mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Software (https://www.mersenneforum.org/forumdisplay.php?f=10)
-   -   Prime95 v30.3 (https://www.mersenneforum.org/showthread.php?t=25823)

kriesel 2020-09-14 13:15

[QUOTE=James Heinrich;556933]For [U]identical worktypes[/U] there is no speed increase. But by eliminating double-checks the overall GIMPS throughput is, while not quite double, it's maybe 70%(?) increased. But only when we get the bulk of users moved to v30 from v29 and older.

And for users like [I]Jwb52z[/I] (and myself) who like to do P-1, there is some slightly optimized bounds selection improvments in v30.[/QUOTE]
A rough estimate of traditional is:
2% TF
2% P-1
100% LL or PRP
100% LL or PRP DC
4% LL TC x 1/3 of exponents ~1.3%
total ~205% of a primality test per exponent

A rough estimate of new is (before adjustment downward of factoring effort to match reduced primality test effort)
2% TF
2% P-1
100% PRP
1% proof generation and verification assuming proof power 8
total ~105%

A rough estimate of reoptimized factoring effort reflecting reduced primality testing is 103% total.

So, about 2:1 effort reduction per exponent.

And the bonus of verification in hours, not years.

PRP/GEC shows up unreliable kit immediately, not up to 9 years later as in LL DC TC.

storm5510 2020-09-14 13:53

[QUOTE=axn;556867]I'm guessing [C]mprime[/C] doesn't have "execute" permission. I'm also guessing mprime's owner is not your login account.

If you do [C]ls -l[/C] and paste it here, we can confirm.

Use [C]sudo chown[/C] to change the ownership of the files in that folder. And [C]chmod[/C] to set the execute bit for [C]mprime[/C][/QUOTE]

[U]Thank you all for your replies.[/U]

I don't use it often enough to remember how, and I need to change that. I did some web searching, which I should have been before coming here.

[QUOTE]sudo chmod +x mprime
sudo chmod +rw on worktodo.txt and results.txt.
sudo chmod +r on the libraries.
[/QUOTE] I used [C]ls -l[/C] after every change to verify that is was correct. Afterwards [C]./mprime[/C] and everything worked properly.

I found a good reference site [URL="https://ss64.com/"]here[/URL]. :smile:

axn 2020-09-14 15:14

[QUOTE=storm5510;556946]Afterwards [C]./mprime[/C] and everything worked properly.[/QUOTE]

:tu:

storm5510 2020-09-14 15:20

[QUOTE=axn;556952]:tu:[/QUOTE]


I still have to use [C]sudo[/C]. If I don't, I get messages about some type of missing certificate. However, mprime will continue to run. Still learning. :smile:

Mark Rose 2020-09-14 15:51

[QUOTE=storm5510;556954]I still have to use [C]sudo[/C]. If I don't, I get messages about some type of missing certificate. However, mprime will continue to run. Still learning. :smile:[/QUOTE]

Could be that you're running a very out of date OS without updated TLS certificates used in the HTTPS connection to mersenne.org?

Prime95 2020-09-14 16:26

mprime does not use https

James Heinrich 2020-09-14 16:34

[QUOTE=storm5510;556954]I get messages about some type of missing certificate[/QUOTE]Maybe quoting the specific error message would provide more answers and less speculation.

storm5510 2020-09-14 16:49

[QUOTE=Mark Rose;556956]Could be that you're running a very out of date OS without updated TLS certificates used in the HTTPS connection to mersenne.org?[/QUOTE]

[I]Windows 10 v2004 Build 19041.508.
[/I]
Just in case you did not read any of the prior, this is [I]Ubuntu on Windows[/I]. It is a console window and nothing more. Providing a password using [C]sudo[/C] solves the problem.

[QUOTE]sudo ./mprime -d[/QUOTE]Perhaps this is the normal way of starting it on the first go-around after opening the console.
-
-

I stopped it, leaving the console window open, and restarted it without [C]sudo[/C]. The exact verbiage is "No certification assignment available." This relates to PRP. With [C]sudo[/C], it doesn't display this. This suggests [I]mprime[/I] is unable to read "prime.txt" and "local.txt" without me providing the password again. Simply, it falls back on it's default settings. I thought I had the permissions set right. Apparently not. Maybe I am expecting too much.

I don't mind using [C]sudo[/C] as long as everything works properly, which it does. Using [I]mprime[/I] is an excellent alternate for [I]Prime95[/I], for me...

kriesel 2020-09-14 19:24

Ubuntu version matters; WSL version matters. It's much more than just a console window on Win10. [url]https://docs.microsoft.com/en-us/windows/wsl/wsl2-index[/url]

kriesel 2020-09-14 21:00

[QUOTE=storm5510;556961]The exact verbiage is "No certification assignment available." This relates to PRP.[/QUOTE]Sometimes a PRP proof is available for verification, and sometimes not. Most of mine get done by Ben Delo.

storm5510 2020-09-14 22:59

[QUOTE=kriesel;556974]Ubuntu version matters; WSL version matters. It's much more than just a console window on Win10. [URL]https://docs.microsoft.com/en-us/windows/wsl/wsl2-index[/URL][/QUOTE]

Ubuntu 2004.2020.812.0. WSL2 requires a manual update, according to the link above, which I have not done. It was not my intent to degrade it. I can only go by what I see.

Xyzzy 2020-09-15 01:50

Feature request: A news button in the client. The user clicks it if there is "something new to read". All it does is opens the home page where the news is posted. The magic is that the Prime95 client checks for news occasionally and maybe puts an icon in the toolbar or something. In any case, this option is implemented in whatever way possible to be non-intrusive.

:mike:

Xyzzy 2020-09-15 01:55

(If we were in charge, we'd have a Twitter-like 144-ish character "news" line printed in the client log window at a user-defined number of iterations, or maybe at startup. With an undoc.txt opt-out switch.)

Viliam Furik 2020-09-15 15:07

[QUOTE=Xyzzy;556989]Feature request: A news button in the client. The user clicks it if there is "something new to read". All it does is opens the home page where the news is posted. The magic is that the Prime95 client checks for news occasionally and maybe puts an icon in the toolbar or something. In any case, this option is implemented in whatever way possible to be non-intrusive.

:mike:[/QUOTE]

:goodposting:

James Heinrich 2020-09-15 15:13

[QUOTE=Xyzzy;556990](If we were in charge, we'd have a Twitter-like 144-ish character "news" line printed in the client log window at a user-defined number of iterations, or maybe at startup. With an undoc.txt opt-out switch.)[/QUOTE]Parallel to this feature could be a version check. Perhaps if there is no recent news it could show [c]Current version: 30.3b6, your version: 24.5 (8.4 years old)[/c] or similar.

storm5510 2020-09-15 16:37

I will simply say that the time estimates in Test -> Status seem to be a lot more accurate than in the past. :smile:

Mark Rose 2020-09-15 20:49

Don't know if this is a bug or not, but I'm getting this from an EC2 instance with AVX-512 FFT length 2800K, Pass1=640, Pass2=4480, clm=1, 12 threads:

[quote]
[Worker #1 Sep 15 20:46] Iteration: 49610000 / 53591537 [92.57%], ms/iter: 1.127, ETA: 01:14:46
[Worker #1 Sep 15 20:46] Possible hardware errors have occurred during the test!
[Worker #1 Sep 15 20:46] 9 ROUNDOFF > 0.4 of which 7 were repeatable (not hardware errors).
[Worker #1 Sep 15 20:46] Confidence in final result is fair.
[/quote]

For what it's worth, mprime choose a length of 2880K for 53903081, so I suspect it may be being too aggressive with the FFT selection?

Prime95 2020-09-15 21:58

Possibly too aggressive. In results.txt, what are the 9 error messages?

