@
Berkeley Open Infrastructure
for Network Computing
|
|
|
Berkeley Open Infrastructure
for Network Computing
|
| Home | Help | Status | Forums | Glossary | Account
|
| log in |
Message boards : Wish list : AMD64
| Author | Message |
|---|---|
|
So far, two projects have added support for AMD64: SETI sends the x86 application and SIMAP, a native x86-64 application. Would you consider supporting windows_amd64 and x86_64-pc-linux-gnu/x86_64-unknown-linux-gnu similarly? | |
ID: 3580 | Rating: -9.9920072216264E-15 | rate:
 /   Reply
Quote
| |
 WormholioCaptain  Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
|
Augustine wrote: So far, two projects have added support for AMD64... Would you consider supporting windows_amd64 and x86_64-pc-linux-gnu/x86_64-unknown-linux-gnu similarly? This sounds like something interesting to investigate, but I do not have a 64bit machine, and I do not know if my compliers can cross-compile for one. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
| ID: 3582 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Wormholio wrote: This sounds like something interesting to investigate, but I do not have a 64bit machine, and I do not know if my compliers can cross-compile for one. You might try http://devcenter.amd.com/ to gain access to AMD64 systems so that you can port Pirates and Einstein applications to AMD64. HTH ____________  | |
| ID: 3583 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Wormholio wrote: This sounds like something interesting to investigate, but I do not have a 64bit machine, and I do not know if my compliers can cross-compile for one. BTW, just let me know if I can be of any help. ;-) ____________ | |
| ID: 3591 | Rating: -1 | rate:
/ Reply
Quote
| |
|
Augustine wrote: So far, two projects have added support for AMD64: SETI sends the x86 application and SIMAP, a native x86-64 application. Would you consider supporting windows_amd64 and x86_64-pc-linux-gnu/x86_64-unknown-linux-gnu similarly? HashClash now supports AMD64 with a 32-bit Linux application and Chess960 with a 64-bit Linux application... ____________ | |
| ID: 3934 | Rating: -0.99999999999999 | rate:
/ Reply
Quote
| |
|
Augustine wrote:
HashClash now supports AMD64 on Windows with a 32-bit application... ____________ | |
| ID: 3967 | Rating: -0.99999999999999 | rate:
/ Reply
Quote
| |
|
FWIW, running two instances of the client, one the 32-bit client, the other, the 64-bit client, on the same 4-core system, but limiting each client to 2 cores, I can compare the relative performance of 32-bit and 64-bit SIMAP's HMMER: the 64-bit version is about 7% faster. | |
| ID: 3974 | Rating: -0.99999999999999 | rate:
/ Reply
Quote
| |
So far, two projects have added support for AMD64: SETI sends the x86 application and SIMAP, a native x86-64 application. Would you consider supporting windows_amd64 and x86_64-pc-linux-gnu/x86_64-unknown-linux-gnu similarly? Guess what? Leiden Classical now supports AMD64 on Linux with a 32-bit application. ;-) ____________ | |
| ID: 4013 | Rating: -1 | rate:
/ Reply
Quote
| |
|
Wormholio wrote:
One can build GCC with the --enable-targets=all (default for x86-64) configuration option and then GCC can compile for either x86 or x86-64 through the -m32 or -m64 compilation options. WRT to Windows, just download the Platform SDK from Microsoft. It contains the AMD64 command-line compiler. Actually, it's a cross-compiler, because it itself is a 32-bit application, so one can do all the building on a 32-bit system. For details, see http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/30887.pdf. If I can be of any help testing the resulting AMD64 applications, just let me know. HTH ____________ | |
| ID: 4074 | Rating: -1 | rate:
/ Reply
Quote
| |
|
Wormholio Captain Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
|
Augustine wrote: WRT to Windows, just download the Platform SDK from Microsoft. It contains the AMD64 command-line compiler. Actually, it's a cross-compiler, because it itself is a 32-bit application, so one can do all the building on a 32-bit system. I have been meaning to update my SDK anyway, so I will see about this when I do, or when I set up another build machine. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
| ID: 4085 | Rating: 0 | rate:
/ Reply
Quote
| |
So far, two projects have added support for AMD64: SETI sends the x86 application and SIMAP, a native x86-64 application. Would you consider supporting windows_amd64 and x86_64-pc-linux-gnu/x86_64-unknown-linux-gnu similarly? Now Malaria, Docking and RieselSieve support Linux AMD64 clients by sending the 32-bit application and ABC ß, a 64-bit application. ____________ | |
| ID: 4906 | Rating: -2 | rate:
/ Reply
Quote
| |
|
Wormholio Captain Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
|
