Thank you. The -d 1 was just what I needed. This is ans example of what I am getting now times eight.

[CODE]got assignment: exp=12828031 bit_min=70 bit_max=71 (18.64 GHz-days)
Starting trial factoring M12828031 from 2^70 to 2^71 (18.64 GHz-days)
k_min = 46016088541500
k_max = 92032177090732
Using GPU kernel "barrett76_mul32"
class | candidates | time | ETA | avg. rate | SievePrimes | CPU wait
2508/4620 | 2.29G | 19.180s | 2h20m | 119.62M/s | 11661 | 4.99%[/CODE]

Trying to run mfakto on Intel HD 4000
 

I have a Core i7-3770 with both an integrated Intel HD 4000 GPU and an add-in AMD Radeon HD 7700 graphics card.

As described in [URL="http://www.mersenneforum.org/showthread.php?t=17538"]this thread[/URL], I'm trying to see if it's possible to run mfakto on both the 4000 and the 7700 at the same time (different instances). I can run it no problem on the 7700, but after enabling the 4000 in UEFI (it does show up now on GPU-Z), mfakto is still giving me an error if I try to select that GPU.

Maybe I'm doing something wrong, or maybe reports that the 4000 supports OpenCL are incorrect, or...?

Guidance and info are welcome!

Rodrigo

Near as I can tell the 590 is giving me just over 601 GHZ-days/day. Does that sound about right to those who would know?

[QUOTE=dbaugh;320968]Near as I can tell the 590 is giving me just over 601 GHZ-days/day. Does that sound about right to those who would know?[/QUOTE]

Well I think it depends on what kind of work you are doing.

According to [URL]http://www.mersenne.ca[/URL]

[URL="http://www.mersenne.ca/mfaktc.php"]TF: 452.4 GHz-days/day[/URL]
[URL="http://www.mersenne.ca/cudalucas.php"]LL: 43.2-55.8 GHz-days/day[/URL]

If you use other programs like mmff, which is doing the sieving on the GPU. I have no idea what you would expect.

To me it seems that you have OCed your GPU very much so its a good idea to take notice of the temperature.

The clocks are 630/864/1260 and the temperature is steady at 87 degrees Celsius. I am doing 70 to 71 bit TF on eight numbers in the area of 13 meg at the same time with a lower priority program using any extra CPU cycles.

[QUOTE=Rodrigo;320915]I have a Core i7-3770 with both an integrated Intel HD 4000 GPU and an add-in AMD Radeon HD 7700 graphics card.

As described in [URL="http://www.mersenneforum.org/showthread.php?t=17538"]this thread[/URL], I'm trying to see if it's possible to run mfakto on both the 4000 and the 7700 at the same time (different instances). I can run it no problem on the 7700, but after enabling the 4000 in UEFI (it does show up now on GPU-Z), mfakto is still giving me an error if I try to select that GPU.

Maybe I'm doing something wrong, or maybe reports that the 4000 supports OpenCL are incorrect, or...?

Guidance and info are welcome!

Rodrigo[/QUOTE]
Hi Rodrigo,

the reports of Intel supporting OpenCL are not wrong. Intel has an own development kit for their Hardware to create "their" OpenCL programs. However, the advantage of OpenCL is supposed to be that no matter which DevKit was used, it should run on any device supporting OpenCL.

From that, what you're trying to do should work. However, as I do not have such a CPU/GPU, you will at least need to tell, which -d combinations you've tried, and what the error was.

In addition, the output of clinfo (now in the system32 directory since the later Catalyst versions) will be helpful to see which capabilities the embedded GPU has. I guess, clinfo will have 3 sections on your computer: for the CPU, the embedded GPU and the discrete GPU. You should be able to run an mfakto instance on any of these three.

Bdot

[QUOTE=Bdot;320976]Hi Rodrigo,

the reports of Intel supporting OpenCL are not wrong. Intel has an own development kit for their Hardware to create "their" OpenCL programs. However, the advantage of OpenCL is supposed to be that no matter which DevKit was used, it should run on any device supporting OpenCL.

