What is the matter with the stats?!

[eatmadustch]

This just can't be right: my stats show about 300k cEM/s, but the log file of JUST ONE of my 7 computers shows 400k cEM/s! Are the stats down or what?!

my log file:


[Sun Mar 09 16:03:08 2003] restarting proth test from cache (k=22699, n=3087262) [23.0%]
[Sun Mar 09 16:08:40 2003] block processing paused
[Sun Mar 09 16:08:42 2003] block processing resumed
[Sun Mar 09 16:13:20 2003] logging into server
[Sun Mar 09 16:13:20 2003] login successful
[Sun Mar 09 16:13:20 2003] n.high = 734412 . 90 blocks left in test
[Sun Mar 09 16:23:57 2003] logging into server
[Sun Mar 09 16:23:57 2003] login successful
[Sun Mar 09 16:23:58 2003] n.high = 760641 . 89 blocks left in test
[Sun Mar 09 16:34:36 2003] logging into server
[Sun Mar 09 16:34:36 2003] login successful
[Sun Mar 09 16:34:37 2003] n.high = 786870 . 88 blocks left in test
[Sun Mar 09 16:44:59 2003] logging into server
[Sun Mar 09 16:44:59 2003] login successful
[Sun Mar 09 16:45:00 2003] n.high = 813099 . 87 blocks left in test
[Sun Mar 09 16:55:13 2003] logging into server
[Sun Mar 09 16:55:13 2003] login successful
[Sun Mar 09 16:55:14 2003] n.high = 839328 . 86 blocks left in test
[Sun Mar 09 17:05:27 2003] logging into server
[Sun Mar 09 17:05:27 2003] login successful
[Sun Mar 09 17:05:28 2003] n.high = 865557 . 85 blocks left in test
[Sun Mar 09 17:15:43 2003] logging into server
[Sun Mar 09 17:15:43 2003] login successful
[Sun Mar 09 17:15:44 2003] n.high = 891786 . 84 blocks left in test
[Sun Mar 09 17:25:57 2003] logging into server
[Sun Mar 09 17:25:58 2003] login successful
[Sun Mar 09 17:25:58 2003] n.high = 918015 . 83 blocks left in test
[Sun Mar 09 17:36:33 2003] logging into server
[Sun Mar 09 17:36:33 2003] login successful
[Sun Mar 09 17:36:34 2003] n.high = 944244 . 82 blocks left in test
[Sun Mar 09 17:48:00 2003] logging into server
[Sun Mar 09 17:48:00 2003] login successful
[Sun Mar 09 17:48:01 2003] n.high = 970473 . 81 blocks left in test
[Sun Mar 09 18:01:08 2003] logging into server
[Sun Mar 09 18:01:08 2003] login successful
[Sun Mar 09 18:01:08 2003] n.high = 996702 . 80 blocks left in test
[Sun Mar 09 18:11:42 2003] temporarily unable to connect -- block added to submit queue
[Sun Mar 09 18:22:09 2003] logging into server
[Sun Mar 09 18:22:09 2003] login successful
[Sun Mar 09 18:22:10 2003] n.high = 1049160 . 78 blocks left in test
[Sun Mar 09 18:32:40 2003] logging into server
[Sun Mar 09 18:32:40 2003] login successful
[Sun Mar 09 18:32:41 2003] n.high = 1075389 . 77 blocks left in test
[Sun Mar 09 18:43:01 2003] logging into server
[Sun Mar 09 18:43:01 2003] login successful
[Sun Mar 09 18:43:02 2003] n.high = 1101618 . 76 blocks left in test
[Sun Mar 09 18:53:19 2003] logging into server
[Sun Mar 09 18:53:19 2003] login successful
