http://img707.imageshack.us/img707/669/ppdpmhz.png
Printable View
Looking to set up a thread here with a really easy way for a new or old cruncher to come in and figure out what they can expect from their machine. But I need some quick blurbs of data from all of you so I can make it more comprehensive!
My thoughts: with all the different CPUs and OSs and BOINC versions we run, it can be kind of confusing to predict what to expect from a machine. I want to simplify it....that way it'll be easier for people to spot if there's a problem, decide what components to buy for a crunching machine, predict the value of a machine to potentially add, etc. Plus, it's just fun to be able to know what your machine will do and see how close you can get to its potential...it's like an efficiency benchmark ;)
Here's how I want to do it: looking for a quick copy and paste from people's "Device Statistics" page on the WCG site. With it, please post your OS, your CPUs and MHz (and whether or not HT is on, if applicable), and BOINC version. Ideally, the data would be from dedicated machines so the numbers are pure....but if you say your system is also your daily-use system, I can take that into account :)
For example, this is my HTPC's data blurb from the Device Statistics page:
0:043:07:56:30 226,691
It's an i5 750 at 3500MHz running Win7 x64 and BOINC 6.10.43.
With this info, I can see I'm doing 20926 PPD at 3500MHz. That works out to 6PPD per MHz.
With that information, the Lynnfield (no HT) entry in the database would be:
"Lynnfield (no HT) | Win7 x64 | BOINC 6.10.43 : 6 PPD per MHz"
And with that (and hopefully with a list of other CPUs from all of you!), anybody can quickly assess what they should get from their CPU, could get from a future CPU, etc.
I will maintain the thread in this section and all I need are the blurbs from your crunchers....if you want to do the math yourself and save me a few seconds, go for it, just do it right! :lol: :cool:
Hopefully this catches on :)
EDIT: if you've been changing clockspeeds...the blurbs from the past few days should be enough (5 days is probably all that's needed for a decent average)
Prelim Database:
Lynnfield (no HT) | Win7 x64 | BOINC 6.10.43 : 6 PPD per MHz
Kentsfield | WinXP x86 | unknown : 4.5 PPD per MHz
0:114:06:20:22 434,042
daily use q6600 G0 @3.5 on xp32....These numbers generated from the last 30 days realtime and daily use for me is surfing for around 8 hours and a gtx 260 running grid
0:216:09:45:20 509,470
24/7 dual opty2356 @2.3 on xp32 again from 30 days realtime and a 9800gtx running grid
0:104:16:16:40 429,553
24/7 x3350 @3.84 on vista64 30days realtime and a 9800gx2 running grid/folding
0:94:3:39:15 490,273
24/7 qx9650 @4.0 on vista 64 30 days realtime and 2* 9800gx2's running grid/folding
as you can see running graphics on grid or folding takes away cpu cycles
EDIT this period includes a power cut day and all are on boinc 6.10.18
Awesome! Got 4.34 PPD per MHz...considering mild usage, we'll call it 4.5 PPD per MHz for now :)
Mini database started :)
I've got a couple others I can't access presently to check details, but here are 3.
E5420 - 4 cores/threads @ 2.27GHz | Win7 x64 | BOINC 6.6.20 | 3:256:10:09:39 4,536,224 7,373
E5520 - 4 cores/8 threads @ 2.27GHz | Server 2008 R2 | BOINC 6.6.20 | 1:308:01:56:56 1,751,060 2,851
2x E5110 - 4 cores/threads @ 1.60GHz | Server 2003 x86 | BOINC unknown | 5:333:05:01:31 3,926,854 7,208
Awesome! We have an increasingly useful database forming :cool:
Wolfdale... 4000MHz
0:016:17:52:53 76,272 145
That's 1.14 ppd per MHz per core... I think. Remember to divide run-time by the number of cores. For example, while my Wolfdale has crunched for about 16.75 days, it's only been up for 8.375 days, in which time it produced 76.3k WCG points. That brings it out to 9107.1 WCG PPD or 2.277 points per MHz... I think. :P
1:007:19:07:56 1,647,992 2,533
however its been the same machine but different chips... currently its an i7 920 @ 4000mhz HT ON
Last points generated was 28,215... so that makes it 7.05PPD/mhz
BOINC 6.10.18 x64 on Win7 64bit
core i5 750 4core @ 4100mhz/boinc 6.10.45/win 7 64bit/0:078:11:26:56/486 404/838
amd turion 64 x2 2 core @ 1900mhz/boinc 6.10.45/win vista 64 bit/0:035:14:01:19/63 136/122
intel core2duo e5200 2 core @2000mhz/boinc 6.10.45/win 7 64 bit/0:007:15:18:28/22 241/36
pentium d dual core 2 core @2800mhz/boinc 6.10.45/win 7 64bit/0:014:02:01:20/23 853/50
there you go vapor
Phenom II X3 720be@x4 2800MHz@3600Mhz
0:053:07:08:59 197,914
53.29 days / 4 cores = 13.32 days running time
197914 points / 13.32 days = 14858.4 WCG PPD
14858.4 PPD / 3600MHz = 4.13 PPD/MHz
Phenom II x4 @ 3600MHz = 4.13 PPD/MHz
BOINC Manager 6.10.43 x64
Win7x64
Catching up with all the submissions! Made it a chart....let me know what you think :up:
actually to get a true reflection on expected performance, you'll actually have to measure similar/same systems at different clocks to properly prepare an accurate formula for predicting point generation. Also the effects of operating system choice and dedication to work also have a significant impact on the points.
