I was actually surprised to hear that anybody still uses CPU for TF in GIMPS range.

[QUOTE=Maciej Kmieciak;520053]If anyone interested, I've just hit 500 factors.

Most of them is in the 900M range, from 2^70 to 2^71, I think the easiest way to find factors under 1G currently.

The 500th one is 1800724098004283562241, the factor of [URL="https://www.mersenne.org/report_exponent/?exp_lo=914797787&full=1"]M914797787[/URL]

Using six 3.8GHz cores of my i5-8400, I'm expecting to dig up 1000th factor by the end of 2019.[/QUOTE]

Congratulations.... that's quite the commitment
If you really want lots more factors look into a GPU.
They are crazy fast at factoring

[QUOTE]I was actually surprised to hear that anybody still uses CPU for TF in GIMPS range.[/QUOTE]
[QUOTE]If you really want lots more factors look into a GPU.[/QUOTE]
I see but I don't know much about this kind of technology. I couldn't understand how to make the software work. :sad:

Should I use mfakto for AMD GPU? I downloaded it from [URL="https://github.com/Bdot42/mfakto"]here[/URL] and I found this in README file.
[QUOTE]#############################
# 1.2 Compilation (Windows) #
#############################

- Install AMD APP SDK >= 2.5
- Use the VS2010 solution to build the 32-bit or 64-bit binary, or
- use the Makefile to build using MinGW and gcc[/QUOTE]
What does it mean? Is there any place where I can learn how to do these steps?

[QUOTE=Maciej Kmieciak;520078]Should I use mfakto for AMD GPU?[/QUOTE]Assuming from your quote that you're on Windows, you don't need to compile anything, just download the precompiled binary from [url]https://download.mersenne.ca/mfakto[/url] (latest version is v0.15pre6). Once unzippped, put your worktodo.txt in the directory and run mfakto.exe and your results will appear in results.txt
Note that there is no integrated automated work-fetch or result-submission, you'll either need to do it manually, or set up something to do it for you. I've heard some people use [url=https://www.mersenneforum.org/forumdisplay.php?f=103]MISFIT[/url] (I haven't used it myself), for my own [url=https://www.mersenne.ca/tf1G.php]>1000M TF project[/url] I provide example scripts to automate work fetching and result submission.

If you have more specific mfakto questions you can ask them in [url=https://www.mersenneforum.org/showthread.php?t=15646]the mfakto thread[/url].

Thanks a lot! Everything working fine. Now I feel like a donkey...

Great you got it working. Now you can free up your CPU for other GIMPS work that is better suited to CPU, such as PRP, or perhaps P-1 if you have a lot of RAM available and want to find factors.

[QUOTE=James Heinrich;520082](...)GIMPS work that is better suited to CPU, such as PRP, or perhaps P-1 if you have a lot of RAM available and want to find factors.[/QUOTE]

Or ECM in very low ranges if you don´t have that much mem available... :whistle:

Hope you´ll find many new factors with the GPU! They are really much, much more efficient at TF than CPUs.

For fetching work and submit results, under windoze, you may install MISFIT as James said. You need to configure it in the beginning, from where to take work, how much, where to report the results. Then you can forget about it, it will do its job by itself.

If you have more than one GPU in your system (or network, whatever you use), then MISFIT is a must to have, it fetches and distributes the work to all GPUs equitable according with their speed, etc.

If you want to continue working in high range (900M or so), select to fetch work directly from gimps. There you will find a lot of factors, there are still a lot of "low hanging fruits" there.

If you want to switch working at the "LL front", (in the 80M+) you should consider making an account with gpu72.com, and set MISFIT to fetch work from there. You will not find so many factors, and assignments will take longer, but your GPU will work more efficient (I assume you have at least a mid-range AMD card, the low-range work about the same in the 900M), and your contribution to the project will be more appreciated, due to the fact that you help the people who do immediate LL/PRP tests.

The 900M range will not be LL/PRP tested for the next 20 years or so, and working there is not so "valuable" (beside of the satisfaction of finding a lot of factors). I use the quotes because "valuing" it is very subjective, ANY work is valuable, and 101km/h is faster than 100km/h, but we don't know what the next 20 years will bring, on the hardware side, or theory side. Maybe the hardware in 20 years will be so fast to do in a day factoring we do now in a year, or maybe some mathematical breakthrough in factoring will make our TF activity futile, who knows? That is why some of us consider working at the LL/PRP front as "more valuable". But that is up to you. It is your hardware, your money, your headache :razz:

another one from ryan 5443
[QUOTE]
[url]https://www.mersenne.org/report_exponent/?exp_lo=5443&full=1[/url]
[/QUOTE]

[URL="https://www.mersenne.ca/exponent/1073741827"]M1,073,741,827[/URL] has a factor: 16084529043983099051873383

This exponent is just outside the range of Primenet. This exponent is relevant to the (trivial) "[URL="http://mprime.s3-website.us-west-1.amazonaws.com/new_mersenne_conjecture.html"]New Mersenne Conjecture[/URL]"

The factor is 84 bits. It was found with mfaktc 99.7% of the way through the 83 to 84 bit range.

Thank you for the advice. I will consider them.

Meanwhile, I found my first P-1 factor: 110778360181007451990785681 of [URL="https://www.mersenne.org/report_exponent/?exp_lo=91909661&full=1"]M91909661[/URL]