[Sun Mar 09 18:53:19 2003] n.high = 1127847 . 75 blocks left in test
[Sun Mar 09 19:03:34 2003] logging into server
[Sun Mar 09 19:03:34 2003] login successful
[Sun Mar 09 19:03:35 2003] n.high = 1154076 . 74 blocks left in test
[Sun Mar 09 19:13:54 2003] logging into server
[Sun Mar 09 19:13:54 2003] login successful
[Sun Mar 09 19:13:54 2003] n.high = 1180305 . 73 blocks left in test
[Sun Mar 09 19:25:16 2003] logging into server
[Sun Mar 09 19:25:16 2003] login successful
[Sun Mar 09 19:25:16 2003] n.high = 1206534 . 72 blocks left in test
[Sun Mar 09 19:36:14 2003] logging into server
[Sun Mar 09 19:36:14 2003] login successful
[Sun Mar 09 19:36:15 2003] n.high = 1232763 . 71 blocks left in test
[Sun Mar 09 19:46:44 2003] logging into server
[Sun Mar 09 19:46:45 2003] login successful
[Sun Mar 09 19:46:45 2003] n.high = 1258992 . 70 blocks left in test
[Sun Mar 09 19:58:16 2003] logging into server
[Sun Mar 09 19:58:16 2003] login successful
[Sun Mar 09 19:58:17 2003] n.high = 1285221 . 69 blocks left in test
[Sun Mar 09 20:09:04 2003] logging into server
[Sun Mar 09 20:09:05 2003] login successful
[Sun Mar 09 20:09:05 2003] n.high = 1311450 . 68 blocks left in test
[Sun Mar 09 20:19:25 2003] logging into server
[Sun Mar 09 20:19:26 2003] login successful
[Sun Mar 09 20:19:26 2003] n.high = 1337679 . 67 blocks left in test
[Sun Mar 09 20:30:06 2003] logging into server
[Sun Mar 09 20:30:07 2003] login successful
[Sun Mar 09 20:30:08 2003] n.high = 1363908 . 66 blocks left in test
[Sun Mar 09 20:40:23 2003] logging into server
[Sun Mar 09 20:40:24 2003] login successful
[Sun Mar 09 20:40:24 2003] n.high = 1390137 . 65 blocks left in test
[Sun Mar 09 20:50:42 2003] logging into server
[Sun Mar 09 20:50:42 2003] login successful
[Sun Mar 09 20:50:42 2003] n.high = 1416366 . 64 blocks left in test
[Sun Mar 09 21:01:00 2003] logging into server
[Sun Mar 09 21:01:00 2003] login successful
[Sun Mar 09 21:01:00 2003] n.high = 1442595 . 63 blocks left in test
[Sun Mar 09 21:11:21 2003] logging into server
[Sun Mar 09 21:11:22 2003] login successful
[Sun Mar 09 21:11:22 2003] n.high = 1468824 . 62 blocks left in test
[Sun Mar 09 21:21:40 2003] logging into server
[Sun Mar 09 21:21:41 2003] login successful
[Sun Mar 09 21:21:45 2003] n.high = 1495053 . 61 blocks left in test
[Sun Mar 09 21:32:53 2003] logging into server
[Sun Mar 09 21:32:53 2003] login successful
[Sun Mar 09 21:32:53 2003] n.high = 1521282 . 60 blocks left in test
[Sun Mar 09 21:43:40 2003] logging into server
[Sun Mar 09 21:43:41 2003] login successful
[Sun Mar 09 21:43:41 2003] n.high = 1547511 . 59 blocks left in test
[Sun Mar 09 21:54:25 2003] logging into server
[Sun Mar 09 21:54:25 2003] login successful
[Sun Mar 09 21:54:26 2003] n.high = 1573740 . 58 blocks left in test
[Sun Mar 09 22:05:13 2003] logging into server
