Some Somewhat Easier n^i Sequences Available for Termination

[EdH]

Sorry for the confusion. I think you made your way through unraveling it. That was the list of sequences that had been run once on my machines.


347^63 was run once, twice, because after my first run, you released it again in post 927.


23^123 was from the index 1 thread. The list I posted isn't separated by projects at all.

[EdH]

I'm going to queue up 337^63 for an overnight single run when 367^65 moves into LA.

I'm going to look at 331^63 and 337^65 tomorrow and see if I want to run those. I need to set up a few that will go through Thursday without intervention, since I'll be busy then.

[gd_barnes]

Are you communicating correctly here? You're running a C161 on 367^65 before doing anything else? You said in the other thread that C149s were taking you ~5.5 hours. Wouldn't a C161 take you ~a day?

Regardless, if that is what you are doing, why are you running a sequence where the base has 2 remaining and the other sequence on the base has a C136? I'm just curious about your thinking here.

I feel like I'm missing something here. Could you perhaps be talking about the C136 on 367^63?

[gd_barnes]

1184^54 terminates

Base 1184 is complete!

I'm happy to be done with that one!

I think it's safe to say that this sequence likely has the project record for opposite-parity sequences that haven't merged. The factor 3 remained for 124 consecutive indexes!

I had hoped it would be the longest opposite-parity without merge sequence by total number of indexes. But it fell short. I've had at least one in the 190s. This one only made it to index 177.

[gd_barnes]

379^51, 53, 55, & 59 terminate

383^51, 53, 57, & 59 terminate

I'm still working on:
379^57 & 61
383^55, 61, & 65

[EdH]

Correct - there's no method to my madness. Yes, the c161 took 21.25 hours with 7.5 of that LA. During the LA, my second factoring host ran the c137 for 2 hours and all was quiet when I arose. I didn't put any time into developing the next set of candidates, expecting to do that today, and I had already gathered some work for the c161, so I let it run. I was also curious how the diminished sieving work force would affect the overall run, which turned out to indicate not much. I suppose that's because such a large portion is LA which didn't change.

ETA: As for the smaller 337^63 (c137), as mentioned, I quickly grabbed something to test my second host script. I'll be trying to figure out some things differently today, so something(s) can run through Thursday.

[EdH]

I've formed my plan for the next couple of days. I'm going to queue three sequences from this sub-project:

367^63: 159/136
331^63: 155/148
200560490130^14: 159/148

followed by the remaining c149s in the index 1 thread:

60^87: 156/149
84^83: 161/149
88^93: 181/149
90^93: 183/149
92^86: 169/149
120^73: 153/149
276^64: 157/149
780^60: 175/149
966^52: 156/149

I left 367^65 for you to ECM and see if it is of interest now. If not, I'll fit it in for another run at some convenient time.

[gd_barnes]

Great! Thanks for the explanations. It's interesting to hear about your process. 21.25 hours seem unusually fast even for you. I know you posted your various timings a few months ago but I'd have to dig awhile for the specific post. Have things been improved since then?

Before bed this morning, I saw that you had split 337^63 so I followed up on it. I ran it to t40 and nothing happened. So it's at 155/154/3/t40 and released. It doesn't look particularly good to me but I'll keep following up on whatever you split. I'll see what's up on 367^65 that I was harassing you about in a little while.

These exponent 63s in the 300s prime bases continue to be irritating. They aren't any good for the newest bases 379 and 383 either.

Edit: I found your timings from late July:
https://mersenneforum.org/showpost.p...&postcount=599

You were previously ~32 hours for a C160. That's a nice improvement!

[EdH]

I suppose there are lots of factors: quality of poly, time of day (how many machines for the sieving), etc., but here are the two c161s I have timing for:

Msieve Filtering: 00:51:22
Msieve Linear Algebra: 05:20:05
Msieve Square Root: 00:16:54
Total time: 22:39:46 (c161)

Msieve Filtering: 00:50:46
Msieve Linear Algebra: 07:22:30
Msieve Square Root: 00:21:03
Total time: 21:12:52 (c161)

The last one was over an hour faster, with less sievers, but it mostly ran during the day.

Actually, (now that you point it out) the change was probably getting adequate cooling for the GPU that runs LA, and that might explain why LA was longer for the last one. To cut down noise, I swapped back to inadequate cooling during the recent cooler weather. The hotter summertime may also influence the sieving machines. We'll see how they all perform in the coming cooler days.


ETA: I think the previous c161 timings was actually prior to me getting the GPU for LA working. That would explain a lot of the timing change.

[gd_barnes]

Very nice timings! Thanks for the detail. Meanwhile...

373^65 terminates

One remaining on base 373. Guess which exponent that is.

[EdH]

How many guesses am I allowed? I suppose I'll have to work that c152 in at some point. . .