[QUOTE=xilman;258502]Ok, you talked me into it. Just grabbed another two exponents to see how well it goes with them. A few minutes in and it looks as if the pair of them will take around 25-30 minutes. Two exponents running serially took 45 minutes yesterday.
[/QUOTE]

Yeah it's awesome isn't it? If you're really keen you can fiddle with the "SievePrimes" variable in the .ini file. To get the most out of the CPU/GPU ratio.

[QUOTE=xilman;258502]
Doesn't seem to have fried the display yet, though the gpu temperature has reached a maximum of 62C with a fan speed of 52% --- well within acceptable limits IMO.

Paul[/QUOTE]

Luxury! Back in my day.....

My GTX460 is at 75degC 72% fan speed. GPU load is hovering around 97%. 29.5degC ambient. 85degC+ is starting to get into danger zone territory.

All I want for xmas is to see how P-1 testing goes on a GPU. :)

-- Craig

We are very confused (but intrigued!) after reading this lengthy thread, so perhaps someone would be willing to answer a few easy questions?
[LIST][*]What hardware (GPU) version is best? Which has the best bang for the buck?[*]What brand for the card, or does this matter? (We use a lot of ASUS products.)[*]The GPU isn't used as a display card, right? (Our box is headless.)[*]We have a [URL="http://www.antec.com/pdf/flyers/EA500D_flyer_EN.pdf"]500W[/URL] power supply. Is this enough power for a card?[*]Is there a HOWTO for not-so-smart people?[/LIST]

[QUOTE=xilman;258569]What does the "Successes" column mean? I thought it may be the number of factors found. [/QUOTE]

It is the number of factors found. I don't see anything obviously wrong when I eyeballed the PHP code.

Is anyone else having this problem??

I'll try to reproduce the problem with one of Paul's factors - so don't be alarmed if it disappears from the database for a while.

[QUOTE=Prime95;258632]Is anyone else having this problem??[/QUOTE]
The 3 factors I found yesterday are all listed and I hope that the 2 I've just reported will show up after the next update :smile:

One unimportant thing: When I entered my results manually I didn't append a new line after the last line and ended up with two result entries like this:

[SIZE=2]Manual testing[/SIZE][SIZE=2] 56521571[/SIZE][SIZE=2] NF [/SIZE][SIZE=2]2011-04-14 12:52[/SIZE][SIZE=2] 0.1 [/SIZE][SIZE=2]no factor for M56521571 from 2^68 to 2^69 [mfaktc 0.16p1 barrett79_m[/SIZE][SIZE=2] 1.0577[/SIZE]
[SIZE=2]Manual testing[/SIZE][SIZE=2] 56521613 [/SIZE][SIZE=2]NF[/SIZE][SIZE=2] 2011-04-14 13:42[/SIZE][SIZE=2] 0.2[/SIZE][SIZE=2] no factor for M56521613 from 2^68 to 2^69 [mfaktc 0.16p1 barrett79_m[/SIZE][SIZE=2] 1.0577

An uncut entry looks like this:

[/SIZE][SIZE=2]Manual testing[/SIZE][SIZE=2] 56492741[/SIZE][SIZE=2] NF[/SIZE][SIZE=2] 2011-04-14 06:13 [/SIZE][SIZE=2]0.0[/SIZE][SIZE=2] no factor for M56492741 from 2^68 to 2^69 [mfaktc 0.16p1 barrett79_mul32] [/SIZE][SIZE=2] 1.0582

[/SIZE]

Hi Xyzzy,

[QUOTE=Xyzzy;258600]What hardware (GPU) version is best? Which has the best bang for the buck?[/QUOTE]

For mfaktc you really want a Fermi generation GPU (Geforce 4xx or 5xx). Performancewise a GTX 470/480/570/580 are the best but GTX 460/560 are a very good deal, too. Those 460/560 need less power and are less noisy.

[QUOTE=Xyzzy;258600]What brand for the card, or does this matter? (We use a lot of ASUS products.)[/QUOTE]
I think they are all very similar, personally I don't like factory-overclocked GPUs and I like "reference designs".

[QUOTE=Xyzzy;258600]The GPU isn't used as a display card, right? (Our box is headless.)[/QUOTE]
Both is possible, mfaktc runs on the command line. For lowend cards the GUI becomes unusable in many cases once mfaktc is running (should be better with the next release)

[QUOTE=Xyzzy;258600]We have a [URL="http://www.antec.com/pdf/flyers/EA500D_flyer_EN.pdf"]500W[/URL] power supply. Is this enough power for a card?[/QUOTE]
Should be OK for non-extreme GPUs.

[QUOTE=Xyzzy;258600]Is there a HOWTO for not-so-smart people?[/QUOTE]
Windows or Linux? Built from source or running precompiled executable?
Did you take a look at the README included in both, tarball and windows executables?