What difference, if any, would there be in running the 32 bit client and 32 bit applications, or running a 64 bit client (someone has to build this) and then running 32 bit applications with it? | |
| ID: 4907 | Rating: 0 | rate:
/ Reply
Quote
| |
What difference, if any, would there be in running the 32 bit client and 32 bit applications, or running a 64 bit client (someone has to build this) and then running 32 bit applications with it? I am not sure I understood your question, but the idea is to run 64-bit applications. Projects such as SIMAP, Chess960 and ABC reported significant performance improvements by doing so. Fine. However, not all projects are willing to port their applications to 64 bits for different reasons. Then, in order that those clients running 64 bits can contribute to other projects too, sending the 32-bit application would allow a project to benefit from their volunteered computing power. I did propose here that a x86-64 client could optionally fall back to requesting the x86 application if a project didn't support it, but it fell on deaf ears. ____________ | |
| ID: 4909 | Rating: -1 | rate:
/ Reply
Quote
| |
So far, two projects have added support for AMD64: SETI sends the x86 application and SIMAP, a native x86-64 application. Would you consider supporting windows_amd64 and x86_64-pc-linux-gnu/x86_64-unknown-linux-gnu similarly? RieselSieve supports AMD64 Windows clients by sending the 32-bit application too. ____________ | |
| ID: 4910 | Rating: -3 | rate:
/ Reply
Quote
| |
|
I'm happy to see that several projects have started this week to support AMD64. Here's an updated list:
| |
| ID: 4911 | Rating: -1 | rate:
/ Reply
Quote
| |
|
Nicolas Volunteer tester  Send message Joined: 6 Apr 05
Argentina
The Final Front Ear Credit: 147.9 RAC: 0.00
Joined: Apr 6, 2005
Verified: NEVER | |
|
Captain, could you PLEASE put the TRIM_THIS_quote feature back? | |
| ID: 4912 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Wormholio Captain Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
Captain, could you PLEASE put the TRIM_THIS_quote feature back? It's on my list. So are other things, but I want that and the other features back too. .-) ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
| ID: 4913 | Rating: 0 | rate:
/ Reply
Quote
| |
|
FYI, the new x86-64 Linux client, version 5.8.11, can be found at boinc_5.8.11_x86_64-pc-linux-gnu.gz. Again, the x64 Windows client, version 5.4.11, by Crunch3r, at boinc_5.4.11_windows_amd64.zip. | |
| ID: 4922 | Rating: -1 | rate:
/ Reply
Quote
| |
|
Wormholio Captain Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
FYI, the new x86-64 Linux client, version 5.8.11, can be found at boinc_5.8.11_x86_64-pc-linux-gnu.gz. Again, the x64 Windows client, version 5.4.11, by Crunch3r, at boinc_5.4.11_windows_amd64.zip. How do we know that we can trust these? How do we know that the benchmarks are correct, or that they claim the "right" amount of credit? What is the "right" amount of credit for 64bit apps? I'm just a bit cautious. After all, you have been known to consort with pirates. .-) ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
| ID: 4924 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Ageless Chief Petty Officer Volunteer tester  Send message Joined: 20 Jul 04
Netherlands
Machinae Supremacy Credit: 1,295.9 RAC: 0.00
Joined: Jul 20, 2004
Verified: Jul 9, 2011 Dubloons: 3 Pieces of Eight: 7 Punishment: Cat o' Nine Tails | |
|
I am not sure he understood your question. ;-) | |
| ID: 4925 | Rating: 0 | rate:
/ Reply
Quote
| |
How do we know that we can trust these? How do we know that the benchmarks are correct, or that they claim the "right" amount of credit? What is the "right" amount of credit for 64bit apps? How to assign credits is an issue that goes beyond the client, as many projects go through hoops on how to define them fairly. On trust, the only way I can think of is through peer review: compile the client yourself with GCC 4.1.1 using the same options I did, "-O3 -ftree-vectorize", and compare the results yourself. Then again, how is anyone supposed to trust your peer review, Capt'n? .-D ____________ | |
| ID: 4927 | Rating: 0 | rate:
/ Reply
Quote
| |
FYI, the new x86-64 Linux client, version 5.8.11, can be found at boinc_5.8.11_x86_64-pc-linux-gnu.gz. Again, the x64 Windows client, version 5.4.11, by Crunch3r, at boinc_5.4.11_windows_amd64.zip. As WCG uses HTTPS in the communications with the client, a file with public encryption keys is needed. Download this new x86-64 Linux client, still version 5.8.11, boinc_5.8.11_x86_64-pc-linux-gnu.tgz and copy all the files in the tar-ball to your system's working BOINC directory. For more information, see BoincStats Forum. HTH ____________ | |
| ID: 4931 | Rating: -1 | rate:
/ Reply
Quote
| |
|
Hello, BOINC World!
