what gpus will be supported in a near future with the new code and/or the new gpu subproject?

I believe all cuda nvidia gpus: geforce 8xxx-today

will you support ati cards that aren't supported at this time? below 38xx cards (like 3650 and the 2xxx series)?
As these support gpu computing too...

what gpus will be supported in a near future with the new code and/or the new gpu subproject?

I believe all cuda nvidia gpus: geforce 8xxx-today

will you support ati cards that aren't supported at this time? below 38xx cards (like 3650 and the 2xxx series)?
As these support gpu computing too...

If the separate GPU project with another code base and using single precision is dead, then these older cards won't be able contribute. Would be a pity, as I have a working version for all supported ATI cards (the complete HD2000, HD3000 and HD4000 series) using that code base for six weeks now. But if the project decides that single precision isn't good enough, we will have to live with that.

Just a bit of speculation, I think the project might need to broaden its individual searches before having a separate less precise 'first pass' makes sense. Whether that's done by looking at larger wedges (to the point where approximating with a cylinder isn't good enough anymore?) or some other means. Until then I think the searches are quick enough that getting server-side support for a two-pass search in place is more effort than it's worth.

For the CUDA application a CUDA 1.3 capable card is required for double precision support, this translates to a NVIDIA GeForce GTX 295, 285, 280, and 260. These cards are required because they are the only ones that support double precision.

have u trying simulate DP on lower cards (esp NVIDIA)?

my gtx 480 not supported???? all i get is computation errors... its driving me nuts.. all this potential n upable to use it..