Mark Rose 2020-09-16 00:38

[QUOTE=Prime95;557086]Possibly too aggressive. In results.txt, what are the 9 error messages?[/QUOTE]

[Tue Sep 15 03:18:13 2020]
Iteration: 614960/53591537, Possible error: round off (0.4305230135) > 0.42188
Continuing from last save file.
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.
Continuing from last save file.
[Tue Sep 15 04:59:01 2020]
Iteration: 5758319/53591537, Possible error: round off (0.4233788043) > 0.42188
Continuing from last save file.
[Tue Sep 15 05:13:30 2020]
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.
Continuing from last save file.
[Tue Sep 15 12:18:46 2020]
Iteration: 28314499/53591537, Possible error: round off (0.4425135837) > 0.42188
Continuing from last save file.
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.
Continuing from last save file.
[Tue Sep 15 13:40:49 2020]
Iteration: 32437359/53591537, Possible error: round off (0.4299476187) > 0.42188
Continuing from last save file.
[Tue Sep 15 14:07:06 2020]
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.
Continuing from last save file.
[Tue Sep 15 14:24:48 2020]
Iteration: 33363026/53591537, Possible error: round off (0.4268056902) > 0.42188
Continuing from last save file.
[Tue Sep 15 14:51:45 2020]
Iteration: 34251107/53591537, Possible error: round off (0.4265658891) > 0.42188
Continuing from last save file.
[Tue Sep 15 14:58:57 2020]
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.
Continuing from last save file.
Iteration: 34476765/53591537, Possible error: round off (0.4396423807) > 0.42188
Continuing from last save file.
[Tue Sep 15 15:08:02 2020]
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.
Continuing from last save file.
[Tue Sep 15 16:37:00 2020]
Iteration: 39188425/53591537, Possible error: round off (0.4290449117) > 0.42188
Continuing from last save file.
[Tue Sep 15 16:59:26 2020]
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.
Continuing from last save file.
[Tue Sep 15 19:06:40 2020]
Iteration: 45470192/53591537, Possible error: round off (0.4226116219) > 0.42188
Continuing from last save file.

For what it's worth, 53484341, 53512867, 53513183, 53539313, and 53565907 also produced reports about possible round off errors, but they each matched an existing result. Additionally, 53851783 and higher did not produce any errors. 53752613 is being run with a 2880K FFT and I suspect it will be fine.

53591537 did up [url=https://www.mersenne.org/report_exponent/?exp_lo=53591537&full=1]matching an existing result[/url], despite the high number of errors.

Prime95 2020-09-16 01:48

Those errors are "normal". The 0.4425135837 error does indicate roundoffs are getting pretty high. I'll change lower the AVX-512 crossover in 30.4 from 53620000 to 53490000.

Prime95 2020-09-16 02:05

I've made 30.3 build 6 the official download version at mersenne.org!

tanaydin 2020-09-16 08:56

Any release plan fro macOS
 
Hi there,

I'm so excited about new version, but still missing macOS version.
Do you have a plan for it?

S485122 2020-09-16 09:40

[QUOTE=Prime95;557095]Those errors are "normal". The 0.4425135837 error does indicate roundoffs are getting pretty high. I'll change lower the AVX-512 crossover in 30.4 from 53620000 to 53490000.[/QUOTE]To give some more data : i9-10920X with AVX-512F
and the following parameter in prime.txt : SoftCrossoverAdjust=-0.012

I got following errors (among runs without errors, the CPU although limited by me to 140 W instead of 165 W returns almost 2 DC results a day)[code]{"status":"C", "exponent":55190467, "worktype":"LL", "res64":"17DF95B829EA628B", "fft-length":2949120, "shift-count":9071492, "error-code":"01000100", "security-code":"6A9D12BA", "program":{"name":"Prime95", "version":"29.8", "build":6, "port":4}, "timestamp":"2020-05-20 22:21:53", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"4A9A8C34C8355B10C8AD93569EFC5E45"}
{"status":"C", "exponent":51473897, "worktype":"LL", "res64":"DBC33E34BE0E032C", "fft-length":2752512, "shift-count":24082686, "error-code":"01000100", "security-code":"F0D8AE05", "program":{"name":"Prime95", "version":"29.8", "build":6, "port":4}, "timestamp":"2020-06-04 08:04:32", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"00214F9A3E0FB26D5DE31C84CD796DC7"}
{"status":"C", "exponent":51461401, "worktype":"LL", "res64":"3B4A76AD7BBB77AB", "fft-length":2752512, "shift-count":3254699, "error-code":"01000100", "security-code":"F11D7C2F", "program":{"name":"Prime95", "version":"29.8", "build":6, "port":4}, "timestamp":"2020-06-08 06:04:17", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"4870B494ECD1B1A6682C9AB6FAF3DA1E"}
{"status":"C", "exponent":52364737, "worktype":"LL", "res64":"7795CD64AC219172", "fft-length":2809856, "shift-count":39243522, "error-code":"02000300", "security-code":"9693848C", "program":{"name":"Prime95", "version":"29.8", "build":6, "port":4}, "timestamp":"2020-06-27 22:57:04", "errors":{"Roundoff":3}, "user":"S485122", "computer":"i9-10920X", "aid":"3DBC1913AC2DE2977C78342D966A8ABA"}
{"status":"C", "exponent":53400911, "worktype":"LL", "res64":"2C6FB1EE1442C004", "fft-length":2867200, "shift-count":43429403, "error-code":"01000100", "security-code":"3CA93AF4", "program":{"name":"Prime95", "version":"30.3", "build":3, "port":4}, "timestamp":"2020-08-30 14:06:23", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"C849AC91C157F25903DF01FAB21CFF6B"}
{"status":"C", "exponent":53420317, "worktype":"LL", "res64":"17510756BF341103", "fft-length":2867200, "shift-count":34672085, "error-code":"01000200", "security-code":"0BDF5987", "program":{"name":"Prime95", "version":"30.3", "build":3, "port":4}, "timestamp":"2020-09-04 06:33:36", "errors":{"Roundoff":2}, "user":"S485122", "computer":"i9-10920X", "aid":"2C8BD32F8C5B13EBF5C5E1ED17EBA9EE"}
{"status":"C", "exponent":53421169, "worktype":"LL", "res64":"4F2AF1B315BEB07D", "fft-length":2867200, "shift-count":29613781, "error-code":"01000100", "security-code":"BF5D83E2", "program":{"name":"Prime95", "version":"30.3", "build":3, "port":4}, "timestamp":"2020-09-04 20:32:39", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"421E1259667AC0B12D88E5B5E5D18366"}
{"status":"C", "exponent":53440421, "worktype":"LL", "res64":"9E870DC045207123", "fft-length":2867200, "shift-count":27998026, "error-code":"01000100", "security-code":"14A0E14E", "program":{"name":"Prime95", "version":"30.3", "build":3, "port":4}, "timestamp":"2020-09-08 20:44:57", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"D5AE4F687E540B76FB7F0FABF4D8C5AB"}
{"status":"C", "exponent":53452613, "worktype":"LL", "res64":"D57B5329BE9E49FA", "fft-length":2867200, "shift-count":52353048, "error-code":"01000100", "security-code":"C783F209", "program":{"name":"Prime95", "version":"30.3", "build":3, "port":4}, "timestamp":"2020-09-10 22:14:30", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"0F02154927C635FB72ED727BE6349256"}
{"status":"C", "exponent":53454073, "worktype":"LL", "res64":"783100F91A32C303", "fft-length":2867200, "shift-count":31074045, "error-code":"02000200", "security-code":"BACD6363", "program":{"name":"Prime95", "version":"30.3", "build":6, "port":4}, "timestamp":"2020-09-11 12:35:10", "errors":{"Roundoff":2}, "user":"S485122", "computer":"i9-10920X", "aid":"472C6B93E4DD376691E09C606371E4F7"}[/code]After 2020-09-11 10:30 I removed the SoftCrossoverAdjust=-0.012 parameter to replace it with : ExtraSafetymargin=0.02 but still got the following errors :[code]{"status":"C", "exponent":53450387, "worktype":"LL", "res64":"3FF78DA89C649E2B", "fft-length":2867200, "shift-count":24602115, "error-code":"01000100", "security-code":"9084119F", "program":{"name":"Prime95", "version":"30.3", "build":6, "port":4}, "timestamp":"2020-09-12 02:41:26", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"22BA5D1071303EE4A14332B1BA271A7D"}
{"status":"C", "exponent":53481889, "worktype":"LL", "res64":"91BCE109FFD2E974", "fft-length":2867200, "shift-count":29308399, "error-code":"01000100", "security-code":"5C42F4E1", "program":{"name":"Prime95", "version":"30.3", "build":6, "port":4}, "timestamp":"2020-09-13 06:31:56", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"B1CAE2FDEF46A5D9C5FCAA8DD27D9C1D"}
{"status":"C", "exponent":53494141, "worktype":"LL", "res64":"193A0B5089007F76", "fft-length":2867200, "shift-count":40860647, "error-code":"01000100", "security-code":"4C392479", "program":{"name":"Prime95", "version":"30.3", "build":6, "port":4}, "timestamp":"2020-09-14 10:29:50", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"DA2D6BD6F611EC3F51B2F3BC238F86A8"}
{"status":"C", "exponent":53494739, "worktype":"LL", "res64":"7CBE9C72B81E2808", "fft-length":2867200, "shift-count":33455163, "error-code":"01000100", "security-code":"D9E46935", "program":{"name":"Prime95", "version":"30.3", "build":6, "port":4}, "timestamp":"2020-09-15 00:44:22", "errors":{"Roundoff":1}, "user":"S485122", "computer":"i9-10920X", "aid":"10804C8ED603CB3012622D1BC466ADD5"}[/code]

