,000 iterations, you will see that they have converged to different values depending on whether 0 or 0.1 is used.

Basically, the convergence uses some values closer to zero than a normal floating point representation dan store, so a trick is used called “denormal numbers or denormalized numbers (now often called subnormal numbers)” as described in Denormal number – Wikipedia:

In a normal floating-point value, there are no leading zeros in the significand; rather, leading zeros are removed by adjusting the exponent (for example, the number 0.0123 would be written as 1.23 × 10−2). Denormal numbers are numbers where this representation would result in an exponent that is below the smallest representable exponent (the exponent usually having a limited range). Such numbers are represented using leading zeros in the significand.

Since a denormal number is a boundary case, many processors do not optimise for this.


Posted in .NET, Algorithms, ARM, Assembly Language, C, C#, C++, Delphi, Development, Software Development, x64, x86 | Leave a Comment »

Some notes on loosing performance because of using AVX

Posted by jpluimers on 2019/03/20

It looks like AVX can be a curse most of the times. Below are some (many) links that lead me to this conclusion, based on a thread started by Kelly Sommers.

My conclusion

Running AVX instructions will affect the processor frequency, which means that non-AVX code will slow down, so you will only benefit when the gain of using AVX code outweighs the non-AVX loss on anything running on that processor in the same time frame.

In practice, this means you need to long term gain from AVX on many cores. If you don’t, then the performance penalty on all cores, including the initial AVX performance, will degrade, often a lot (dozens of %).

Tweets and pages linked by them

Kelly raised a bunch of interesting questions and remarks because of the above:

I collected the above links because of [WayBack] GitHub – maximmasiutin/FastMM4-AVX: FastMM4 fork with AVX support and multi-threaded enhancements (faster locking), where it is unclear which parts of the gains are because of AVX and which parts are because of other optimizations. It looks like that under heavy loads on data center like conditions, the total gain is about 30%. The loss for traditional processing there has not been measured, but from the above my estimate it is at least 20%.

Full tweets below.

Read the rest of this entry »

Posted in Assembly Language, Development, Software Development, x64, x86 | Leave a Comment »

performance – Why is this C++ code faster than my hand-written assembly for testing the Collatz conjecture? – Stack Overflow

Posted by jpluimers on 2019/02/28

Geek pr0n at [WayBackperformance – Why is this C++ code faster than my hand-written assembly for testing the Collatz conjecture? – Stack Overflow

Via: [WayBack] Very nice #Geekpr0n “Why is C++ faster than my hand-written assembly code?” The comments are of high quality i… – Jan Wildeboer – Google+


Posted in Assembly Language, C, C++, Development, Software Development, x64, x86 | Leave a Comment »

A refefernce to 6502 by “Remember that in a stack trace, the addresses are return addresses, not call addresses – The Old New Thing”

Posted by jpluimers on 2018/09/11

On x86/x64/ARM/…:

It’s where the function is going to return to, not where it came from.


Bonus chatter: This reminds me of a quirk of the 6502 processor: When it pushed the return address onto the stack, it actually pushed the return address minus one. This is an artifact of the way the 6502 is implemented, but it results in the nice feature that the stack trace gives you the line number of the call instruction.

Of course, this is all hypothetical, because 6502 debuggers didn’t have fancy features like stack traces or line numbers.

Source: [WayBackRemember that in a stack trace, the addresses are return addresses, not call addresses – The Old New Thing

Which resulted in these comments at [WayBack] CC +mos6502 – Jeroen Wiert Pluimers – Google+:

  • mos6502: And don’t forget the crucial difference in PC on 6502 between RTS and RTI!
  • Jeroen Wiert Pluimers: +mos6502 I totally forgot about that one. Thanks for reminding me
    <<Note that unlike RTS, the return address on the stack is the actual address rather than the address-1.>>


[ Tutorials and Aids – RTI

RTI retrieves the Processor Status Word (flags) and the Program Counter from the stack in that order (interrupts push the PC first and then the PSW).

Note that unlike RTS, the return address on the stack is the actual address rather than the address-1.

[ Tutorials and Aids – RTS

RTS pulls the top two bytes off the stack (low byte first) and transfers program control to that address+1. It is used, as expected, to exit a subroutine invoked via JSR which pushed the address-1.

RTS is frequently used to implement a jump table where addresses-1 are pushed onto the stack and accessed via RTS eg. to access the second of four routines.


Posted in 6502, 6502 Assembly, Assembly Language, Development, History, Software Development, The Old New Thing, Windows Development, x64, x86 | Leave a Comment »

How to check if a binary is 32 or 64 bit on Windows? – Super User

Posted by jpluimers on 2018/08/17

It seems there are a few, but only loading the binary is the sure method to know what the process will be using: [WayBackHow to check if a binary is 32 or 64 bit on Windows? – Super User and [WayBack] How do I determine if a .NET application is 32 or 64 bit? – Stack Overflow.

Details in the answers of these questions, here are a few highlights:

  • The first few characters in the binary header reveal what it was originally designed for.
  • A .NET executable might still have an x64 header for bootstrapping.
  • The Windows SDK has a tool dumpbin.exe with the /headers option.
  • You can use sigcheck.exe from SysInternals.
  • The file utility (e.g. from cygwin, which comes with msysgit) will distinguish between 32- and 64-bit executables.
  • Use the command line 7z.exe on the PE file (Exe or DLL) in question which gives you a CPU line.
  • Virustotal File detail is a way to find out if a binary is 32 bit or 64 bit.
  • Even an executable marked as 32-bit can run as 64-bit if, for example, it’s a .NET executable that can run as 32- or 64-bit. For more information see, which has an answer that says that the CORFLAGS utility can be used to determine how a .NET application will run.


Search terms: win64, win32, x64, x86_64, x86

Posted in Assembly Language, Development, Power User, Windows, x64, x86 | Leave a Comment »

%d bloggers like this: