mersenneforum.org  

Go Back   mersenneforum.org > Search Forums

Showing results 1 to 25 of 1000
Search took 0.83 seconds.
Search: Posts Made By: ATH
Forum: GpuOwl 2020-09-17, 06:37
Replies: 2,470
Views: 145,807
Posted By ATH
Added the info about copying the libstdc* files...

Added the info about copying the libstdc* files back in each notebook to the original post #33:

https://mersenneforum.org/showpost.php?p=555507&postcount=33
Forum: Hardware 2020-09-17, 00:02
Replies: 4
Views: 145
Posted By ATH
Sounds a bit like this stuff: ...

Sounds a bit like this stuff:

https://www.digikey.ca/en/product-highlight/p/panasonic/pyrolytic-graphite-sheets

Except it wasn't really meant for CPU cooling so did not turn out so well:
...
Forum: Data 2020-09-15, 19:21
Replies: 3,319
Views: 210,166
Posted By ATH
Now there are TF assignments in the DC Cat 0 and...

Now there are TF assignments in the DC Cat 0 and Cat 1 again:

https://www.mersenne.org/assignments/?exp_lo=53000000&exp_hi=54000000&execm=1&exdchk=1&exfirst=1&exp1=1

This should not happen, it...
Forum: Lounge 2020-09-15, 16:49
Replies: 1,071
Views: 46,334
Posted By ATH
Lockdown seems to have an effect. US numbers...

Lockdown seems to have an effect.

US numbers went down after lockdown and then back up when you reopened too early.

Italy locked down completely and numbers went down dramatically with a slight...
Forum: Lounge 2020-09-13, 17:35
Replies: 5
Views: 295
Posted By ATH
Ben Delo surpassed CurtisC lifetime total...

Ben Delo surpassed CurtisC lifetime total sometimes mid July and he started around Oct 1st last years, so it took 9.5 months.

Only stats left above him is ANONYMOUS lifetime results and...
Forum: PrimeNet 2020-09-13, 17:22
Replies: 6
Views: 177
Posted By ATH
The PRP-CF "wavefront" is at 10.5M, but as soon...

The PRP-CF "wavefront" is at 10.5M, but as soon as new factors are found for exponents <10.5M they are assigned to PRP-CF.
Forum: GpuOwl 2020-09-09, 14:12
Replies: 58
Views: 1,949
Posted By ATH
First-time test on cofactors after a new factors...

First-time test on cofactors after a new factors is found, compared to double checking the same results afterwards (which is not needed anymore if proof files are used).

But PRP-CF and PRP-CF-DC...
Forum: GpuOwl 2020-09-09, 01:35
Replies: 58
Views: 1,949
Posted By ATH
I think they consider PRP-CF tests as extra...

I think they consider PRP-CF tests as extra feature which would be announced if it was added, but not something they have to say is not included.

It was only added to Prime95 around 2 years ago,...
Forum: GpuOwl 2020-09-08, 12:41
Replies: 58
Views: 1,949
Posted By ATH
Gpuowl can only do PRP tests, not PRP-CF test...

Gpuowl can only do PRP tests, not PRP-CF test where 1 or more factors are known and testing the remaining cofactor.
Forum: GpuOwl 2020-09-07, 01:10
Replies: 2,470
Views: 145,807
Posted By ATH
Mine has often been stuck at 96%. When I try...

Mine has often been stuck at 96%. When I try later to upload it again, it says it was already uploaded, and the certificate has already been approved.
Forum: Marin's Mersenne-aries 2020-09-05, 20:03
Replies: 8
Views: 26,268
Posted By ATH
Yes 53M was fine for me as well. When you reach...

Yes 53M was fine for me as well. When you reach 54M you will probably have the same problem.

The "-use STATS" command which I did not know previously shows this:
Roundoff: N=100000, mean...
Forum: GpuOwl 2020-09-05, 17:30
Replies: 2,470
Views: 145,807
Posted By ATH
Ok, I guess it was the old version...

Ok, I guess it was the old version v6.11-278-ga39cc1a I was still using, because it is slightly faster than the newest one (311 µs vs 322 µs at 3M FFT on V100). It
seems this old version have a...
Forum: Marin's Mersenne-aries 2020-09-05, 15:48
Replies: 697
Views: 26,027
Posted By ATH
Did you try 54M exponents? I have done 350+ DC...