From that, what you're trying to do should work. However, as I do not have such a CPU/GPU, you will at least need to tell, which -d combinations you've tried, and what the error was.

In addition, the output of clinfo (now in the system32 directory since the later Catalyst versions) will be helpful to see which capabilities the embedded GPU has. I guess, clinfo will have 3 sections on your computer: for the CPU, the embedded GPU and the discrete GPU. You should be able to run an mfakto instance on any of these three.

Bdot[/QUOTE]
Hi Bdot,

Thanks very much!

I have an output file for clinfo.exe. Should I post it here or send it by PM?

I've tried various values for -d, and the only one that doesn't give an error message is 1. For example, if I type:

[QUOTE]mfakto-x64 -d 2 -st[/QUOTE]

to try the selftest, I get the following message:

[QUOTE]
Select device - Error: Only 1 devices found. Cannot use device 2 <bad parameter to option -d>.
init_CL<3, 2> failed
[/QUOTE]

The same error occurs with every number that I have tried for -d, except 1 or 11.

Oh, wait -- I just tried 21, and instead of giving that error message, it actually started to do something. Examining the output, it said:

[QUOTE]
Select device - GPU not found, fallback to CPU.
[/QUOTE]
and then it started working on the 3770.

Then when I tried using 31, it said the following:

[QUOTE]Select device - Error: Only 2 platforms found. Cannot use platform 3 <bad parameter to option -d>.[/QUOTE]

I'm wondering if I have to install the [URL="http://software.intel.com/en-us/articles/intel-opencl-sdk/"]Intel OpenCL SDK[/URL]? That looks like it is for people who want to write OpenCL programs rather than simply use them, but maybe the package is necessary in order to do what we want.

Rodrigo

MISFIT now support file locking support for MFAKTO
 

Please see the MISFIT thread for more information.

Thanks to BDOT for explaining how the '.LCK' locking system works.


Scott

[QUOTE=Rodrigo;321001]Hi Bdot,

Thanks very much!

I have an output file for clinfo.exe. Should I post it here or send it by PM?

I've tried various values for -d, and the only one that doesn't give an error message is 1. For example, if I type:



to try the selftest, I get the following message:



The same error occurs with every number that I have tried for -d, except 1 or 11.

Oh, wait -- I just tried 21, and instead of giving that error message, it actually started to do something. Examining the output, it said:


and then it started working on the 3770.

Then when I tried using 31, it said the following:



I'm wondering if I have to install the [URL="http://software.intel.com/en-us/articles/intel-opencl-sdk/"]Intel OpenCL SDK[/URL]? That looks like it is for people who want to write OpenCL programs rather than simply use them, but maybe the package is necessary in order to do what we want.

Rodrigo[/QUOTE]

Hi Rodrigo,

you can post the clinfo here as an attachment. I wonder if the option -d 22 gives access to the embedded GPU.

There should be no need for the Intel SDK - AMD ships all required files in the driver. I guess you have an Intel driver for the eGPU?

Bdot

[QUOTE=Bdot;321015]Hi Rodrigo,

you can post the clinfo here as an attachment. I wonder if the option -d 22 gives access to the embedded GPU.

There should be no need for the Intel SDK - AMD ships all required files in the driver. I guess you have an Intel driver for the eGPU?

Bdot[/QUOTE]
Hi Bdot,

OK, here's the attached clinfo file.

Yes, the Intel GPU shows up in Device Manager, and it's up to date (October 2012 driver version). GPU-Z also lists the Intel HD 4000, but the check box for OpenCL is blank (all of those boxes at the bottom are unchecked).

[QUOTE=Rodrigo;321016]Hi Bdot,

OK, here's the attached clinfo file.

Yes, the Intel GPU shows up in Device Manager, and it's up to date (October 2012 driver version). GPU-Z also lists the Intel HD 4000, but the check box for OpenCL is blank (all of those boxes at the bottom are unchecked).[/QUOTE]

On my system, GPU-Z is not completely reliable in this regard. It shows all boxes checked for GTX 570, which is the primary adapter, but drops CUDA and Direct Compute for the secondary 460. This is in spite of the fact that the 460 is running CuLu.