- Status
EXT64
Splendid
- Mar 1, 2007
- 8,069
- 10
- 31,165
This is a bit out of date: http://foldingforum.org/viewtopic.php?f=38&t=23440
I guess we could always start our own (we have a decent number of GPUs here) 😀
I guess we could always start our own (we have a decent number of GPUs here) 😀
Unolocogringo
Illustrious
- Dec 31, 2007
- 8,028
- 814
- 38,390
GTX 460 800/1600/1800
open cl -F@H
ES= 10.71ns day
IS= 50.5904ns day
Don't know what they represent except for the nanoseconds of folding achieved per day on each test.
Unolocogringo
Illustrious
- Dec 31, 2007
- 8,028
- 814
- 38,390
Also had 16-18 percent CPU usage during the test Pheno2 960 6@3.5g a lot of cpuuse for graphics calculations.
EXT64
Splendid
- Mar 1, 2007
- 8,069
- 10
- 31,165
Well, the first test is explicit solvent. In that test the individual water molecules are modeled along with the protein. Traditionally explicit has only been run on the CPU (the GPUs couldn't do them). Core 17 fixed this. The current beta project (7662) is an explicit problem.
In implicit the water is not modeled as individual molecules but rather a continuum or something like that (it is more of an approximation). This is what all traditional GPU units were.
Obviously, even though the new core can run explicit, it is still substantially faster at implicit. I imagine you will still see implicit units now and then for that reason, but the current focus I think is prove explicit is working then optimize (speed it up - there should be some big improvements in the somewhat near future).
For NVidia, both tests may use and entire CPU core (I know it does for explicit).
For AMD, Explicit should not use a core but I think that Implicit does.
Not really sure what is causing this - and sometimes strange changes can make it go away.
In implicit the water is not modeled as individual molecules but rather a continuum or something like that (it is more of an approximation). This is what all traditional GPU units were.
Obviously, even though the new core can run explicit, it is still substantially faster at implicit. I imagine you will still see implicit units now and then for that reason, but the current focus I think is prove explicit is working then optimize (speed it up - there should be some big improvements in the somewhat near future).
For NVidia, both tests may use and entire CPU core (I know it does for explicit).
For AMD, Explicit should not use a core but I think that Implicit does.
Not really sure what is causing this - and sometimes strange changes can make it go away.
lilcinw
Distinguished
- Jan 25, 2011
- 833
- 0
- 19,010
I just downloaded FAHBench and ran it on my GTX 560 Ti 825/1650/2004.
Results while SMP is running:
ES = 12.13705 ns/day
IS = 67.09348 ns/day
With SMP paused:
ES = 13.77007 ns/day
IS = 67.98907 ns/day
The results are an average of multiple test runs. I ran each test three times initially then ran a fourth test with SMP active because the deviation was so high.
I did notice that the GPU running an actual core 17 WU reported much more consistent usage while SMP was paused. There were no dips below 20% when it finished a frame and it stayed pegged at 99% instead of oscillating between 99 and 93 like it typically does.
I noticed the FAHBench web page showed results for an Intel HD 4000. Does anyone have an AMD APU they could test this on?
Results while SMP is running:
ES = 12.13705 ns/day
IS = 67.09348 ns/day
With SMP paused:
ES = 13.77007 ns/day
IS = 67.98907 ns/day
The results are an average of multiple test runs. I ran each test three times initially then ran a fourth test with SMP active because the deviation was so high.
I did notice that the GPU running an actual core 17 WU reported much more consistent usage while SMP was paused. There were no dips below 20% when it finished a frame and it stayed pegged at 99% instead of oscillating between 99 and 93 like it typically does.
I noticed the FAHBench web page showed results for an Intel HD 4000. Does anyone have an AMD APU they could test this on?
- Jul 24, 2006
- 23,093
- 357
- 57,490
Using OpenCL with all of the default options checks/unchecked...
Explicit: 15.9201 ns/day
Implicit: 73.4624 ns/day
Using a GTX 560 Ti 448 at default clocks.
Explicit: 15.9201 ns/day
Implicit: 73.4624 ns/day
Using a GTX 560 Ti 448 at default clocks.
Howdy Everyone!
I see Tom's has changed a few things around, but new members are finding their way here still - very good. Lots of folders I recognize from last year.
A heads up note for those who want to race in The Great Chimp Challenge. It will be starting in mid April this year, and will again be running 24/7 for 10 days. OCF warmly invites any 40051 folders to join us in the fun and, as before, we'll make sure your team loses nothing as a result of a temporary drop in points for Tom's.
I still say it's the most fun you can have folding, with your clothes on. 😉
It's also a natural way to invite some friends to join folding, as well.
"I'm in a race, and I'm looking for teammates to join us. Got a computer? ..."
With the new v7 folding client, it's easier to recruit than ever. Not that it's easy, because it never has been easy to recruit good folders.
The race posts will be on the OCF folding team forum:
http://www.overclockers.com/forums/forumdisplay.php?f=21
and on the race website, as well. (website under construction).
Our team goals remain:
*to melt Stanford's servers with finished work units
*to fling monkey poo and work units faster than ever
*to have a good time, race hard, and extend awareness of Folding@Home
Since we won last year, a repeat is unlikely, but we WILL defend the lucky Jaded Monkey, right down to the last CPU!