[Sun Mar 09 22:05:14 2003] login successful
[Sun Mar 09 22:05:14 2003] n.high = 1599969 . 57 blocks left in test
[Sun Mar 09 22:15:49 2003] logging into server
[Sun Mar 09 22:15:50 2003] login successful
[Sun Mar 09 22:15:50 2003] n.high = 1626198 . 56 blocks left in test
[Sun Mar 09 22:26:10 2003] logging into server
[Sun Mar 09 22:26:10 2003] login successful
[Sun Mar 09 22:26:10 2003] n.high = 1652427 . 55 blocks left in test
[Sun Mar 09 22:36:33 2003] logging into server
[Sun Mar 09 22:36:33 2003] login successful
[Sun Mar 09 22:36:33 2003] n.high = 1678656 . 54 blocks left in test
[Sun Mar 09 22:47:06 2003] logging into server
[Sun Mar 09 22:47:06 2003] login successful
[Sun Mar 09 22:47:07 2003] n.high = 1704885 . 53 blocks left in test
[Sun Mar 09 22:58:15 2003] logging into server
[Sun Mar 09 22:58:16 2003] login successful
[Sun Mar 09 22:58:16 2003] n.high = 1731114 . 52 blocks left in test
[Sun Mar 09 23:09:00 2003] logging into server
[Sun Mar 09 23:09:00 2003] login successful
[Sun Mar 09 23:09:01 2003] n.high = 1757343 . 51 blocks left in test
[Sun Mar 09 23:19:35 2003] logging into server
[Sun Mar 09 23:19:35 2003] login successful
[Sun Mar 09 23:19:36 2003] n.high = 1783572 . 50 blocks left in test
[Sun Mar 09 23:30:19 2003] logging into server
[Sun Mar 09 23:30:19 2003] login successful
[Sun Mar 09 23:30:19 2003] n.high = 1809801 . 49 blocks left in test
[Sun Mar 09 23:40:48 2003] logging into server
[Sun Mar 09 23:40:49 2003] login successful
[Sun Mar 09 23:40:50 2003] n.high = 1836030 . 48 blocks left in test
[Sun Mar 09 23:51:13 2003] logging into server
[Sun Mar 09 23:51:17 2003] login successful
[Sun Mar 09 23:51:17 2003] n.high = 1862259 . 47 blocks left in test
[Mon Mar 10 00:02:10 2003] logging into server
[Mon Mar 10 00:02:10 2003] login successful
[Mon Mar 10 00:02:10 2003] n.high = 1888488 . 46 blocks left in test
[Mon Mar 10 00:12:50 2003] logging into server
[Mon Mar 10 00:12:50 2003] login successful
[Mon Mar 10 00:12:51 2003] n.high = 1914717 . 45 blocks left in test
[Mon Mar 10 00:23:25 2003] logging into server
[Mon Mar 10 00:23:25 2003] login successful
[Mon Mar 10 00:23:25 2003] n.high = 1940946 . 44 blocks left in test
[Mon Mar 10 00:33:45 2003] logging into server
[Mon Mar 10 00:33:45 2003] login successful
[Mon Mar 10 00:33:46 2003] n.high = 1967175 . 43 blocks left in test
[Mon Mar 10 00:44:01 2003] logging into server
[Mon Mar 10 00:44:02 2003] login successful
[Mon Mar 10 00:44:03 2003] n.high = 1993404 . 42 blocks left in test
[Mon Mar 10 00:54:37 2003] logging into server
[Mon Mar 10 00:54:37 2003] login successful
[Mon Mar 10 00:54:37 2003] n.high = 2019633 . 41 blocks left in test
please note: time in CET (SOB-Time +6h I think). The client has been showing almost 400k ever since I switched it on ...

