Author |
Message |
rroonnaalldd Volunteer developer Volunteer tester
 Send message
Joined: 3 Jul 09 Posts: 1213 ID: 42893 Credit: 34,634,263 RAC: 0
                 
|
I got Iain's boinc-sources for Genefer.
In the moment i test GeneferX64-boinc in workunit #247616643.
[add]
link to GeneferX64-boinc
____________
Best wishes. Knowledge is power. by jjwhalen
|
|
|
rroonnaalldd Volunteer developer Volunteer tester
 Send message
Joined: 3 Jul 09 Posts: 1213 ID: 42893 Credit: 34,634,263 RAC: 0
                 
|
Why we have another and new source code for 'Genefer'?
Iains aim was to generate one source code for mac, linux and (maybe) windows.
I have done a recompilation of his sources. All linux-apps are now static apps (except the Cuda-apps).
All executables were shrinked with UPX3.08 and packed in tar.bz2-packages for boinc & nonboinc...
genefer-apps_boinc.tar.bz2
genefer-apps_nonboinc.tar.bz2
PS: Iains linux-version of GeneferCuda causes ever a 100% cpu-load on one cpu-core. I hope Iain patches this with the next version.
____________
Best wishes. Knowledge is power. by jjwhalen
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14036 ID: 53948 Credit: 475,998,733 RAC: 222,579
                               
|
genefer-apps_boinc.tar.bz2
genefer-apps_nonboinc.tar.bz2
There should be no reason to distribute a non-boinc build. The boinc build works in both environments. The only reason a non-boinc version of the source exists is so that developers who don't have a boinc environment can build the code. But if you do have a boinc dev environment, (i.e., you can build a boinc version), there's no reason to build a non-boinc executable. The boinc build fulfills both roles.
We're trying to get the minimum different versions of the code and the builds as possible.
____________
My lucky number is 75898524288+1 |
|
|
rroonnaalldd Volunteer developer Volunteer tester
 Send message
Joined: 3 Jul 09 Posts: 1213 ID: 42893 Credit: 34,634,263 RAC: 0
                 
|
The workunit #247616643 was completed after ~44 ksec on a Core2-cpu and validated.
____________
Best wishes. Knowledge is power. by jjwhalen
|
|
|
|
Boinc is not sending cpu work (windows). Is says that there is no work available.
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14036 ID: 53948 Credit: 475,998,733 RAC: 222,579
                               
|
Boinc is not sending cpu work (windows). Is says that there is no work available.
In the unlikely event you're requesting work for Long (WR) workunits, that's correct. You shouldn't be sent WR tasks for CPUs, ever. If someone has a CPU that can complete a long WU within the 21 day deadline, send me a PM and tell me. Actually, tell everyone -- that would be quite an accomplishment!
For the Short WUs -- I thought the Windows CPU app was turned back on, but it looks like it's not. I'll look into it and see if we can't get it going again.
____________
My lucky number is 75898524288+1 |
|
|
|
Boinc is not sending cpu work (windows). Is says that there is no work available.
In the unlikely event you're requesting work for Long (WR) workunits, that's correct. You shouldn't be sent WR tasks for CPUs, ever. If someone has a CPU that can complete a long WU within the 21 day deadline, send me a PM and tell me. Actually, tell everyone -- that would be quite an accomplishment!
For the Short WUs -- I thought the Windows CPU app was turned back on, but it looks like it's not. I'll look into it and see if we can't get it going again.
I only selected short wu's. I got some (for cpu) a couple of days ago (no app_info).
Under windows, my cpu took 76000 seconds (roughly 21 hours) to crunch a short task (n=19, b= 175k). That's about 4 times the time my gtx 550ti takes to do the same tasks. I do not know how long a WR task would take on that card, but I suspect that current ranges would take more than 5 days. If the ratio gpu/cpu is the same, my cpu would miss the 21 days deadline with n=22 tasks, but not for much, if running 24/7.
I have a linux VM set up, that is getting short genefers for cpu, so the issue is only with windows. Running on a vm takes longer, though, so it would be nice to have windows version running again.
____________
676754^262144+1 is prime |
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14036 ID: 53948 Credit: 475,998,733 RAC: 222,579
                               
|
According to the apps page, the Windows CPU app is back. I can't test it myself because I'm not going to mess with my app_info with 1 hour to go on my Long WU. I've learned THAT lesson!
So if you want to try it now, give it a shot.
____________
My lucky number is 75898524288+1 |
|
|
|
According to the apps page, the Windows CPU app is back. I can't test it myself because I'm not going to mess with my app_info with 1 hour to go on my Long WU. I've learned THAT lesson!
So if you want to try it now, give it a shot.
The app is in the official download page.
But still no joy without app info (I mean, no work sent by server - "no tasks available for selected work", it said). linux version works fine, though.
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14036 ID: 53948 Credit: 475,998,733 RAC: 222,579
                               
