mersenneforum.org > News The Next Big Development for GIMPS
 Register FAQ Search Today's Posts Mark Forums Read

2020-07-05, 06:32   #89
Prime95
P90 years forever!

Aug 2002
Yeehaw, FL

7,151 Posts

Quote:
 Originally Posted by preda And a full-sized proof can be found here: https://drive.google.com/file/d/1vqp...ftY7nJG0l72p5J
Code:
Certifying proof for M97898863.  Proof power = 8, TopK = 97899008, hash length = 64
Prover did 145 excess squarings
Certifying a negative PRP test
h0 = 5A532D61D8CA121A
h1 = A8F263B98C58E6E0
h2 = B6E1C3415D782661
h3 = 17DC37D9709E9384
h5 = 4CFD2CFB61D17C2E
h6 = FCFEFF2A9D9A878D
Random exponent = 254357798350921
Hash B = 73619C891332B067CBFC62A4EC5F18672950A398BC720AF48F0A8EFFC5980ABE
Server certification cost 1395 squarings
Hash A = FA2D8DB74D30BC7BD84F37C4F022B9023047B322C91C2C41EEC061C3C81D11BD
Max roe during calculations: 0.149459
FAIL
User certification cost 382418 squarings
Preda -- please check the hashes.
The failed verification took 3.6 hours on a single core of my laptop.

2020-07-05, 10:22   #90
preda

"Mihai Preda"
Apr 2015

2×599 Posts

Quote:
 Originally Posted by Prime95 Code: Certifying proof for M97898863. Proof power = 8, TopK = 97899008, hash length = 64 Prover did 145 excess squarings Certifying a negative PRP test Root hash = 7F5AD60E97A24704452043D28C4EEB405094AC6F14C323BA8E0E7D847C3745A5 h0 = 5A532D61D8CA121A h1 = A8F263B98C58E6E0 h2 = B6E1C3415D782661 h3 = 17DC37D9709E9384 h4 = D4632194BADFADB2 h5 = 4CFD2CFB61D17C2E h6 = FCFEFF2A9D9A878D Random exponent = 254357798350921 Hash B = 73619C891332B067CBFC62A4EC5F18672950A398BC720AF48F0A8EFFC5980ABE Server certification cost 1395 squarings Hash A = FA2D8DB74D30BC7BD84F37C4F022B9023047B322C91C2C41EEC061C3C81D11BD Max roe during calculations: 0.149459 FAIL User certification cost 382418 squarings Preda -- please check the hashes. The failed verification took 3.6 hours on a single core of my laptop.
Yep, I fixed a bug on my side, and updated the fixed proof

The hashes I expect now are:
Code:
8e0e7d847c3745a5
5a532d61d8ca121a
dae5da6f8ac0b069
7c48bec7ef1f40d9
3047fa02f1c341bc
ec14677d2a04b012
5832f965f2903c39

2020-07-05, 10:23   #91
M344587487

"Composite as Heck"
Oct 2017

3·11·19 Posts

Quote:
 Originally Posted by Prime95 Code: Certifying proof for M97898863. Proof power = 8, TopK = 97899008, hash length = 64 Prover did 145 excess squarings Certifying a negative PRP test Root hash = 7F5AD60E97A24704452043D28C4EEB405094AC6F14C323BA8E0E7D847C3745A5 h0 = 5A532D61D8CA121A h1 = A8F263B98C58E6E0 h2 = B6E1C3415D782661 h3 = 17DC37D9709E9384 h4 = D4632194BADFADB2 h5 = 4CFD2CFB61D17C2E h6 = FCFEFF2A9D9A878D Random exponent = 254357798350921 Hash B = 73619C891332B067CBFC62A4EC5F18672950A398BC720AF48F0A8EFFC5980ABE Server certification cost 1395 squarings Hash A = FA2D8DB74D30BC7BD84F37C4F022B9023047B322C91C2C41EEC061C3C81D11BD Max roe during calculations: 0.149459 FAIL User certification cost 382418 squarings Preda -- please check the hashes. The failed verification took 3.6 hours on a single core of my laptop.
I'm not following the maths that goes into a proof so pinch of salt, but judging by the format page shouldn't topk be 382418 in this case?

