GPU to 72 status...

[DrobinsonPE]

When you are rested up from your real work, I am still getting this repeating over and over in colab.

Beginning GPU Trial Factoring Environment Bootstrapping...
Please see https://www.gpu72.com/ for additional details.

20210111_151440 ( 0:03): GPU72 TF V0.423 Bootstrap starting (now working single assignments at a time)...
20210111_151440 ( 0:03): Working as "user notebook"...

20210111_151440 ( 0:03): Installing needed packages...
20210111_151500 ( 0:03): GPU not found.  Make sure GPU is enabled (Runtime menu -> Change runtime type).
20210111_151500 ( 0:03): Running as CPU only...

20210111_151500 ( 0:03): GPU72 CPU Wrapper V0.05 starting...
20210111_151500 ( 0:03): Working as "user notebook"...
20210111_151500 ( 0:03): 
20210111_151500 ( 0:03): Sending Telemetry for user notebook (user notebook)
20210111_151500 ( 0:03): Fetching payload...
20210111_151505 ( 0:03): CPU Payload starting...
20210111_151506 ( 0:03): [Main thread Jan 11 15:15] Mersenne number primality test program version 29.8
20210111_151506 ( 0:03): [Main thread Jan 11 15:15] Optimizing for CPU architecture: Core i3/i5/i7, L2 cache size: 256 KB, L3 cache size: 45 MB
20210111_151506 ( 0:03): [Comm thread Jan 11 15:15] Updating computer information on the server
20210111_151506 ( 0:03): [Main thread Jan 11 15:15] Starting worker.
20210111_151507 ( 0:03): [Work thread Jan 11 15:15] Worker starting
20210111_151507 ( 0:03): [Comm thread Jan 11 15:15] Exchanging program options with server
20210111_151507 ( 0:03): [Comm thread Jan 11 15:15] Exchanging program options with server
20210111_151507 ( 0:03): [Comm thread Jan 11 15:15] Exchanging program options with server
20210111_151507 ( 0:03): [Main thread Jan 11 15:15] Restarting all worker threads.
20210111_151507 ( 0:03): [Comm thread Jan 11 15:15] Successfully quit GIMPS.
20210111_151507 ( 0:03): [Comm thread Jan 11 15:15] Done communicating with server.
20210111_151512 ( 0:03): [Work thread Jan 11 15:15] Worker stopped.
20210111_151512 ( 0:03): [Main thread Jan 11 15:15] Execution halted.
20210111_151512 ( 0:03): [Main thread Jan 11 15:15] Choose Test/Continue to restart.
20210111_151512 ( 0:03): CPU process finished.  Launching checkpointer to send last results (if interrupted)
20210111_151512 ( 0:03): 
20210111_151512 ( 0:03): Exiting.

[storm5510]

Originally Posted by chalsall

...Sorry for any wasted cycles. Seriously bad timing...

My auto fetch batch process is working fine now. The results are being accepted by Primenet without warnings.

[ZFR]

Originally Posted by chalsall

And, I can now talk a /little/ bit about what's been keeping me so busy lately... Please see https://bimsafe.gov.bb/

I didn't write the App, I but was responsible for a bunch of the back-end stuff that

Hey, that's nice. Thanks for sharing.

[ixfd64]

I noticed that the GPU to 72 spider has reserved a lot of assignments ahead of the current DC wavefront: https://mersenne.org/assignments/?ex...xp_hi=60100000

However, no assignments in that range are available on the GPU to 72 website. Is there a particular reason for this?

[chalsall]

Originally Posted by ixfd64

Is there a particular reason for this?

Something changed on the Primenet server around the beginning of the year, such that whenever GPU72 was told by Primenet that a candidate was available for reservation, the slaved mprime would also be given three DC candidates.

I only noticed (and fixed) this yesterday. Finally had some cycles for some house-keeping, log reviews, etc.

George &/| Aaron... Sorry for this, but if you could please unresearve any DC assignments held by GPU72, that would help clean up that area. Either that, or else wait the six months for them to expire.

I'd do it myself if I could, but the DB has no knowledge of the AIDs.

[ixfd64]

You can still log into the GPU Factoring account on the GIMPS website and unreserve active assignments. However, this will probably take a long time to do manually.

[Prime95]

Originally Posted by chalsall

George &/| Aaron... Sorry for this, but if you could please unresearve any DC assignments held by GPU72, that would help clean up that area. Either that, or else wait the six months for them to expire..

Easy, peasy. Should be done within a few minutes.

[chalsall]

Originally Posted by Prime95

Easy, peasy. Should be done within a few minutes.

Coolness. Thanks.

And, wow! I see about 100,000 DC candidates returned to the pool.

When the Humans are distracted, the 'Bots will do the most annoying things...

[ATH]

There are many old TF assignments mostly from March to July 2020 in the Cat 1 range.

Granted most were Cat 3 when assigned, but did something go wrong?

https://www.mersenne.org/assignments...hk=1&exfirst=1

EDIT: Noticed the range of old assignments continues up to 104,897,000 but it can only show 1000 assignments, and there are more from 105,124,000 and up to 105,790,000, and even older from Feb 2020 above 106M. So maybe this is intended?


EDIT2: Some more specific numbers:

							 		Total	Before Sep 1st 2020 (>~6 mdr)
104280000-104897000 + 105000000-105538009 (Cat1)			15,140  	14,394 (95.1%)
105538010-105790000 + 106000000-106838000 + 107000000-107628848 (Cat2)	30,604		28,994 (94.7%)

[Chuck]

I haven't worked for several months, but my Colab results are not being auto-submitted. My results page says "automatic submission of results is activated for your account".

[chalsall]

Originally Posted by ATH

Granted most were Cat 3 when assigned, but did something go wrong?

Not /wrong/ exactly...

The TF'ing is, currently, only a bit ahead of the FC Cat 1 wavefront (and that buffer is getting smaller every day).

Thanks, Ben et al! (Seriously and sincerely!!! )

We still have about 60 days or so of buffer. As in, we're not currently blocking anything. In fact, we're still also strategically feeding the other Cats optimally.

We'll either find some way to pull ahead (a couple of paths are modeled, but they take (my) human cycles), or we'll release as required with the TF'ing sub-optimally done.

Please trust me on this: I understand this is GIMPS, not GIMFS. GPU72 will not block any GIMPS progress nor milestones.

Having said all of the above, if anyone has some GPUs they could throw our way doing TF'ing, it would be appreciated...