Gave running the cache down a try last week. Both rigs are doing ap-only right now. The single-core machine usually only has 3 at a time (4-day cache), but the main rig has 18-24. Been doing pretty well with r339. It is faster on my CPUs than r168 was by about 10%.

Ran the credits/sec math on AP vs MB, and it turns out that AP gives me 0.001 credits/sec more (about 10%) than MB does, and I seem to have run into a batch of 30 repetitive pulse units, and those end up giving about 0.004 credits/sec more than MB (nearly a 50% gain).

I was actually surprised to see how easy it was to get APs assigned. Almost every single work request ends up getting one.

We do appreciate the Beta testing of r339, and barring any serious problem cropping up it'll probably go to release status fairly soon.

You've had a fair number which got to the repetitive limit early but ran full length, those are the best for credit rate. The worst are heavily blanked tasks, and I didn't see any of those in your completed AP tasks. If you get a bunch of them they'll make the credit comparison less favorable. Matt's software blanking may make them less likely, but I've seen up to 88% blanking recently so they're not totally gone.

                                                              Joe

All my OpenCL precessed tasks had blanking of 0%. Some were overflowed by rep pulses, some almost clean ones.

Gave running the cache down a try last week. Both rigs are doing ap-only right now. The single-core machine usually only has 3 at a time (4-day cache), but the main rig has 18-24. Been doing pretty well with r339. It is faster on my CPUs than r168 was by about 10%.

Ran the credits/sec math on AP vs MB, and it turns out that AP gives me 0.001 credits/sec more (about 10%) than MB does, and I seem to have run into a batch of 30 repetitive pulse units, and those end up giving about 0.004 credits/sec more than MB (nearly a 50% gain).

I was actually surprised to see how easy it was to get APs assigned. Almost every single work request ends up getting one.

We do appreciate the Beta testing of r339, and barring any serious problem cropping up it'll probably go to release status fairly soon.

You've had a fair number which got to the repetitive limit early but ran full length, those are the best for credit rate. The worst are heavily blanked tasks, and I didn't see any of those in your completed AP tasks. If you get a bunch of them they'll make the credit comparison less favorable. Matt's software blanking may make them less likely, but I've seen up to 88% blanking recently so they're not totally gone.

                                                              Joe

Yeah, so far with 323 or 339, the most blanking I've run across is 5.96%. Almost all of them are zero, but a small handful were 1 or 2%. I guess the software blanking is doing a pretty decent job.

As before with all previous builds for the various AP apps, I have my excel spreadsheet with all the APs I get, and the PDFs of the taskid for said WUs.

---

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

FOUR astropulse WU running at once!
Just got a new laptop with an I7 720 chip in in.

Looked at my BOINC manager last night and saw that I had 4 AP units all running at once.
20+ hours to complete looks like.
I'll get a screen shot up later today

---

I tried running AP only and all I ended up with is a PC with no cache. I'll take them if I get them but I can't see having my computers idling when they can be crunching MB

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I have THREE AP running now at the same time with completion time of about 13 hours and another 10 APs in the pipeline.

---

Looks like the AP splitters are disabled.. and the fourth one is now shown as being on bambi (I do not recall if it was previously). I know the first one was on a different server. I'm hoping they're working on recoding/recompiling the splitter process for the new OS version so we can get more than one splitter going.

---

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

My second rig the i7 just bagged itself 2 AP's.
First I've had since the 505's came out.
Can't wait to see how long they take to crunch.

Whoopie Do.

Dave

---

Well, a new splitter turned on a little while ago, and now there's a second one running. I'm guessing we'll have all four of them running soon.

---

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