2020-07-05, 10:50   #92
preda

"Mihai Preda"
Apr 2015

2×599 Posts

Quote:
 Originally Posted by M344587487 I'm not following the maths that goes into a proof so pinch of salt, but judging by the format page shouldn't topk be 382418 in this case?
topK is defined as the lowest multiple of 2^POWER (multiple of 256 in our case) that is larger than E, the mersenne exponent (which is 97898863 in our case).

 2020-07-05, 11:01 #93 M344587487     "Composite as Heck" Oct 2017 3×11×19 Posts Ah, when working out topk I missed out the last step of multiplying by 256. Reading is hard. For the header, does case matter? If it does the github page is in all caps but the files contain "PRP Proof". Does order of the fields matter?
2020-07-05, 11:39   #94
preda

"Mihai Preda"
Apr 2015

2×599 Posts

Quote:
 Originally Posted by M344587487 Ah, when working out topk I missed out the last step of multiplying by 256. Reading is hard. For the header, does case matter? If it does the github page is in all caps but the files contain "PRP Proof". Does order of the fields matter?
Thanks for reviewing! I'm fixing gpuowl's header to mach the case in the spec (all caps) -- I hope George doesn't mind that change.

About the order of the fields, I would say yes the order matters, just to keep it simple (also because GpuOwl, being lazy, only parses this particular order :)

2020-07-05, 13:11   #95
kriesel

"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

32×491 Posts

Quote:
 Originally Posted by preda Thanks for reviewing! I'm fixing gpuowl's header to match the case in the spec (all caps) -- I hope George doesn't mind that change. About the order of the fields, I would say yes the order matters, just to keep it simple (also because GpuOwl, being lazy, only parses this particular order :)
Examples are fine and useful, but do not a specification make.
Explicitly stating the header is all to be upper case would be good.
Also stating clearly that the format requires a particular order.
A little redundancy in a specification, and a specification description, are preferred over a little omission.
Hmm, no contents CRC at the end of the file or end of the header?
(The utility of a CRC would be detection of the occasional case, file mangled, no point in performing the verification computations.) An incomplete transfer could be more easily detected if the header is intact, by computing expected file size and comparing to actual file size. No point in attempting a verification if one of the required middle residues is missing or truncated. I feel both the server and client should make that size check. When the size is wrong, generate an error message and request retransmit.

Header fields are expressed in 8-bit ASCII, 1 character per byte. Some header records are variable in length.

All header fields shown in the example are required. Format and order must be as described. Except for the first header line, the structure is descriptor, "=", value, newline character (0x0A). Numerical values are expressed as unsigned base-ten integers in ASCII. <variable> below means insert value of variable. Alpha characters in header descriptors shall be upper case.

First header record is file type: PRP PROOF\n
Second header record is version: VERSION=<version>\n
Third header record is hashsize: HASHSIZE=<hashsize>\n
Fourth header record is power: POWER=<power>\n
Fifth header record is number: NUMBER=<numbertype><exponent>\n

Only hashsize=64 is currently supported.
Power 8, 9 or 10 are allowed values. Power 8 is preferred.
Only Mersenne type numbers are currently supported, denoted by numbertype="M"

Following the header's last \n, residues B and Middle(0), Middle(1),...,Middle(power-2),Middle(power-1) follow consecutively, in that order, without separator fields. Each residue is expressed as an unsigned binary multibyte integer, in least-significant-byte-first byte order, in a whole number of bytes, and its MSB is zero-extended. Byte count per residue is not present in the header, and is computed from the NUMBER field of the header. The residues' first byte position offsets are computable from header length, exponent, and which residue is sought.