Dedicated Machine list:
All 920's have HT on...
1: i7 920 @ 3.8ghz Win XP x64 Boinc 6.10.43 64 bit 0:101:08:58:42 402,015
2: i7 920 @ 3.8ghz Win XP x64 Boinc 6.10.43 64 bit 0:097:12:31:54 389,044
3: i7 920 @ 3.8ghz Win XP x64 Boinc 6.10.43 64 bit 0:049:17:38:12 194,922
4: QX9650 @ 4.0ghz Win XP x64 Boinc 6.10.43 64 bit 0:071:05:43:07 366,752
5: Q9650 @ 4.05ghz Win XP x64 Boinc 6.10.43 64 bit 0:070:16:20:51 354,876
6: Q6600 @ 2.4ghz Win XP x64 Boinc 6.10.43 64 bit 0:053:09:58:13 161,400
Daily use machine:
I7 920 @ 3.8ghz Win XP x32 Boinc 6.10.43 32 bit 0:152:04:40:45 598,787
:)
SR2 rig..Tough to show a single day but at 4152MHz with X5680's it does average 105,000PPD
X5670's on the SMDA3 showing app 75,000PPD but thats climbing,just been on 3 days.
Looks like 25.5 PPD per MHz from a Dual Westmere....sheesh. If I put that on the graph, it messes everything up :lol:
Alright, added Dave's Dual Westmere systems and make the whole chart larger (and it still goes WAYYYY off the chart if I let it)....
Alright, might be able to take this out of 'beta' the next day or so...:cool:
Looking for more AMD data (Kuma/Rana, Deneb X3, Deneb X4, Thuban, etc), Gulftown, Clarkdale, more multi-CPU data, and Sossaman data. I think we've got enough here on Bloomfield/Lynnfield/Kentsfield/Yorkfield--I'll probably be trimming the current graph down to simplify things a bit too :)
E6600 @ 2.4Ghz |0:024:17:04:17|73,745| BOINC 6.10.43 |win7 x64 - it's been crunching 24/7 untouched except when I check on it
Okay, I'll try..
X5680 (12 Threads) @ 4,5Gghz; Win7 x64; warning, changes CPUs a lot - also used for gaming etc; 3:005:10:10:26 4,806,383 7,677
Quad Opteron 8347HE (16 Threads) @ 2Ghz, Win Server 2k8 x64; 9:123:06:01:54 6,444,458 9,779
i7 920 (8 Threads) @ 3,8Ghz; Win7 x64; 1:225:10:03:13 1,972,080 3,773
Dual Gainestown X5570 (16 Threads)@2,93Ghz; Win Server 2k8 x64; 10:081:05:39:22 11,260,471 17,392
Dual X5650 (24 Threads) @ 2,9Ghz; Win7 x64; 4:219:09:33:35 5,653,308 9,502
Dual X5450 (8 Threads) @ 3,6Ghz; WinXP x64; 3:273:00:01:42 6,422,064 9,326 <-- this machine has over 15 years of runtime on it, just had to reinstall because BOINC was corrupted somehow
Dual Sossaman (4 Threads) @ 2Ghz; some sort of Unix 32Bit (no idea what, loaned out to a friend as a 24/7 fileserver); 1:306:22:44:00 1,336,191 2,155
Is that what you need? :shrug:
Whoa! Lotsa juicy info there....great stuff jcool :up: I'll get to adding it soon!
(just updated the graph in post 2 with Conroe, thanks Red Maw!)
EDIT: for the X5680 system....do you think you could post the data from the past 5-7 days? if the overall results have been tainted by other CPUs, then just the recent results are probably more accurate :)
Guess I can't :p:Quote:
Message
The page you requested is currently unavailable because the statistics are being updated.
The statistics update will finish in about 62 minutes 9 seconds.
Checked a few days ago though, the machine is averaging around 60k/d.