but the attached pic shows MUCH less work done!

[Troodon]

I quote my post from the other thread:


Since I'm in this project, the stats of my daily production aren't accurate. For example, today my computer has done (and reported) 12 blocks. That's 3 GcEMs (250 McEMs per block). But my stats say 766.99 McEMs.
The total production is measured accurately. Anyone else is having this little problem?

The last day's speed calculation is accurate (last day's production/86400), but the problem is with the last day's production.

[Lagardo]

As best as I can figure, they calculate ALL of the data since yesterday (1a.m. EST this morning, to be precise) and divide by 86400 seconds to give your score -- which means you start out with zero at the beginning of the day when you haven't submitted anything yet, and over the course of the day the ENTIRE day's production "rises" and by the end of the day you see the numbers you've been doing all day long.

I can only guess why they're doing things this way (dividing by a full days worth of seconds instead of dividing by the number of seconds between 1a.m. and "now"). But it seems to be a new "feature" - I hadn't seen it before yet...

[Troodon]

They're calculating your work in the last 86400 seconds (24 hours) BUT yesterday I did those 3 GcEMs in 12 hours, so in the worst case it would be 3 GcEMs in the last 86400 seconds. Instead,the stats showed only 767 McEMs.

[eatmadustch]

How is this possible? I have 7 computers running sob, 2 of them PIV 2.53Ghz. Yet my personal stats (here (http://www.seventeenorbust.com/stats/users/user.mhtml?userID=594) ) show a poor 150k cEM/s rating continuously for last day, and they've shown such a low rate for quite a while. I've tried clearing the cache on some of the machines, but that doesn't seem to work. What's wrong? Is it the stats system that isn't working? Is the work of some of PCs just not being accepted? Is the client or some of my pcs faulty? The majority of my tests don't even seem to be getting transferred ... according to the stats I didn't submit a single test yesterday, even though I have 7 computers running 24/7 and the two fastest are capable of completing a test in about 17 hours each!
Please can someone tell me what is going on here, I really don't want to have to chane to another dc project!!

[jjjjL]

your stats are correct. the reason it appears incorrect is because the rate is averaged over the entire life of tests. that means that if a computer isn't running for any amount of time, the overall rate for that test may slide a bit.

another way to look at it is that since the average is done over the whole test, and your computer runs at a constant speed, the "rate" can only go down each time your computer doesn't run SB for a period of time.

since you have 7 tests, graphically, this corresponds to 7 rectangles on the graph. if they are all over a day old, the graph will just be 7 stacked rectangular areas so your graph will look flat with some fixed "rate". each time you report a new block, the progress is updated and that single rectangle is redrawn. the start time is the same as it was before so one edge of the rectangle is fixed. so lets say you play UT for a few hours and SB doesn't run. when the next block report comes in, it will update the end time and the total work done on that test, but it averages the work out evenly across the whole time you've been working on the entire test, so it will make it appear as though your rate is a little below your peak rate for the whole time.

back in the day, blocks were discrete so you could see each one sperately, so during that hour on UT, your production would go to zero but then it would shoot back up afterwards. now it doesn't track things the same way so it doesn't behave like that. your current rate will be getting averaged into your avg rate on all your tests until those tests finish.

so i'm saying, it's not surprising that your stats look the way they do and i'm sure they are correct.

-Louie

[eatmadustch]

I'm sure they're not correct! I have two computers on 24/7(PIV 2.53GHz) that produce 800 k cEM/s together, yet the stats show 150 k cEM/s for the entire day for all my computers! Also they show that I haven't completed a single test in the last 24h, but these two computers finish one test in about 18h each, so theoretically I have to complete at least 2 tests a day! I really get the feeling that some of my work is being rejected!

EDIT: I know how the stats work, and that they sometimes look wrong. But here they don't just look wrong, they are wrong too! Also I don't play UT enough for it to make such a drastic change!

[Troodon]

Louie, the problem is the last day's production isn't computed correctly. Please (re)read my previous posts.

[MathGuy]

Originally posted by eatmadustch
I'm sure they're not correct! I have two computers on 24/7(PIV 2.53GHz) that produce 800 k cEM/s together, yet the stats show 150 k cEM/s for the entire day for all my computers! Also they show that I haven't completed a single test in the last 24h, but these two computers finish one test in about 18h each, so theoretically I have to complete at least 2 tests a day! I really get the feeling that some of my work is being rejected!

Do you have logs that show completed tests in the last 24 hours? The logs should show if the server accepted or rejected them. If they were rejected, then we know that's what's going on...if they were accepted, maybe Louie could look those blocks up and see what user got credit for them...perhaps somehow some of your boxen have an incorrect (but valid) userid set and your work is being credited to someone else (just guessing...)

[eatmadustch]

Troodon, I did read your posts, I got 150 k cEM/s for the whole time my 2.53GHz (about 400 k cEM/s) computer was running -- non stop, without any UT, any other processor time consuming apps or any shutdowns.

I think it's more like something MathGuy suggests

[Troodon]

No, what suggests MathGuy isn't happening because:
- All my blocks were accepted
- The overall production is computed correctly.
For example, I've checked after sending one block and the total production increased 250 M cEM, while the day's productions increased only ~36 M cEM!

Anyway, for I me is just a "cosmetic" problem - I what I really want is a prime, not the last day's stats :D .