Perhaps they are cleaning out the data storage and area. Tape 22mr13aa is 0.25 GB in size the smallest tape I have seen in a long while.

---

Seems the forums and other pages are taking ages to load!!

---

Not here Bernie. Seems OK in France ! :p

---

Likewise- no problems here with the sites at the moment.

---

Grant
Darwin NT

No problems here either.

---

[/quote]

Old James

No problem here eithier.

Claggy

No worries here either, although there have been some transient problems from time to time which I was not here to witness.

All quiet on the eastern front at present.

Meow.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Having just returned home I find it is as speedy as ever.

Must have been something my end. Panic over!!

---

Cool Bernie :)

---

Bunch of shorties going through?
Ready-to-send buffer is very, very slowly shrinking- as usual the splitters struggle to keep up.

---

Grant
Darwin NT

Bunch of shorties going through?

Yup - 14 out of 15 from my last fetch.

Bunch of shorties going through?

Yup - 14 out of 15 from my last fetch.

We were up to about 80k coming back in for a bit there.
It's dropped back down a bit since.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Another stuck tape?
Splitters have been struggling more than usual for the last few days.

---

Grant
Darwin NT

Another stuck tape?
Splitters have been struggling more than usual for the last few days.

Look pretty nominal to me: http://setistats.haveland.com/

Creation rate (green)
Ready to send (green)

Just the usual catch-up dip after maintenance, as we are warned every week.

Another stuck tape?
Splitters have been struggling more than usual for the last few days.

Look pretty nominal to me: http://setistats.haveland.com/

Creation rate (green)
Ready to send (green)

Just the usual catch-up dip after maintenance, as we are warned every week.

Checkout the weekly graph.
Normally the splitters will run, top up the work, shut down, etc- resulting in a squarewave form result creation graph, with a saw tooth for the ready-to-send buffer. The last few days the splitters have been running continuously & the ready-to-send buffer, while it has recovered, is almost flat.
A sign of struggling splitters.

---

Grant
Darwin NT

Another stuck tape?
Splitters have been struggling more than usual for the last few days.

Look pretty nominal to me: http://setistats.haveland.com/

Creation rate (green)
Ready to send (green)

Just the usual catch-up dip after maintenance, as we are warned every week.

Checkout the weekly graph.
Normally the splitters will run, top up the work, shut down, etc- resulting in a squarewave form result creation graph, with a saw tooth for the ready-to-send buffer. The last few days the splitters have been running continuously & the ready-to-send buffer, while it has recovered, is almost flat.
A sign of struggling splitters.

A sign of a lot of shorties on recent tapes. We request the same amount of work (by time), and get allocated more tasks. We're sucking harder, and the splitters ramp up to meet demand. Agreed, they may not be able to fully keep up if all tapes are shorties, and the storm goes on for a long time.

Edit - I've just done a work fetch. Got 34 new tasks. Every single one of them a shorty.

Question:

Do shorties affect overall credit? One would think that doing more of smaller WUs would balance out, but my machines have been getting a lot less credit the last few days, and my RAC has been dropping (about 5% so far).

---

Yes, but not in a positive sense. I think the reason is that there is an overhead that is pretty much the same for all WU, and that isn't included in the credits calculation :-(

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Question:

Do shorties affect overall credit? One would think that doing more of smaller WUs would balance out, but my machines have been getting a lot less credit the last few days, and my RAC has been dropping (about 5% so far).

The adjustment of estimates for the added Autocorrelation search in S@H v7 was an increase of about 48% for VHAR shorties. That was a compromise, the stock CPU application needed about 20%, and the stock GPU applications needed something like 100%. The net result is that shorties may give more credit per hour than midrange when processed on CPU, but the opposite when processed on GPU.

As long as those splitter estimates in rsc_fpops_est can be kept reasonably representative of all hosts, they're fair, but short term variation in what ARs are being split will shift RAC up or down.

                                                                   Joe