:slapass:
Ah well, you can get it in an hour if you're still around...no hurry :)
Updated again....we now have three systems that have the bar going off the chart (I have it at 15 PPD/MHz). :D
Core Duo (no idea what is was coded) T7400 @ 2.16 (2 threads) | OS X 10.5.8 Boinc 6.2.18 | 0:013:18:51:21 39,802 83 Sit's running free 23 hours a day
Dual Woodcrest 5110 @ 3.0 (4 threads) | OS X 10.5.8 Boinc 6.2.18 | 0:079:05:06:52 310,069 575 Daily user but always running boinc
Great work Vapor, I'll get ya Gulftown single socket data and Lynnfield w/ HT by the end of the week
Updated again with mysticmerlin's numbers (good to see OS X still rocking out and doing so well, brings back some fond and not-so-fond memories of my Q6600 hackintosh)
Looking forward to the Gulftown and Lynnfield numbers (though Lynnfield should be within spitting distance of Bloomfield I'd think). I might be able to get some Thuban numbers by the end of the week, will be interesting to see if it can match Lynnfield without HT :lol: And for that matter, I think Clarkdale might be able to keep up with Deneb X4, clock-for-clock :cool:
definately interested in someone sharing for thisQuote:
Originally Posted by Vapor
will take a while to get meanginfull data for Thuban.. mines only been crunching full time for just under 3 days, and the first days result was only 4000pts, so it's going to through everything for a bit.
days 2 and 3 have been 24-28kppd though (3.9Ghz), Will see how it averages out and contrubute to thread..
Has yours been at 3900 the whole time? Is it a dedicated cruncher or a daily use rig?
Whats your total CPU time and points so far? 3 Days is what I'd consider the lower limit of 'useful' for finding PPD/MHz, but it'll give a good hint :)
Can you take a screen shot of the WCG page where it shows the output by days with the hours,etc for just that machine..like this:Quote:
Originally Posted by mAJORD
sorry guys, i'm still a bit nob at this, just realised the 1st days low pts are acounted for the in the total time stats, so maybe it's almost enough..
tell me if I'm doing this calc wrong.
Phenom II X6 3920Mhz win7x64 BOINC v6.10.18 (latest recomended)
Device stats:
5/4/10 0:012:23:55:02 56,882 98
12.99 days /6 core
=2.166days
56882/2.166 = 26261PPD
26261/3920mhz
=6.70PPD/Mhz
Device stats breakdown:
5/4/10 0:006:10:40:55 28,036 40
5/3/10 0:005:10:41:21 24,368 45
5/2/10 0:001:02:32:46 4,477 13
im at work so can't really do screenshots atm
0:012:23:55:02 is within minutes of 13 days, not 12.38....otherwise you got it right :D
Works out to 6.7 PPD per MHz on the first two days....we'll see if that number holds up (hopefully it does, then we can get PPD/MHz numbers faster :lol: )
EDIT: and based off the numbers from Dave's screenshot....his SR-2 is doing 24.9 PPD / MHz (using 4175 as his MHz) :slobber:
hey vapor, ive got a few quad socket 8356 optis for ya @2.3ghz
they are all running win server03 r2 64bit
tyan-2, and tyan-3 are running 8gb of ram
tyan-1 is running 16 gb
Device Statistics
Statistics Last Updated: 5/5/10 00:06:01 (UTC) [4 hour(s) ago]
Agent Type Device Name Last Results Returned
(UTC) Total Run Time
(y:d:h:m:s) Points
Generated Results Returned
BOINC skycrane-tyan3 5/5/10 1:364:15:20:54 1,585,813 2,742
BOINC skycrane-tyan1 5/5/10 0:112:15:13:50 255,205 793
BOINC skycrane-tyan2 5/5/10 0:180:08:09:06 401,626 1,095
Between the three you're averaging 15.3 PPD per MHz :up:
Between your three and jcool's Quad Barcelona, it looks like that number will settle at 15.2 PPD per MHz....with over 11 years of Quad Barcelona runtime, I think we're now set for data on this CPU :D
i would like to see a screen shot as well, and would like to keep an eye on this. if this info is correct; these could produce in the same ballpark as a 920.Quote:
Originally Posted by mAJORD
my 920's along with gpugrid vary on a daily basis. its always up and down, but i think on avg they are around 30kppd
Vapor,Quote:
Originally Posted by Vapor
Here are some clarksdale numbers... There are 2 users on this machine so It's not dedicated (as you can see)
PHEONIX 5/5/10 0:173:07:15:08 725,788 1,078
5/4/10 0:003:19:58:49 14,841 21
5/3/10 0:006:18:36:40 26,327 35
5/2/10 0:004:16:07:39 16,445 28
5/1/10 0:002:22:11:11 9,713 17
4/30/10 0:003:11:24:29 12,244 19
4/29/10 0:001:14:39:48 5,518 10
4/28/10 0:001:20:23:00 7,416 10
4/27/10 0:004:06:29:31 18,207 26
4/26/10 0:004:03:11:14 16,453 28
4/25/10 0:004:01:41:40 17,183 28
4/24/10 0:003:20:40:17 15,898 29
4/23/10 0:003:19:18:50 17,489 26
4/22/10 0:003:19:34:03 16,229 26
4/21/10 0:003:21:45:59 16,896 31
Currently at 4.37gHz (lowered for crunching), win7 x64
If you need me to pull something else off let me know...