Jacob

(And the manual assignments pages should NOT give out exponents that are in the double check CAT 0 and CAT 1categories [noparse];-)[/noparse]

pmart 2020-09-16 12:44

Hi,

I think [URL="https://www.mersenne.org/download/"]download[/URL] section currently lists the wrong checksum

[CODE]
$ sha256sum p95v303b6.linux64.tar.gz
ee54b56062feb05c9f80963a4e3ae8555d0e59ca60ddbcba65ce05225c9b9a79 p95v303b6.linux64.tar.gz
[/CODE]

storm5510 2020-09-16 14:37

I am having some difficulties with [I]mprime (30.3 B6)[/I] and I am fairly certain this is the result of the way I am stopping it. Using [C]-m[/C] on the command line brings up the menu at startup. Once I allow it to run, it doesn't respond to anything other than [C]Ctrl-C[/C]. I looked through the documents in the archive. Beside the reference to the command for startup, I didn't see anything about a command to properly stop it or bring the menu up. Using [C]Ctrl-C[/C], has resulted in emptied [I]worktodo[/I] files, and hidden results files which are sometimes deleted. Trying other keys and combinations also had no affect. There must be a way...

Prime95 2020-09-16 15:49

[QUOTE=tanaydin;557117]Hi there,

I'm so excited about new version, but still missing macOS version.
Do you have a plan for it?[/QUOTE]

My Mac died. I might can steal some CPU cycles from my wife's Mac. Would a GUI-less version suffice?

DrobinsonPE 2020-09-16 18:15

[QUOTE=storm5510;557134]I am having some difficulties with [I]mprime (30.3 B6)[/I] and I am fairly certain this is the result of the way I am stopping it. Using [C]-m[/C] on the command line brings up the menu at startup. Once I allow it to run, it doesn't respond to anything other than [C]Ctrl-C[/C]. I looked through the documents in the archive. Beside the reference to the command for startup, I didn't see anything about a command to properly stop it or bring the menu up. Using [C]Ctrl-C[/C], has resulted in emptied [I]worktodo[/I] files, and hidden results files which are sometimes deleted. Trying other keys and combinations also had no affect. There must be a way...[/QUOTE]

I have not downloaded 30.3 B6 but am using 30.3 B4 and B5 on multiple computers. The menu includes a number at the start of each line. I have always just typed in the number and hit enter.

Enter - brings up the menu.
4 Enter - starts and stops the workers.
5 Enter - quits mprime.

Those are what I use the most.

Mark Rose 2020-09-16 18:20

[QUOTE=storm5510;557134]I am having some difficulties with [I]mprime (30.3 B6)[/I] and I am fairly certain this is the result of the way I am stopping it. Using [C]-m[/C] on the command line brings up the menu at startup. Once I allow it to run, it doesn't respond to anything other than [C]Ctrl-C[/C]. I looked through the documents in the archive. Beside the reference to the command for startup, I didn't see anything about a command to properly stop it or bring the menu up. Using [C]Ctrl-C[/C], has resulted in emptied [I]worktodo[/I] files, and hidden results files which are sometimes deleted. Trying other keys and combinations also had no affect. There must be a way...[/QUOTE]

You just want to run it, I use `./mprime -d`. Then ctrl-c will cause it to quit.

jwnutter 2020-09-16 18:28

[QUOTE=tanaydin;557117]Hi there,

I'm so excited about new version, but still missing macOS version.
Do you have a plan for it?[/QUOTE]

I second this request for a macOS version of the new app. Two of my clients currently run macOS. These are older macs that generally don’t run anything other than p95.

Prime95 2020-09-16 19:31

[QUOTE=tanaydin;557117]
I'm so excited about new version, but still missing macOS version.
Do you have a plan for it?[/QUOTE]

[QUOTE=jwnutter;557152]I second this request for a macOS version of the new app. Two of my clients currently run macOS. These are older macs that generally don’t run anything other than p95.[/QUOTE]

Please try this:

[url]https://www.dropbox.com/s/xo7psmmjfkb52pr/p95v303b6.MacOSX.tar.gz?dl=0[/url]

No GUI. Not tested much. If it works, I'll make it the official Mac release

jwnutter 2020-09-16 23:23

[QUOTE=Prime95;557159]Please try this:

[url]https://www.dropbox.com/s/xo7psmmjfkb52pr/p95v303b6.MacOSX.tar.gz?dl=0[/url]

No GUI. Not tested much. If it works, I'll make it the official Mac release[/QUOTE]

Thanks, I will wait for the current assignments to complete - which will take a few weeks - and then attempt the to transition these systems.

kriesel 2020-09-17 00:43

[QUOTE=jwnutter;557171]Thanks, I will wait for the current assignments to complete - which will take a few weeks - and then attempt the to transition these systems.[/QUOTE]
Consider trying out the new version in a separate test folder, soon, on something quick; P-1, prp-cf. Then if there's an issue George can have time to look into it.

jwnutter 2020-09-17 00:47

[QUOTE=kriesel;557175]Consider trying out the new version in a separate test folder, soon, on something quick; P-1, prp-cf. Then if there's an issue George can have time to look into it.[/QUOTE]


:tu:

jwnutter 2020-09-17 15:04

I now have some P-1 work running on a macOS catalina (10.15) box and DC work running on a MacOS high sierra (10.13) box.

I'll let the P-1 work run for a few days, but the DC work will take about 10 days to complete.

The only problem I've noticed so far is that I asked the high sierra machine to request P-1 work (option 4), but it received DC work (option 151). Interestingly, the DC work type is the 4th work type in the list. That said, I also asked the catalina box to request work type 4 and all seems to have gone well here.

In any event, I'm not sure if this is a bug or more of a missunderstanding on my part. If it is a bug, I'm not sure that it needs to be fixed as I highly doubt there are many p95 users still running high sierra.

jwnutter 2020-09-19 14:00

[QUOTE=Prime95;557159]Please try this:

[url]https://www.dropbox.com/s/xo7psmmjfkb52pr/p95v303b6.MacOSX.tar.gz?dl=0[/url]

No GUI. Not tested much. If it works, I'll make it the official Mac release[/QUOTE]

Here's what I can report back after a few days of testing:

- The macOS high sierra (10.13) box was assigned DC work and crashed at some point along the way. Here's what was shown in terminal.

[CODE][Work thread Sep 17 21:32] Iteration: 3090000 / 57755491 [5.35%], ms/iter: 15.852, ETA: 10d 00:42
[Work thread Sep 17 21:34] Iteration: 3100000 / 57755491 [5.36%], ms/iter: 13.721, ETA: 8d 16:18
[Work thread Sep 17 21:37] Iteration: 3110000 / 57755491 [5.38%], ms/iter: 13.724, ETA: 8d 16:19
[Work thread Sep 17 21:39] Iteration: 3120000 / 57755491 [5.40%], ms/iter: 13.703, ETA: 8d 15:58
[Work thread Sep 17 21:41] Running Jacobi error check. dyld: lazy symbol binding failed: Symbol not found: ____chkstk_darwin
Referenced from: XXXXXXX
Expected in: /usr/lib/libSystem.B.dylib

dyld: Symbol not found: ____chkstk_darwin
Referenced from: XXXXXXX
Expected in: /usr/lib/libSystem.B.dylib

Abort trap: 6
logout
Saving session...
...copying shared history...
...saving history...truncating history files...
...completed.
Deleting expired sessions...13 completed.

