Posts by valterc
log in
1) Message boards : Number crunching : Windows SSE 32-bit Computation errors (Message 1344)
Posted 8 days ago by Profile valterc
OK. It's difficult to figure out the problem, there are many XP computers that compute correctly (like hostid=29204). The Windows x32 SSE2 version of the application was actually compiled using one of the latest Cygwin (gcc MingW 6.4.0).
2) Message boards : Number crunching : Windows SSE 32-bit Computation errors (Message 1342)
Posted 8 days ago by Profile valterc
It seems that your OS/CPU doesn't fully support SSE2 (that's the meaning of 'illegal instruction'). If I remember correctly XP needs SP3 for this, but it seems that you already have it installed. Please check with some tools (like CPU-Z) if SSE2 is correctly enabled on your host.
3) Message boards : Number crunching : Validation Inconclusive (Message 1339)
Posted 3 Jul 2018 by Profile valterc
The computational speed is strange indeed. It could be because of many reasons, the exe or the data files, the input and the big one (in this case hgnc_data_mat.csv) that have been modified. As far as I know BOINC should prevent tampering with those files but I'm not sure about this (there are some strange config options like <dont_check_file_sizes> that I do not fully understand). Why someone would like to intentionally do this (just waste computational power for zero credits)?. The output file is a big document and there is no way to guess it without doing the computation.

If a user would like to find a way to speed up the calculation (like Daniel did in the past) he can just play with the source code and then use the anonymous platform, but, first, he have to be sure that the calculation is correct, otherwise there is a very very low probability of getting a match at the validation stage and get some credits.
4) Message boards : Number crunching : Validation Inconclusive (Message 1337)
Posted 2 Jul 2018 by Profile valterc
Both hosts belong to one of the gridcoin pools, so it's impossible for me to just warn the owner. The hosts seem obviously completely un-managed (no one cares about them). The uploaded output files are somewhat 'garbled' and clearly invalid, so rejected by the server. It's a waste of computational power but if the number of such hosts remains low this should not be a big issue.

There is a large number of hosts that download a lot of workunits and return them (correctly computed) after weeks (too late for validation). This is also a waste of power, probably a side effect of having such a large number of hosts that are not managed.

Anyway, thank you for the information, it's a 'behavior' that should be monitored.
5) Message boards : Science : New experiment: gene network expansion of human pathologically relevant genes (Message 1331)
Posted 7 Jun 2018 by Profile valterc
I just wanted to add some technicalities: The dataset we are using right now is very big, a floating point matrix 87554x1829, it's size on the hard disk is ~0.5Gb and contains, for many genes, different isoforms/transcripts. The computational time for an algorithm's iteration, because of its 'complexity', is longer than the previous experiments.

Any single gene expansion is packed up into 294 workunits. Any workunit you receive has a name like this: 142041_Hs_T155402-MAP1B_wu-(1 to 294), a counter, Hs, an internal T code, a mnemonic for the gene/isoform. The T code is just a shortcut for the gene coordinates, for example T155402 is chr5:71403265..71403276,+

At this moment we don't know how many experiments we will make with this dataset, that's why I didn't update the 'Science status' page with the usual counters.
6) Message boards : Number crunching : Validation Inconclusive (Message 1327)
Posted 6 Jun 2018 by Profile valterc
I've just checked that host and I agree that it's behavior is quite strange (a lot of invalids with very short computation time). However, I checked some of the valid returned results, they were returned very fast, nevertheless they seem 'good'. For any workunit, sent to at least two different users, we check the results, the two output files are compared byte by byte and declared valid if identical (this should definitely avoid any kind of cheating).

The BOINC server has also some kind of mechanism that should stop sending workunits to hosts that produce errors or invalid results. I will continue to monitor that host.

Be also aware that we have some kind of a bug in the application (and we weren't able to find it), in some cases when a workunit is started, stopped before the first checkpoint and re-started again the output file will become garbled and the final computation will be eventually declared invalid. However, this is a very rare event.
7) Message boards : Number crunching : sse2 vs avx (Message 1324)
Posted 31 May 2018 by Profile valterc
An issue on my end is a pretty high error rate on sse2 tasks with my ryzen pc. avx runs flawless now.

