That the gpu runs 2 after download i didn´t know.That dazed me :).
not too long ago, when we had to use an app_info.xml file to change the count parameter, the change would work right away. now that the count parameter is manipulated through our Einstein@Home web preferences (server-side instead of host-side), it takes some time for the change to filter through.
What do you think about the new GTX 650 or 660? I think the 650 is very costefficient in its tdp from 64 watts especially for people which have an GTX 550Ti or older its interesting and the price around 100 Dollar/Euro looks ok.
If someone tested it pls post the times and how many work you simultaniously compute i know the cards are out now on 13th but hey "I can wait" :)
What do you think about the new GTX 650 or 660? I think the 650 is very costefficient in its tdp from 64 watts especially for people which have an GTX 550Ti or older its interesting and the price around 100 Dollar/Euro looks ok.
If someone tested it pls post the times and how many work you simultaniously compute i know the cards are out now on 13th but hey "I can wait" :)
I am running both the 550Ti OC and 660Ti SC here (2 cuda's at a time on each) so go ahead and take a look at my hosts.
One thing for sure is that the 550Ti OC is one 3rd the cost of the 660Ti SC
The GTX 660 (w/o Ti) might be a different story tho, it's brand new (released yesterday as mentioned above).
What is strange about the GTX 650 is the power connector that this thing seems to have (at least in the reference design). With 65 W TDP, it usually would not need an extra connector. But this might give it some headroom for overclocking?
The GTX 660 (w/o Ti) might be a different story tho, it's brand new (released yesterday as mentioned above).
What is strange about the GTX 650 is the power connector that this thing seems to have (at least in the reference design). With 65 W TDP, it usually would not need an extra connector. But this might give it some headroom for overclocking?
HB
I guess the headroom is there for OC but not only for that.
AFAIK, the max TDP its an average, but the chip is going to have short peeks above that and the memory and voltage regulators will need also some power... not sure how much but, it seems wise to not rely on the motherboard beeing so close to the 75W limit, specially with a card that supports SLI.
I am running both the 550Ti OC and 660Ti SC here (2 cuda's at a time on each) so go ahead and take a look at my hosts.
One thing for sure is that the 550Ti OC is one 3rd the cost of the 660Ti SC
As I can see your host is spending approximate 22.5 minutes for one task - 2.700 secs / 60(number of seconds in minutes) / 2(number of simultaneous tasks)
I'm running 4 WUs on 560 Ti and it gives me 17,5 minutes for one task.
Can't believe 560 Ti is faster then 660 Ti
I am running both the 550Ti OC and 660Ti SC here (2 cuda's at a time on each) so go ahead and take a look at my hosts.
One thing for sure is that the 550Ti OC is one 3rd the cost of the 660Ti SC
As I can see your host is spending approximate 22.5 minutes for one task - 2.700 secs / 60(number of seconds in minutes) / 2(number of simultaneous tasks)
I'm running 4 WUs on 560 Ti and it gives me 17,5 minutes for one task.
Can't believe 560 Ti is faster then 660 Ti
you're not comparing apples to apples though...
firstly, MAGIC's GTX 660 Ti has enough VRAM to run 4 simultaneous BRP4 tasks too, but he chooses to only run 2 at a time. so until he starts running 4 simultaneous tasks like you are (or alternatively, until you start running only 2 simultaneous tasks like he is), we really can't accurately conclude that your GTX 560 Ti is faster than his GTX 660 Ti, let alone make the assertion that the GTX 560 Ti is better than the GTX 660 Ti for Einstein@Home (or anything else for that matter).
secondly, we have no idea what other DC projects MAGIC's GTX 660 Ti host might be running, and therefore we have no idea what his BRP4 CUDA tasks are contending w/ regarding CPU resource allocation. if his CPU is even mildly over-allocated, it could put his GTX 660 Ti far off its true potential compute capability. in any event, the chances that all other factors (such as OS, CPU & other hardware, other currently running DC projects, etc.) are truly equal between your host and his (save for the difference in GPUs) is slim to none. in other words, the difference in GPUs is probably only one of several differences between your host and his...and as such, we can only compare the run times of your 560 Ti and his 660 Ti so accurately.
