I get tons of VLAR tasks which usually take ~4hrs on my i7 using hyperthreading. They are consistently about 367Kb on my disk. I checked this morning and I have a VLAR task that's burned 7.5 hrs and it's at 92% completion. I have 7 of these things and instead of having the usual .vlar extension they have .vlar_1_0 and take up ~27Kb of disk space. They are all crunching now and all are taking a large amount of time. I checked the wingman on one and he/she finished it in 4.75 hours on a core2 duo.

I feel like I've run into a neutron star. I hope it's not just my computer.

I get tons of VLAR tasks which usually take ~4hrs on my i7 using hyperthreading. They are consistently about 367Kb on my disk. I checked this morning and I have a VLAR task that's burned 7.5 hrs and it's at 92% completion. I have 7 of these things and instead of having the usual .vlar extension they have .vlar_1_0 and take up ~27Kb of disk space. They are all crunching now and all are taking a large amount of time. I checked the wingman on one and he/she finished it in 4.75 hours on a core2 duo.

I feel like I've run into a neutron star. I hope it's not just my computer.

The *****.vlar_1_0 file is the file that will upload to the server once complete, you'll only have those files if you're started it's task,

Claggy

Files like .vlar_1_0 are the result files being prepared for return to the server.

They are created, and a non-trival amount of data put into them, as soon as any task starts running - that's the 27KB you're seeing. More is added as the run progresses. Exactly the same thing happens every time - MB or AP, CPU or GPU, VLAR or not.

Don't delete them, or mess with them in any way. They'll be deleted when the task has finished uploading at the end of the run.

So I guess it creates that when it starts the tsak, not on completion. Here's my 94% 7.75 hr task. http://setiathome.berkeley.edu/result.php?resultid=2348276444 Anything I should be concerned about. It seems way out of the norm.

Weird. I ran CPU benchmarks and got 439 fp mips and 1756 integer mips. Bad, bad. I stopped Boinc, restarted and core dumped. Reboot, rerun CPU benchmarks. Now they're 1728 fp mips 9486 integer mips. Anybody run into this or is it just me? I hadn't rebooted in a day or so. I've noticed this slowdown 3 or 4 times recently which a reboot corrects.

Weird. I ran CPU benchmarks and got 439 fp mips and 1756 integer mips. Bad, bad. I stopped Boinc, restarted and core dumped. Reboot, rerun CPU benchmarks. Now they're 1728 fp mips 9486 integer mips. Anybody run into this or is it just me? I hadn't rebooted in a day or so. I've noticed this slowdown 3 or 4 times recently which a reboot corrects.

While benchmarks can be a useful thing they are not something to fret over. With Seti@home tasks you want to compare tasks be their AR(Angle Range). The one you referred to lists "WU true angle range is : 0.013978". The lower that number the longer tasks take. I have seen having nothing but VLAR tasks on a machine slow it down even more. If I remember correctly it has something to do with lower angle range doing more memory access or something along those lines.

I was having issues on my i7-860 HTPC if I had 8 VLAR's running at a time. Videos would become choppy, pause, or the video would go out of sync. I ended up cutting it down to only running 4 tasks at once. Comparing the RAC to my other five i7-860 boxes it did not seem to have a a great effect on the RAC of the machine. Sometimes it would be as high as 4th out of the six.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Thanks. I'll shut off the hyperthreading and that will cut the CPU tasks to 3. I ran in that configuration for months with no problems. It was just fun seeing 7 CPU tasks getting knocked off at once (although they took nearly twice as long.) I save most of the other CPU to feed the 2 GPU tasks. We'll see if that helps. Thanks again!

If you run more than one task per GPU, you don't need to cut hyperthreading.
In fact, what I see on my new i7-3820, is that HT gives me 1/3 more tasks output and GPU processing speed is virtually not affected, with even not dedicated logical(or physical) core to feed the GPU.
GPU apps have higher priority anyway too.

---

Thanks. I'll shut off the hyperthreading and that will cut the CPU tasks to 3. I ran in that configuration for months with no problems. It was just fun seeing 7 CPU tasks getting knocked off at once (although they took nearly twice as long.) I save most of the other CPU to feed the 2 GPU tasks. We'll see if that helps. Thanks again!

I leave HT on, but some do turn it off when also doing GPU tasks. I just limit BOINC to running x number of tasks. You can do that in the BOINC GUI setting "On multiprocessor systems, use at most 38% of the processors" for 3/8. You could also set another venue/location with those setting on the webpage if you desired. I use the entry in my cc_config.xml file and set it to 4. As that was the best option for me. YMMV.

Setting to limit to 3 CPU would also limit the machine to 150 CPU tasks. As the limits ate 50 per CPU core and 400 per GPU.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Thanks. I'll shut off the hyperthreading and that will cut the CPU tasks to 3. I ran in that configuration for months with no problems. It was just fun seeing 7 CPU tasks getting knocked off at once (although they took nearly twice as long.) I save most of the other CPU to feed the 2 GPU tasks. We'll see if that helps. Thanks again!

I leave HT on, but some do turn it off when also doing GPU tasks. I just limit BOINC to running x number of tasks. You can do that in the BOINC GUI setting "On multiprocessor systems, use at most 38% of the processors" for 3/8. You could also set another venue/location with those setting on the webpage if you desired. I use the entry in my cc_config.xml file and set it to 4. As that was the best option for me. YMMV.

Setting to limit to 3 CPU would also limit the machine to 150 CPU tasks. As the limits ate 50 per CPU core and 400 per GPU.

What's the XML tag for tasks in cc_config? I'll try your setup.

Thanks. I'll shut off the hyperthreading and that will cut the CPU tasks to 3. I ran in that configuration for months with no problems. It was just fun seeing 7 CPU tasks getting knocked off at once (although they took nearly twice as long.) I save most of the other CPU to feed the 2 GPU tasks. We'll see if that helps. Thanks again!

I leave HT on, but some do turn it off when also doing GPU tasks. I just limit BOINC to running x number of tasks. You can do that in the BOINC GUI setting "On multiprocessor systems, use at most 38% of the processors" for 3/8. You could also set another venue/location with those setting on the webpage if you desired. I use the entry in my cc_config.xml file and set it to 4. As that was the best option for me. YMMV.

Setting to limit to 3 CPU would also limit the machine to 150 CPU tasks. As the limits ate 50 per CPU core and 400 per GPU.

What's the XML tag for tasks in cc_config? I'll try your setup.

If you want to do it that way make sure you keep a handy reference to the instructions.

This is an example that you can use to start from if you are not already using a cc_config.xml file.

<cc_config>
	<log_flags>
	</log_flags>
	<options>
		<ncpus>4</ncpus>
	</options>
</cc_config>

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Thanks again. I created a cc_config.xml file a week or so ago after we had the problems reporting to many completed tasks. It's nice to have it around. Also thanks for the link. I'll have to see what else I can do in there.