Colab question - Page 27

bayanne · 2022-01-22, 09:42

Quote:

Originally Posted by chalsall

As usual with the free Colab, we can't know the future, but currently...

Yup, for the last couple of days I'm getting T4's about 75% of the time, lasting a good four hours or so each.

At about the three-hour mark some ask me to do the Robot thing. A fair trade for the compute.

P.S. I *really* wish Colab Pro was available here in little Bimshire. Google, I'm willing to pay!

ditto :)

kriesel · 2022-01-22, 13:06

Here, K80, T4, or most commonly no-gpu. At least one account has not gotten a gpu since 2022-01-16 despite multiple sessions per day launched, leaving incomplete a TF assignment 82% done needing 14.5 more hours of T4 gpu time. K80 is faster DP (PRP, LLDC or P-1), T4 ~4x faster SP (TF).

slandrum · 2022-01-22, 22:47

Quote:

Originally Posted by kriesel

At least one account has not gotten a gpu since 2022-01-16 despite multiple sessions per day launched, leaving incomplete a TF assignment 82% done needing 14.5 more hours of T4 gpu time.

This happens to me regularly. Colab will randomly decide that one of my accounts can't get a GPU, and then 1-3 weeks later, it suddenly can again.

mrk74 · 2022-01-25, 22:16

First let me say I'm not worried about not getting credit. I'm doing it for the cause but I am wondering what's going on.

To explain: I had my last TF in May. My ID was attached to it. I started getting them again in December. I just happened to look all the ones since December have been marked Anon. I haven't changed anything as far as I can remember so I don't know why the change would have happened. As far as I can tell my ID IS attached to P-1's.

chalsall · 2022-01-26, 18:01

Quote:

Originally Posted by mrk74

I just happened to look all the ones since December have been marked Anon. I haven't changed anything as far as I can remember so I don't know why the change would have happened. As far as I can tell my ID IS attached to P-1's.

Sorry about that...

In order for the "Auto Colab Submission" spider to work, a "slaved" mprime is used to create, and then keep up to date, the "virtual machine" called GPU72_TF for each account.

Unfortunately, Primenet sometimes comes back with a set of messages which cause the VM to be moved to ANONYMOUS. In your particular case, I didn't notice this, and so lately your results have been submitted thusly.

I have corrected this, so moving forward things should be "sane". Also, you did receive all of the credits as appropriate on GPU72, but not on Primenet. There was a unique CID for each submission, so I guess George, Aaron, and/or James could fix the historical data, but what I'm hearing is you don't really care all that much about credits.

Again, my apologies for the SPE. And thanks for bringing this forward. I will keep an eye out for this kind of thing more closely moving forward.

mrk74 · 2022-01-26, 22:50

Quote:

Originally Posted by chalsall

Sorry about that...

In order for the "Auto Colab Submission" spider to work, a "slaved" mprime is used to create, and then keep up to date, the "virtual machine" called GPU72_TF for each account.

Unfortunately, Primenet sometimes comes back with a set of messages which cause the VM to be moved to ANONYMOUS. In your particular case, I didn't notice this, and so lately your results have been submitted thusly.

I have corrected this, so moving forward things should be "sane". Also, you did receive all of the credits as appropriate on GPU72, but not on Primenet. There was a unique CID for each submission, so I guess George, Aaron, and/or James could fix the historical data, but what I'm hearing is you don't really care all that much about credits.

Again, my apologies for the SPE. And thanks for bringing this forward. I will keep an eye out for this kind of thing more closely moving forward.

No problem. More wanted to let somebody know in case the credit was a big deal to them. Honestly I don't look too often at but I decided to on one and that's how I noticed it.

Uncwilly · 2022-02-19, 20:09

This exponent's status is stuck, even though a later exponent on that worker is reporting in.
109240601
It did complete according to the worker window.

chalsall · 2022-02-19, 20:30

Quote:

Originally Posted by Uncwilly

This exponent's status is stuck, even though a later exponent on that worker is reporting in. 109240601 It did complete according to the worker window.

