[QUOTE=SELROC;504854]For what is worth I don't trust windows as a secure and stable operating system. Sorry, it is just personal experience.[/QUOTE]

Friends don't let friends run WinBlows... :wink:

Although I sometimes have to support WinDoze machines for clients, I'll never run it for any mission-critical servers. I just don't trust it....

[QUOTE=SELROC;504854]You are still reading old messages, but you didn't run Quake 3 :-)[/QUOTE]This is a Quake - free zone and will stay that way. And it is irrelevant to the driver performance difference.

@rodrigo: As symptoms described, we believe that the issue is not from mfaktX, but from another program/driver installed after. Like a PDF viewer for example (yes, this is not a joke). You have to identify it and decide if it is more important to you than mfaktX.

If not, kill it. Mind that it can also be an OS upgrade. New things they put into it, related to [URL="https://en.wikipedia.org/wiki/Meltdown_(security_vulnerability)#Mitigation"]Spectre and Meltdown[/URL], for example, got installed into your computer after Oct. 2018 either if you wanted or not, if you did any upgrade of the OS after that date. So, it may not be easy to find, or get rid of it, unless you are some "expert" in doing that. All these things make your computer run slower and lag more. We did all in our power to avoid installing them, but they still tricked us for few of our systems...

If yes (or you can't find it), then the solution would be to adjust the mfaktX settings in such a way to let your system free for longer time, or to use a much lower priority. One can adjust the settings for mfaktX in the ini file, to have less threads, etc., and only use 98% or 96% of the stuff you have there. This gives your system a better usability, i.e. no lag. You will get a more usable system, and consume less electricity, for a small (3-5%) penalty on mfaktX output. Another way is to lower the priority of the process to idle (from task manager) and see what is happening.

[QUOTE=R. Gerbicz;504831]Hm, reading the code and still don't know where do you sieve by p>11; confirmed that it is in SegSieve in gpusieve.cl, but as I can see we don't even call this in the runs:[/QUOTE]

I'm not sure if anyone is maintaining mfakto sources that can answer your question.

Since mfakto was "ported" from mfaktc, you might need to ask questions in that thread.

BTW, when I played with mfakto six or so years ago, I found a bug in Intel's OpenCL implementation and added some work-around code to the mfakto source. If Intel ever fixed their bug you might can improve performance ever-so-slightly by removing said work-around code.

Bdot's last activity was July 4, 2018, and his last post is from Jan 2018. So, most probably he found other things to do...

[QUOTE=kriesel;504862]This is a Quake - free zone and will stay that way. And it is irrelevant to the driver performance difference.[/QUOTE]


I was not talking about the driver but properly the gpu hardware involved.


It is unfortunate for you to be in a Quake free zone. To save yourself you can try with Doom and see if it helps :-)


[url]https://it.wikipedia.org/wiki/Doom[/url]

[QUOTE=R. Gerbicz;504831]Hm, reading the code and still don't know where do you sieve by p>11; confirmed that it is in SegSieve in gpusieve.cl, but as I can see we don't even call this in the runs:[/QUOTE]

Are you referring to the source from [url]https://github.com/Bdot42/mfakto/tree/master/src[/url] ?

[QUOTE=Prime95;504873]
Since mfakto was "ported" from mfaktc, you might need to ask questions in that thread.
[/QUOTE]

Thanks, trying that.

[QUOTE=axn;504882]Are you referring to the source from [url]https://github.com/Bdot42/mfakto/tree/master/src[/url] ?[/QUOTE]

Yes, I've also given that link.

[QUOTE=R. Gerbicz;504887]Yes, I've also given that link.[/QUOTE]
In there, mfakto.cpp is calling SegSieve -- search for CL_SIEVE.

[QUOTE=axn;504890]In there, mfakto.cpp is calling SegSieve -- search for CL_SIEVE.[/QUOTE]

I misread that part.

In same file (mfakto.cpp) there is a tf_class_opencl which calls gpusieve (defined in gpusieve.cpp) which calls run_cl_sieve (mfakto.cpp) which calls segsieve.

tf_class_opencl itself is called from mfaktc.c which is the main.

Intel(R) UHD Graphics 630
 

I have a Intel UHD Grapics 630 video card and I am running mfakto-0.15pre5 and got the message to post here to add the card.

Here is the OpenCL info output:

OpenCL device info
name Intel(R) UHD Graphics 630 (Intel(R) Corporation)
device (driver) version OpenCL 2.1 NEO (24.20.100.6136)
maximum threads per block 256
maximum threads per grid 16777216
number of multiprocessors 24 (1536 compute elements)
clock rate 1200MHz


Plus I have set the parameter "GPUType=INTEL" and just trying to run one TF test now. Anything else i should try or post?