[Process completed][/CODE]

I have retreated back to p95 v29.4,build 7 on this machine. As I had mentioned before, I highly doubt many (I would be shocked to learn that the number is greater than 1 / 1000) p95 users are running macOS high sierra (10.13).

- The macOS catalina (10.15) box has completed a P-1 assignment ([URL="https://www.mersenne.org/report_exponent/?exp_lo=100156723&full=1"]100156723[/URL]) and is now working on an ECM assignment. This box either crashed or installed an OS update at some point during the first P-1 assignment, but in all fairness an OS crash isn't out of the ordinary for this machine.

On both boxes I'm having some difficulty selecting the preferred work type, but I believe these issues are hardware related (i.e. not enough ram for some assignment types).

Please let me know if you'd like to see the results of any specific test cases.

James Heinrich 2020-09-20 19:53

I have 3 workers, #1 is doing a PRP that will be done in about a month, #2 and #3 are doing small P-1 that take about an hour each. Whenever the client gets Cert work it is inserted into only worker #1, never #2 or #3. In [i]undoc.txt[/i] there is reference to [c]CertWorker=n[/c] to force cert work to a specific single worker, I think it would be better if this could take a [i]list[/i] of worker(s) for which Cert work should be assigned (if the setting is absent then it is allowed on any/all workers).

retina 2020-09-21 02:20

I have no problem doing cert verification work. But I am unhappy that it interrupts current work in progress.

Can there be, or is there, a setting to only start cert verifications when a previous assignment has finished? IOW tell it not to interrupt anything.

Prime95 2020-09-21 05:12

[QUOTE=retina;557448]I have no problem doing cert verification work. But I am unhappy that it interrupts current work in progress.

Can there be, or is there, a setting to only start cert verifications when a previous assignment has finished? IOW tell it not to interrupt anything.[/QUOTE]

There is no such setting. Just disable getting CERT work. There is no problem getting CERT work completed at this time.

Flaukrotist 2020-09-21 19:34

Download rate not sufficient with cloud drive
 
After changing to prime95 v30.3 b6, I was eager to perform my first CERT check only to see it failing to download over and over again despite having thought to have a quite good download rate.

[CODE][Sun Sep 20 12:52:47 2020 - ver 30.3]
Server assigned CERT work.
Got assignment xxx: CERT Mxxx
[Sun Sep 20 14:00:02 2020 - ver 30.3]
CURL library error: Operation timed out after 180001 milliseconds with 10668203 out of 12482167 bytes received
[Sun Sep 20 15:00:02 2020 - ver 30.3]
CURL library error: Operation timed out after 180020 milliseconds with 11127045 out of 12482167 bytes received
[Sun Sep 20 16:00:01 2020 - ver 30.3]
CURL library error: Operation timed out after 180008 milliseconds with 11213237 out of 12482167 bytes received
[Sun Sep 20 17:00:01 2020 - ver 30.3]
CURL library error: Operation timed out after 180011 milliseconds with 10997930 out of 12482167 bytes received
[Sun Sep 20 18:00:02 2020 - ver 30.3]
CURL library error: Operation timed out after 180013 milliseconds with 10763795 out of 12482167 bytes received
[Sun Sep 20 19:00:02 2020 - ver 30.3]
CURL library error: Operation timed out after 180016 milliseconds with 11735694 out of 12482167 bytes received
[Sun Sep 20 20:00:00 2020 - ver 30.3]
CURL library error: Operation timed out after 180005 milliseconds with 11807160 out of 12482167 bytes received
[Sun Sep 20 21:00:02 2020 - ver 30.3]
CURL library error: Operation timed out after 180002 milliseconds with 11605479 out of 12482167 bytes received
Unreserving assignment xxx[/CODE]

I think, the explanation is me being lazy and putting prime95 into a cloud drive directory. When it started downloading, it simultaneously uploaded the new files to the cloud and thus limited the download bandwidth drastically by the upload.

In fact, the solution was easy: put prime95 into a non-cloud directory and everything works fine.:brian-e::blush:

kriesel 2020-09-21 23:29

[QUOTE=Prime95;557454]There is no such setting. Just disable getting CERT work. There is no problem getting CERT work completed at this time.[/QUOTE]Ben Delo is taking cert work. QED.

tanaydin 2020-09-22 12:47

[QUOTE=Prime95;557159]Please try this:

[url]https://www.dropbox.com/s/xo7psmmjfkb52pr/p95v303b6.MacOSX.tar.gz?dl=0[/url]

No GUI. Not tested much. If it works, I'll make it the official Mac release[/QUOTE]

thank you very much, I'll run it a couple of days and share the results.

ixfd64 2020-09-22 17:32

Is there a way to increase the timeout for downloading files?

tanaydin 2020-09-22 22:53

Hi, I've two mac computers,

application works on Catalina (10.15.6)
Darwin tanaydins-MBP-4 19.6.0 Darwin Kernel Version 19.6.0: Thu Jun 18 20:49:00 PDT 2020; root:xnu-6153.141.1~1/RELEASE_X86_64 x86_64

but crashes on

High Sierra (10.13.6)
Darwin Tanaydins-iMac 17.7.0 Darwin Kernel Version 17.7.0: Thu Jun 18 21:21:34 PDT 2020; root:xnu-4570.71.82.5~1/RELEASE_X86_64 x86_64

I got this as error.


[Worker #2 Sep 23 00:46] Running Jacobi error check. dyld: lazy symbol binding failed: Symbol not found: ____chkstk_darwin
Referenced from: /Applications/Prime95/mprime
Expected in: /usr/lib/libSystem.B.dylib

dyld: Symbol not found: ____chkstk_darwin
Referenced from: /Applications/Prime95/mprime
Expected in: /usr/lib/libSystem.B.dylib

[1] 63331 abort /Applications/Prime95/mprime -d -m -w/Users/tanaydin/Prime95

MagnusLarsson 2020-09-23 06:24

CERT result "n/a"
 
Hello GIMPS experts,

I have upgraded to p95v303b6.linux64 on Linux Debian Testing (Bullseye), from previous version p95v298b6.linux64.

uname -a
Linux debian 5.8.0-1-amd64 #1 SMP Debian 5.8.7-1 (2020-09-05) x86_64 GNU/Linux

Issue:
I get "n/a" in the Result column even if the p95v303b6 log file states success, for work type CERT. I have received 2 CERT jobs and both became n/a.
The [URL]https://www.mersenne.org/results/[/URL] table and the log file is copied below.
I expected "success", "verified" or similar.

*** Result table ***
KML_Linux_Debian[URL="https://www.mersenne.org/report_exponent/?exp_lo=99770291&full=1"] 99770291[/URL] CERT 2020-09-22 21:180.1 n/a 1.4864
KML_Linux_Debian[URL="https://www.mersenne.org/report_exponent/?exp_lo=99773281&full=1"] 99773281[/URL] CERT 2020-09-23 05:180.1 n/a 1.4864

*** Log file ***
[Work thread Sep 22 23:18] M99770291 certification hash value 972E9D617C74E4E661A81F3E3926A00E6AB129EE4C6EEF6F2854876C3AE6175A. Wh8: 139439D8,63264065,00000000
[Comm thread Sep 22 23:18] Sending result to server: UID: Karl_Magnus_Larsson/KML_Linux_Debian, M99770291 certification hash value 972E9D617C74E4E661A81F3E3926A00E6AB129EE4C6EEF6F2854876C3AE6175A. Wh8: 139439D8,63264065,00000000, AID: D7A049D92268242724
[Comm thread Sep 22 23:18] URL: http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=ar&g=aca4a9d7eea5a9a9ad61d8a6f228a062&k=D7A049D9226824272443AE0C7CC4871E&m={"worktype":"Cert",+"exponent":99770291,+"sha3-hash":"972E9D617C74E4E661A81F3E3926A00E6AB129EE4C6EEF6F2854876C3AE6175A",+"fft-length":5505024,+"shift-count":63264065,+"error-code":"00000000",+"security-code":"139439D8",+"program":{"name":"Prime95",+"version":"30.3",+"build":6,+"port":8},+"timestamp":"2020-09-22+21:18:08",+"user":"Karl_Magnus_Larsson",+"computer":"KML_Linux_Debian",+"aid":"D7A049D9226824272443AE0C7CC4871E"}%0A&r=200&d=1&A=1&b=2&n=99770291&c=-1&s3=972E9D617C74E4E661A81F3E3926A00E6AB129EE4C6EEF6F2854876C3AE6175A&ec=00000000&sc=63264065&fftlen=5505024&ss=45172&sh=B3158D4FB38261142C221B09EA6AE613
[Comm thread Sep 22 23:18] RESPONSE:
[Comm thread Sep 22 23:18] pnErrorResult=0
[Comm thread Sep 22 23:18] pnErrorDetail=Certification successfully completes double-check of M99770291 --
[Comm thread Sep 22 23:18] CPU credit is 1.4864 GHz-days.
[Comm thread Sep 22 23:18] ==END==