Did you try 54M exponents? I have done 350+ DC in 50M-53M without any incorrect results with the version I'm using, I just think I reached near the end of the 2.75M FFT that gpuowl chooses.
Forum: GpuOwl 2020-09-05, 14:42
Replies: 2,470
Views: 145,807
Posted By ATH
No, it is weird. I will do an entire exponent...

No, it is weird. I will do an entire exponent with 2.75M FFT and -use STATS on.

I had about 11 DC mismatches (not all confirmed yet) with P100 and V100 cards, and now none since I started using...
Forum: GpuOwl 2020-09-04, 23:21
Replies: 2,470
Views: 145,807
Posted By ATH
Since I started to force 3M FFT all 15+ results...

Since I started to force 3M FFT all 15+ results have been good, so gpuowl is most likely using a too aggressive FFT selection around 54M.

I tested a 54.1M exponent in Prime95 and it chose 2880K...
Forum: Marin's Mersenne-aries 2020-09-03, 12:20
Replies: 697
Views: 26,027
Posted By ATH
Thanks. I kept getting those mismatches even on...

Thanks. I kept getting those mismatches even on the new V100 cards. I'm pretty sure now it is because gpuowl chooses a too low FFT size for those 54M exponents: 2.75M FFT.

I'm now forcing it to...
Forum: PrimeNet 2020-09-03, 04:04
Replies: 7
Views: 303
Posted By ATH
The expiration rules only depends on the category...

The expiration rules only depends on the category at assignment, not what category it becomes later.
Forum: GpuOwl 2020-09-02, 23:51
Replies: 2,470
Views: 145,807
Posted By ATH
I just tried the newest version...

I just tried the newest version v6.11-380-g79ea0cc and it still chose 2.75M FFT for a 54.1M exponent, and I did not use ORIG_SLOWTRIG.

I forgot that this version I run LL tests on is several...
Forum: GpuOwl 2020-09-02, 13:48
Replies: 2,470
Views: 145,807
Posted By ATH
I have been getting failed LL DC tests with...

I have been getting failed LL DC tests with gpuowl on Colab P100 and V100 even though they have ECC RAM.

I think it is because gpuowl is choosing too small FFT size: 2.75M for a 54.1M exponent.
...
Forum: GPU to 72 2020-09-01, 13:21
Replies: 192
Views: 10,967
Posted By ATH
I just got 1 V100 on 1 of my instances: 4.4...

I just got 1 V100 on 1 of my instances: 4.4 hours for 54M DC compared to 7.1 hours on a P100.


4.4 hours for a 54M exponent is amazing to think about. 33M exponent took me 211 days in 2003, that...
Forum: GpuOwl 2020-08-31, 12:35
Replies: 56
Views: 4,610
Posted By ATH
Sorry I forgot about those files. On the instance...

Sorry I forgot about those files. On the instance you compiled it on (or compile it again) add a line after compiling:
!cp /usr/lib/x86_64-linux-gnu/libstdc* '/content/drive/My Drive/'

Now those...
Forum: GpuOwl 2020-08-31, 10:12
Replies: 56
Views: 4,610
Posted By ATH
Are your worktodo.txt lines of this format?: ...

Are your worktodo.txt lines of this format?:
PRP=<AID>,1,2,<exponent>,-1,75,0


You can also try this notebook which compiles gmp-6.2.0 and gpuowl with gcc9 in /root folder and copies the...
Forum: Marin's Mersenne-aries 2020-08-31, 09:27
Replies: 697
Views: 26,027
Posted By ATH
Thanks. That is very weird since they were run on...

Thanks. That is very weird since they were run on a Tesla P100 on Google Colab Pro, and those Tesla cards have ECC RAM.
But I did suspect something went wrong with that instance since 3 did not...
Forum: Marin's Mersenne-aries 2020-08-29, 20:27
Replies: 697
Views: 26,027
Posted By ATH
TC: (Cannot use gpuowl with 0 shift, since I...

TC: (Cannot use gpuowl with 0 shift, since I already used it)
54091019
54091039
54091097
Forum: PrimeNet 2020-08-28, 10:20
Replies: 59
Views: 2,220
Posted By ATH
This is the original question: When he tires of...

This is the original question: When he tires of mfaktc and wants to run CUDALucas, should he run first time LL tests with CUDALucas or stick to LL DC on CUDALucas.

While he can run first time LL...
Showing results 1 to 25 of 1000

 
All times are UTC. The time now is 07:18.

Sun Sep 20 07:18:32 UTC 2020 up 10 days, 4:29, 0 users, load averages: 1.05, 1.31, 1.34

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.