Re: using DCompute
On Friday, 28 July 2017 at 00:23:35 UTC, Nicholas Wilson wrote: On Thursday, 27 July 2017 at 21:33:29 UTC, James Dean wrote: I'm interested in trying it out, says it's just for ldc. Can we simply compile it using ldc then import it and use dmd, ldc, or gdc afterwards? The ability to write kernels is limited to LDC, though there is no practical reason that, once compiled, you couldn't use resulting generated files with GDC or DMD (as long as the mangling matches, which it should). This is not a priority to get working, since the assumption is if you're trying to use the GPU to boost your computing power, then you like care enough to use LDC, as opposed to DMD (GDC is still a bit behind DMD so I don't consider it) to get good optimisations in the first place. Yes, but dmd is still good for development since LDC sometimes has problems. Can we compile kernels in LDC and import them in to a D project seamlessly? Basically keep an LDC project that deals with the kernels while using dmd for the bulk of the program. I mean, is it a simple import/export type of issue?
using DCompute
I'm interested in trying it out, says it's just for ldc. Can we simply compile it using ldc then import it and use dmd, ldc, or gdc afterwards? --- a SPIRV capable LLVM (available here to build ldc to to support SPIRV (required for OpenCL)). or LDC built with any LLVM 3.9.1 or greater that has the NVPTX backend enabled, to support CUDA. --- Is the LDC from the download pages have these enabled? Also, can DCompute or any GPU stuff efficiently render stuff because it is already on the GPU or does one sort of have to jump through hoops to, say, render a buffer? e.g., suppose I want to compute a 3D mathematical function and visualize it's volume. Do I go in to the GPU, do the compute, back out to cpu, then to the graphics system(opengl/directX) or can I just essentially do it all from the gpu?
Randomed/encoded code
I would like to encode code in such a way that each compilation produces "random" code as compared to what the previous compilation produced, but ultimately the same code is ran each time(same effect). Basically we can code a function that does a job X in many different ways. Each way looks different in binary but does the same job(Same effect). I'd like a way to sort of randomly sample/generate the different functions that do the same job. The easiest way to wrap your head around this is to realize that certain instructions and groups of instructions can be rearranged, producing a binary that is different but the effect is the same. Probably, ultimately, that is all that can be done(certain other tricks could possibly be added to increase the sampling coverage such as nop like instructions, dummy instructions, etc). The main issue is how to take an actual D function and transform it in to a new D function, which, when ran, ultimately does the same thing as the original but is not the same "binary". Encrypting is a subset of this problem as we can take any string and use it to encode the code then decrypt it. And this may be usable, but then the encryption and decryption instructions must somehow be randomizable, else we are back at square one. It might be easier though, to use the encryption method to randomize the original function since the encryption routine is known while the original function is not(as it could be any function). I'm not looking for a mathematical solution, just something that works well in practice. i.e., The most skilled human reading the disassembly would find it very difficult to interpret what is going on. He might be able to figure out one encryption routine, say, but when he sees the "same code"(same effect) he will have to start from scratch to understand it because its been "randomized". The best way I can see how to do this is to have a list of well known encoding routines that take an arbitrary function, encrypt it. Each routine can be "randomized" by using various techniques to disguise it such as those mentioned earlier. This expands the list of functions tremendously. If there are N functions and M different ways to alter each of those functions then there are N*M total functions that we can use to encrypt the original function. If we further allow function composition of these functions, then we can get several orders of magnitude of complexity with such as a few N. The goal though, is to do this efficiently and effectively in a way that can be amended. It will be useful in copy protection and used with other techniques to make it much more effective. Ultimately the weak point with the encryption techniques is decryption the functions but by composing encryption routines makes it stronger. Any ideas how to achieve this in D nicely?