<snip>

Work thread Sep 23 07:18] M99773281 certification hash value A718DA9DA72ED427236B92C522A9190F3317D993C035816DD4F5D7B09FBEB05D. Wh8: 2AF05134,2663376,00000000
[Comm thread Sep 23 07:18] Sending result to server: UID: Karl_Magnus_Larsson/KML_Linux_Debian, M99773281 certification hash value A718DA9DA72ED427236B92C522A9190F3317D993C035816DD4F5D7B09FBEB05D. Wh8: 2AF05134,2663376,00000000, AID: 970A3F374D319DE9798
[Comm thread Sep 23 07:18] URL: http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=ar&g=aca4a9d7eea5a9a9ad61d8a6f228a062&k=970A3F374D319DE979895201A75B994A&m={"worktype":"Cert",+"exponent":99773281,+"sha3-hash":"A718DA9DA72ED427236B92C522A9190F3317D993C035816DD4F5D7B09FBEB05D",+"fft-length":5505024,+"shift-count":2663376,+"error-code":"00000000",+"security-code":"2AF05134",+"program":{"name":"Prime95",+"version":"30.3",+"build":6,+"port":8},+"timestamp":"2020-09-23+05:18:10",+"user":"Karl_Magnus_Larsson",+"computer":"KML_Linux_Debian",+"aid":"970A3F374D319DE979895201A75B994A"}%0A&r=200&d=1&A=1&b=2&n=99773281&c=-1&s3=A718DA9DA72ED427236B92C522A9190F3317D993C035816DD4F5D7B09FBEB05D&ec=00000000&sc=2663376&fftlen=5505024&ss=10790&sh=D56313BF8D4A13DD39A9A667399519D5
[Comm thread Sep 23 07:18] RESPONSE:
[Comm thread Sep 23 07:18] pnErrorResult=0
[Comm thread Sep 23 07:18] pnErrorDetail=Certification successfully completes double-check of M99773281 --
[Comm thread Sep 23 07:18] CPU credit is 1.4864 GHz-days.
[Comm thread Sep 23 07:18] ==END==


Best regards,


Magnus Larsson

James Heinrich 2020-09-23 12:19

[QUOTE=MagnusLarsson;557622]I get "n/a" in the Result column even if the p95v303b6 log file states success, for work type CERT. I have received 2 CERT jobs and both became n/a.[/QUOTE]This is normal. It still needs fixing, but it's a website display issue and not a problem on your end. Using [M]M99,770,291[/M] as an example, your Cert verifies the PRP run by Ben Delo, therefore the PRP section shows "verified".

The "Cert... n/a" is a known display issue. It's not a trivial fix, but Aaron is aware of it and will (when time permits) work some magic to relate the Cert to the PRP and make it display nicely.

ATH 2020-09-23 12:22

[QUOTE=MagnusLarsson;557622]Issue:
I get "n/a" in the Result column even if the p95v303b6 log file states success, for work type CERT. I have received 2 CERT jobs and both became n/a.
The [URL]https://www.mersenne.org/results/[/URL] table and the log file is copied below.
I expected "success", "verified" or similar.
[/QUOTE]

The PRP test itself changed to status "Verified" which means your CERT was ok.
This is all still very new, they did not implement a status for the CERT on the server.

MagnusLarsson 2020-09-23 18:36

@James Heinrich, @ATH,


Aha, OK, I see. Thank you for the clarification. I did not pay attention to the status change. I see the same thing, both [URL="https://www.mersenne.org/report_exponent/?exp_lo=99770291&full=1"]99770291[/URL] and [URL="https://www.mersenne.org/report_exponent/?exp_lo=99773281&full=1"]99773281[/URL] show "verified" in the PRP status field.

jwnutter 2020-09-24 03:00

[QUOTE=tanaydin;557601]Hi, I've two mac computers,

application works on Catalina (10.15.6)
Darwin tanaydins-MBP-4 19.6.0 Darwin Kernel Version 19.6.0: Thu Jun 18 20:49:00 PDT 2020; root:xnu-6153.141.1~1/RELEASE_X86_64 x86_64

but crashes on

High Sierra (10.13.6)
Darwin Tanaydins-iMac 17.7.0 Darwin Kernel Version 17.7.0: Thu Jun 18 21:21:34 PDT 2020; root:xnu-4570.71.82.5~1/RELEASE_X86_64 x86_64

I got this as error.


[Worker #2 Sep 23 00:46] Running Jacobi error check. dyld: lazy symbol binding failed: Symbol not found: ____chkstk_darwin
Referenced from: /Applications/Prime95/mprime
Expected in: /usr/lib/libSystem.B.dylib

dyld: Symbol not found: ____chkstk_darwin
Referenced from: /Applications/Prime95/mprime
Expected in: /usr/lib/libSystem.B.dylib

[1] 63331 abort /Applications/Prime95/mprime -d -m -w/Users/tanaydin/Prime95[/QUOTE]

I stand corrected on my earlier assessment.

[URL="https://www.mersenneforum.org/showpost.php?p=557357&postcount=326"]Post # 326[/URL]

jwnutter 2020-09-25 21:06

Wanted to report back that I've had no other issues with the Mac version of P95 v30.3 - with the exception of the error (lazy symbol binding failed: Symbol not found: ____chkstk_darwin) reported by tanaydin and myself regarding a MacOS High Sierra (10.13) instance.

ixfd64 2020-09-26 00:40

I noticed that the latest version still uses the older "Wh4" identifier in the [C]results.txt[/C] file. Is this intentional?

Prime95 2020-09-26 02:10

[QUOTE=ixfd64;557918]I noticed that the latest version still uses the older "Wh4" identifier in the [C]results.txt[/C] file. Is this intentional?[/QUOTE]

The Wh4 code is not really used anymore. Over time it loosely relates to gwnum FFT library version number. The middle letter gets bumped when there are major changes to the FFT code.

ZFR 2020-09-28 13:59

What's the best way of upgrading from 29.8 to 30.3 of mprime on Linux, so as not to lose existing work? Just unpack everything into mprime, and overwrite any existing files?

Prime95 2020-09-28 14:52

[QUOTE=ZFR;558102]What's the best way of upgrading from 29.8 to 30.3 of mprime on Linux, so as not to lose existing work? Just unpack everything into mprime, and overwrite any existing files?[/QUOTE]

That should work. Make a backup if you're leery (or prudent).

ZFR 2020-09-28 23:22

[QUOTE=Prime95;558119]That should work. Make a backup if you're leery (or prudent).[/QUOTE]

Thank you, Sir. Looks like everything went well.

ZFR 2020-09-28 23:30

Actually, I'm getting this when I select 2. Test / Worker Threads

[CODE]
Number of workers to run (4):

Options for worker #1

Type of work to get (0):
CPU cores to use (multithreading): 2

Options for worker #2

Type of work to get (0):
CPU cores to use (multithreading): 2

Options for worker #3

Type of work to get (0):
CPU cores to use (multithreading): 2

Options for worker #4

Type of work to get (0):
CPU cores to use (multithreading): 2
Get occasional proof certification work (Y):

Accept the answers above? (Y):
You have allocated more cores than are available. This is likely to
GREATLY REDUCE performance. Do you want to correct this? (Y):
[/CODE]


My CPU should have 2 threads per 4 cores

Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
CPU speed: 4000.04 MHz, 4 hyperthreaded cores
CPU features: Prefetch, SSE, SSE2, SSE4, AVX, AVX2, FMA
L1 cache size: 4x32 KB, L2 cache size: 4x256 KB, L3 cache size: 8 MB

ZFR 2020-09-29 00:14

I'm getting the "You have allocated more cores than are available." message even when I run 29.8 that I've backed up. So it looks like it's not a v30.3 issue. Still, I wonder why I get this. I'm pretty sure I haven't come across this message when I first set up mprime.

