processed dc and tc posts

[Uncwilly]

Originally Posted by kriesel

Substitute "at least 0.1% of completion" in place of the phrase highlighted in red.

You needed to quote the whole thing to change 1 clause?

[kriesel]

Originally Posted by Uncwilly

You needed to quote the whole thing to change 1 clause?

As a convenience to readers, all the gpu related information was preserved in one post. I suppose I could have done it not as a quote, but as an amended repost. Lacking moderator privileges here, I could not do as you did, modify post 5. I was not counting on a moderator to change post 5. Post 6 was intended for readers with gpus. I don't understand why you object to a full quote with color highlighting that makes it very easy to see where the difference lies.
(Do you prefer extremely terse expression of such, for example "@post 5, s/a couple dozen iterations/at least 0.1% of completion/", which makes finding the change location as inconvenient as practical?)

If conciseness is your goal, please remove posts 6 7 and 8 (this one), and perhaps also the color highlighting in 5.

[kriesel]

Originally Posted by GP2

PRP checks
...
(Type-1 residues, use mprime, don't use gpuOwL because most versions can't do Type-1)
...
Please use gpuOwL to test these

Gpuowl V0.7 to V3.9 does PRP residue type 1; Gpuowl V4.x to V6.5 does PRP residue type 4 (and in v4.x-5.0, has PRP-1 residue type 0 option). See https://www.mersenneforum.org/showpo...3&postcount=15 The switch to type 4 occurred when Preda was adding P-1 capability. I believe all PRP-capable gpuowl versions to date do offset zero only.

[kriesel]

I'm tackling Doublecheck=50878981,74,1 (LL in gpuowl v0.6 with Jacobi check)
Re my post 7, I'm not sure there was a gpuowl version designated v0.7 (or 0.8 or 0.9). I've confirmed a v1.0 was announced by Preda.
https://www.mersenneforum.org/showpo...&postcount=186

[kriesel]

Gpuowl V0.7 to 1.1 and perhaps later does PRP residue type 4; v1.5 & perhaps earlier to V3.9 does PRP residue type 1; Gpuowl V4.x to V6.5 does PRP residue type 4 (and in v4.x-5.0, has PRP-1 residue type 0 option). See https://www.mersenneforum.org/showpo...3&postcount=15 The switch to type 4 occurred when Preda was adding P-1 capability. I believe all PRP-capable gpuowl versions to date do offset zero only.

[moebius]

Originally Posted by kriesel

Gpuowl V0.7 to 1.1 and perhaps later does PRP residue type 4; v1.5 & perhaps earlier to V3.9 does PRP residue type 1; Gpuowl V4.x to V6.5 does PRP residue type 4 (and in v4.x-5.0, has PRP-1 residue type 0 option). See https://www.mersenneforum.org/showpo...3&postcount=15 The switch to type 4 occurred when Preda was adding P-1 capability. I believe all PRP-capable gpuowl versions to date do offset zero only.

It now seems to be again residue-type 1.

https://mersenneforum.org/showpost.p...postcount=1273

[kriesel]

Originally Posted by moebius

It now seems to be again residue-type 1.

Yes. Content at my link posted earlier was updated. The plan is to continue to update blog content in place in the future, so the links scattered about remain valid pointers to current and updated reference data. Feel free to PM me with any link for which the content becomes stale.

[Prime95]

Can someone please DC this one: M89657587
For some reason prime95 calculated an incorrect checksum in the results line. Very weird. I've got several more like it if this one fails DC.

[GP2]

Originally Posted by Prime95

Can someone please DC this one: M89657587
For some reason prime95 calculated an incorrect checksum in the results line.

Which version of prime95 returns a zero shift count??

[kriesel]

Originally Posted by Uncwilly

Can you post a single post that handles all of the data that is needed for the GPUs. And if you have some updates on GP2's post, can you spell it out all nice and neat? That way I can update the posts (or replace them with the new one.) That way we can keep the thread tidy. We can even break out my how-to info into the same post as the GPU how to. I tried to keep it short and not worry about covering -everything-, just enough for normal cases.

There's a revised somewhat streamlined version at https://www.mersenneforum.org/showpo...4&postcount=21. Just link to that. Something like:

For gpu runs, see https://www.mersenneforum.org/showpo...4&postcount=21 and follow links it contains as needed.

The gpu directions are longer than for the prime95/mprime case, partly because the gpu situation is more complicated than the prime95/mprime case. I may reduce its size by splitting off the optional extension portion to a separate linked post.

Feel free to paste the above quote line containing a link into your leading post in place of "If someone can post tips for the GPU's that would help." Then perhaps remove my posts in this thread about it.

It would be good also to modify GP2's post to specify selecting a gpuowl version matching the PRP residue type required, and indicate which residue type is needed. "V4.x or higher" probably gave residue type 4 at the time it was written, but that's no longer the case, because while earlier commits of v6.5 do type 4, v6.5-84-g30c0508 does type 1. You can see residue type switches a few times in gpuowl's history, as well as a switch from LL to PRP. (attachment at https://www.mersenneforum.org/showpo...3&postcount=15)

[Uncwilly]

Originally Posted by kriesel

It would be good also to modify GP2's post to specify selecting a gpuowl version matching the PRP residue type required, and indicate which residue type is needed. "V4.x or higher" probably gave residue type 4 at the time it was written, but that's no longer the case, because while earlier commits of v6.5 do type 4, v6.5-84-g30c0508 does type 1. You can see residue type switches a few times in gpuowl's history, as well as a switch from LL to PRP. (attachment at https://www.mersenneforum.org/showpo...3&postcount=15)

Can you PM GP2 about this and work it out? My current knowledge about residue types, etc. is nearly non-existent.