I won't clutter the thread with more updates from the HD 530 (it's still running), but while we wait, I've started a run on the UHD 620. It's running the same v1.70 'newer' app as the 530.
If I may ? : FWIW and IIRC ( ! ) with the cases of even and odd N ( the formula result ) will differ by one frequency bin because the FFT is a discrete transform. The exact logic escapes me at the moment, but I will look up a reference text that I have for further clarity. So probably the formula in the source code is correct ( as is your analysis of it ), but the preceding comment in the code doesn't fully describe the cases, which is a 'bug' of sorts and definitely misleading.
Cheers, Mike.
( edit ) For N = odd : the output frequency bins skip over the Nyquist frequency ( the maximum allowable -> to avoid aliasing ) and thus perhaps it is better to add one extra input value ( say, let it be a zero value* ) and hence make it an even transform ( thus with an extra frequency bin that hits the Nyquist ). At N = 222, which is typical for E@H, I don't think this is a biggie. ;-)
* A better choice might be the mean value of the others I think.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Only seen the Windows one so far, and yes, it's NVidia. Got them on three machines so far, and they seem to be running well. About to try on 0.5 of a GTX 1660/1650
The Meerkat are not ticked by default in the settings sometimes. My computers with everything ticked got them automatically, but those with only some apps ticked didn't get them. So some people may not be aware of their existence.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
they also (in some cases) are resending older WUs that had issues with cross validation between nvidia/AMD. one of mine just validated with a Radeon VII task completed on Aug 3rd.
Sometime pretty recently the applications section of the project preferences page for my account came up with a new line:
Binary Radio Pulsar Search (MeerKAT)
Back in early August, the few I got presumably were enabled by my permission to run test applications under the Beta settings header. Just now I've ticked the MeerKAT box for the three locations I actually use.
I don't expect to get anything until the project turns an AMD GPU application running on Windows on again.
they also (in some cases) are resending older WUs that had issues with cross validation between nvidia/AMD. one of mine just validated with a Radeon VII task completed on Aug 3rd.
Just got one where a v0.01 BRP7-opencl-ati validated with a v0.03 BRP7-cuda55 (both Linux), and tossed my v0.03 Windows out of the window.
I won't clutter the thread
)
I won't clutter the thread with more updates from the HD 530 (it's still running), but while we wait, I've started a run on the UHD 620. It's running the same v1.70 'newer' app as the 530.
If I may ? : FWIW and IIRC (
)
If I may ? : FWIW and IIRC ( ! ) with the cases of even and odd N ( the formula result ) will differ by one frequency bin because the FFT is a discrete transform. The exact logic escapes me at the moment, but I will look up a reference text that I have for further clarity. So probably the formula in the source code is correct ( as is your analysis of it ), but the preceding comment in the code doesn't fully describe the cases, which is a 'bug' of sorts and definitely misleading.
Cheers, Mike.
( edit ) For N = odd : the output frequency bins skip over the Nyquist frequency ( the maximum allowable -> to avoid aliasing ) and thus perhaps it is better to add one extra input value ( say, let it be a zero value* ) and hence make it an even transform ( thus with an extra frequency bin that hits the Nyquist ). At N = 222, which is typical for E@H, I don't think this is a biggie. ;-)
* A better choice might be the mean value of the others I think.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
On another front - the
)
On another front - the MeerKATs are back! Just completed four under Windows with v0.03, successful and valid.
https://einsteinathome.org/host/1001562/tasks/4/57
And four more downloaded.
Richard Haselgrove wrote: On
)
Thanks.
The applications page:
https://einsteinathome.org/apps.php
currently shows one Linux and one Windows application, both labelled with 0.03 (BRP7-cuda55).
From the cuda component, I'll guess these two are NVidia only.
Only seen the Windows one so
)
Only seen the Windows one so far, and yes, it's NVidia. Got them on three machines so far, and they seem to be running well. About to try on 0.5 of a GTX 1660/1650
The Meerkat are not ticked by
)
The Meerkat are not ticked by default in the settings sometimes. My computers with everything ticked got them automatically, but those with only some apps ticked didn't get them. So some people may not be aware of their existence.
If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.
they also (in some cases) are
)
they also (in some cases) are resending older WUs that had issues with cross validation between nvidia/AMD. one of mine just validated with a Radeon VII task completed on Aug 3rd.
so far the v0.03 cuda55 app is working well.
_________________________________________________________________________
Sometime pretty recently the
)
Sometime pretty recently the applications section of the project preferences page for my account came up with a new line:
Binary Radio Pulsar Search (MeerKAT)
Back in early August, the few I got presumably were enabled by my permission to run test applications under the Beta settings header. Just now I've ticked the MeerKAT box for the three locations I actually use.
I don't expect to get anything until the project turns an AMD GPU application running on Windows on again.
Works on gtx 2080 super it
)
Works on gtx 2080 super it seems https://einsteinathome.org/da/host/12870120/tasks/4/57
Ian&Steve C. wrote: they
)
Just got one where a v0.01 BRP7-opencl-ati validated with a v0.03 BRP7-cuda55 (both Linux), and tossed my v0.03 Windows out of the window.
Workunit 661244882