Prepare the way! SRBase is coming for the next bit level.

[rebirther]

Originally Posted by chalsall

It's just TF'ing on low ranges, to get each 0.1M range below 2,000 unfactored candidates. Please see this thread where Wayne lays out the idea. Coordination happens there, so no "toes are stepped on".

I'm sure Wayne et al would appreciate the assistance. But please keep in mind that although the bit levels are low, the runtimes are high.

It looks like you want to cleanup 0-100M. If you could provide me a link as some times ago to get the assignments for GPU for each bitlevel choosen from the lowest available work that would be fine.

[chalsall]

Originally Posted by rebirther

It looks like you want to cleanup 0-100M.

Actually, specific sub-ranges below 49.7M.

Originally Posted by rebirther

If you could provide me a link as some times ago to get the assignments for GPU for each bitlevel choosen from the lowest available work that would be fine.

I would be happy to reconfigure the API I built for you a couple of years ago to feed you curated work (coordinating with others to avoid "toe-stepping"). This API has always been available, actually.

However, please understand what is involved here... Each WU will take much more time to complete than your TF'ers are currently used to (even though the bit-levels are much lower).

As to your specific request, there remain 228 candidates at only 65 "bits". But it can be argued that because of the ECM work that has already done, it doesn't really make sense to TF these.

But... I'm just a facilitator. I (usually) do what I'm told... 9-)

Thanks for considering assisting with this. Even having your workers doing the less expensive 70 and 71 to 72 bit work would help Waynes (let's face it, silly (but fun!)) project.

I would welcome Wayne's thoughts on this as well, since it's his sub-project.

[LaurV]

Originally Posted by chalsall

It's just TF'ing on low ranges, to get each 0.1M range below 2,000 unfactored candidates. Please see this thread where Wayne lays out the idea. Coordination happens there, so no "toes are stepped on".
I'm sure Wayne et al would appreciate the assistance. But please keep in mind that although the bit levels are low, the runtimes are high.

Reb has already been there, last time no longer than here, and it seems he was not satisfied by my reply two posts after...
(you mainly repeat what I told him about the assignment duration, etc).

[petrw1]

Originally Posted by chalsall

I would welcome Wayne's thoughts on this as well, since it's his sub-project.

Wow the help would be wonderful.

The most TF help required is actually the relatively lower bit levels.
Where this project is at now extensive TF is becoming less useful then extensive P1.
But a little more TF would be a great help.
And yes, LaurV had the right idea.

If reb (or anyone else for that matter) could take all the remaining 2x.xM ranges to 73 bits and the 1x.xM to 72 bits that would be huge. I don't think it would take them more than a couple months.
GPU72 can best coordinate this effort.
There are a few ranges currently spoken for but we can work that out.

After this much TF I suggest P1 could clear the majority of these ranges with a little exceptional TF that would be identified at a later time.

I don't know if the above arbitrary cut off from 73 to 72 bits is the correct one....maybe someone has a more educated idea.
Maybe it should be 3 bits above the yellow here: https://www.mersenne.ca/status/tf/0/0/3/2000.

Thanks again.

[masser]

Originally Posted by petrw1

I don't know if the above arbitrary cut off from 73 to 72 bits is the correct one....maybe someone has a more educated idea.
Maybe it should be 3 bits above the yellow here: https://www.mersenne.ca/status/tf/0/0/3/2000.

72 in the 1xM ranges and 73 in the 2xM ranges would be a very good start. After that dust clears, it might be easier to tell what else should be TF'd.

[chalsall]

Originally Posted by masser

72 in the 1xM ranges and 73 in the 2xM ranges would be a very good start. After that dust clears, it might be easier to tell what else should be TF'd.

OK... GPU72 is currently managing (but, importantly, not "holding" (read: they're not reserved from Primenet)) several ranges in 2xM. The specifics can be reviewed at this report shared earlier.

rebirther... What do you want to start with? Candidates going to 71, 72, or 73 "bits"?

I'll rejig your custom API script tomorrow (Saturday), and will PM you the details. I'll also bring in some additional ranges, so there's an appropriate buffer available.

Thanks for the help! Hopefully, your TF'ers won't be too annoyed by the longer run times. The credits will remain (approximately) the same, though.

[rebirther]

Originally Posted by chalsall

rebirther... What do you want to start with? Candidates going to 71, 72, or 73 "bits"?

I'll rejig your custom API script tomorrow (Saturday), and will PM you the details. I'll also bring in some additional ranges, so there's an appropriate buffer available.

Thanks for the help! Hopefully, your TF'ers won't be too annoyed by the longer run times. The credits will remain (approximately) the same, though.

yes 71-73 would be fine, as before max 50k per request and choosing bitlvl, looking forward :)

[chalsall]

Originally Posted by rebirther

yes 71-73 would be fine, as before max 50k per request and choosing bitlvl, looking forward :)

OK. Dataset configured. API IPC ready (I've sent you a PM with the URL).

You can have up to 200,000 assignments at any one time, but each API call is limited to 10,000 candidates. If you really want me to bump this up to 50,000, I can.

But... Again... Please understand the runtimes are going to be higher than what you're used to. You might want to start by going to 71, to begin with, to get a feel for the characteristics down here.

Thanks again for helping out. I'm sure Wayne et al will appreciate the help!

[Uncwilly]

Has the warm fuzzies.

[tuckerkao]

Originally Posted by rebirther

If I can see the first post Iam currently in this 73-74bit range.

It'll take around a full year before SRBase finish up all the way to M999M and come back down again for 2^74 to 2^75.

[chalsall]

Originally Posted by tuckerkao

It'll take around a full year before SRBase finish up all the way to M999M and come back down again for 2^74 to 2^75.

Unless other arrangements have been made...

Do please try to keep up.