EDIT: it's an i3-530 ($99 from MC), I call her Pheonix.
EDIT: peak average has been right under 2200ppd Boinc (~2190)
Here you go:)
Note: Italicized ones are all laptops.
Athlon II 240 Regor 2 Cores @ 3.6GHz
0:110:00:03:49 428,455
Windows 7 Ultimate 64bit
6.10.18 64bit
Xeon X3210 Kentsfield 4 Cores @ 2.13GHz
0:235:12:34:58 560,881
Windows XP Professional 32bit
6.10.18 32bit
C2D E6750 Conroe 2 Cores @ 2.66GHz
0:103:06:13:18 305,882
Windows XP Professional 32bit
6.10.18 32bit
Pentium T3200 Merom 2 Cores @ 2GHz
0:013:16:39:37 25,244
Vista Home Premium 32bit
6.10.18 32bit
P4 Prescott HT @ 3.0GHz
0:137:13:00:33 151,427
XP Home 32bit
6.10.18 32bit
Athlon X2 TK-53 Tyler 2 Cores @ 1.7GHz
0:275:10:07:57 435,970
Vista Basic 32bit
6.10.18 32bit
Athlon X2 QL-65 Lion 2 Cores @ 2.1GHz
0:044:15:52:07 38,287
Vista Home Premium 32bit
6.10.18 32bit
Celeron M 370 Dothan 1 Core @ 1.5GHz
0:026:17:50:30 36,924
XP Home 32bit
6.10.18 32bit
I'll ss tonight, or maybe after the 12hr update
BTW.. I'm crunching all prjects except HP2.. should we include what projects ppl have selected in the results? Does it make a big difference?
EDIT: oops, thanks for th corection VAPOR.. heh
Here are my two rigs for last few days:
i7 920 8 cores @ 3.6 Win 7 x64 BOINC 6.10.43 (x64)
04/05/10 0:007:17:01:01 30,467 50
03/05/10 0:006:09:05:20 25,607 51
02/05/10 0:009:05:21:38 36,693 69
01/05/10 0:007:00:47:20 27,537 45
30/04/10 0:006:16:08:30 26,662 43
7.7/8=0.9625
30,467/0.9625=31,654 PPD/core
31,654/3600=8.79 PPD/Mhz
Xeon E5520 (x2) 16 cores @ 2.267 Server 08R2 BOINC 6.10.43 (x64)
04/05/10 0:013:14:33:34 33,323 65
03/05/10 0:014:06:36:09 33,952 63
13.58/16=0.84875
33,323/0.84875=39,261 PPD/core
39,261/2267=17.32 PPD/Mhz
C'mon people, do the math yourself ;)
E6750@2.66GHz@WinXP 32bit
0:040:03:53:03 120,249 -> 2.25PPD/MHz
E7500@2.93GHz@WinXP 32bit
0:037:00:44:14 132,613 -> 2.44PPD/MHz
W3503@2.4GHz@WinXP 32bit
0:037:21:38:07 138,791 -> 3.05PPD/MHz
All of these running BOINC Manager 6.10.43
Vapor,Quote:
Originally Posted by Vapor
my single X5680 may be a Westmere-EP, but it's performance is identical to that of a Gulftown, so you can use that for single CPU for now ;)
Last 12 days on it, backed down to 4,4Ghz though. This is my daily driver used for gaming and video rendering, so roughly 80% of time BOINC is running (pretty accurate estimation, so you can factor that in).
04.05.10 0:011:20:56:59 54,254 136
03.05.10 0:012:15:22:45 58,354 122
02.05.10 0:010:23:21:42 51,097 102
01.05.10 0:009:11:17:31 43,000 79
30.04.10 0:009:19:24:49 44,870 84
29.04.10 0:006:09:19:30 29,702 65
28.04.10 0:013:05:22:59 61,640 113
27.04.10 0:010:05:05:46 46,969 80
26.04.10 0:012:05:29:20 56,984 110
25.04.10 0:007:14:13:53 35,610 74
24.04.10 0:007:07:06:16 34,770 74
23.04.10 0:013:19:28:40 64,716 137
Averages to around 48700.. / 0.8 gives the 60k :)
On the chart can you put the corresponding platform in the name, i.e. I7 x58, I7 P55, I7 x58 Xeon, etc.? It would be easier to figure out the platform for "Average Joe". Also, I am assuming these are based on 100% usage.
