This yields 37k RAC from the GTX750Ti, whereas your current GTX550Ti should produce 19k RAC if ran 24/7.
20 nm Maxwells will undoubtly be more energy efficient crunchers, but we don't know when they'll arrive in mass quantities (expect end of year) and when they'll reach sane prices (expect an initial markup for the new stuff - there's lot's of demand for more efficient GPUs).
20 nm Maxwells will undoubtly be more energy efficient crunchers, but we don't know when they'll arrive in mass quantities (expect end of year) and when they'll reach sane prices (expect an initial markup for the new stuff - there's lot's of demand for more efficient GPUs).
MrS
I just read a pretty good argument that both Nvidia and AMD will stick with 28 nm until the FinFET's arrive at 16 nm (don't hold your breath for that one). If so, then the GTX 750 Ti is not an anomaly, but just the first in a series. http://techsoda.com/no-20nm-graphics-amd-nvidia/
Everything you say is otherwise undoubtedly correct, except that maybe they could arrive a little sooner.
This yields 37k RAC from the GTX750Ti, whereas your current GTX550Ti should produce 19k RAC if ran 24/7.
Filipe's host is PCIe2 so the 750Ti won't produce results at the level of Alex's host. The Q8400 is more of a bottleneck than the 550Ti in RAC production.
I run 2 550Ti single project hosts. My Q9400 host at Albert@H runs x2 GPU(BRP4, BRP5, GR3) + 1 CPU BRP4 stream, averages 29k RAC w/ GPU loading ~83%. This host will be comparable to Filipe's. The E@H host runs x4 GPU BRP5 only and averages 32k RAC.
There might be room in Filipe's system for additional performance gains without any card change...
The Q8400 is more of a bottleneck than the 550Ti in RAC production.
Yes the 8400 is pretty slow at einstein. (I believe because of low amount of L2 cache)
I've a Q9000 at only 2Ghz i'ts a lot faster than the Q8400 2.66Ghz at S6 Search. So, i'don't really use the cpu because it's not eficient and use a lot of electricity.
I'm searching for a GPU +/-40000RAC and not power hungry. (The Ati's are really fast here but waste a lot of power.
Anyone with a cpu with a lot of L2 cache? maybe an i7? Who can confirm that the S6 Search is heavily dependent on the amount of the L2 cache?
I think you can go into Task Manager and then under the processes tab find the unit and how much memory it is taking. If it is more then your L2 cache then you are in trouble as the units is constantly swapping out to the harddrive instead of being able to keep the whole thing in memory. I have an i7 running those units doing 5 at a time and the largest one right now is taking 273,604K of memory and taking just short of 8 hours each to finish. I was running only 4 units at once but did not notice a change in time when I added a 5th unit. Obviously I DO have HT turned on on my laptop.
Anyone with a cpu with a lot of L2 cache? maybe an i7? Who can confirm that the S6 Search is heavily dependent on the amount of the L2 cache?
I think you can go into Task Manager and then under the processes tab find the unit and how much memory it is taking. If it is more then your L2 cache then you are in trouble as the units is constantly swapping out to the harddrive instead of being able to keep the whole thing in memory.
Mikey, the L2 cache is a fast built in memory in the CPU it's got nothing to do with how much RAM one have installed in the computer. Usually CPUs have L1 (level 1) cache for both instructions and data, then they have L2 cache for data and sometimes also L3 cache for data. The L2 cache on my i7 3770K is only 4x256KB so 1 MB, the L3 on the other hand is 8 MB.
S6 tasks take ~9.5 hours, I run the CPU @ 4.2 GHz HT=on and load 5 tasks on it along with one on the built in HD 4000 and 2 on the GTX 660TIi.
I have an Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz [Family 6 Model 58 Stepping 9]
(8 processors) running on Ubuntu with a GTX 770 (2047MB). My cache sizing is: L1 32 kb, L2 256 kb, and L3 8192 kb.
It takes around 8 hours for an S6 to complete. This machine is dedicated to E@H and runs 24/7. It processes 3 GPU jobs and 5/6 CPU jobs depending on the mix.