| |
| ID: 4932 | Rating: -1 | rate:
/ Reply
Quote
| |
|
Wormholio Captain Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
How do we know that we can trust these? ... One test we can try is to run some workunits with the 32 bit version sent to the 64 bit clients and see if the credit claims are reasonable. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
| ID: 4933 | Rating: 0 | rate:
/ Reply
Quote
| |
One test we can try is to run some workunits with the 32 bit version sent to the 64 bit clients and see if the credit claims are reasonable. So, after 366 WUs, how's it looking like? Aye? ____________ | |
| ID: 4934 | Rating: -1 | rate:
/ Reply
Quote
| |
|
Wormholio Captain Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
So, after 366 WUs, how's it looking like? Aye? I have some concerns. At first glance it appears that workunits run on the 64 bit client are systematically claiming more credit for the same amount of work. But this first test may have some problems. While it's the same application code, and the 32 bit version to boot, the two I've used for comparison where not built with the same BOINC libraries. And the workunits are very short. If there is some extra startup time for the 64 bit machine to run a 32 bit application that could make the workunit seem to take longer. So a real test would be to use an app version which has been compiled exactly the same for both platforms, and to use longer workunits to make a better comparison. We'll see if we cannot work that into the schedule. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
| ID: 4944 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Correction: the x86-64 Linux client, version 5.8.11, can be downloaded from boinc_5.8.11_x86_64-pc-linux-gnu.tgz (make sure to copy both files to the BOINC working directory). The new x64 Windows client, version 5.8.11, by Crunch3r, can be found at boinc_5.8.11_windows_amd64.zip.
| |
| ID: 4967 | Rating: -0.99999999999999 | rate:
/ Reply
Quote
| |
|
An updated x86-64 Linux client, version 5.8.15, can be downloaded from boinc_5.8.15_x86_64-pc-linux-gnu.tgz (make sure to read the file "README.x86_64-pc-linux-gnu" in it). | |
| ID: 5012 | Rating: -2 | rate:
/ Reply
Quote
| |
|
Here's the new recommended version for the x86-64 Linux client:
| |
| ID: 5291 | Rating: -2 | rate:
/ Reply
Quote
| |
|
Here's a development version of the x86-64 Linux client:
| |
| ID: 6416 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Even though there's an official AMD64 client for Linux, it refers to too many dynamic libraries and requires a fairly recent Linux setup to run on.
| |
| ID: 6900 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Wormholio Captain Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
|
Augustine wrote: Even though there's an official AMD64 client for Linux, it refers to too many dynamic libraries and requires a fairly recent Linux setup to run on. What I would be more interested in are instructions on how to make such a "portable" Linux client. I've tried a few tricks, but not all of it has worked, and I've yet to find clear instructions on how to do this. Not just the steps, mind you, but an explanation of the steps. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
| ID: 6902 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Wormholio wrote: What I would be more interested in are instructions on how to make such a "portable" Linux client. I've tried a few tricks, but not all of it has worked, and I've yet to find clear instructions on how to do this. Not just the steps, mind you, but an explanation of the steps. Here's a variation of what I posted elsewhere:
| |
| ID: 6903 | Rating: 0 | rate:
/ Reply
Quote
| |
|
Wormholio Captain Send message Joined: 6 Jun 04
United States
Away Credit: 4,009.8 RAC: 0.00
Joined: Jun 6, 2004
Verified: Mar 13, 2008 Dubloons: 3 Pieces of Eight: 10 Punishment: Aztec curse | |
|
Augustine wrote:
Thank you. I get the idea. Actually, I have a 64bit machine on the way, so at some point I should just be able to build on that.
Arrrg! and this is the bit I object to. I know Einstein@Home built this way, and using an old version of Linux seems to the the recommended method, but I really find this distasteful. It should be possible to build backward-compatible apps on a newer release of Linux. You may have to get an older version of GLIBC to link against, but still. Having to build on an old machine is in my mind about as elegant as having to reboot to fix a small problem. You may have to do it from time to time to get work done, but it's not very elegant.
This part also worries me. Why is CUPS a dependency? It's not listed in the software prerequisites, so maybe it needs to be listed too? Ideally, you would not need CUPS to build the client software, since CUPS is a printer _server_ system. Nevermind my ranting, thank you very much for the info. ____________ -- Eric Myers "Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats | |
| ID: 6908 | Rating: 0 | rate:
/ Reply
Quote
| |
I tried badly to go around this, but even library versioning wouldn't work. And building an older version of GLIBC would not work for dynamic linking, unless it's shipped with the client or linked statically, when the whole thing gets enormous.
Beats me too, but I suspect that it's an indirect dependency. HTH ____________ | |
| ID: 6909 | Rating: 0 | rate:
/ Reply
Quote
| |
Message boards : Wish list : AMD64
| Home | Help | Status | Forums | Glossary | Account
|