Code Construction
Heres a module I just started working on, completely incompletely, but demonstrates an ideas that might be very useful in D: Code Construction. The idea is very simple: We have code strings like "class %%name%% { }" and %%name%% is replaced with the name of a type T. The idea is that we can map a type to a code string and have the type "fill in the blanks" rather than having to do it all in D, which is far more verbose and terse, we can do it using a different method. Eventually the idea is that we can take any type T, map it in to a code string, then map that string to a new type that relates to T. What I use this for is to simplify generating related members. enum eState { Ready, Starting, Running, Pausing, Paused, Unpausing, Stopping, Stopped, Finished, } Can be used to generate new stuff: mixin(sCodeConstruction.MemberMap!("\t\tsMultiCallback!callbackSignature Callback_%%name%%;\n", eState)); Generates the following code: sMultiCallback!callbackSignature Callback_Ready; sMultiCallback!callbackSignature Callback_Starting; sMultiCallback!callbackSignature Callback_Running; sMultiCallback!callbackSignature Callback_Pausing; sMultiCallback!callbackSignature Callback_Paused; sMultiCallback!callbackSignature Callback_Unpausing; sMultiCallback!callbackSignature Callback_Stopping; sMultiCallback!callbackSignature Callback_Stopped; sMultiCallback!callbackSignature Callback_Finished; and if the enum changes, one does not have to regenerate the callbacks. (Obviously this is a simple case and D can handle this reasonably easily, but the more complex cases are far more verbose than what can be done by substitutions) Using a substitution grammar is a lot easier IMO and D needs something like. Unfortunately, I won't be finishing it anytime soon so I thought I'd mention the idea and maybe someone else could tackle it. module mCodeConstruction; import std.meta, std.traits, std.string, std.algorithm, std.array, std.conv; /* Code construction that simplifies common tasks. */ struct sCodeConstruction { //[ For examples, let X in myModule.myClass.X represents the member `int X = 4;` and Y in myModule.myClass.Y be `bool Y(double q)`] public static: enum Tokens : string { Name= "%%name%%", // The member name (X => "X", Y => "Y") FullName = "%%fullName%%",// The full member name (X => "myClass.X", Y => "myClass.Y") CompleteName = "%%completeName%%", // The complete member name (X => "myModule.myClass.X", Y => "myModule.myClass.Y") Value= "%%value%%", // The default value of the member (X => 4, Y => "") Type= "%%type%%", // The type of the member (X => int, Y => bool delegate(double)) Module= "%%module%%", // The module the member is in (X => "myModule", Y => "myModule") ParentType = "%%parentType%%",// The type of the parent (X => "class", Y => "class") MethodReturn = "%%method/return%%", // The return type of a member if it has one (X => "", Y => "bool") MethodParamNName = "%%methodParam/N/Name%%", // The Nth parameter's name (Y => "q") MethodParamNType = "%%methodParam/N/type%%", // The Nth parameter's type (Y => "double") DoublePercent = "%%%", // the %% symbol } /* A Resolver is a function takes a grammar symbol and returns the corresponding element for it from T. The following list of resolvers are provided for common use: MemberResolver: Resolves member information: MemeberResolver("%%name%%", myClass.X) = "X". */ auto MemberResolver(string S, alias T)() { mixin("import "~moduleName!(T)~";"); mixin("alias val = " ~ fullyQualifiedName!(T) ~ ";"); switch(S) { case Tokens.Name : return to!string( __traits(identifier, T)); } //pragma(msg, name); static if (is(typeof(T) == function) || is(typeof(*T) == function)) { // Get the parameters in a (w)string array. enum p = split(Parameters!(member).stringof[1..$-1], ",").map!(n => strip(n)); Tuple!(S,S)[] params; foreach(a; aliasSeqOf!(p)) { } } return ""; } auto StandardResolver(string S, alias
teething troubles