Hardware upgrade
)
This pc runs einstein on a GTX 750ti
http://einsteinathome.org/host/6801076/tasks
2 nVidia gpu wu's, two igpu wu's and 5 cpu wu's are running together. No overclocking.
RE: This pc runs einstein
)
This yields 37k RAC from the GTX750Ti, whereas your current GTX550Ti should produce 19k RAC if ran 24/7.
20 nm Maxwells will undoubtly be more energy efficient crunchers, but we don't know when they'll arrive in mass quantities (expect end of year) and when they'll reach sane prices (expect an initial markup for the new stuff - there's lot's of demand for more efficient GPUs).
MrS
Scanning for our furry friends since Jan 2002
RE: 20 nm Maxwells will
)
I just read a pretty good argument that both Nvidia and AMD will stick with 28 nm until the FinFET's arrive at 16 nm (don't hold your breath for that one). If so, then the GTX 750 Ti is not an anomaly, but just the first in a series.
http://techsoda.com/no-20nm-graphics-amd-nvidia/
Everything you say is otherwise undoubtedly correct, except that maybe they could arrive a little sooner.
RE: This yields 37k RAC
)
Filipe's host is PCIe2 so the 750Ti won't produce results at the level of Alex's host. The Q8400 is more of a bottleneck than the 550Ti in RAC production.
I run 2 550Ti single project hosts. My Q9400 host at Albert@H runs x2 GPU(BRP4, BRP5, GR3) + 1 CPU BRP4 stream, averages 29k RAC w/ GPU loading ~83%. This host will be comparable to Filipe's. The E@H host runs x4 GPU BRP5 only and averages 32k RAC.
There might be room in Filipe's system for additional performance gains without any card change...
Gord
RE: The Q8400 is more of a
)
Yes the 8400 is pretty slow at einstein. (I believe because of low amount of L2 cache)
I've a Q9000 at only 2Ghz i'ts a lot faster than the Q8400 2.66Ghz at S6 Search. So, i'don't really use the cpu because it's not eficient and use a lot of electricity.
I'm searching for a GPU +/-40000RAC and not power hungry. (The Ati's are really fast here but waste a lot of power.
Thanks for the answers.
Anyone with a cpu with a lot
)
Anyone with a cpu with a lot of L2 cache? maybe an i7? Who can confirm that the S6 Search is heavily dependent on the amount of the L2 cache?
RE: Anyone with a cpu with
)
I think you can go into Task Manager and then under the processes tab find the unit and how much memory it is taking. If it is more then your L2 cache then you are in trouble as the units is constantly swapping out to the harddrive instead of being able to keep the whole thing in memory. I have an i7 running those units doing 5 at a time and the largest one right now is taking 273,604K of memory and taking just short of 8 hours each to finish. I was running only 4 units at once but did not notice a change in time when I added a 5th unit. Obviously I DO have HT turned on on my laptop.
RE: taking just short of 8
)
8hours? Mine is taking close to 20 hours!!!
That's a lot of diferrence
RE: RE: Anyone with a cpu
)
Mikey, the L2 cache is a fast built in memory in the CPU it's got nothing to do with how much RAM one have installed in the computer. Usually CPUs have L1 (level 1) cache for both instructions and data, then they have L2 cache for data and sometimes also L3 cache for data. The L2 cache on my i7 3770K is only 4x256KB so 1 MB, the L3 on the other hand is 8 MB.
S6 tasks take ~9.5 hours, I run the CPU @ 4.2 GHz HT=on and load 5 tasks on it along with one on the built in HD 4000 and 2 on the GTX 660TIi.
I have an Intel(R) Core(TM)
)
I have an Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz [Family 6 Model 58 Stepping 9]
(8 processors) running on Ubuntu with a GTX 770 (2047MB). My cache sizing is: L1 32 kb, L2 256 kb, and L3 8192 kb.
It takes around 8 hours for an S6 to complete. This machine is dedicated to E@H and runs 24/7. It processes 3 GPU jobs and 5/6 CPU jobs depending on the mix.