Back to the REAL Server Problems....

It would appear the 'FIX' for the Windows ATI AstroPulse Fiasco has been Broken.
It's common knowledge the Newer Windows drivers Break the ATI AsrtoPulse App resulting in very few Signals being reported. This was supposedly Fixed by NOT assigning a Windows ATI Host as a Tiebreaker or even matching two Windows ATI GPUs on an AstroPulse task.
Well, they are back to cross validating and giving Hosts producing the CORRECT results FALSE Invalids. Why can't this be solved? it's been like this for well over a couple of Years, long enough to totally pollute the AP Database.
This is the Wrong Result;

   Task      Computer	          Sent	                   Time reported                    Status	        Run time  CPU time   Credit                    Application
6648837963   7769537  17 May 2018, 23:19:16 UTC  18 May 2018, 3:33:53 UTC 	Completed, marked as invalid    1,737.41   222.33      0.00  AstroPulse v7 Anonymous platform (ATI GPU)
6648837964   8119708  17 May 2018, 23:19:16 UTC  18 May 2018, 13:16:51 UTC 	Completed and validated         2,481.45    51.31    531.55  AstroPulse v7 v7.09 (opencl_ati_100) windows_intelx86
6650504186   6845948  18 May 2018, 13:17:04 UTC  20 May 2018, 17:32:12 UTC 	Completed and validated         1,126.88   153.52    531.55  AstroPulse v7 v7.09 (opencl_ati_100) windows_intelx86

This Host has Numerous Invalid APs, https://setiathome.berkeley.edu/results.php?hostid=8119708&state=5
This Host only Validates when the Signal count is Low, or matched with another Windows ATI Host, https://setiathome.berkeley.edu/results.php?hostid=6845948&appid=20
This Only happens with the recent Windows drivers, somewhere after AMD App 1800. The Linux and Mac ATI Hosts don't have this problem.

Not that this is anything server related. I noticed I had a number of tasks with the date of the 19th then I looked & realised that the data was recorded only 2 days ago19my18aa.23825.885.14.41.98_1 Now that's what I call real time processing.

I have also noticed that this morning I had a good number probably 5 BLC tasks that exited after about 22 seconds

---

Back to the REAL Server Problems....

It would appear the 'FIX' for the Windows ATI AstroPulse Fiasco has been Broken.
It's common knowledge the Newer Windows drivers Break the ATI AsrtoPulse App....

Last time I spent a weekend working with Raistmer on a suspected driver problem (intel_gpu app on 6th generation HD 630 iGPU), he tried everything he could think of - driver, Intel OpenCL FFT libraries, the whole shooting match.

It turned out to be a compiler optimisation (fused multiply+add) on his build machine - the compiler was selecting a quicker, but less precise, code path.

Somebody needs to revisit that app with a similarly open mind.

So, what happened with the outage today?
The site came back long time ago, but the important servers and DB's are still down.

. . Yep, I just got a feeder not running message ...

Stephen

? ?

All finished work returned, now just waiting for fresh work to arrive.

[edit] And they now begin to arrive.

Cheers.

Downloads are stalled on both boxes.

---

Downloads are stalled on both boxes.

I just came here to see if it was just me.

Looks like it isn't.

---

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

Caches are full, but you just have to get them to download.

Cheers.

Any special trick used?

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Any special trick used?

Crack a can of beer per download!!

Hic---Hic---Hic :))

Any special trick used?

Crack a can of beer per download!!

Hic---Hic---Hic :))

Ha Ha . . . I can get with that program LOL. Gin&Tonics for me.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

;-)P

Downloads are stalled on both boxes.

. . Yep, I've been having stalled and very slow downloads since the outage too, but it seems to be improving.

Stephen

:(

Any special trick used?

Crack a can of beer per download!!

Hic---Hic---Hic :))

Ha Ha . . . I can get with that program LOL. Gin&Tonics for me.

. . Make mine a Margherita ....

Stephen

:)

Any special trick used?

Crack a can of beer per download!!

Hic---Hic---Hic :))

Ha Ha . . . I can get with that program LOL. Gin&Tonics for me.

. . Make mine a Margherita ....

Stephen

:)

Bugger it, have all 3 types..

;-) End of the day we will think we have found ET..

A couple more hours & we'll be out of work.
Since the outage, the splitters just haven't managed to really get going- the best they've been able to do is to meet present demand (which isn't all that high). Most of the time they've been well below the level of demand. The Ready-to-send buffer continues to empty.
Maybe once the backlog of AP & MB WU deletions clears they'll be able to pump out some work (that particular malaise seems to have returned as well).

---

Grant
Darwin NT

I see that the RTS is almost empty. Splitters never did take off after the outage. I wonder if it has anything to do with the Resend Lost Tasks, the first two tasks that I got were the two Ghosts that I had. Of course they could just need a size twelve applied to the right spot. The guys will get them going when they come in.

---

Bruce

RTS still going down (around 70K at the moment). I'm surprised it has lasted this long. All the numbers look good except the creation rate. Hopefully someone will soon be in to give the machine some percussive maintenance, as was previously suggested.

It's showing 54/sec and 101K available now - I'd say dawn has broken in California.

The splitters are still struggling to keep up with demand and having a hard time building the RTS buffer.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)