Last fiddled with by kriesel on 2020-07-05 at 13:31

2020-07-05, 19:35   #96
kriesel

"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

114316 Posts

Quote:
 Originally Posted by preda For GpuOwl, the default proof power will be 8. This would result in temporary disk requirements of 3GB per 100M exponent. Users with plenty of disk would have an option to use power 9, doubling the temporary disk requirements.
Help output from gpuowl v6.11-335-gff60b08 says in part:
Code:
-proof [<power>]   : enable PRP proof generation. Default <power> is 9.

Last fiddled with by kriesel on 2020-07-05 at 19:35

2020-07-05, 21:57   #97
preda

"Mihai Preda"
Apr 2015

2·599 Posts

Quote:
 Originally Posted by kriesel Help output from gpuowl v6.11-335-gff60b08 says in part: Code: -proof [] : enable PRP proof generation. Default is 9.
Yep, fixed (changed default power to 8) now.

2020-07-06, 12:56   #98
kriesel

"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

10001010000112 Posts

Quote:
 Originally Posted by preda Yep, fixed (changed default power to 8) now.
Thanks.
I have a -proof 8 running on gpuowl-win v6.11-335-gff60b08 for M136000091 now which will take about 10 days to complete on a GTX1080 (with summer power limit set down to 90 watts). Gpuowl is creating a 17MB temporary file about every hour in that exponent's proof subdirectory. Here's what that looks like so far. topk = 136000256. topk/2^power = 531251. File names are successive multiples of that value.
Code:
 Directory of C:\Users\User\Documents\gpuowl-v6.11-335-gff60b08\136000091\proof

07/06/2020  07:45 AM    <DIR>          .
07/06/2020  07:45 AM    <DIR>          ..
07/05/2020  04:28 PM        17,000,016 1062502
07/05/2020  05:29 PM        17,000,016 1593753
07/05/2020  06:30 PM        17,000,016 2125004
07/05/2020  07:31 PM        17,000,016 2656255
07/05/2020  08:32 PM        17,000,016 3187506
07/05/2020  09:33 PM        17,000,016 3718757
07/05/2020  10:35 PM        17,000,016 4250008
07/05/2020  11:36 PM        17,000,016 4781259
07/05/2020  03:26 PM        17,000,016 531251
07/06/2020  12:37 AM        17,000,016 5312510
07/06/2020  01:38 AM        17,000,016 5843761
07/06/2020  02:39 AM        17,000,016 6375012
07/06/2020  03:40 AM        17,000,016 6906263
07/06/2020  04:42 AM        17,000,016 7437514
07/06/2020  05:43 AM        17,000,016 7968765
07/06/2020  06:44 AM        17,000,016 8500016
07/06/2020  07:45 AM        17,000,016 9031267
17 File(s)    289,000,272 bytes
2 Dir(s)  27,380,432,896 bytes free
When and how will PRP-proofs be reportable for manual assignments?

When does a test version of proof-enabled prime95 or mprime become available?

Last fiddled with by kriesel on 2020-07-06 at 13:48

2020-07-06, 16:18   #99
Prime95
P90 years forever!

Aug 2002
Yeehaw, FL

7,151 Posts

Quote:
 Originally Posted by kriesel When and how will PRP-proofs be reportable for manual assignments? When does a test version of proof-enabled prime95 or mprime become available?

We've just started brainstorming how to upload these files. Be prepared to archive your proofs locally for quite a while.

I'll roll out a proof-enabled mprime to my local machines later this week.

 Similar Threads Thread Thread Starter Forum Replies Last Post airsquirrels Hardware 313 2019-10-29 22:51 mathwiz GMP-ECM 0 2019-05-15 01:06 Jean Penné Software 0 2011-06-16 20:05 Jean Penné Software 6 2011-04-28 06:21 Jean Penné Software 4 2010-11-14 17:32

All times are UTC. The time now is 15:26.

Tue Sep 22 15:26:17 UTC 2020 up 12 days, 12:37, 1 user, load averages: 1.68, 1.77, 1.73

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.