Optimal projects for older hardware

Moderators: Site Moderators, PandeGroup

Optimal projects for older hardware

Postby inoculate86 » Sun Jan 31, 2016 5:30 am

I understand that there is a continuous need to push new hardware to its limits with more science, but what about people with older hardware? I have a GTX 590 and I'm wondering should I set max-packet-size to small? medium? or what? I don't see the point in running work units that are benchmarked for newer hardware when there should be work units designed for this hardware. To get optimal output for this hardware, how should I proceed?
inoculate86
 
Posts: 27
Joined: Tue Nov 29, 2011 1:04 am

Re: Optimal projects for older hardware

Postby bruce » Sun Jan 31, 2016 7:39 am

No.

Each half of a GTX 590 contains a Fermi chip. FAH has full support for Fermi, Kepler, and Maxwell hardware. The "older hardware" you're referring to is the Tesla architecture (e.g. G8x, G9x, GT20x). Support for Tesla (and earlier) is more limited and may require the special settings.

NVidia has done a good job of maintaining drivers which support FAH's computational requirements.
bruce
Site Admin
 
Posts: 20454
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: Optimal projects for older hardware

Postby 7im » Sun Jan 31, 2016 4:39 pm

bruce wrote:No.

Each half of a GTX 590 contains a Fermi chip. FAH has full support for Fermi, Kepler, and Maxwell hardware. The "older hardware" you're referring to is the Tesla architecture (e.g. G8x, G9x, GT20x). Support for Tesla (and earlier) is more limited and may require the special settings.

NVidia has done a good job of maintaining drivers which support FAH's computational requirements.

Yes, but this assumes folding on a 24/7 basis or near to it. Folding the newest work units will always be optimal for the project. Sometimes Fah can offer work that is less urgent for older, slower, or part time folders. But not always, and as mentioned before, the core_15 work units are under an End of Life notice, and can be pulled at any time.
How to provide enough information to get helpful support
Tell me and I forget. Teach me and I remember. Involve me and I learn.
User avatar
7im
 
Posts: 15179
Joined: Thu Nov 29, 2007 4:30 pm
Location: Arizona

Re: Optimal projects for older hardware

Postby mmonnin » Tue Feb 02, 2016 1:03 am

I can still get ~50k PPD from a GTX570. A 590 is still completely relevant.
mmonnin
 
Posts: 365
Joined: Wed Dec 05, 2007 1:27 am


Return to General GPU client issues

Who is online

Users browsing this forum: Yahoo [Bot] and 2 guests

cron