I noticed it, exceptions due to 'illegal instructions'... The strange thing (at least for me) is that I do not understand this behavior after such a long computational time (it would make much more sense if the exception were raised at the very beginning, after just some seconds)
8) Message boards : Number crunching : sse2 vs avx (Message 1322)
Posted 30 May 2018 by Profile valterc
But why would the sse2 version be faster than avx? Is this just a quirk of the particular application? Or is there a problem with the avx implementation on this particular cpu?


With this new app, avx seems to be faster than sse2...

The new Windows app is just a recompilation of the original one with gcc 6.4.0, the source code was not modified.
9) Message boards : News : New organism (Message 1318)
Posted 25 May 2018 by Profile valterc
i know this is low prio right now, but when do you update the science status page?

http://gene.disi.unitn.it/test/gene_science.php

Right now we are doing the experiments on selected genes (chosen by the biologists), varying the parameters and also trying to figure out which modifications (filtering) we need to do on the input dataset. This is a preliminary phase, when we'll switch to the extensive analysis of all the genes of the selected dataset (the OneGenE experiment) I'll update the science page.
10) Message boards : News : New organism (Message 1316)
Posted 24 May 2018 by Profile valterc
Ok, after a lot of problems we are almost ready to start again (even not at full speed, by now). The new workunits (starting with id 141380_) are 10% longer, with a much smaller output file.
11) Message boards : News : Server down this weekend (Message 1315)
Posted 22 May 2018 by Profile valterc
Some of the tasks I've had waiting to upload have done so. Other are still failing with the message

2018-05-22 02:38:33 | TN-Grid Platform | [error] Error reported by file upload server: can't open file


I am getting this too. Server site is really slow too.

I had to move the upload directory to a nfs mounted storage (which is slower than the previous one), also a lot of people is uploading their results, the server is overwhelmed by work...
[edit] but it is getting better...
12) Message boards : News : Server down this weekend (Message 1311)
Posted 22 May 2018 by Profile valterc
I just stopped the boinc server....

I moved the upload directory to another storage and restarted the system, we have to find a way to reduce the size of the output file.
You may see that a lot of workunits are erroring out very quick. The reason is a bad input file (zero bytes or similar) that was made while the disk was full. They should auto-abort after some tries.
13) Message boards : News : Server down this weekend (Message 1310)
Posted 20 May 2018 by Profile valterc
I just stopped the boinc server....
14) Message boards : News : Server down this weekend (Message 1305)
Posted 19 May 2018 by Profile valterc
I noticed the problem. Unfortunately I cannot do anything to try to fix it until Monday. For now I was able just to stop the work generator.
15) Message boards : News : New organism (Message 1299)
Posted 17 May 2018 by Profile valterc
I could not get one :-(

Be patient, at the moment we want to go forward slowly...
16) Message boards : News : New organism (Message 1297)
Posted 16 May 2018 by Profile valterc
From the checks I just made it seems that the new applications work correctly. There might be some issues while cross-validating against ARM applications, please tell me if you noticed this.
Also remember please to REMOVE the anonymous platform if you use it on Windows.
17) Message boards : News : New organism (Message 1296)
Posted 16 May 2018 by Profile valterc
It looks like Windows app versions 1.10 and 1.11 might be giving conflicting results on Hs tasks, similar to the Linux/Windows issue before

Well, it should, that's why I have "deprecated" all the 1.10 Windows apps, I don't know why they are still around....
18) Message boards : News : New organism (Message 1294)
Posted 15 May 2018 by Profile valterc
I just made available the new Windows x64 sse2 application (other Windows applications will follow if everything works well)
19) Message boards : Number crunching : Application List (Message 1293)
Posted 14 May 2018 by Profile valterc
It seems to me that the php script is not cached, so all the numbers are calculated when the script is loaded. The 'average computing' column should be calculated by the validator at the end of each scan.
20) Message boards : News : New organism (Message 1290)
Posted 11 May 2018 by Profile valterc
The experiments on Vitis vinifera are almost finished, thank you all. We will soon switch to a new organism. We made some tests some time ago but we found that the outputs of the Windows and Linux applications didn't match each other. The new forthcoming Windows applications should behave correctly.
The next Monday we will distribute them and (slowly) start creating new work.


Next 20

Main page · Your account · Message boards


Copyright © 2018 CNR-TN & UniTN