AMD 1055T 6 core CPU @ 3.2Ghz running 30% to 80% utilization on WHS x32 bit and WCG 6.2.28, 4Gb DDR2 ram @ 920Mhz
5/5/10 (0:002:05:40:59) (7,480) (13)
5/4/10 (0:002:12:13:20) (8,230) (15)
5/3/10 (0:003:10:56:04) (11,806) (27)
5/2/10 (0:000:08:11:09) (1,086) (5)
I7 920 @ 4.2Ghz HT on running 50% to 100% utilization on W7 Ultimate x64 and WCG 6.2.28, 6Gb ram @ 1600Mhz
5/5/10 (0:004:17:20:24) (20,645) (35)
5/4/10 (0:004:08:48:49) (19,203) (32)
5/3/10 (0:001:18:42:52) (8,457) (14)
5/2/10 (0:000:20:24:35) (3,986) (5)
4/23/10 (0:000:05:32:03) (873) (3)
Not sure how good this is, but it appears to be about 4000 per day on the AMD 6 core and 5000 per day on the Intel I7. Based on the other numbers posted here it seems pretty low, but these are multi-functioning machines, so maybe it's ok.
Here is what I have for my dual xeon e5410 rig (2.33 MHz), dedicated cruncher, average of the last three full days (28,000 ppd ave):
Harpertown x2 (no HT) | WinServerStandard03 x32 | BOINC 6.2.28 : 12 PPD per MHz
or, for one of the cpus:
Harpertown (no HT) | WinServerStandard03 x32 | BOINC 6.2.28 : 6 PPD per MHz
Catching up with adding the new data :up: No need to do the math yourself...once I saw the explosion of results, I set up an Excel that I could just copy and paste the data blurbs into (and tell it MHz and threads) and it tells me PPD, PPD/MHz, and days of runtime :)
Looks like jcool was the first to the punch with Gulftown numbers.
I like your idea utnorris....I'll see if and how I can do that. I agree code names aren't ideal, but it was the shortest nomenclature for groupings that I could think of....brand series name might just work though :up:
those SS i promised. nother days results, no change in avg so far.
http://img571.imageshack.us/img571/5949/ph2x6.png
hmm,interesting..Quote:
Originally Posted by mAJORD
So if I could get a dual socket G34 board with 2-12 core MC chips in it to 3920 I could pop 100,000 PPD with it..:D
Time to make a call to shamino ??????Quote:
Originally Posted by Movieman
Quote:
Originally Posted by Vapor
Some AMD info, all running Ubuntu 9.10 x64 using client from "apt-get install boinc" (6.4.5) . Both X2-550 are unlocked.
CPU cores Mhz Days Points Results
X3-720BE 3 2800Mhz 0:013:02:06:12 42,611 117
X2-550BE 4 3400Mhz 0:018:23:01:42 67,351 165
X2-550BE 4 3400Mhz 0:018:23:42:22 68,045 184
which should make the ppd per mhz around :
X3-720 ~2.55
X2-550@4x3.4g ~4.18
X2-550@4x3.4g ~4.21
D
I'll play, too :)
Dual Xeon SL8WT (Sossaman) @ 2GHz | Win. Vista 32-bit SP2 | BOINC 6.10.18
5/4/10 | 0:004:02:55:28 | 8,996 | 15
5/3/10 | 0:004:10:39:26 | 9,044 | 16
5/2/10 | 0:004:06:35:47 | 8,711 | 15
5/1/10 | 0:004:03:12:55 | 8,280 | 14
E8400 (Wolfdale) @ 3.6GHz | Win. Vista 64-bit SP2 | BOINC 6.10.43 (computer sees occasional light use)
5/5/10 | 0:003:01:38:02 | 14,691 | 23
5/4/10 | 0:001:15:06:29 | 7,797 | 14
5/3/10 | 0:002:02:35:37 | 10,415 | 18
5/2/10 | 0:002:00:23:38 | 9,348 | 18
5/1/10 | 0:001:20:26:31 | 8,815 | 14
Dual-Core Pentium D 820 (Smithfield) @ 2.8GHz | Win. XP Pro 32-bit SP2 | BOINC 6.10.43
5/5/10 | 0:002:14:06:09 | 4,122 | 8
5/4/10 | 0:000:23:27:24 | 1,608 | 3
5/3/10 | 0:002:18:10:43 | 4,214 | 8
5/2/10 | 0:001:19:33:46 | 2,826 | 5
5/1/10 | 0:001:04:56:57 | 1,725 | 4
4/30/10 | 0:001:14:32:04 | 2,492 | 4
Triple-Core Phenom 8450 (Toliman) @ 2.1GHz | Win. Vista 64-bit SP2 | BOINC 6.10.43
5/5/10 | 0:002:20:43:04 | 6,089 | 11
5/4/10 | 0:003:06:11:02 | 6,942 | 13
5/3/10 | 0:001:22:03:19 | 4,086 | 8
5/2/10 | 0:002:10:08:08 | 5,055 | 9
5/1/10 | 0:002:20:58:26 | 6,129 | 13
Athlon64 X2 3800+ (Manchester) @ 2.0GHz | Win. XP Pro 32-bit SP3 | BOINC 6.10.43
5/5/10 | 0:001:21:06:27 | 3,741 | 6
5/4/10 | 0:002:08:10:59 | 4,653 | 8
5/3/10 | 0:001:10:48:43 | 2,874 | 7
5/2/10 | 0:002:10:02:59 | 4,607 | 8
5/1/10 | 0:001:02:02:08 | 2,078 | 4
4/30/10 | 0:001:13:59:18 | 3,114 | 9
E6600 (Conroe) @ 2.4GHz | Win. Vista 64-bit SP2 | BOINC 6.10.43
5/5/10 | 0:002:22:50:39 | 7,851 | 18
5/4/10 | 0:001:05:05:34 | 3,395 | 8
5/3/10 | 0:001:11:16:35 | 4,117 | 8
5/2/10 | 0:000:13:34:13 | 1,489 | 3 <-- Dip due to being pulled offline for a rebuild
5/1/10 | 0:001:16:42:47 | 4,685 | 8
4/30/10 | 0:001:19:41:46 | 4,745 | 13
4/29/10 | 0:001:12:58:01 | 4,037 | 9
I also have some Apple laptops crunching, but they're all over the place due to BIONC halting (when on battery), quite a lot of use, and throttling down the BOINC usage due to noise & heat concerns. I'm thinking of running my second Sossaman system on a Linux distro... not sure yet (could always use something like Rocks and cluster the two Sammys, too).