James Heinrich 2020-09-29 00:21

Prime95 has explicit GUI settings that show that hyperthreading is specifically not-recommended (except in the case of TF which itself is not-recommended). Just allocate 4 workers with 1 core per worker. 2 hyperthreads per worker may actually produce less throughput than a single worker per real core.

jwnutter 2020-09-29 00:28

[QUOTE=James Heinrich;558173]Prime95 has explicit GUI settings that show that hyperthreading is specifically not-recommended (except in the case of TF which itself is not-recommended). Just allocate 4 workers with 1 core per worker. 2 hyperthreads per worker may actually produce less throughput than a single worker per real core.[/QUOTE]

@ZFR - Here's an example from a recent ~24 hour benchmark I completed on a i9-9900KF. That said, there are many other benchmarks in this same thread and you might consider benchmarking your system assuming you haven't already.

[url]https://www.mersenneforum.org/showpost.php?p=556817&postcount=821[/url]

ZFR 2020-09-29 00:53

I'm using mprime, so no GUI... But I've moved this from Prime95 when I switched from Windows to Linux. I must have missed the warning because as far as I remember I've been using 4 worker threads ever since I got this PC 5 years ago...

So what's the best way to change this to 1 worker without losing any progress. Can I just change to 1 worker in Settings in the menu and then manually change the worktodo file?

James Heinrich 2020-09-29 00:55

You can still keep 4 workers, that's not a problem. Just use one thread per worker, not 2.[quote][COLOR="SeaGreen"]Number of workers to run (4):[/COLOR]
[COLOR="Orange"][STRIKE]CPU cores to use (multithreading): 2[/STRIKE][/COLOR]
[COLOR="SeaGreen"]CPU cores to use (multithreading): [B]1[/B][/COLOR][/quote]

ZFR 2020-09-29 01:00

[QUOTE=James Heinrich;558178]You can still keep 4 workers, that's not a problem. Just use one thread per worker, not 2.[/QUOTE]

OK, I see.

When I run the the mprime menu "2. Test/Worker threads" it doesn't give me the option to use 1 thread per worker. I only get to choose "Number of Workers to run" and if I choose 4, then for "CPU cores to use (multithreading)": 2 gets selected automatically.

Prime95 2020-09-29 01:11

Running 4 workers is fine. In Test/Workers, have each worker use just one CPU core.

Note that running 4 PRP tests at a time will quadruple the amount of temporary disk space the program uses. At default settings, you are letting mprime use up to 24GB of disk space.

ZFR 2020-09-29 01:20

[QUOTE=Prime95;558182]Running 4 workers is fine. In Test/Workers, have each worker use just one CPU core.

[/QUOTE]

It doesn't allow me to use 1 CPU core.

When I run "2. Test/Worker thread" and select 4 Workers, then 2 CPU cores for each worker are automatically selected.

EDIT: I even changed local.txt "CoresPerTest=2" to "CoresPerTest=1" but as soon as I run it it gets changed to 2 again.

Prime95 2020-09-29 03:10

Please PM or email to me the prime.txt and local.txt files.

Is hours-per-day you expect to run mprime set to something other than 24?
If so, set it to 24 and try again.

ZFR 2020-09-29 08:22

[QUOTE=Prime95;558192]
Is hours-per-day you expect to run mprime set to something other than 24?
If so, set it to 24 and try again.[/QUOTE]

Yes, this was the issue. I had it set to 10hrs. If I set it to anything above 12 and run "2. Test/Worker Threads" I get this:

[CODE]
Options for worker #1

Type of work to get (0):
CPU cores to use (multithreading): 1

Options for worker #2

Type of work to get (0):
CPU cores to use (multithreading): 1

Options for worker #3

Type of work to get (0):
CPU cores to use (multithreading): 1

Options for worker #4

Type of work to get (0):
CPU cores to use (multithreading): 1
Get occasional proof certification work (Y):
[/CODE]

But after I set it to 12 hours or less, I get
CPU cores to use (multithreading): 2

CoresPerTest in local.txt changes accordingly.

Is this behaviour correct?
I've changed it to 13 for the time being. But 10 hours is closer to what I really use. Does hours-per-day affect anything except the expected completion time on PrimeNet?

tha 2020-09-29 10:29

Now that we have the new functionality in place and can double the throughput of Primenet, it might be a good idea to monitor the upgrades on the local machines by all the users. Traditional first time LL tests now become pretty much a waste of time. Even all exponents that have been tested once but are still waiting to be double checked can better be 'double checked' with a new PRP test and cogenerated proof file than just traditionally double checked.

The next step would than be to see if there is a way to reach participants that still run outdated clients and not regularly visit the GIMPS website. But being able to monitor the work done by the different versions would be first. Anyone?

moebius 2020-09-29 10:57

[QUOTE=Prime95;558192]..[/QUOTE]
Without wanting to offend you, I can't understand why your own (presumably gpuowl) results are still without proof? gpuowl v6.11.364 to v6.11.380 runs without errors under Linux at least Ubuntu 18.04. Or do you upload the proof files later in one batch?

James Heinrich 2020-09-29 11:42

[QUOTE=moebius;558213]I can't understand why your own (presumably gpuowl) results are still without proof?[/QUOTE]Can you perhaps provide some example(s) of which exponents without proof concern you?

moebius 2020-09-29 11:49

[QUOTE=James Heinrich;558216]Can you perhaps provide some example(s) of which exponents without proof concern you?[/QUOTE]
The problem seems to have been resolved by now, when I looked through the exponents till yesterday, none of his PRP results on the recent cleared page had proofs assigned or confirmed.

Viliam Furik 2020-09-29 12:45

[QUOTE=moebius;558218]The problem seems to have been resolved by now, when I looked through the exponents till yesterday, none of his PRP results on the recent cleared page had proofs assigned or confirmed.[/QUOTE]

I have noticed that proof's existence doesn't show up publicly until it's assigned for certification. That means you will not know, whether the proof was uploaded until it is assigned to certifier, and that is not instantaneous.

Prime95 2020-09-29 17:49

[QUOTE=ZFR;558207]
CoresPerTest in local.txt changes accordingly.

Is this behaviour correct?
I've changed it to 13 for the time being. But 10 hours is closer to what I really use. Does hours-per-day affect anything except the expected completion time on PrimeNet?[/QUOTE]

I'd say the behavior is pretty close to a bug. It certainly is unexpected.

The hours-per-day setting affects very little. Some estimated completion dates, work assignments if you use "What makes sense" work preference.

I'll look into what I need to fix in the next release.

Prime95 2020-09-29 17:52

[QUOTE=moebius;558213]Without wanting to offend you, I can't understand why your own (presumably gpuowl) results are still without proof? gpuowl v6.11.364 to v6.11.380 runs without errors under Linux at least Ubuntu 18.04. Or do you upload the proof files later in one batch?[/QUOTE]

No offense taken. I have scripts to upload my proofs at night (move the gpuowl proofs to prime95 directory).

ATH 2020-09-29 18:19

I got the new message on "Manual Results" page:

[QUOTE]Your software is outdated.
Your L-L result has been accepted for now, but please upgrade your software to at least v30.3 as soon as possible.[/QUOTE]

But I was handing in LL DC results. I understand the message for new LL results but for LL DC?
Do we want everyone to switch to 30.3 even for LL DC, so we are doing PRP + CERT negating the original LL results?

kriesel 2020-09-29 18:53

[QUOTE=ATH;558259]I understand the message for new LL results but for LL DC?
Do we want everyone to switch to 30.3 even for LL DC, so we are doing PRP + CERT negating the original LL results?[/QUOTE]Maybe. On average it will take ~1.04 LL to confirm a first LL, due to error rate, but only ~1.01 PRP equivalent with the effort of a CERT.

