I have two GPU , one is GT 640 and another is Radeon 280.
If I use on Primegrid both card, then I always get 99% utilization of GPU and it is clearly that Radeon 280 is around 2 x faster then Gt 640 in sieve tasks.
I use ( for my project) program I get from https://sites.google.com/site/kenscode/prime-programs
After one hour of experimenting with values and switches, I got 88 Mp/s on GT 640 and 97 % GPU load
Syntax is -p166729534762177 -P200000000000000 -i sr_2.abcd -f factors.txt -r 3600 -m 127 -B 65535 -C 65535
In meantime CPU utilization is around 5% for that CUDA APP
But not even after hours and countless combination of switches I cannot get more then 30 % GPU utilization on Radeon 280. Top speed is 28 Mp/s and with that condition it uses nearly full CPU core.
Does anybody here know how to increase speed on sieving but using only one CPU core ( so I cannot start more threads since that will use more then one CPU core I can spent.
Thanks fro any reply
92*10^1439761-1 NEAR-REPDIGIT PRIME :) :) :)
4 * 650^498101-1 CRUS PRIME
314187728^131072+1 GENERALIZED FERMAT
Proud member of team Aggie The Pew. Go Aggie!