*Edit* ...and before I get grief for only the E8400 being overclocked (and rather mildly, at that): Can't overclock the Sammy (Intel server mobo), the E8400's water cooling system can't take much more heat (all internal, P180 case, built for silence, also has two OC'd HD4850's in the loop and a P45 NB, CPU currently loading at ~63C with the GPUs at idle), the Pentium D's mobo doesn't clock well and the memory is junk (blame my wife, she was on a budget when she built it years ago... also, heat is an issue since it's geared toward silence), the Phenom is our "media PC" that is used for little more than occasionally watching stuff on Hulu (also geared toward silence... but the mobo and memory sure could handle it), the only spare heatsink I had for the A64 was a stocker ('nuff said... the DFI LP NF4 Ultra-D and Geil One memory (rated for DDR400 @ 1.5-2-2-5 & DDR600 @ 2.5-4-4-7) probably want to strangle me), and I have no excuse for the E6600 other than the undervolted (5V) NMB fans it's running.
Nice numbers, is it possible to get it to 4.2?Quote:
Originally Posted by mAJORD
not a lot of info but heres mine.100% used for WCG:D
i7-920@4200 w/HT on DDR3-1600 6-9-6-21 1t
http://i39.tinypic.com/rk7tyd.jpg
EDIT:forgot to add,im running vista x64 and HT on
* * * * * *
I know I've fallen behind on updating....promise I'll get it to done in the next few hours :)
Looks like the database is getting filled out well :D
Radaja, that is VERY high granted credit for a 920 at that speed. Which projects do you run?
the first few days i ran all of them and two days ago i changed it to HCC,HFCC and HCMD.
but today it finally started doing those three.i had set my thing to cache three days of work.
9 PPD/MHz is pretty darn good for a Bloomfield at that speed (PPD/MHz does decline slowly at higher MHz). skinnee has a stock E5520 at 9.2 PPD/MHz whereas jcool's 920's (at 3800) are doing 8.3-8.4 PPD/MHz. Probably boils down to projects run and secondary settings (uncore and RAM namely).
A further investigation into the two settings you guys run might be pretty fruitful :up:
i manually update throughout the day also.so maybe this has something to do with it?
after i have about 6-10 WU 's finished i send them away?
I don't think that would have anything to do with it.
I think it's projects being run....I have my 930 at 3850MHz with a scant 2800MHz uncore and RAM at 1400 9-9-9-24 and I'm doing 8.8-8.9 PPD/MHz. I run only HCC.
EDIT: I'll switch my 750 to a new account and enable it for all non-HPF2 projects and run the PPD/MHz equation on the various projects after a few weeks, hopefully that will shed some light on which are the most gainful for PPD :)
yea it must be the projects because nothing special here with the overclock.
21x200@4200 HT on uncore at 3200 1600 6-9-6-21
This is an awesome chart. What would be cool to add is watts used vs. PPD vs. MHZ or something like that.