Oliver

[QUOTE=Ralf Recker;258635]The 3 factors I found yesterday are all listed and I hope that the 2 I've just reported will show up after the next update[/QUOTE]

Paul's factors were recorded and credited too. The problem is the they did not show up as "successes" in the top producers report.

[QUOTE=Prime95;258641]Paul's factors were recorded and credited too. The problem is the they did not show up as "successes" in the top producers report.[/QUOTE]
I know. My number of "successes" was correctly updated a few minutes ago (from 7 to 9). The other thing I mentioned in the post above was just an observation that I've made when copying the results from the results.txt file into the results box. I opened the file with gedit, did a Ctrl+A/C/V to copy and paste the results into the entry form and submitted them. The last entry in my results lists was cut off ("[SIZE=2][mfaktc 0.16p1 barrett79_m[/SIZE]" instead of "[SIZE=2][mfaktc 0.16p1 barrett79_mul32][/SIZE]" but counted nonetheless) probably because entering the results this way removed the newline character after the last line. I did this a second time to verify this behaviour afterwards. When I uploaded the results file directly this behaviour did not occur, likewise when I added a(n empty) line after copying the results into the form.

[QUOTE]For mfaktc you really want a Fermi generation GPU (Geforce 4xx or 5xx). Performancewise a GTX 470/480/570/580 are the best but GTX 460/560 are a very good deal, too. Those 460/560 need less power and are less noisy.[/QUOTE]We read this thread again and were able to get a better handle on the various models. We are not yet sure how they all are related performance-wise, but we suspect, like processors, that the most expensive ones are probably not the best bang-for-the-buck.

Based on our decade+ positive experience with ASUS, and their very neato cooling system, we just ordered two [URL="http://www.asus.com/Graphics_Cards/NVIDIA_Series/ENGTX570_DCII2DIS1280MD5/"]GTX-570[/URL] cards. The GTX 570 seems like a good price/performance compromise.

[quote]I think they are all very similar, personally I don't like factory-overclocked GPUs and I like "reference designs".[/quote]We assumed all of the cards were the same reference design wrapped up in different clothes and sometimes overclocked. We are not interested in overclocking, either. The ASUS seems relatively conservative.

[quote]Both is possible, mfaktc runs on the command line. For lowend cards the GUI becomes unusable in many cases once mfaktc is running (should be better with the next release)[/quote]SSH CLI if fine with us.

[quote]Should be OK for non-extreme GPUs.[/quote]Hopefully the GTX 570 is not extreme. A little time on Google today led us to several sites that indicated 20A on the 12V rail would be enough, especially since the boxes are minimalist boxes. Even the drives (SSD) are low power. Our watt-meter shows ~120W wide open with Prime95.

We calculate 403W, with the GTX 570, at [URL="http://www.antec.outervision.com/PSUEngine"]this site[/URL]. We put in 90% for CPU and GPU utilization.

[quote]Windows or Linux? Built from source or running precompiled executable? Did you take a look at the README included in both, tarball and windows executables?[/quote]Linux, possibly built from source.

We didn't think to look for a README.

:redface:

[QUOTE=xilman;258569]A few factors have been discovered and reported through the manual reporting page but I have no "Successes"[/QUOTE]

Alright, I have a working theory. Does this sound possible:

1) You submitted the results using the manual form but forgot to login to the web pages.
2) Maybe because you noticed the exponents were still assigned to you, you resubmitted them after logging in.
3) The server didn't record the duplicate results, but since you had a valid assignment ID, it gave you the CPU credit. This whole credit for duplicate results mess was added to avoid outrage about losing CPU credit because someone poached an assignment.

If this scenario is correct, then the "bug" is that you didn't get a "success" for finding an already known factor with your valid assignment ID.

[QUOTE=Xyzzy;258647]Hopefully the GTX 570 is not extreme. A little time on Google today led us to several sites that indicated 20A on the 12V rail would be enough, especially since the boxes are minimalist boxes. Even the drives (SSD) are low power. Our watt-meter shows ~120W wide open with Prime95.[/QUOTE]

GTX 570s are really fast, I hope that you've noticed that you still need some CPU power. Any once you're bored running TF you can try msft's CUDALucas and run LL tests on GPU without the need for much CPU power.

[QUOTE=Xyzzy;258647]Linux, possibly built from source.

We didn't think to look for a README.

:redface:[/QUOTE]

Depending on your Linux skills it shouldn't be a big deal. You have to setup your CUDA environment followed by compiling mfaktc. Since mfaktc doesn't depend on many external libraries it should be relative easy.

Oliver

This sets my decision: GTX470, though I'm going to get mine from Galaxy since I like the GEForce 210. This will be another xubuntu box, this time with intel sandy bridge i7 on it, and I'm wondering if the 32 Gigs of Ram option is worth considering.