Just wondering if anyone has changed their cache size due to the recent problems? I ran out of GPU work but managed to keep crunching CPU units.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

Haven't changed caches (run at 8 days), but because one of my machines had a flood of VLARs just before the semi-outage began, I couldn't reschedule anything to GPU, so I ran out of GPU WUs on that machine.

BTW - I'm running 6.10.18, and I find that it has been asking only for GPU WUs, so it is refilling cache s-l-o-w-l-y...is there any way I can force it to ask for CPU WUs, or put it in a "don't care" state, so I can get more CPU WUs to reschedule to GPU? (SETI/BOINC doesn't seem to generate near as many GPU WUs as CPU ones, as far as I can tell, which makes my cache recovery slower than it ought to be).

---

(SETI/BOINC doesn't seem to generate near as many GPU WUs as CPU ones, as far as I can tell, which makes my cache recovery slower than it ought to be).

SETI@home doesn't generate either type. The general-purpose WUs are assigned to CPU/GPU as requested.

The clou for you is:

I find that it has been asking only for GPU WUs...

So, it's the 'fault' of your client. Are you using the rescheduling tool? That could explain the imbalance.

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours

my cash size has changed...
during/after the outage i was thinking of
spending more money on computers for Seti.
mmmmm not any more.

but to your original question
my cache of work remains the same.
although i have flipped the "no new tasks" switch on
a few computers.

i'll see how i feel about it later.

---

Haven't changed caches (run at 8 days), but because one of my machines had a flood of VLARs just before the semi-outage began, I couldn't reschedule anything to GPU, so I ran out of GPU WUs on that machine.

BTW - I'm running 6.10.18, and I find that it has been asking only for GPU WUs, so it is refilling cache s-l-o-w-l-y...is there any way I can force it to ask for CPU WUs, or put it in a "don't care" state, so I can get more CPU WUs to reschedule to GPU? (SETI/BOINC doesn't seem to generate near as many GPU WUs as CPU ones, as far as I can tell, which makes my cache recovery slower than it ought to be).

Nope, still set to 4 days.


@JRavin:
You may be able to get more CPU tasks by setting Reschedule to 100% and letting it move everything (it won't move the VLARs) over to the GPU. Uncheck the VHAR and VHAR+VLAR options, too, if need be. The idea is to reduce the CPU count as much as possible for now. This usually works on my machine. I can sometimes repeat this until I fill my cache to the 350-400 I have when "full". I then set NNT, put the slider back to 50% and run Reschedule one more time. RE-checkmark your options if you unchecked anything.

Normally running with NNT set helps to prevent a VLAR unexpectedly showing up and going to the GPU.

Martin

How much work your machine caches depends on how much it has actually done in the recent past.

So if it's been idle for a few days the daily average is lower, and it will cache less work units.

Likewise a machine that is only running 8 hours a day will cache less work than one running 24/7 if the number of days is set to the same value.

Once you get some work returned and get your average up again the cache will return to normal.

Ian

Nope. My slowest machines run a 10 day cache so they can have a task other then the one they are processing. (60-100hr processing times) Otherwise I use a 5 day cache. Which even over the stuff we had going on last week worked just fine for me.

---

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

my cash size has changed...
during/after the outage i was thinking of
spending more money on computers for Seti.
mmmmm not any more.

but to your original question
my cache of work remains the same.
although i have flipped the "no new tasks" switch on
a few computers.

i'll see how i feel about it later.

Then Hiamps will beat you... LOL

Realistically it not about what you buy to find ET. It is the Journey and the Friends you find along the way.

Regards

Edit: I have met Hiamps, and know that he has respect for you and the competition you share.

Al

---

Please consider a Donation to the Seti Project.

my cash size has changed...
during/after the outage i was thinking of
spending more money on computers for Seti.
mmmmm not any more.

but to your original question
my cache of work remains the same.
although i have flipped the "no new tasks" switch on
a few computers.

i'll see how i feel about it later.

Then Hiamps will beat you... LOL

Realistically it not about what you buy to find ET. It is the Journey and the Friends you find along the way.

Regards

Ha, I've been playing catch up to Phud for a long time. Everytime I think I have him he just plugs another in....LOL

---

Official Abuser of Boinc Buttons...
And no good credit hound!

