![]() |
|
|
#12 |
|
Einyen
Dec 2003
Denmark
61278 Posts |
These large vectors are only a performance improvement for multithreading MPI=1 right?
|
|
|
|
|
|
#13 |
|
Tribal Bullet
Oct 2004
3,541 Posts |
They apply with or without MPI; regular mulithreaded runs can use large vectors too.
|
|
|
|
|
|
#14 |
|
"Carlos Pinho"
Oct 2011
Milton Keynes, UK
115238 Posts |
Is your 15% speed up on LA phase? I’ve tested Tom’s example on a GNFS130 and I only had a speed up on the filtering stage of 40%. Left the LA stabilise at 10%.
Last fiddled with by pinhodecarlos on 2018-02-10 at 20:50 |
|
|
|
|
|
#15 |
|
Tribal Bullet
Oct 2004
1101110101012 Posts |
Yes, that was the speedup I get for a SNFS C160 LA job on a Sandy Bridge laptop. I also see it for smaller problems.
|
|
|
|
|
|
#16 |
|
Jul 2003
So Cal
40728 Posts |
I haven't had time to work on this in a while, but I hope to get back to it soon. There's still a problem with the MPI version. The dependencies file created isn't valid, so sqrt doesn't run. There were changes here for the CUDA version to prevent having the full vector on the first GPU, and there's breakage somewhere.
|
|
|
|
![]() |
| Thread Tools | |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Resume linear algebra | Timic | Msieve | 35 | 2020-10-05 23:08 |
| Has anyone tried linear algebra on a Threadripper yet? | fivemack | Hardware | 3 | 2017-10-03 03:11 |
| Linear algebra at 600% | CRGreathouse | Msieve | 8 | 2009-08-05 07:25 |
| Linear algebra crashes | 10metreh | Msieve | 3 | 2009-02-02 08:34 |
| Linear algebra proof | Damian | Math | 8 | 2007-02-12 22:25 |