I see the web site is back, as well as the server status data; the graphs are updating again.

---

Grant
Darwin NT

You missed all of the fights and (intoxicated) slurs over on the BOINC forums. Well, or not, of course. :)

At least he didn't go into a total meltdown.

---

This site will crash again pretty soon.
No doubt about that.
More tea thank you...burp...

That's a givin ain't it nothing's 100% reliable , seti website will " Go down " again and work will either keep flowing like this time or stop , or maybe even be intermittant , one thing i do know is the " real " world will continue and events happen that are far more important than a website / project :-)

So shal it be :-)

( ohh poo that's blasphomy what i just posted bug#er ;-) )

---

Life is what you make of it :-)

When i'm good i'm very good , but when i'm bad i'm shi#eloads better ;-) In't I " buttercups " p.m.s.l at authoritie !!;-)

Wasn't expecting it to be back any time today tbh, so the person/s that did it early on a Sunday morning should be thanked :-)

Just for grins, based on the stated times for running 100 WUs on GPUs up thread, I have decided to try an experiment on my test cruncher Sabertooth to see if in fact I get about the same RAC using 1 WU/GPU instead of the current 3/GPU (It's running 1 GTX 980 and 1 GTX 1080). I should be able to get a real sense of this in about 24 hours. I made the change a few minutes ago.

At present, the only thing I can see in in SIV, which showed the GPUs as running around a steady 95% GPU utilization with 3 WUs, and is now showing them at ~60-80% and highly variable . Don't know what that really implies going forward, but lets find out what happens to the RAC.

---

How many hours does it take for a GTX 1080 to go trough the 100WU limit?

My GTX1080 does guppi vlar in 150 seconds and a shortie in 46 seconds so a 100 of them would take 4600 - 15000 seconds (1hr 17 min - 4hr 10min).
My GTX1080Ti does guppi vlar in 110 seconds and a shortie in 37 seconds so a 100 of them would take 3700 - 11000 seconds (1hr 2 min - 3hr 4min).

Petri

---

To overcome Heisenbergs:
"You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones

I found several hundreds of teams and user profiles with spam in them, varying from pornography through illegal download torrents to (online) casinos. Having that in the database couldn't have made it easier to load either, especially not if these were linked to and read by thousands from outside. There is a script running in the background now ridding the database of all that crud, which is causing some slowdowns here as it's quite resource intensive.

---

There is a script running in the background now ridding the database of all that crud, which is causing some slowdowns here as it's quite resource intensive.

Excellent.

---

+1

---

Just for grins, based on the stated times for running 100 WUs on GPUs up thread, I have decided to try an experiment on my test cruncher Sabertooth to see if in fact I get about the same RAC using 1 WU/GPU instead of the current 3/GPU (It's running 1 GTX 980 and 1 GTX 1080). I should be able to get a real sense of this in about 24 hours. I made the change a few minutes ago.

At present, the only thing I can see in in SIV, which showed the GPUs as running around a steady 95% GPU utilization with 3 WUs, and is now showing them at ~60-80% and highly variable . Don't know what that really implies going forward, but lets find out what happens to the RAC.

. . I would expect the RAC to decrease from the usage figures you have said. What crunching app are you running ??

. . I find that with SoG r3557 I am getting near 99 percent usage running doubles (on a GTX950 in an i5 rig with sleep ON). As the SoG apps have been developed they have become more and more efficient requiring less concurrent tasks to achieve near full GPU utilisation. Most guys achieving high utilisation running singles also run with sleep OFF and have a CPU core dedicated to each task. I don't know if that can help you at all, but I thought I would raise the point :)

Stephen

.

Am running Beta 6 SoG.

I have run with sleep off for quite a while now, as I found that the CPU usage accompanying GPU WUs was basically bogus before making the change. That is, with sleep off my CPU usage/WU is <20% for Arecibo, and ~30% for GBT, as opposed to ~100% without it, and that completion times are about the same - BUT I can run a couple of more CPU threads now because of the more available CPU time.

---

Use_sleep with a 1080?? I would drop that and use -hp instead. You have to remember that a 5% increase in GPU times will be more than your entire CPU can do for RAC.

EDIT: And I would update to r3584 as well.

