Smaug: a new chess engine based on glaurung

[bob]

The Intel C++ compiler is available for a limited time free trial if you give them some marketing data,

Yes, but there are some very severe restrictions even during the limited time free trial: You are not allowed to distribute the binaries you compile, and you are not even allowed to use them on your own computer after the trial time runs out.

Otherwise it's US$599 plus US$240/year for support.

Is it really that expensive? I don't remember how much i payed a couple of years back, but it was nowhere near that much. Intel must have increased their prices a lot since then.

I tried the free version some time ago. I don't recall any earthshaking improvements over g++, but I do remember that the Intel compiler did supply a few more useful diagnostics about questionable coding.

I think the improvements depend a lot on the program. The Intel compiler (if you use the -ipo switch) is able to inline function calls across file boundaries, which GCC currently cannot (at least not version 4.2, I'm not sure about 4.3 and 4.4). I've also found that the Intel compiler is much faster at doing 64-bit arithmetics on 32-bit CPUs. For a bitboard program like Glaurung, this means that the Intel compiler gives a huge performance improvement on x86-32 CPUs, but a smaller improvement on x86-64.

Another advantage of the Intel compiler is that profiler guided optimization actually works. With GCC, I usually get executables that crash immediately at startup when I try to use PGO.

Tord

I don't see these "severe restrictions" nor any mention of a trial period for the latest download I have which is version 11.0 of the compiler. I downloaded it as I always do, my old license had expired (this is the license that lets me download stuff from intel, not the compiler license file itself) and I applied for a new one which was granted instantly. I asked for a non-commerical license and nothing I have received from them mentions any 30 day trial. I did see a "download 30-day trial versoin" but that is a different animal from what I downloaded, which was the free-for-linux version. I have had 11.0 since about Jan 20, and it has not expired, which it would had I been given a 30 day key to use...

In short, I don't see any difference between now and the last time I downloaded something which was late in the 10.x series...

[bob]

The Intel C++ compiler is available for a limited time free trial if you give them some marketing data, Otherwise it's US$599 plus US$240/year for support.

I tried the free version some time ago. I don't recall any earthshaking improvements over g++, but I do remember that the Intel compiler did supply a few more useful diagnostics about questionable coding.

I just looked at version 11.0 which I have had since the middle of January. Nothing I have mentions any sort of 30-day trial period. They have always had that as an option, but they have also always had a "free download for non-commercial linux users" that you have to hunt around for, but it is there. I have tried some commercial versions from Intel on our cluster, and there a trial version will automatically stop working at the end of the 30 day trial based on the key you are given to activate the thing. But I'm not using that on my laptop, although we do have a commercial license on our clusters.

[Tord Romstad]

I don't see these "severe restrictions" nor any mention of a trial period for the latest download I have which is version 11.0 of the compiler. I downloaded it as I always do, my old license had expired (this is the license that lets me download stuff from intel, not the compiler license file itself) and I applied for a new one which was granted instantly. I asked for a non-commerical license and nothing I have received from them mentions any 30 day trial. I did see a "download 30-day trial versoin" but that is a different animal from what I downloaded, which was the free-for-linux version. I have had 11.0 since about Jan 20, and it has not expired, which it would had I been given a 30 day key to use...

In short, I don't see any difference between now and the last time I downloaded something which was late in the 10.x series...

The difference is that you are using the Intel compiler under Linux, while Steven and I are using it under Mac OS X. Under Linux, there is a complete free version of the Intel compiler without any restrictions whatsoever, except that you can't make commercial programs with it. Under Mac OS X, unfortunately, there is only the 30-day trial version with the restrictions I described.

Tord

[Sven]

The Intel C++ compiler is available for a limited time free trial if you give them some marketing data, Otherwise it's US$599 plus US$240/year for support.

I tried the free version some time ago. I don't recall any earthshaking improvements over g++, but I do remember that the Intel compiler did supply a few more useful diagnostics about questionable coding.

I just looked at version 11.0 which I have had since the middle of January. Nothing I have mentions any sort of 30-day trial period. They have always had that as an option, but they have also always had a "free download for non-commercial linux users" that you have to hunt around for, but it is there. I have tried some commercial versions from Intel on our cluster, and there a trial version will automatically stop working at the end of the 30 day trial based on the key you are given to activate the thing. But I'm not using that on my laptop, although we do have a commercial license on our clusters.

While you are referring to Linux and to the non-commercial version of Intel C++ that is available for Linux only, Steven might have referred to either MacOS or Windows. There you have that 30-day trial version that "ceases to function" after 30 days.
Sven
[Edit: Tord was a few seconds faster with the same answer ]

[sje]

The difference is that you are using the Intel compiler under Linux, while Steven and I are using it under Mac OS X. Under Linux, there is a complete free version of the Intel compiler without any restrictions whatsoever, except that you can't make commercial programs with it. Under Mac OS X, unfortunately, there is only the 30-day trial version with the restrictions I described.