😉
I'm qualifying my passkey on the OCFthanksTomsHardware folding handle. Then we'll see what "Tanker" (the 4P) can do on it, briefly.
See you on your top 20 producers list, and I'll be back with the details, if that's OK.
I see Tom's has changed a few things around, but new members are finding their way here still - very good. Lots of folders I recognize from last year.
A heads up note for those who want to race in The Great Chimp Challenge. It will be starting in mid April this year, and will again be running 24/7 for 10 days. OCF warmly invites any 40051 folders to join us in the fun and, as before, we'll make sure your team loses nothing as a result of a temporary drop in points for Tom's.
I still say it's the most fun you can have folding, with your clothes on. 😉
It's also a natural way to invite some friends to join folding, as well.
"I'm in a race, and I'm looking for teammates to join us. Got a computer? ..."
With the new v7 folding client, it's easier to recruit than ever. Not that it's easy, because it never has been easy to recruit good folders.
The race posts will be on the OCF folding team forum:
http://www.overclockers.com/forums/forumdisplay.php?f=21
and on the race website, as well. (website under construction).
Our team goals remain:
*to melt Stanford's servers with finished work units
*to fling monkey poo and work units faster than ever
*to have a good time, race hard, and extend awareness of Folding@Home
Since we won last year, a repeat is unlikely, but we WILL defend the lucky Jaded Monkey, right down to the last CPU!
😉
I'm qualifying my passkey on the OCFthanksTomsHardware folding handle. Then we'll see what "Tanker" (the 4P) can do on it, briefly.
See you on your top 20 producers list, and I'll be back with the details, if that's OK.
- Jul 24, 2009
- 4,169
- 5
- 22,965
Welcome back to tom's Adak1!
Yeah, not too long ago our forums got updated but were still not in our own area yet. Although, according to our mods, something is in the work now that we've got this update.
As for the chip challenge, I almost forgot that was going to start here soon. Still thinking it's going to be later (I been listening off and on that the chip challenge was being scheduled for the cooler spring). So thanks for the notice!
Anyways, You'll have my support like last year. At least this time I'm packing a bit more power than my laptop (core i7 920 supplied by EXT).
Yeah, not too long ago our forums got updated but were still not in our own area yet. Although, according to our mods, something is in the work now that we've got this update.
As for the chip challenge, I almost forgot that was going to start here soon. Still thinking it's going to be later (I been listening off and on that the chip challenge was being scheduled for the cooler spring). So thanks for the notice!
Anyways, You'll have my support like last year. At least this time I'm packing a bit more power than my laptop (core i7 920 supplied by EXT).
- Jul 24, 2009
- 4,169
- 5
- 22,965
Not sure if they have a page with all the details yet but here is the thread with the discussion on it.
http://www.overclockers.com/forums/showthread.php?t=724007
This year the CC will be a bit different:
* Starting in Mid April (13th is a good bet right now)
* Different point system. Hopefully, using my handicapping system
* A new (and small) team is competing
* New website for the race
* Only whole teams are racing. There is no name change required, and no sign-ups, either. If you are on a racing team, you WILL be racing, and all your points count. The only thing you need to change, is your team number to 32 - you don't need to change your folding name or handle.
This last point has many advantages. (As does my handicapping system.)
Hopefully the new team will work out well with this set up. If it does, the CC will be looking for more teams to add to the race. I'd like you guys (and gals), to consider racing in the CC as the Tom's Hardware Team, next year.
No pressure, just a fun suggestion you can run up the flagpole.
I've kept quiet on most of this stuff because at least one team has a small, but very vocal group of complainers, on this (and many other), topics. Without a doubt, they are THE most ego-centric and negative whiners ... suffice it to say the team Captains can't work around them, and get much done.
So we meet in our own forum - thank heavens!
Thanks for the welcome back, and I'll keep you updated until we get things ironed out. At that point, I'll have to ask you to keep up on the CC news, in the big thread "Chimp Challenge 2013" on Overclockers.com. I haven't started it yet this year, but it's always full of info, general jib-jab, and updates aplenty.
I'll post a link to it for you, after it's up and running.
king smp
Splendid
- Jul 16, 2010
- 9,623
- 7
- 31,515
well I upgraded to the latest drivers (still running 310 series)
my scores now are
default settings on F@H bench
IMP - 15.75
EXP - 81.50
still seems low
my 660Ti superclocked is at stock settings
I am going to try an OC and see how it looks
edit - well with OC i got this score
IMP 16.26
EXP - 83.61
well this is a real n00b question
is higher or lower better?
my scores now are
default settings on F@H bench
IMP - 15.75
EXP - 81.50
still seems low
my 660Ti superclocked is at stock settings
I am going to try an OC and see how it looks
edit - well with OC i got this score
IMP 16.26
EXP - 83.61
well this is a real n00b question
is higher or lower better?
- Jul 24, 2006
- 23,093
- 357
- 57,490
- Status
TRENDING THREADS
-
-
-
Discussion What's your favourite video game you've been playing?- Started by amdfangirl
- Replies: 4K
-
Question Cant set new nvme as my boot device despite it fully working
- Started by gary_murphy101
- Replies: 10
-
-
Question Do I need a heat management system for a 990 pro in a laptop?
- Started by IJustWantMoreStorage
- Replies: 3
Facebook
Twitter
Instagram