Use_sleep with a 1080?? I would drop that and use -hp instead. You have to remember that a 5% increase in GPU times will be more than your entire CPU can do for RAC.

EDIT: And I would update to r3584 as well.

1) I am running dual 8 core E5-2670s, so 32 threads. Considering the GPU threads, actually running 24 or 25 CPU threads on my 2 main crunchers, and they come close to ~1 GPU. Or at least not completely left in the dust like a desktop CPU would be.

2) As far as R3584 goes, I am loathe to do anything not in stock or Lunatics, since it is so easy to screw up the app_info.

3) use_sleep seems to work fine for me, as I mentioned above.

---

2) As far as R3584 goes, I am loathe to do anything not in stock or Lunatics, since it is so easy to screw up the app_info.

Tisk ... me too ... I still have "ghosts" out there from the (next to) last time I tried. The last time I got no ghosts ...'cause there were no WU's in (anybodies) queue ...

EdF

There is a script running in the background now ridding the database of all that crud, which is causing some slowdowns here as it's quite resource intensive.

Glad to hear the script is running to get rid of the crud. Sounds like some house cleaning will do a lot of good.

---

~Chris

Am running Beta 6 SoG.

I have run with sleep off for quite a while now, as I found that the CPU usage accompanying GPU WUs was basically bogus before making the change. That is, with sleep off my CPU usage/WU is <20% for Arecibo, and ~30% for GBT, as opposed to ~100% without it, and that completion times are about the same - BUT I can run a couple of more CPU threads now because of the more available CPU time.

. . How strange, that is the complete opposite of my experience. With sleep ON under SoG r3554 on the i5 with GTX950, and previously on the Core2 Duo with the GYX1050ti, I was/am able to run multiple WUs per GPU and only need to allocate a single CPU core. ON my rig with the 1060s I was running SoG with sleep OFF and I could only run singles as each GPU WU was taking a whole CPU core. Now I am running Linux and CUDA80 on the latter 2 machines and that is singles only regardless. And I wouldn't have it any other way :)

. . But I agree that the completion times do not vary much between the two options.

Stephen

.

Use_sleep with a 1080?? I would drop that and use -hp instead. You have to remember that a 5% increase in GPU times will be more than your entire CPU can do for RAC.

EDIT: And I would update to r3584 as well.

. . I would use r3584 too if there was a Lunatics release with it included. I am loathe to attempt editing app_info.xml manually, I have heard too many scary stories.

. . Stephen

.

Use_sleep with a 1080?? I would drop that and use -hp instead. You have to remember that a 5% increase in GPU times will be more than your entire CPU can do for RAC.

EDIT: And I would update to r3584 as well.

1) I am running dual 8 core E5-2670s, so 32 threads. Considering the GPU threads, actually running 24 or 25 CPU threads on my 2 main crunchers, and they come close to ~1 GPU. Or at least not completely left in the dust like a desktop CPU would be.

2) As far as R3584 goes, I am loathe to do anything not in stock or Lunatics, since it is so easy to screw up the app_info.

3) use_sleep seems to work fine for me, as I mentioned above.

The current stock NV & ATI apps are r3584.

---

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

The current stock NV & ATI apps are r3584.

But with crappy CPU apps which offset the r3584 gains.

OTOH Lunatics Beta 6 installer uses the slightly older GPU r3557 app which is kilometres better than any of the older SoG apps (which were quite frankly crappy and invalid prone) and only a millimetre short of the current r3584 app (without invalids), but with kilometres better CPU apps, and most of us would rather wait for the next Lunatics Installer to correct that instead of trying to jump through hoops and trashing work trying to alter our app_info.xml to suit the r3584 app.

Suitable horses for suitable courses and K.I.S.S. comes into play here as well HAL. ;-)

Cheers.

At present, the only thing I can see in in SIV, which showed the GPUs as running around a steady 95% GPU utilization with 3 WUs, and is now showing them at ~60-80% and highly variable .

Using GPUz my GTX 1070s show 90-95% GPU load with 70-90% Memory Controller load (some drops down to 20% depending on where it is in the WU) and Bus Interface load generally around 18% and Power Consumption around 62% running just 1WU at a time.
Running SoG with aggressive settings and 1 CPU core reserved per WU.

---

Grant
Darwin NT