Not possible....there's a non-linear response between watts and MHz (considering a voltage increase accompanies most clock changes on this forum...assuming max MHz scales linearly with voltage, you actually have watts scaling with the cube of MHz when overclocking past stock voltage! [not a hard-and-steady rule, just a quick rule of thumb]) and there's also secondary components, individual differences between CPUs, and so so much else that just muddies it to impossibility.Quote:
Originally Posted by Angmaar
If you want the most efficient cruncher possible...undervolt, overclock, run minimal secondary components, etc. When it's all said and done, it's very likely the most efficient is either a P55 i7 or a Gulftown system, but that's just my educated guess :shrug:
Here are a few more for you:
980x - 4.0GHz - Win 7 - x64 - daily usage
5/5/10 0:012:04:34:59 49,198 58
5/4/10 0:012:03:48:28 48,633 58
5/3/10 0:010:01:30:58 39,664 52
5/2/10 0:014:17:42:24 58,747 77
5/1/10 0:009:19:42:46 39,395 50
980X - 3.33GHz - Win 7 x64 - dedicated (haven't had time to tweak yet)
5/5/10 0:011:22:29:13 40,160 51
5/4/10 0:013:08:18:05 44,754 57
5/3/10 0:011:06:07:15 37,915 53
5/2/10 0:018:14:13:58 62,403 71
5/1/10 0:017:22:35:07 57,276 70
Phenom II - BE550 - unlocked 4 cores - 3.2GHz - media center light usage
5/5/10 0:003:14:42:18 11,475 16
5/4/10 0:004:07:04:43 13,756 19
5/3/10 0:003:18:49:01 12,143 14
5/2/10 0:004:22:41:33 15,426 19
5/1/10 0:004:16:43:36 14,963 18
http://img707.imageshack.us/img707/669/ppdpmhz.png
Massive chart, eh?
The 920 was running 200*19, 3200 Uncore, 1600 9-9-9 Triple Channel DDR3. :shrug:Quote:
Originally Posted by Vapor
Running HCC and HFCC, like all my machines do.
Average scores over the last 18days:
Opteron 185
2.6GHz
5103 average PPD
Linux 64bit (daily drive)
Opteron 170
2GHz
3763.9 average PPD
Linux 64bit (boy's daily drive)
Dual Sossaman
2GHz
6301 average PPD
Linux 32bit (also running caching web proxy, content filter and DNS forwarder for whole network)
Vapour: ... how did you work out the 4.0 for jcool's Sossa? The numbers in his post give me 0.99PPD/MHz From your first example you appear to be multiplying by the number of threads. If I do that, my Sossaman scores a 12.6 on your chart.
4.0 x 2000MHz = 8000ppd ...and that's about what mine has been putting out per calendar day for the past 4 or 5 days (it's new, numbers are in post #48). So, 4ppd/MHz doesn't seam unreasonable to me.Quote:
Originally Posted by D_A
If 8000 points were produced in one calendar day, it would show up as ~4 CPU days on a sammy... so you'd be looking at 1.0 points-per-CPU-day/MHz, or 4.0 points-per-calendar-day/MHz, no? I'm way too tired to be doing this.... :zombie:
Great chart... I'm a little upset with the Thuban numbers though. I expected higher.
From this chart, I'm going to try and figure up 'best bang for buck' for my next machine. (Sossaman looks difficult to beat)
I don't know how many people have a killawatt meter, but it would be interesting to get power consumption for these as well, like Angmaar said. I guess the problem would be that it's not linear w.r.t vcore and vcore is not linear w.r.t MHz.
jcool's averages ~8000 points per 96hr of runtime....which comes to 4.0 PPD/MHz. Yours comes in at 3.15 PPD/MHz, but you're running a bunch of other stuff on it, so its numbers will be diluted.Quote:
Originally Posted by D_A
I'm not surprised by Thuban...it's actually doing better than 1.5x a Deneb X4, but that could just be because of the small sample sizes and the taint of non-dedication.
Of all the numbers I think are undershooting real performance, I think it's Clarkdale. I'd expect Clarkdale to pull 4.1-4.5 PPD/MHz on a dedicated system (based on what other Nehalem derivatives do for PPD/MHz/core). The 3.8 PPD/MHz on the chart is from a single system used by multiple users....I hope I can get more data on Clarkdale soon :)
In terms of cost effectiveness, a Sossaman is going to be hard to beat ($120 for CPUs+board+ram+HSFs?), but that's largely because it uses outdated parts. Microcenter's occasional deal of $99 for board + Athlon X4 would vie for that title as well if you can find cheap RAM. As for power....it's something that's way too difficult to chart for all of this, nevermind obtaining comparable data. It's a project I'd be interested in seeing the results of, but it would have to be done by a single individual that knows scientific method (and overclocking and computer and electrical engineering fundamentals) that has a lot of time and doesn't mind sacrificing production to accurately chart low-clock, high-efficiency output.
I'm not volunteering for this... :para:Quote:
Originally Posted by Vapor
I have Lynnfield, Bloomfield, and Thuban here....but I'm not volunteering either. Too much of my watercooling/TIM testing to do using Bloomfield/Thuban....and my Lynnfield is already running a WCG test session to see which projects are the most gainful in terms of PPD/MHz.
I'll see if I can replace the Clarksdale and set it aside for a few days... This is not going to be an easy sale ;) .
Whats the minimum sample period, in days, that you would need?
3.5 is the lowside of what I'm comfortable with....5 is satisfactory. I don't think you should set it aside though....with the quorum points system, it'll be really hard to tell when 'dedicated' kicks in (unless you run it for like 10 days dedicated, which is too much to take away from the people that actually use it :p: ).
