Were they giving different times to the 0.23 clients?

---

Go away, I was asleep

One question is what are they designated as so I can set up an app_info file correctly.

I tried it as milkyway3, but that did not seem to work.

<app_info>
 <app>
  <name>milkyway3</name>
 </app>
 <file_info>
  <name>milkyway3_0.02_x86_64-apple-darwin</name>
  <executable/>
 </file_info>
 <app_version>
  <app_name>milkyway3</app_name>
  <version_num>02</version_num>
  <file_ref>
   <file_name>milkyway3_0.02_x86_64-apple-darwin</file_name>
   <main_program/>
  </file_ref>
 </app_version>
</app_info>

---

Were they giving different times to the 0.23 clients?

0.23 shouldn't be able to run those workunits.

---

Question, is V 3 the one where there is no output files that I heard you talk about? Or is that a feature still to come?

V3 is the one with no output files (and no input parameter file), yes.

Ok, thanks... way cool ... I have 6-7 run all waiting validation ... :)

Not sure if there is any more to get ... I guess I will just have to wait events ... :)

Thanks for the info... I am so excited to see v3 get going ... faster server! Yea! :)

Presently I have 7 WU's running with several hours on each with no progress bar activity showing. (non-CUDA apps)

Mike

On windows?

Yes, Windows 7

Mike

I have one WU trying to download itself:

de_new_test_1605_1271651869_search_parameters

which has been trying to download for the better part of a day. I don't pretend to know which WU's are of which family, but I can say that this one is not getting to me. My Messages tab tells me that there's nothing wrong with my end or my connections...

---

I have one WU trying to download itself:

de_new_test_1605_1271651869_search_parameters

which has been trying to download for the better part of a day. I don't pretend to know which WU's are of which family, but I can say that this one is not getting to me. My Messages tab tells me that there's nothing wrong with my end or my connections...

I'd just kill that workunit.

---

new version Milkyway 3 does not work on fermi-cards!
Correct?
When can i use my new cards for Milkyway again?

Ton (ftpd)

---

Ton (ftpd) Netherlands

I have just tried to reset the project without successfully pulling down the new V3 stock GPU client.

I will try a project detach and reattach from BAM

---

Go away, I was asleep

new version Milkyway 3 does not work on fermi-cards!
Correct?
When can i use my new cards for Milkyway again?

Ton (ftpd)

If it can't do double precision math you can't use it. Sorry, but we need more precision than single precision :(

---

I have just tried to reset the project without successfully pulling down the new V3 stock GPU client.

I will try a project detach and reattach from BAM

We still don't have all versions of the application up and going yet, so that might be part of your problem.

---

If it can't do double precision math you can't use it. Sorry, but we need more precision than single precision :(

The Fermi cards can all do double precision, even if Nvidia -have- artificially capped them. The problem, as I recall, is that the application checks for CUDA compute capability 1.3, and the Fermi cards have compute capability 2.0. So it should test for >= 1.3 instead.

If it can't do double precision math you can't use it. Sorry, but we need more precision than single precision :(

The Fermi cards can all do double precision, even if Nvidia -have- artificially capped them. The problem, as I recall, is that the application checks for CUDA compute capability 1.3, and the Fermi cards have compute capability 2.0. So it should test for >= 1.3 instead.

That might be the issue. I'll pass it on to Anthony.

---

There is a also a bug in the versions description code in some versions of BOINC. I do not recall which of the versions gets it right. You may have to up BOINC to v 6.10.45 or .46 to get that change.

I discovered that the if I download the file with my browser on the iMac, it does not have the execute bit set on it.

Once I detached and reattached it downloaded the new app and it did have the execute bit set.

---

Well, decided to allow another milkyway WU to be crunched through. This WU is massive at 115 estimated hours to crunch.

I'm currently running Mandriva 2010.1 beta so I tend to see a lot of patching and updates in the beta stage. Some patches require restarts, while others require only logouts/logins.

What I've noticed is that before I shut down this last time, I had 23hours done and 102hours to go. did a reboot due to the last patch, took a look and I see I'm back again at the initial value of 115 hours to go again (23 hours done).

My opinion is this 3.0 workunit is not desktop friendly because users are likely going to shut down their computers at night and start up the next day, which means that with checkpoints this far apart, I don't think the average desktop user is going to accomplish a workunit (sempron 2+GHz machine with only a basic built-in graphics card).

At 115hours to be done before Apr29th, this looks like something you would run on a server that's waiting for something to do 24x7, not a desktop that is only run sporadicly when needed.

Another thing to consider is if the average default boinc setup is set to switch every 1 hour between science projects, then we have a problem with this app if it is going to reset back if the save points are too far apart.

(right now this computer is set to change projects after 4000+ minutes due to the out of ordinary save points for cosmos, so I don't know if that is affecting milkyway's savepoints).

---

http://www.joescat.com/boinc/

I am running test 3 for two ours but it shows no progress so what can i do to fix it?

I have been running de_mw3_s11_test_4532_1271886235_0 using milkyway3 version 1 for 10:13:40 with 0.000% Progress and "---" To Completion ... My machine usually takes about 10 hours for Milkyway WU. Based on the other Posts, I will let it keep running (at least for today).

---

The version of the CUDA milkyway3 app my client received doesn't run on Centos 5.1 (the regular 0.24 application does). It seems to have been linked with a very recent libstdc++:

[david@n0008 milkyway.cs.rpi.edu_milkyway]$ ldd ./milkyway3_0.02_x86_64-pc-linux-gnu__cuda23
./milkyway3_0.02_x86_64-pc-linux-gnu__cuda23: /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.9' not found (required by ./milkyway3_0.02_x86_64-pc-linux-gnu__cuda23)
linux-vdso.so.1 => (0x00007fff4a1fe000)
libcudart.so.2 => not found
libcuda.so.1 => /usr/lib64/libcuda.so.1 (0x00000000006d0000)
libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x0000000000b6c000)
libm.so.6 => /lib64/libm.so.6 (0x0000000000e6c000)
libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x0000000003069000)
libc.so.6 => /lib64/libc.so.6 (0x000000000525f000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x00000000010ef000)
libz.so.1 => /usr/lib64/libz.so.1 (0x0000000001309000)
libdl.so.2 => /lib64/libdl.so.2 (0x000000000151d000)
/lib64/ld-linux-x86-64.so.2 (0x0000000000110000)

That build has been running OK with Ubuntu 9.04 for a couple of days now.

I am running test 3 for two ours but it shows no progress so what can i do to fix it?

It might be a problem with how it was compiled. I'm assuming this is on windows?

---