I really do not know how to create an app_info.xml for running 2 WU on a single capable card. I wonder if I can use any app_info.xml too, and also in which folder to store it. I need help, lots of them.
I really do not know how to create an app_info.xml for running 2 WU on a single capable card. I wonder if I can use any app_info.xml too, and also in which folder to store it. I need help, lots of them.
I wonder how the new low-end Kepler cards would fit in here, especially the retail GT 640. I understand that there are completely different cards sold under the GT 640 label: those based on the Fermi chip, and newer ones using Kepler chips. If you are not sure which is which: if core and shader clock are the same ==> Kepler, otherwise it's a Fermi.
I think they are sold for under 100€, consume only around 65 W and still should be able to crunch a BRP4 unit in little more than 2100 sec or so if you run two in parallel. Can someone confirm this? This would make an excellent price/performance ratio.
I wonder how the new low-end Kepler cards would fit in here, especially the retail GT 640. I understand that there are completely different cards sold under the GT 640 label: those based on the Fermi chip, and newer ones using Kepler chips. If you are not sure which is which: if core and shader clock are the same ==> Kepler, otherwise it's a Fermi.
I think they are sold for under 100€, consume only around 65 W and stil should be able to crunch a BRP4 unit in little more than 2100 sec or so if you run two in parallel. Can someone confirm this? This would make an excellent price/performance ratio.
Any experience with this card?
CU
HBE
I would also be interested to know how the GT 640 handles with this project. In particular, I like this card that EVGA has designed:
I like the small form factor and the high flow bracket on the back.The card has two SMX units (384 cores) and PCI-E 3.0 support which should be beneficial for the project here.
The main thing that I do not like at this stage is that the newer Linux NVIDIA drivers (> 295.33) which support this card have PCI-E 3.0 disabled with no way that I could find to enable the link speed at 8 GT/s. I sent in a support ticket with NVIDIA for this issue and I am hoping that they will have some kind of solution for this. NVIDIA did release a patch program for Windows to set the link speed at 8 GT/s.
Over at anandtech this card was tested and the result is: good as htpc but horrible at gaming.
This card has DDR3 and starving for bandwidth.
In folding it is below a GTS450.
Over at anandtech this card was tested and the result is: good as htpc but horrible at gaming.
This card has DDR3 and starving for bandwidth.
In folding it is below a GTS450.
Thanks for the link!
That's kind of disappointing, but still, to be sure, I'd like to see performance figures over here at E@H. It's obvious that a 65 W card can't do wonders in terms of performance But I would have hoped that it outperforms my GTS 450 on BRP4 tasks.
When I´m runnig it with 2 tasks it will take around 2h to finish both. CPU @ 3.2GHz + only PCI-E 1.0. Maybe thats the problem. But the powerconsumption is way worse than with the newer cards.
I really do not know how to
)
I really do not know how to create an app_info.xml for running 2 WU on a single capable card. I wonder if I can use any app_info.xml too, and also in which folder to store it. I need help, lots of them.
RE: I really do not know
)
check your pm's...
RE: Here you go Bikeman - A
)
I wonder how the new low-end Kepler cards would fit in here, especially the retail GT 640. I understand that there are completely different cards sold under the GT 640 label: those based on the Fermi chip, and newer ones using Kepler chips. If you are not sure which is which: if core and shader clock are the same ==> Kepler, otherwise it's a Fermi.
I think they are sold for under 100€, consume only around 65 W and still should be able to crunch a BRP4 unit in little more than 2100 sec or so if you run two in parallel. Can someone confirm this? This would make an excellent price/performance ratio.
Any experience with this card?
CU
HBE
RE: I wonder how the new
)
I would also be interested to know how the GT 640 handles with this project. In particular, I like this card that EVGA has designed:
http://www.evga.com/products/moreInfo.asp?pn=02G-P4-2643-KR
I like the small form factor and the high flow bracket on the back.The card has two SMX units (384 cores) and PCI-E 3.0 support which should be beneficial for the project here.
The main thing that I do not like at this stage is that the newer Linux NVIDIA drivers (> 295.33) which support this card have PCI-E 3.0 disabled with no way that I could find to enable the link speed at 8 GT/s. I sent in a support ticket with NVIDIA for this issue and I am hoping that they will have some kind of solution for this. NVIDIA did release a patch program for Windows to set the link speed at 8 GT/s.
Jeroen
Over at anandtech this card
)
Over at anandtech this card was tested and the result is: good as htpc but horrible at gaming.
This card has DDR3 and starving for bandwidth.
In folding it is below a GTS450.
RE: Over at anandtech this
)
Thanks for the link!
That's kind of disappointing, but still, to be sure, I'd like to see performance figures over here at E@H. It's obvious that a 65 W card can't do wonders in terms of performance But I would have hoped that it outperforms my GTS 450 on BRP4 tasks.
Cheers
HBE
the specs for the GT640 look
)
the specs for the GT640 look similar to my old GTX260.
http://en.wikipedia.org/wiki/Comparison_of_Nvidia_graphics_processing_units#GeForce_200_Series
http://en.wikipedia.org/wiki/GeForce_600_Series
When I´m runnig it with 2 tasks it will take around 2h to finish both. CPU @ 3.2GHz + only PCI-E 1.0. Maybe thats the problem. But the powerconsumption is way worse than with the newer cards.