I wonder what it does with CUDALucas results or Gpuowl LLDC. PRP via Gpuowl is not an option for some gpu models, whose drivers don't support a high enough OpenCL version. It would be a shame to push the older Quadros, Teslas, etc with decent SP/DP ratio down to TF or recyclers. (The models most likely to have the OpenCL incompatibility are also the most likely to have limited ram for P-1 in CUDAPm1. And CUDAPm1 is still alpha software with issues.)

ZFR 2020-09-29 18:53

[QUOTE=Prime95;558255]I'd say the behavior is pretty close to a bug. It certainly is unexpected.

The hours-per-day setting affects very little. Some estimated completion dates, work assignments if you use "What makes sense" work preference.

I'll look into what I need to fix in the next release.[/QUOTE]

Thanks. For the time being I'll just keep hours-per day greater than 12 then.

James Heinrich 2020-09-29 19:04

[QUOTE=ATH;558259]Do we want everyone to switch to 30.3 even for LL DC, so we are doing PRP + CERT negating the original LL results?[/QUOTE]I think that would be good. A PRP+Cert will clear the exponent just as well as a successful LL-DC would, and be quicker if the DC doesn't match.

There will no doubt be lots of old installations that won't get updated quickly, if ever. I would suggest that pre-30 installations soon (if not immediately) be relegated to DC-only work, even if they asked for first-time LL assignments. I know George has a much more lenient and tolerant approach to users who refuse to upgrade than I would so it might not happen, but I think we need to encourage as many users as possible to upgrade as soon as possible to avoid unneccesary effort being wasted on two checks when one will suffice.

Uncwilly 2020-09-29 19:13

[QUOTE=ATH;558259]Do we want everyone to switch to 30.3 even for LL DC, so we are doing PRP + CERT negating the original LL results?[/QUOTE]
A broader install base of v30 will also spread the load of running the certs around. This will help. Let slower machines do the certs while faster machines do more of the PRP work.

Prime95 2020-09-29 20:29

[QUOTE=ATH;558259]I got the new message on "Manual Results" page:



But I was handing in LL DC results. I understand the message for new LL results but for LL DC?
Do we want everyone to switch to 30.3 even for LL DC, so we are doing PRP + CERT negating the original LL results?[/QUOTE]


I think encouraging the upgrade is a good idea regardless.

I'm ambivalent about DCes being done as LL or PRP+CERT. Cost is about the same. LL-DC provides feedback on the original computer which is of some minor value.

Uncwilly 2020-09-29 20:44

[QUOTE=Prime95;558272]I think encouraging the upgrade is a good idea regardless.

I'm ambivalent about DCes being done as LL or PRP+CERT. Cost is about the same. LL-DC provides feedback on the original computer which is of some minor value.[/QUOTE]
How about LL-DC's for assignments less than 2 years old. Anything much longer and the data of a bad LL is not actionable.

Also, start giving any machine that is returning first time LL's, 25% DC's then in 2 months move it to 35%, then 45%, etc. If they want to avoid the DC's they need to upgrade.

richs 2020-09-30 14:41

I searched through this thread and was not able to find how to stop getting cert work with the new version. One of my older laptops that I upgraded to the latest version yesterday has been receiving cert work but keeps getting the message "Error getting CERT starting value." The program tries twice then aborts the assignment. I would prefer to stop getting cert work on this laptop.

James Heinrich 2020-09-30 14:46

[QUOTE=richs;558363]I searched through this thread and was not able to find how to stop getting cert work with the new version.[/QUOTE]Prime95 GUI has the checkbox under [c]Test > Worker Windows... > [ ] Get occasional proof certification work[/c].
Alternately, in [i]local.txt[/i] put [c]CertWork=0[/c] to disable Cert work.

jwnutter 2020-09-30 15:56

[QUOTE=James Heinrich;558265]I think that would be good. A PRP+Cert will clear the exponent just as well as a successful LL-DC would, and be quicker if the DC doesn't match.

There will no doubt be lots of old installations that won't get updated quickly, if ever. I would suggest that pre-30 installations soon (if not immediately) be relegated to DC-only work, even if they asked for first-time LL assignments. I know George has a much more lenient and tolerant approach to users who refuse to upgrade than I would so it might not happen, but I think we need to encourage as many users as possible to upgrade as soon as possible to avoid unneccesary effort being wasted on two checks when one will suffice.[/QUOTE]

Your suggestion wouldn't impact me in any way and I can't disagree, but if the name of the game is maximizing total available clock cycles to the GIMPS project a forced upgrade may do more to harm the cause than help. But I would agree, there is a trade off between maximizing the total number of available clock cycles and the efficiency of each cycle. Regardless, encouraging users to upgrade as soon as possible would be of great benefit.

James Heinrich 2020-09-30 16:05

[QUOTE=jwnutter;558373].. a forced upgrade may do more to harm the cause than help.[/QUOTE]I doubt there would ever be any "forced" upgrade (as far as I know there's still 15+ year old v24 installations chugging along contributing their little bit to the project). But an [i]encouraged[/i] upgrade would make the cycles of those who do upgrade more useful.

jwnutter 2020-09-30 16:13

[QUOTE=James Heinrich;558375]I doubt there would ever be any "forced" upgrade (as far as I know there's still 15+ year old v24 installations chugging along contributing their little bit to the project). But an [i]encouraged[/i] upgrade would make the cycles of those who do upgrade more useful.[/QUOTE]

I would also add that "DC-only work" could be viewed by some as equivalent to a forced update.

kriesel 2020-09-30 16:27

[QUOTE=Prime95;558272]I think encouraging the upgrade is a good idea regardless.

I'm ambivalent about DCes being done as LL or PRP+CERT. Cost is about the same. LL-DC provides feedback on the original computer which is of some minor value.[/QUOTE]It's worth thinking about the pros and cons.
Reliability data versus exponent on LL will show the effectiveness of the current error detection methods including Jacobi check.

Version upgrade is good for those systems new enough to run wavefront assignments, including the relatively quick Cert work. The XP systems need an OS upgrade to allow it.

The timeliness of a prompt GEC check or Cert is of much greater value in my opinion than LL reliability feedback on a computer that ran LL first test several years ago. Many of those systems that produced existing LLDC candidates' first tests (54M and higher) are likely to no longer be in operation or running primality tests by the time the LL reliability feedback is obtained, if not already replaced by now.

There is a very slight ~2% throughput advantage to PRP/GEC/CERT over LLDC, and a large reliability advantage. Approx 2% x 506K DC to Mp51 adds up (~10,120 tests).

There is no great harm in having a mixed situation, with some LLDC and some PRP/CERT in place of LLDC (& ~4% LLTC, ~0.04% LLQC, ~0.0008% LL5C, ~16E-8 LL6C).

PrimeNet continuing to automatically issue first time LL assignments ought cease sometime soon, since each one commits the project to future DC in some form, at at least 100% of first test cost, at higher more costly exponents. July 2021, a year after PRP proof introduced? Wait till mlucas supports PRP proof?
A separate question is whether to also cease issuing first time LL assignments as manual assignments. Some gpus can't run gpuowl, so can't run PRP, with proof or not.

Aramis Wyler 2020-10-01 06:29

[QUOTE=richs;558363]I searched through this thread and was not able to find how to stop getting cert work with the new version. One of my older laptops that I upgraded to the latest version yesterday has been receiving cert work but keeps getting the message "Error getting CERT starting value." The program tries twice then aborts the assignment. I would prefer to stop getting cert work on this laptop.[/QUOTE]


Make sure you're not using GPU72 as a proxy. It messes with the Certs.

tha 2020-10-01 08:41

[URL="https://www.mersenne.org/report_exponent/?exp_lo=98000101&exp_hi=&full=1"]https://www.mersenne.org/report_exponent/?exp_lo=98000101&exp_hi=&full=1[/URL]

On the line "Type : CERT" under status it says "n/a" suggesting something is to come later. I would expect something like "successfully verified"

James Heinrich 2020-10-01 11:12

[QUOTE=tha;558436][m]M98000101[/m] On the line "Type : CERT" under status it says "n/a" suggesting something is to come later. I would expect something like "successfully verified"[/QUOTE]As previously noted, it's being worked on (but Aaron lacks spare cycles at the moment):[QUOTE=James Heinrich;557639]This is normal. It still needs fixing, but it's a website display issue and not a problem on your end. Using [M]M99,770,291[/M] as an example, your Cert verifies the PRP run by Ben Delo, therefore the PRP section shows "verified".