Ha, I've been playing catch up to Phud for a long time. Everytime I think I have him he just plugs another in....LOL







Well I won't get in your way when you pass me in the next day or 2. Go get him : )

---

my cash size has changed...
during/after the outage i was thinking of
spending more money on computers for Seti.
mmmmm not any more.

but to your original question
my cache of work remains the same.
although i have flipped the "no new tasks" switch on
a few computers.

i'll see how i feel about it later.

Then Hiamps will beat you... LOL

Realistically it not about what you buy to find ET. It is the Journey and the Friends you find along the way.

Regards

Ha, I've been playing catch up to Phud for a long time. Everytime I think I have him he just plugs another in....LOL

Tommorow, I need to drive down to MaChord. Can I borrow a spare GPU? GRIN

Regards

---

Please consider a Donation to the Seti Project.

@JRavin:
You may be able to get more CPU tasks by setting Reschedule to 100% and letting it move everything (it won't move the VLARs) over to the GPU. Uncheck the VHAR and VHAR+VLAR options, too, if need be. The idea is to reduce the CPU count as much as possible for now. This usually works on my machine. I can sometimes repeat this until I fill my cache to the 350-400 I have when "full". I then set NNT, put the slider back to 50% and run Reschedule one more time. RE-checkmark your options if you unchecked anything.

Normally running with NNT set helps to prevent a VLAR unexpectedly showing up and going to the GPU.

Martin

Martin - thanks for the thought, but it won't help me as ALL the CPU WUs I have now (~400) are VLARs. I have about 100 GPU WUs, and it is still asking only for GPU WUs, and I'm getting 1 or 2 at a time, whereas my other machine gets 30-40 per request (for CPU)... I guess I will have to wait until the VLARs run down enough for BOINC to request CPU WUs.

---

Martin - thanks for the thought, but it won't help me as ALL the CPU WUs I have now (~400) are VLARs. I have about 100 GPU WUs, and it is still asking only for GPU WUs, and I'm getting 1 or 2 at a time, whereas my other machine gets 30-40 per request (for CPU)... I guess I will have to wait until the VLARs run down enough for BOINC to request CPU WUs.

Thankfully, the tasks I've received lately are not all VLARs. It does seem like that was all I was getting earlier this month. Keep crunching! The VLAR storm will pass.

Martin

my cash size has changed...
during/after the outage i was thinking of
spending more money on computers for Seti.
mmmmm not any more.

but to your original question
my cache of work remains the same.
although i have flipped the "no new tasks" switch on
a few computers.

i'll see how i feel about it later.

Then Hiamps will beat you... LOL

Realistically it not about what you buy to find ET. It is the Journey and the Friends you find along the way.

Regards

Ha, I've been playing catch up to Phud for a long time. Everytime I think I have him he just plugs another in....LOL

Tommorow, I need to drive down to MaChord. Can I borrow a spare GPU? GRIN

Regards

Well I still never saw what this will do with 3 cards but if you need it and it will help things you could borrow my GTX 260. Let me know. I also have a 9600 GSO that is real easy to get to if that will work. Which ever you need I will make work.

---

Official Abuser of Boinc Buttons...
And no good credit hound!

I'm still set for 4 days. One cruncher actually went idle for an hour or two and then managed to squeeze a successful work request in and got 14 normal-length WUs, so that 4-day cache is filled.

Main cruncher was down to about 30 shorties left (about 80 minutes each, and 4 cores), and finally got a successful 1.8M-second work request and got all the WUs it needed to top off the cache. Smooth sailing now.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

I had a 4 day cache and ran out of Cuda on Rig on a Bench but still had 800 VLAR's to crunch on.
The i7 never ran out of work with 4 day cache.
I have set cache to 6 days now.

Dave

---

Cache? Now, why would I need much of a cache? Most of my machines are attached to a few dozen projects so an outage at one just means that some of the others get a bit of extra CPU time.

---

BOINC WIKI

Cache? Now, why would I need much of a cache? Most of my machines are attached to a few dozen projects so an outage at one just means that some of the others get a bit of extra CPU time.

John, I take this as a Server Side update affecting preferences or why the Boinc Core ignored server side settings. Then things like a single DCF affects many things.

We will let it develop from there.

For the record 6.10.32 and a couple of projects that have not updated "server code" now get more work than my venues. That is a Boinc Alpha/Dev thing.

Regards

---

Please consider a Donation to the Seti Project.