Re: What Delphi users really want (other than bugfixes)

From: Eric Grange (egrangeNO_at_SPAMglscene.org)
Date: 08/31/04 

Date: Tue, 31 Aug 2004 18:26:14 +0200

> Maybe you can't directly access SIMD from .NET code but what do you make
> of this:

Still waiting for any of that "JIT-only optimization" to happen
in the real world and have it beat an AOT compiler... not that
this wouldn't be theoretically impossible... just quite anti-JIT
philosophy, given that modern C/C++ compilers f.i. can do
whole-application optimizations.

> * Processor-Specific Optimizations— At run time, the JIT knows
> whether or not it can make use of SSE or 3DNow instructions.

Yes, it knows it, but just doesn't (can't?) come up with anything
useful out of it (or maybe in hyper-specific cases noone has ever
seen?).
Reason: base data types do not map to SSE or 3DNow! types.
As long as you don't have support for 2-4 floats vector types
(and matrix), right there in the language alongside integer
or classic float type, compiler can't use SIMD efficiently,
because code won't be layed out in a way that would make it
efficient.

> * Your executable will be compiled specially for P4, Athlon or
> any future processor families. You deploy once, and the same
> code will improve along with the JIT and the user's machine.

I would be more impressed if the JIT was able to overtake ye olde
Delphi compiler codegen ;)
We're speaking of 2004 technology failing to edge what is essentially
a 386-era compiler with no instruction scheduling, no support for
Pentium instructions, local optimizations only, no inlining support,
archaic FPU codegen, etc.
Somehow, I can't reconcile the inferior performance of the JIT codegens
given their superior optimization support list...

So, I don't have much faith in promises of "superior optimization"
when it comes from the same people that made a JIT codegen that
can generate such an hideous ASM as the .Net JIT does (just look
at calls and how they are implemented)... seems
to be quite a different team than the one behind VC++ compiler f.i.,
and the words they use don't carry the same weight/meaning.

Maybe it's a case of over-optimizing the transmission and chassis
but forgetting about tyres and engine.

> I compared Delphi binaries to C# .NET binaries and the .NET code
> ran twice as fast.

Please post samples, should such a case exist, it could make
it a world's first! ;)

> I could come to was that .NET was doing something like SIMD to accomplish
> the same thing. I could be wrong [..]

I'm pretty sure it is not using SIMD, at best it will use scalar SSE2
on P4 and A64, but that's about it: ie. use SSE2 to manipulate one
value at a time, but unable to use it to add two values at a time.

Eric

Relevant Pages

  • Re: What Delphi users really want (other than bugfixes)
    ... Maybe you can't directly access SIMD from .NET code but what do you make ... Since the JIT is activated at run time, there is a lot of information ... around that a compiler isn't aware of. ... looking at the assembly language Delphi 7 produced the only conclusion I ...
    (borland.public.delphi.non-technical)
  • Re: Can you write code directly in CIL ???
    ... >> and here is what the JIT compiler actually generated from this (!! ... > performance between C++ and C# for something as simple as nested loops. ... > optimization disabled can be at least an order of magnitude. ...
    (microsoft.public.dotnet.languages.csharp)
  • Re: C++ /clr vs. unmanaged C++
    ... The JIT can do more cross-module optimization than the native C++ compiler, but the C++ compiler considers a lot more optimizations, only some of which are used when generating IL. ... Also, when compiling /clr without using .NET features, you don't take advantage of the .NET memory management scheme which usually compensates for the reduced optimization. ...
    (microsoft.public.dotnet.framework.interop)
  • Re: Java performance, CRTJVAPGM vs JIT
    ... JIT compiler compiles each Java method every time it sees it until the ... Sun's HotSpot JIT, but the points it makes are valid for various other ... performance improvement for a small investment in optimization time. ...
    (comp.sys.ibm.as400.misc)
  • Re: Redundant IL generated by the compiler
    ... The compiler is doing what you told it to do. ... So if you are that concerned with doing it right first time and not wasting ... to optimization; however, I'm a firm believer of doing things right the ... unrolling) because the JIT would do a better job of it seeing that it's ...
    (microsoft.public.dotnet.framework.clr)
Loading