I’ve been chugging along at more than 8M cEMs/sec, but around Jan. 1st my rate went down to around 6 cEMs/sec; I thought one of my machines went offline (judging from the rate decrease), so I rebooted it but the rate continued to plummet. It’s now a little less than 5M cEMs/sec.
Looking in Task Manager, it shows two SB processes running, both at 50% (this is a dual core). Even if my other machines are offline, my dual core should be crunching at 6-7 M cEMs/sec.
Any ideas what’s going on here?
No definity answer, but it is possible we have just passed another FFT boundry.
did your machines maybe pick up some secondpass tests?
You can tell by checking the n values and if they’re exceptionally low (firstpass n is currently > 12,000,000)
It looks like someone dropped some secondpass tests and I think all dropped-tests are delegated if they’re less than the firstpass queue…
see here:
http://www.seventeenorbust.com/secret/
due to an "error" in the cEMs/sec calculation, higher n values generated higher cEMs/sec - this has been discussed in the SoB forums quite a bit, and there was a vague promise that they would correct the problem with the 3.x client (and readjust all work done to be fair)…
Update: my mistake - those are in the 50M and greater range (boy, someone was screwing around!)
still, you may have ended up with a lower n - since i see the dropped test queue grew by 44M in the last 24 hours. That suggests there were some ~5M n values in there recently
Another possibility is that one or more of your clients all of a sudden are failing to report their intermediate blocks - you can check the log files and make sure they didn’t fail during a block submission (this happened a lot when the project was having DNS issues)
That’s all foreign tongue to me… blocks and intermediate gigapasses, etc. 
But here is some of the latest logged actions:
[Thu Jan 04 17:54:20 2007] resolving hostname [Thu Jan 04 17:54:20 2007] opening connection [Thu Jan 04 17:54:21 2007] logging into server [Thu Jan 04 17:54:21 2007] login successful [Thu Jan 04 17:54:21 2007] n.high = 12212574 . 311 blocks left in test [Thu Jan 04 17:55:54 2007] resolving hostname [Thu Jan 04 17:55:54 2007] opening connection [Thu Jan 04 17:55:54 2007] logging into server [Thu Jan 04 17:55:54 2007] login successful [Thu Jan 04 17:55:54 2007] n.high = 3805038 . 5783 blocks left in test [Thu Jan 04 17:56:06 2007] resolving hostname [Thu Jan 04 17:56:06 2007] opening connection [Thu Jan 04 17:56:06 2007] logging into server [Thu Jan 04 17:56:06 2007] login successful [Thu Jan 04 17:56:06 2007] n.high = 12214125 . 310 blocks left in test [Thu Jan 04 17:57:37 2007] resolving hostname [Thu Jan 04 17:57:37 2007] opening connection [Thu Jan 04 17:57:38 2007] logging into server [Thu Jan 04 17:57:38 2007] login successful [Thu Jan 04 17:57:38 2007] n.high = 3806581 . 5782 blocks left in test [Thu Jan 04 17:57:50 2007] resolving hostname [Thu Jan 04 17:57:50 2007] opening connection [Thu Jan 04 17:57:50 2007] logging into server [Thu Jan 04 17:57:50 2007] login successful [Thu Jan 04 17:57:50 2007] n.high = 12215676 . 309 blocks left in test [Thu Jan 04 17:59:27 2007] resolving hostname [Thu Jan 04 17:59:28 2007] opening connection [Thu Jan 04 17:59:28 2007] logging into server [Thu Jan 04 17:59:28 2007] login successful [Thu Jan 04 17:59:28 2007] n.high = 3808124 . 5781 blocks left in test [Thu Jan 04 17:59:34 2007] resolving hostname [Thu Jan 04 17:59:34 2007] opening connection [Thu Jan 04 17:59:34 2007] logging into server [Thu Jan 04 17:59:34 2007] login successful [Thu Jan 04 17:59:34 2007] n.high = 12217227 . 308 blocks left in test [Thu Jan 04 18:01:13 2007] resolving hostname [Thu Jan 04 18:01:14 2007] opening connection [Thu Jan 04 18:01:14 2007] logging into server [Thu Jan 04 18:01:15 2007] login successful [Thu Jan 04 18:01:15 2007] n.high = 3809667 . 5780 blocks left in test
Does that tell anything interesting?
Does that tell anything interesting?
All’s good, nothing to see here.
(Did you mean to have two clients running btw?)
Looks good - yes.
Can see from the log that that machine is at least working on one 12M n value - and based on the remaining block count of the other one, it probably is also.
But what are the n values that your other clients/machines are running?
Easiest way to tell is to go to the seventeenorbust.com website, login, check your preferences, and look at the "pending test management" page there.
That will show which k/n pairs your machines are working on, when they last reported a block, what percentage complete they are, and what the average cEMs/sec each is producing.
Might help identify if one or more of your machines are failing to submit blocks, or running a low n value test.
if you’re already logged into the site (using the “Log in” link at the top left), you can go here for that info:
http://www.seventeenorbust.com/account/testsPending.mhtml
here are 2 of my tests (sorry for the horrible copy/paste):
749002 21181ââ¬Â¢2^6920948+1 66.92.x.x Mon Dec 11 05:52:24 2006 Fri Jan 5 00:19:36 2007 4879765 70 % 109221 750050 19249ââ¬Â¢2^12643178+1 66.92.x.x Fri Dec 15 19:24:46 2006 Fri Jan 5 00:35:43 2007 12088242 95 % 1106284
You can see that my first one there is a secondpass test (with an n value of 6920948) while the second one is a first pass test (with an n value of 12643178)
I have four currently listed, and they are all in the 12M range.
(Did you mean to have two clients running btw?)
Dual core.
Everything looks normal, but I too am experience slower than normal rate. If you will look Here you will notice the steady decline of the projects overall rate across the board since late December.
This leads me to believe we just crossed or are nearing a FFT Boundry. The FFT size/boundry is slightly different for Intel vs. AMD. The closer one gets to the change over point the better AMD’s perform compared to Intel, once reached, Intel’s again take the advantage.
If memory serves, and past experince holds true, there will be another change over in the vicinity of 18.5M to 19M.
This leads me to believe we just crossed or are nearing a FFT Boundry. The FFT size/boundry is slightly different for Intel vs. AMD. The closer one gets to the change over point the better AMD's perform compared to Intel, once reached, Intel's again take the advantage.
I guess that’s likely then… I had attributed my recent drop to running BOINC side-by-side with half my SoB clients – but maybe it’s more than that 
Okay, I just checked my stats and I’m at around 7M cEMs/sec, still not what I was doing but not nearly as bad as I was doing either.
Maybe it is the FFT boundary thing. I’m only deficient about 1M cEMs/sec. What about you guys?
Edit:
Also, according to my 24-hour graph, I had a 2M cEMs/sec spike around midnight last night; it hasn’t went down yet. Very strange as I wasn’t using the computer at the time or running any software on it to account for that sort of rate change.