Hi all, I need a little helping hand with dmd on a 32 bit Debian box. I installed dmd from http://d-apt.sourceforge.net/ i) First trial: $cat test.d import std.stdio; void main() { writeln(hello); } $ time dmd test.d real0m2.355s user0m1.652s sys 0m0.364s $./test hello $ dmd -v test.d | wc -l 84 Seems to be working. My only concern is whether 2.35s for compiling such a trivial file is normal. ii) 2nd trial == I installed gdc. Now I get $ time gdc test.d real0m6.286s user0m3.856s sys 0m0.884s == Given the dmd and gdc timings, it seems I am doing something wrong. iii) 3rd trial I installed tango from http://d-apt.sourceforge.net/ I know this can ruffle feathers. Please assume good faith. I am just trying to learn from the tango book. === $ls /usr/include/dmd/tango core io math net stdc sys text time util $cat test.d import tango.io.Stdout; void main() { Stdout (hello).newline; } $dmd -I/usr/include/dmd/tango -v test.d binarydmd version v2.065 config/etc/dmd.conf parse test importall test importobject (/usr/include/dmd/druntime/import/object.di) importtango.io.Stdout (tango/io/Stdout.d) test.d(1): Error: module Stdout is in file 'tango/io/Stdout.d' which cannot be read import path[0] = /usr/include/dmd/tango import path[1] = /usr/include/dmd/phobos import path[2] = /usr/include/dmd/druntime/import It seems inspite of specifying usr/include/dmd/tango it cannot import tango.io.Stdout. Here is my /etc/dmd.conf [Environment32] DFLAGS=-I/usr/include/dmd/phobos -I/usr/include/dmd/druntime/import -L-L/usr/lib/i386-linux-gnu -L--export-dynamics What am I doing wrong. Thanks for the help. -- Dean
Re: teething troubles
On a Windows 32 bit that little program compiles in 0.74 seconds (warmed up time) using and old CPU. Modern CPUs should take about 0.5 seconds. Thanks for checking the timings wait I am not alone in using a 32 bit box ! Mine is an Athlon 1045.456 MHz. The minimum of 3 consecutive compilation runs that I get is 2.3 seconds. Keep in mind that writeln and std.stdio are lot of stuff. Just to be clear I am not complaining dmd is slow. I am just concerned that I am doing something wrong. I have also tried this C++ version: #include iostream int main() { std::cout hello std::endl; return 0; } With gcc 4.8.0 it takes me 0.48 seconds to compile (warmed up time). On my box this compiles in 1.2 seconds. So it seems somewhat consistent (as in 3 times slower for both). I got worried because I expected dmd to compile hello world substantially faster than g++. I have heard that dmd is instantaneous. I am still at a loss about tango for D2 problem. Shouldnt providing the -I option with the path to tango work. Does any other magic need to happen. I dont know the internal mechanics of importing modules.
Re: teething troubles
On Thursday, 17 July 2014 at 09:32:24 UTC, bearophile wrote: Dean: Mine is an Athlon 1045.456 MHz. Didn't notice that before hitting send. compilation runs that I get is 2.3 seconds. Then I think your timings could be OK, I am using an old 2.3 GHz CPU. Glad to know that I am not doing something stupid, yet. dmd compiles very quickly, but to compile writeln D has to digest a good amount of Phobos code. Are the reasons for this similar to why C++ STL is not an object code library ? I am still at a loss about tango for D2 problem. I think I've never used Tango with dmd. Not even sure if its a tango problem. Dmd doesn't seem to be picking up the path that I specify with -I. Perhaps the mechanics of module loading is not as simple as I imagine. I initially thought its a permission problem, but that is not the case.
Re: teething troubles
On Thursday, 17 July 2014 at 10:13:46 UTC, bearophile wrote: Dean: dmd compiles very quickly, but to compile writeln D has to digest a good amount of Phobos code. Are the reasons for this similar to why C++ STL is not an object code library ? The reasons for the large amount of code compiled for a writeln are that: writeln is more powerful, Phobos modules import each other a lot. And several parts of Phobos are not compiled because there are templates everywhere. Take a look at Phobos sources and you will see. Bye, bearophile Apologies, I wasnt clear. I was talking about the reason behind compiling the code from source as opposed to linking precompiled objects. I was speculating wether the reasons are similar to that of STL, i.e. specializing to the type as late as possible.
Re: teething troubles
On Thursday, 17 July 2014 at 11:08:16 UTC, Mike Parker wrote: On 7/17/2014 7:01 PM, Dean wrote: Not even sure if its a tango problem. Dmd doesn't seem to be picking up the path that I specify with -I. Perhaps the mechanics of module loading is not as simple as I imagine. I initially thought its a permission problem, but that is not the case. What does your tango source tree look like? Is it a) or b)? a) /usr/include/dmd/tango/io/Stdout.d b) /usr/include/dmd/tango/tango/io/Stdout.d When you pass -I/usr/include/dmd/tango, then it needs to look like b). If it's a), then you should pass -I/usr/include/dmd. The reason is that 'tango' is the top-level package directory. Its *parent* directory needs to be on the import path. Hi Mike, that was it. Thanks a lot.