The "Cert... n/a" is a known display issue. It's not a trivial fix, but Aaron is aware of it and will (when time permits) work some magic to relate the Cert to the PRP and make it display nicely.[/QUOTE]

richs 2020-10-01 16:08

[QUOTE=Aramis Wyler;558430]Make sure you're not using GPU72 as a proxy. It messes with the Certs.[/QUOTE]

Thanks for the info. I am using GPU72 as a proxy. I only do double checks so I'll leave the certs to others.

James Heinrich 2020-10-01 16:11

[QUOTE=Aramis Wyler;558430]Make sure you're not using GPU72 as a proxy. It messes with the Certs.[/QUOTE]Chris is aware of the issue but is so overloaded with Real Life work this month he has not had any opportunity to look into the fix. It will come eventually.

S485122 2020-10-01 17:45

[QUOTE=kriesel;558378]...
There is no great harm in having a mixed situation, with some LLDC and some PRP/CERT in place of LLDC (& ~4% LLTC, ~0.04% LLQC, ~0.0008% LL5C, ~16E-8 LL6C). [/QUOTE]Just for the record the error rate is decreasing for higher (completed) ranges. Based on LL tests it is more like 2% of all tests in error (this includes the necessary triple or quadruple tests. The 51M range has 558 bad tests compared to 39 424 verified tests, but that 1,4% is a record. For the kind of calculations you make 2% is probably better. (I am sure the "special" ranges like 332-332M range will have a much higher bad LL's compared to verified LL's. But we will know once they are completed.)

Jacob

kriesel 2020-10-01 18:37

[QUOTE=S485122;558508]Just for the record the error rate is decreasing for higher (completed) ranges. Based on LL tests it is more like 2% of all tests in error (this includes the necessary triple or quadruple tests. The 51M range has 558 bad tests compared to 39 424 verified tests, but that 1,4% is a record. For the kind of calculations you make 2% is probably better. (I am sure the "special" ranges like 332-332M range will have a much higher bad LL's compared to verified LL's. But we will know once they are completed.)

Jacob[/QUOTE]Do you mean I should have used 2% [U]per LL first test and DC pair[/U]? Perhaps what I wrote earlier was unclear.

I used 2% probability [U]per LL test[/U]; 2% for the first test, 2% for the DC, which gives 4% probability of needing a TC, and continued to use 2% probability of error for any later LL retest that may occur, although there may be a factor of 2 missing for the quad and higher checks. George has stated 1.5%. I've seen 2% commonly used elsewhere. I've seen 1.5 & 2%/LLtest in my own cpu & gpu tests. For given hardware and software, the rate is expected to climb with run time and hardware age. The introduction of the Jacobi check should halve the figures at some point, where applicable (prime95, gpuowl, mlucas, not CUDALucas). Phaseout of LL/Jacobi in favor of PRP/GEC will lower the average of PRP & LL combined test error rate.
Assuming 1% chance of uncorrected error per LL test would make the computation time about a wash. Being able to perform a proof of correctness remains an advantage for PRP.

Primality tests via PRP or LL cost about the same; the GEC and occasional Jacobi are ~0.2-0.3% of a primality test, and both George and Mihai IIRC have stated there's no difference in cost between bare LL and bare PRP.

Assuming power 8 PRP, total cost of a PRP with proof & cert is ~1.01 primality tests.
First LL has cost 1 and error rate e, so cost of a correct test is 1+sum from i=1 to infinity, e^i to obtain a correct res64. LLDC has the same cost. Cost of two correct tests is then for e=.02, 2.04081632...

After obtaining a first LL res64, we don't know whether it's right. The chance of a mismatch with a DC is the sum of the probabilities of the first test being wrong, or the second test being wrong (including the case of both being wrong but differently); ~2e.

Uncwilly 2020-10-01 21:59

[QUOTE=richs;558501]Thanks for the info. I am using GPU72 as a proxy. I only do double checks so I'll leave the certs to others.[/QUOTE]
IIRC Chris has indicated that using GPU72 to proxy for any LL, DC, or PRP is only useful to move up your stats on GPU72. GPU72 does not keep any assignments for those work types.

A minor other reason might be to get assignments that your machine does not qualify for normally.

kriesel 2020-10-02 17:15

Daytime and nighttime P-1/ECM stage 2 memory question
 
Options, Resource Limits, Advanced...,
Daytime P-1/ECM stage 2 memory (GB):
Nighttime P-1/ECM stage 2 memory (GB):

On a multiple worker configuration, are these allowed memory settings per worker, or total for the prime95 application?

The readme does not say either way.
Treating it as per-worker is conservative but suboptimal if it is actually total for the prime95 application.

Setting available memory was introduced in v20.0 for P-1, and ECM supported this memory limit beginning at V25.5.
Multiple ll test workers support was introduced at V25.5.
It's also possible to run P-1 on multiple workers, including overlapping stage 2. And presumably ECM also.
I sometimes run a P-1 on each Xeon in a system, among the 2 or 4 workers per system.

whatsnew.txt says
[CODE][New features in Version 25.7 of prime95.exe
-------------------------------------------

1) Time= in ini files no longer supported. A during/else syntax can be
unsed instead for some ini file options.
2) PauseWhileRunning enhanced to pause any number of workers.
3) LowMemWhileRunning added.
4) Ability to stop and start individual workers added.
5) DayMemory and NightMemory in local.txt replaced with a single Memory
setting.
6) [B]Memory can be set for each worker thread.[/B]
7) Scheme to distribute available memory among workers needing a lot of
memory has been completely revamped.
8) MaxHighMemWorkers replaces delayStage2Workers option.
9) The executable now defaults to talking to the PrimeNet v5 server. To
use the executable with the old v4 server, add "UseV4=1" to the top
of prime.txt.[/CODE]undoc.txt describes both cases:[CODE]The Memory=n setting in local.txt refers to the total amount of memory the
program can use. You can also put this in the [Worker #n] section to place
a maximum amount of memory that one particular worker can use.[/CODE]
Local.txt memory allowance contents set by the menus, and memory allocations by multiple P-1 simultaneous runs, appear consistent with the allowable value being total, not per worker. The first stage 2 to launch gets a lot of memory but not all of the allowed, and the second gets essentially what's left if it would benefit from at least that much. So for example a 4 worker system with 128GB total ram and a setting of 32GB allowed gave 21GB to the first launched stage 2, and 11 to the second. Another system with 128G installed, 2 workers, 48G allowed, allocated 40.5G to one and 7.5G to another.

James Heinrich 2020-10-02 17:41

[QUOTE=kriesel;558631]are these allowed memory settings per worker, or total for the prime95 application?[/QUOTE]The main [c]Memory=[/c] setting in [i]local.txt[/i] controls the overall amount of RAM Prime95 can use for stage2 workers. If nothing else is specified, it's a first-come-first-served pool: first worker takes as much RAM as it wants, second worker takes the lesser of what it wants or what's available, etc. When the first worker is done with stage2 that RAM is released to the pool, and some/all stage2 workers might restart if they're currently using less RAM than they wanted to. When the amount of RAM for stage2 is small (e.g. P-1 in the 10M range) it's not a huge deal, but on wavefront P-1 on older machines (like mine) it can take 10 mins or more to re-initialize stage2, so stage2-thrashing should be avoided. In my case I'm allocating 40GB for stage2, but manually specifying that each of two workers can have up to 20GB, by putting another [c]Memory=[/c] line under each [c]Worker #x[/c] section.

kladner 2020-10-02 18:10

Thanks for the note on individual worker memory, James.

chalsall 2020-10-02 18:25

[QUOTE=Uncwilly;558545]IIRC Chris has indicated that using GPU72 to proxy for any LL, DC, or PRP is only useful to move up your stats on GPU72. GPU72 does not keep any assignments for those work types. A minor other reason might be to get assignments that your machine does not qualify for normally.[/QUOTE]

That is correct. And GPU72 does absolutely no caching of [FC | DC] candidates. All requests are simply passed onto Primenet for all assignments except P-1. (Read: each proxied machine receives the usual Primenet assignment rules filtered work.)

There was a new field added for the Cert work which breaks the comms. WAY too busy to fix ATM. I wouldn't mind deprecating the GPU72 Proxy, actually.


All times are UTC. The time now is 02:22.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.