|
According to the apps page, the Windows CPU app is back. I can't test it myself because I'm not going to mess with my app_info with 1 hour to go on my Long WU. I've learned THAT lesson!
So if you want to try it now, give it a shot.
The app is in the official download page.
But still no joy without app info (I mean, no work sent by server - "no tasks available for selected work", it said). linux version works fine, though.
The Windows CPU app is now live -- but I suspect you already know that: The very first one went out to you about 15 minutes ago. :)
____________
My lucky number is 75898524288+1 |
|
|
|
The Windows CPU app is now live -- but I suspect you already know that: The very first one went out to you about 15 minutes ago. :)
Well, actually you saw it before I did...
Thanks once more. |
|
|
|
How do I run it?
Do I just select the box in the PG subproject preferences and leave the rest to BOINC as usual?
I tried that a few days ago and got the no work message as above, if that is all I have to do then I will try again now it is working.
____________
Member team AUSTRALIA
My lucky number is 9291*2^1085585+1 |
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14036 ID: 53948 Credit: 475,998,733 RAC: 222,579
                               
|
How do I run it?
Do I just select the box in the PG subproject preferences and leave the rest to BOINC as usual?
Essentially yes, but there's a few details you should know:
Go to the preferences screen where you select the PrimeGrid sub-projects, and check the Generalized Fermat Prime Search check box in the CPU column. Make sure the task size selection box is set to "Short tasks"
Finally, at top, make sure that "Use CPU" is checked.
Note that since you don't have a GPU that is compatible with Genefer, you can only run this project on your CPU, and only the shorter tasks.
____________
My lucky number is 75898524288+1 |
|
|
|
Thanks but I have tried it and get the "no work available" message.
Could this be because I also have my AMD GPU on this machine in use? I did not select the GPU column, only the CPU. I left the short tasks selected and block size as zero.
____________
Member team AUSTRALIA
My lucky number is 9291*2^1085585+1 |
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 14036 ID: 53948 Credit: 475,998,733 RAC: 222,579
                               
|
Thanks but I have tried it and get the "no work available" message.
Could this be because I also have my AMD GPU on this machine in use? I did not select the GPU column, only the CPU. I left the short tasks selected and block size as zero.
FYI, block size only applies to the CUDA application; it has no affect on the CPU version of the app.
OH -- I see what the problem is.
The application being distributed is a 64 bit program -- and both of your computers are running 32-bit Vista.
You have the following options:
1) Upgrade to a 64 bit operating system. If you go to Windows 64 bit, that's NOT an upgrade -- it's a brand new install. There's no upgrade path from 32 to 64 bits.
2) Install 64 bit Linux in another partition and run BOINC in there.
3) Run WUBI -- it's essenially Linux running inside a Windows file, so it's like option 2 except you don't need a separate partition.
4) Run a 64-bit OS inside a Virtual Machine. There's at least 2 free VMs you can use, VMWare Player and Virtual Box.
5) Download the 32-bit version of Genefer (Genefer-boinc.exe) and run that using app-info. The 32-bit version of Genefer is about 70% slower than the 64-bit versioin, and your CPUs aren't all that fast to begin with, but the deadlines on the short WUs seem to be set WAY too long, so you should be able to finish the WUs in time. (The WUs take about 5 hours on an average GPU and 2 days on an average CPU, and the deadline is 2 weeks.)
Option 4 is what I would do (and, in fact, HAVE done in the past when this computer was running 32-bit Vista.)
The selection web page needs to be updated. It says "32-bits, 64-bits" for Windows, but that's for the GPU app. The CPU app is 64 bits only. (This now applies to Windows, Linux, and Mac).
Edit: There's also a sixth option: in theory the server could be configured to send out the 32-bit version of Genefer to 32-bit hosts. I'll try and find out if that's possible. No promises.
____________
My lucky number is 75898524288+1 |
|
|
|
It is not worth getting a 64 bit OS for these old machines no matter how it is run, my wife uses my laptop and doorstop for uni work so linux is out.
I prefer option 7. Be patient and wait until I get a new machine with 64 bit OS (and GTX ???). Hopefully soon.
____________
Member team AUSTRALIA
My lucky number is 9291*2^1085585+1 |
|
|