True, except I'm also testing under Linux as well; that's how I found the optimization bug in g++-v4.3 in the Ubuntu distribution for 32 bit Intel hosts (a Pentium4 in my case). I'm also testing using a rather geriatric version of Red Hat on an old dual Pentium3 rackmount.

I rather doubt that the Red Hat machine could host the Intel compiler, but I might give it a shot on the Ubuntu box.

I don't have any AMD processors, but still I wonder if the Intel compiler works as well for them as it does for the Genuine Intel CPUs.

----

Oh, and g++-v4.3 is also emitting spurious warnings about 32 bit constant expression evaluation overflow. The same code compiles without complaint elsewhere.

The Cygwin version of g++ works fine although there are slight differences in the contents of the standard C include files.

I've tried using the Windows Express C++ IDE, but it gives a bogus diagnostic about a preprocessor conditional expression involving logical negation of a boolean literal.

[mhull]

I don't have any AMD processors, but still I wonder if the Intel compiler works as well for them as it does for the Genuine Intel CPUs.

It did for me 6-7 years ago on my AMD Duron 900MHz machine (crafty builds). Beat gcc by 10-15 percent. But I haven't messed with it since then.

[bob]

The Intel C++ compiler is available for a limited time free trial if you give them some marketing data, Otherwise it's US$599 plus US$240/year for support.

I tried the free version some time ago. I don't recall any earthshaking improvements over g++, but I do remember that the Intel compiler did supply a few more useful diagnostics about questionable coding.

I just looked at version 11.0 which I have had since the middle of January. Nothing I have mentions any sort of 30-day trial period. They have always had that as an option, but they have also always had a "free download for non-commercial linux users" that you have to hunt around for, but it is there. I have tried some commercial versions from Intel on our cluster, and there a trial version will automatically stop working at the end of the 30 day trial based on the key you are given to activate the thing. But I'm not using that on my laptop, although we do have a commercial license on our clusters.

While you are referring to Linux and to the non-commercial version of Intel C++ that is available for Linux only, Steven might have referred to either MacOS or Windows. There you have that 30-day trial version that "ceases to function" after 30 days.
Sven
[Edit: Tord was a few seconds faster with the same answer ]

I have absolutely no sympathy for non-Linux users.

You get what you pay for...

[bob]

I don't see these "severe restrictions" nor any mention of a trial period for the latest download I have which is version 11.0 of the compiler. I downloaded it as I always do, my old license had expired (this is the license that lets me download stuff from intel, not the compiler license file itself) and I applied for a new one which was granted instantly. I asked for a non-commerical license and nothing I have received from them mentions any 30 day trial. I did see a "download 30-day trial versoin" but that is a different animal from what I downloaded, which was the free-for-linux version. I have had 11.0 since about Jan 20, and it has not expired, which it would had I been given a 30 day key to use...

In short, I don't see any difference between now and the last time I downloaded something which was late in the 10.x series...

The difference is that you are using the Intel compiler under Linux, while Steven and I are using it under Mac OS X. Under Linux, there is a complete free version of the Intel compiler without any restrictions whatsoever, except that you can't make commercial programs with it. Under Mac OS X, unfortunately, there is only the 30-day trial version with the restrictions I described.

Tord

What better reason to install Linux?

[bob]

The difference is that you are using the Intel compiler under Linux, while Steven and I are using it under Mac OS X. Under Linux, there is a complete free version of the Intel compiler without any restrictions whatsoever, except that you can't make commercial programs with it. Under Mac OS X, unfortunately, there is only the 30-day trial version with the restrictions I described.

True, except I'm also testing under Linux as well; that's how I found the optimization bug in g++-v4.3 in the Ubuntu distribution for 32 bit Intel hosts (a Pentium4 in my case). I'm also testing using a rather geriatric version of Red Hat on an old dual Pentium3 rackmount.

I rather doubt that the Red Hat machine could host the Intel compiler, but I might give it a shot on the Ubuntu box.

Red Hat is what I run (Fedora, actually). I have it running on a PIV and other boxes, so you should have no problems I can think of). I have it running on a 700 mhz Sony Vaio laptop as well.

I don't have any AMD processors, but still I wonder if the Intel compiler works as well for them as it does for the Genuine Intel CPUs.

----

Oh, and g++-v4.3 is also emitting spurious warnings about 32 bit constant expression evaluation overflow. The same code compiles without complaint elsewhere.

The Cygwin version of g++ works fine although there are slight differences in the contents of the standard C include files.

I've tried using the Windows Express C++ IDE, but it gives a bogus diagnostic about a preprocessor conditional expression involving logical negation of a boolean literal.

[bob]

I don't have any AMD processors, but still I wonder if the Intel compiler works as well for them as it does for the Genuine Intel CPUs.

It did for me 6-7 years ago on my AMD Duron 900MHz machine (crafty builds). Beat gcc by 10-15 percent. But I haven't messed with it since then.

For the record, I have had a few issues on Opterons, but with significantly older versions of Intel's icc. Last time I used a quad dual-core from AMD in a CCT, I installed the latest icc and the code was a little slower than that from gcc. Although the amd guys work on gcc and other things themselves, which might have explained it. I've not tried this with version 10/11...