Hmmm... Weird. Thanks for pointing this out. Sorry for any wasted cycles.

LOBES · 2022-03-14, 21:09

I started up a Colab session to test out the updated P-1 code. Only got through Stage 1 @ 38.78% before it stopped on me.

Anyway, I just "moved" the assignment over to my home PC and finished it off since I didn't want it hanging out there. Over a week later its still assigned to me as a Colab assignment.

110479219

I'm hoping that will eventually clear itself out? From past experience, if I were to start a new Colab session, I would get a new P-1 assignment anyway, so I am not aware of any way to actually finish that assignment off using Colab.

chalsall · 2022-03-14, 21:31

Quote:

Originally Posted by LOBES

I'm hoping that will eventually clear itself out? From past experience, if I were to start a new Colab session, I would get a new P-1 assignment anyway, so I am not aware of any way to actually finish that assignment off using Colab.

There is an old and wise saying... No good deed goes unpunished.

You present an "edge-case"; serious programmers welcome these.

In this particular case, a race condition between what was assumed to be automated and what was eventually completed under the direct direction of a human.

It's all cool. Actually... beyond that... Thank you sincerely for demonstrating just how fragile some code paths can be!

Uncwilly · 2022-03-15, 13:36

Went to run GPU72 colab today and got the following in the window:

Code:

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


gzip: stdin: unexpected end of file
tar: Child returned status 1
tar: Error is not recoverable: exiting now
---------------------------------------------------------------------------
FileNotFoundError                         Traceback (most recent call last)
<ipython-input-1-522a61ed5a02> in <module>()
     24   stdout=subprocess.PIPE,
     25   universal_newlines=True,
---> 26   bufsize=0)
     27 
     28 try:

1 frames
/usr/lib/python3.7/subprocess.py in _execute_child(self, args, executable, preexec_fn, close_fds, pass_fds, cwd, env, startupinfo, creationflags, shell, p2cread, p2cwrite, c2pread, c2pwrite, errread, errwrite, restore_signals, start_new_session)
   1549                         if errno_num == errno.ENOENT:
   1550                             err_msg += ': ' + repr(err_filename)
-> 1551                     raise child_exception_type(errno_num, err_msg, err_filename)
   1552                 raise child_exception_type(err_msg)
   1553 

FileNotFoundError: [Errno 2] No such file or directory: './bootstrap.pl': './bootstrap.pl'

2022-01-25, 22:16	#290
mrk74 Jan 2020 3F₁₆ Posts	Not getting credit for TF's First let me say I'm not worried about not getting credit. I'm doing it for the cause but I am wondering what's going on. To explain: I had my last TF in May. My ID was attached to it. I started getting them again in December. I just happened to look all the ones since December have been marked Anon. I haven't changed anything as far as I can remember so I don't know why the change would have happened. As far as I can tell my ID IS attached to P-1's.

2022-01-22, 13:06	#288
kriesel "TF79LL86GIMPS96gpu17" Mar 2017 US midwest 1100101010011₂ Posts	Here, K80, T4, or most commonly no-gpu. At least one account has not gotten a gpu since 2022-01-16 despite multiple sessions per day launched, leaving incomplete a TF assignment 82% done needing 14.5 more hours of T4 gpu time. K80 is faster DP (PRP, LLDC or P-1), T4 ~4x faster SP (TF).

2022-02-19, 20:09	#293
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 24441₈ Posts	This exponent's status is stuck, even though a later exponent on that worker is reporting in. 109240601 It did complete according to the worker window.

2022-03-14, 21:09	#295
LOBES Mar 2019 USA 5·17 Posts	I started up a Colab session to test out the updated P-1 code. Only got through Stage 1 @ 38.78% before it stopped on me. Anyway, I just "moved" the assignment over to my home PC and finished it off since I didn't want it hanging out there. Over a week later its still assigned to me as a Colab assignment. 110479219 I'm hoping that will eventually clear itself out? From past experience, if I were to start a new Colab session, I would get a new P-1 assignment anyway, so I am not aware of any way to actually finish that assignment off using Colab.