[QUOTE=VBCurtis;482548]IMO, setting a record is grounds for proceeding to test-sieve and GGNFS. I'll be stopping at 6M, and I don't think having multiple candidates over 1.45e-13 is in any way under-searched.

I'll have spent about 100 GPU hours, and I assume you two have spent more. 10 GPU-days is a bit light for this size, but Max did a CADO search too and we got great results.

I have a GNFS179 next in my personal queue, and had planned on 14 GPU days minimum (with more if I don't get a score I like). That's consistent with what we did here.[/QUOTE]

Agreed. The new record(s) give me confidence we’ve driven this poly as far as necessary.

What is your GNFS179? I’ll help drive it to ground.

C177
 

I spent 21.5 calendar hours on a 6-core CPU box before I noticed breaking 1.4 mark. Then through manual tweaking I was lucky to get it to 1.527. Now we have 4 candidates above 1.45 to test-sieve.
I'll help you with any of your projects, as usual.
My request, coming sooner or later, will be C200. Anybody is up for the challenge? I'll be looking for a 4-handle.

C177
 

I have about 44 GPU hours before moving the data files to a non-GPU box to finish up -nps & -npr. Add a couple hours for that and I'm at 46-48 hours into it.

The C205 from AS 276 is finishing up the last 100 curves (@ t70) at yoyo@Home.

[QUOTE=Max0526;482556]My request, coming sooner or later, will be C200. Anybody is up for the challenge? I'll be looking for a 4-handle.[/QUOTE]

Bring it on! Is that a sextic poly or still quintic? I’ve been wanting to search for a sextic. In the past I had problems running -npr on the .ms file but I think the default min_escore parameter was set too high.

Regardless of degree and difficulty, I’m down for some cycles.

[code]
C205 from AS 276
[/code]
Ditto. This demands what, 6+ months of GPU searching? Bring it!

C177 work
 

I’ve put ~120 GPU-hrs in so far, with another ~120 GPU-hrs to go. Total search time seems commensurate with a C177.

C205 is still 5th degree; 6th is possibly competitive around GNFS-210, though I think it's unlikely to win. CADO has better degree-6 code, from what I've read; I'll probly play a bit with CADO deg6 when this search gets going.

The C179 is 13*2^914-1:
[code]21738648296306845572021049564329346839114956866953308223382471881448692536700443342760950147619499684163176459093243365690755842713995905470870499048667020929598436481828027622759[/code]

I won't start poly select for a week or so on it. I'll be sieving on personal hardware, so I'll put a little extra effort into poly select- GPU time is cheap, compared to CPU cores. Once I put my two GPU-weeks into this poly, I'll be happy to join the C205 and later the C200 searches.

Max's magic seems to set a new score record every time we do a group search in this thread- he's single-handedly reducing these big projects by 5% of total computation!

[QUOTE=swellman;482561]I’ve put ~120 GPU-hrs in so far, with another ~120 GPU-hrs to go. Total search time seems commensurate with a C177.[/QUOTE]

I wonder if your binary is slower than the newest- RichD sent me a linux binary that does GPU searching almost 3 times as fast as the 1.52 pre-compiled binary I found publicly. With stage 1 norm at 4.7e(whatever is one less than default on this C177), I got through 5-6M in about 24 hours on a 750ti.

My GPU speed is related to the command line I use. My hardware is very old - a 660M - and too old to use higher cc. But if I use one script, it runs @1000/min, or 1M in ~17 hours. My “deeper” script takes 5x longer. But then the deeper script has found a couple of records...

Yes, I agree on Max and his wizardry. Impressive number of records shows working smarter > working harder, though a bit of luck can shift things too.

I’ll help with your C179 poly search once I finish up the C177 in a couple of days. Give me a range of 2M and I’ll run it down. The C205 and C200 come after that.

Sean-
The "M" GPU explains the time taken for your searches.

I ran 5-6.75M on the C177, turning up a 1.38 and nothing else better than 1.30. This completes my effort. Running 3-6.75M produced 32 -nps hits below 1e23.

I decided to delay a round of GPU-ECM in favor of poly select on 13*2^914-1. Reserving 0-1M.
Here's the candidate:
[code]21738648296306845572021049564329346839114956866953308223382471881448692536700443342760950147619499684163176459093243365690755842713995905470870499048667020929598436481828027622759[/code]

I'm using stage1norm=1.1e26, which is default / 9.

[QUOTE=VBCurtis;482644]Sean-
The "M" GPU explains the time taken for your searches.

I ran 5-6.75M on the C177, turning up a 1.38 and nothing else better than 1.30. This completes my effort. Running 3-6.75M produced 32 -nps hits below 1e23.[/QUOTE]

If you post your 1.38, I promise to do my very best to raise it above 1.40. And with enough luck there is a chance for swellman to test-sieve one more candidate.