OpenCL is a port of Cuda, which is why nvidia cards work better. once AMD has had time to modify their drivers and cards, their performance will be a lot better. (on a side note, I am curious how the upcoming Intel GPU will perform with OCL)
Printable View
OpenCL is a port of Cuda, which is why nvidia cards work better. once AMD has had time to modify their drivers and cards, their performance will be a lot better. (on a side note, I am curious how the upcoming Intel GPU will perform with OCL)
the geforce 570 should be added to the list, and the ATI 69xx are double presision, so they should be bold.
also, drugdiscovery has apps for both nvidia and ATI.
CUDA on The Lattice Project too.
I have a question about PCI Express x16 versions.
The slot for the card, is it different between the original version
of PCIe x16 and version 2.0 and 2.1?
I have a computer that has the PCIe x16 slot, but most of the cards
are now version 2.0 or 2.1. I want to make sure that when I buy the
GPU card, I do not have a problem with it not being able to plug in
properly. I hope to save enough money to buy a GPU card for BOINC
crunching, but I do not want to buy a card that will not fit my motherboard.
Any help????
edit:
I see that the following is on sale at Newegg.com for $59.99
plus $9.99 for 1 year warranty and $6.98 shipping.
PowerColor AX4830 512MD3 Radeon HD 4830 512MB 256-bit
GDDR3 PCI Express 2.0 x16 HDCP Ready CrossFireX Support Video Card
There use to be a thread on the old board that pointed to a chart on
how many credits per day each GPU card does. On that chard, which
I printed out awhile ago, there is no reference to the 4830, but it
did have a 4770 at 47k and 4850 at 54k so hopefully the 4830 would
be between those two. I want the double percision for projects like
Milkyway.
PCI-e 1/2/3 are all cross-compatible. youll be able to use any mix of card/slot you want.
Thanks, I hoped that was true.
--------------------------------------
Does anyone know of some chart or graph that shows the relitive
word/credits done by the different GPU boards?
How does each ATI and nVidia GPU card compare to each other?
Which cards are better within you price level?
Months ago, I looked at an ATI HD4770 for around $100 as the lowest
price card with double precision. Now there is a 4830 for about $60, but
my info I have does not tell me how it stacked up wit the other cards
that I have info on.
I wonder if maybe we can gather info within our team about which
cards we have and how many credits per day these GPU cards are
making for their owners and which project that they are running.
Maybe if people would PM or email me, I could collect the info together
and make a chart, or graph, to help enlighten people with which cards
are better for each project that is run with them.
I could use the chart that was created/maintained by Fire$torm and then
add the info on how many credits it does daily, or how fast it takes to complete
a task for the project that is run. That way, since I cannot find the info anywhere
anymore, we can have a reference to what each card can do and then give
more knowledgeable info about the cards and which projects work better
for them.
Anyone willing to help? I am willing to collect the data to make a spreadsheet
and graph. All I need is the input in a "standardized" format that can be used
to compare the cards properly.
I am willing, but are you?
My email for SETI stuff is seti@krackedpress.com . We can do it off the
forum or we can creat a thread for this "project".
This is the type of graph I would like to do. The figures are gleaned from the Collatz
forums.
Click on this image to see the simple chart of nVidia and ATI cards.
How many minutes does it take to do a WU for the Collatz project
Attachment 102