*EDIT* - just to give you a point of reference, see my previous post. you'll see that i'm running 3 at a time on each of my GTX 560 Ti's, and that i'm averaging 1200s (20 minutes) per task. but you'll also notice that my CPU crunches some combination of LHC@Home SixTrack, theSkyNet POGS, and Test4Theory@Home (bi-threaded tasks only) in addition to servicing 6 GPU tasks at any given time. if i weren't running other projects, i'm sure my Einstein BRP4 run times would be substantially lower.
My English isn´t the best
)
My English isn´t the best but i´m laughing thanks for your fast answers.
That the gpu runs 2 after download i didn´t know.That dazed me :).
RE: That the gpu runs 2
)
not too long ago, when we had to use an app_info.xml file to change the count parameter, the change would work right away. now that the count parameter is manipulated through our Einstein@Home web preferences (server-side instead of host-side), it takes some time for the change to filter through.
What do you think about the
)
What do you think about the new GTX 650 or 660? I think the 650 is very costefficient in its tdp from 64 watts especially for people which have an GTX 550Ti or older its interesting and the price around 100 Dollar/Euro looks ok.
If someone tested it pls post the times and how many work you simultaniously compute i know the cards are out now on 13th but hey "I can wait" :)
RE: What do you think about
)
I am running both the 550Ti OC and 660Ti SC here (2 cuda's at a time on each) so go ahead and take a look at my hosts.
One thing for sure is that the 550Ti OC is one 3rd the cost of the 660Ti SC
The GTX 660 (w/o Ti) might be
)
The GTX 660 (w/o Ti) might be a different story tho, it's brand new (released yesterday as mentioned above).
What is strange about the GTX 650 is the power connector that this thing seems to have (at least in the reference design). With 65 W TDP, it usually would not need an extra connector. But this might give it some headroom for overclocking?
HB
Yes over that i read
)
Yes over that i read yesterday too.
Maybe it overtakes it´s big brother the 660 :)
I´ve an gts250 one task more
)
I´ve an gts250 one task more doesn´t go so i was thinking over a new model and the 650 could it perhaps be.I´ve not done my choice at this time.
RE: The GTX 660 (w/o Ti)
)
I guess the headroom is there for OC but not only for that.
AFAIK, the max TDP its an average, but the chip is going to have short peeks above that and the memory and voltage regulators will need also some power... not sure how much but, it seems wise to not rely on the motherboard beeing so close to the 75W limit, specially with a card that supports SLI.
RE: I am running both the
)
As I can see your host is spending approximate 22.5 minutes for one task - 2.700 secs / 60(number of seconds in minutes) / 2(number of simultaneous tasks)
I'm running 4 WUs on 560 Ti and it gives me 17,5 minutes for one task.
Can't believe 560 Ti is faster then 660 Ti
RE: RE: I am running
)
you're not comparing apples to apples though...
firstly, MAGIC's GTX 660 Ti has enough VRAM to run 4 simultaneous BRP4 tasks too, but he chooses to only run 2 at a time. so until he starts running 4 simultaneous tasks like you are (or alternatively, until you start running only 2 simultaneous tasks like he is), we really can't accurately conclude that your GTX 560 Ti is faster than his GTX 660 Ti, let alone make the assertion that the GTX 560 Ti is better than the GTX 660 Ti for Einstein@Home (or anything else for that matter).
secondly, we have no idea what other DC projects MAGIC's GTX 660 Ti host might be running, and therefore we have no idea what his BRP4 CUDA tasks are contending w/ regarding CPU resource allocation. if his CPU is even mildly over-allocated, it could put his GTX 660 Ti far off its true potential compute capability. in any event, the chances that all other factors (such as OS, CPU & other hardware, other currently running DC projects, etc.) are truly equal between your host and his (save for the difference in GPUs) is slim to none. in other words, the difference in GPUs is probably only one of several differences between your host and his...and as such, we can only compare the run times of your 560 Ti and his 660 Ti so accurately.
*EDIT* - just to give you a point of reference, see my previous post. you'll see that i'm running 3 at a time on each of my GTX 560 Ti's, and that i'm averaging 1200s (20 minutes) per task. but you'll also notice that my CPU crunches some combination of LHC@Home SixTrack, theSkyNet POGS, and Test4Theory@Home (bi-threaded tasks only) in addition to servicing 6 GPU tasks at any given time. if i weren't running other projects, i'm sure my Einstein BRP4 run times would be substantially lower.