I'm sure there's another Clarkdale around here somewhere :)
8000 points / 96hrs(4days)/2000MHz = 1 (based on calendar days, which my averages are)Quote:
Originally Posted by Vapor
I may have misunderstood how you wanted the original figures though. My numbers based on hours runtime instead of calender days are:
Attachment 104018
"new-bruce" is the Opty 185 @ 2600 linux 64bit
"boyz-desktop" is the Opty 170 @ 2160 linux 64bit
"sausageman" is the Sossa @ 2000 linux 32bit
"girlz-desktop" is a XP2600+ @ 2133 linux 32bit
"node1" is a XP 3000+ @ 2166 winXP home 32bit
"iCom" is a XP 2400 @ 2000 linux 32bit
i3-540 @ 4.0GHz (HT on), Windows 7 x64, Boinc Manager 6.10.43 X64 (daily usage)Quote:
Originally Posted by Vapor
0:075:19:19:56 - 306,357
i7-860 @ 3.6GHz (HT on), Windows 7 x64, Boinc Manager 6.10.43 X64 (daily usage)
07-05-2010 0:006:06:25:02 25,449 51
06-05-2010 0:006:15:14:17 26,615 47
05-05-2010 0:007:21:52:04 30,787 53
04-05-2010 0:008:03:17:43 30,082 40
03-05-2010 0:010:18:00:23 40,258 47
02-05-2010 0:008:12:53:07 33,624 48
01-05-2010 0:005:22:01:05 24,238 40
30-04-2010 0:007:20:56:37 31,759 46
29-04-2010 0:010:02:27:39 38,003 71
28-04-2010 0:007:20:47:17 29,004 53
27-04-2010 0:006:17:47:00 24,799 42
26-04-2010 0:005:19:59:28 22,038 40
25-04-2010 0:008:19:11:09 30,457 48
Q6600 @ 3.2GHz, Windows 7 x64, Boinc Manager 6.10.43 X64 (daily usage)
0:072:20:08:47 - 273,957
Here are my main 3 crunchers, they are all running Win7 64 and boinc 64:
i7 920 @ 4.5 HT on
http://www.straferight.com/photopost...1/920_4515.JPG
i7 920 @ 4.2 HT on
http://www.straferight.com/photopost...1/920_4200.JPG
Q9550 @ 4.0
http://www.straferight.com/photopost...q9550_4000.JPG
Are those dedicated or do you use them for other stuff, too?
what projects are these running? I never get consistent results like that. I run HFCC & HCC. each project on a different machine. I get anywhere from the mid 20's to in the 40's:shrug: I do not manually update though, just let the system run. these 2 machines are not used very much either; just some occasional surfing, but I do run gpugridQuote:
Originally Posted by sRHunt3r
edit: I have 1 920 and 1 w3520 ea 2 4ghz.
Nah I use all of them intermittently for gaming/ media server/ encoding.
I run HCC, HFCC and HCMD.
could someone please tell me how to change the name of the device
or give it a custom name?like what i circled in red in the pic below.
http://i39.tinypic.com/x4ekxd.jpg
The names listed are the computer names as you have assigned them via the os, I am unsure if there is a way to change the way WCG displays them but you could just rename the pc:shrug:
ok thats what i thought but,was just curious when i saw these cool names.
i guess i have to be more creative when i do my installs.but also mine always
has the name of my computer and the PC,like this bigtop-PC.and thats what
im trying to get rid of,the PC part
If you are using Windows 7Quote:
Originally Posted by radaja
*go to control panel system
*on the right click on change settings
*once that opens click on change
*then type whatever you want the computer name to be
*click ok and restart you PC
ACEkombatkiwi,thanks now i can change the i7-920 rig.
mind the question, though, why not simply taking few days in a row from the points statistics, to see how the PPD goes,
the PPD per MHz is a nice gimmick, though it would be interesting to see how many point each system does usually..
it seems some were posting BOINC points and some were posting WCG points,
it is quite misleading.
What's so gimmicky about PPD/MHz? :confused:
Take any of the numbers in my chart, multiply by system MHz, and it's a good estimate of what to expect from that machine for PPD.
Tracking PPD of systems and having a database of it is something we (almost) all want to do, but considering all the different settings we all run (hardware level and software level), the easiest way to distill the PPD numbers into usable information is to take systems with similar software settings (i.e., dedicated and OS/BOINC intalls) and divide by MHz so anyone can 'reconstitute' it into usable numbers for their own system :rolleyes:
And it looks to me like everyone in this thread has been posting WCG points...and I know for certain that no BOINC points were used in my calculation for my chart. :stick:
Quote:
Originally Posted by radaja
Yes, they're just the names I gave each of the PCs. It's nothing to do with WCG, that's what they're called in the operating system.Quote:
Originally Posted by sRHunt3r
as an idea, a nice idea, (maybe better said?).Quote:
What's so gimmicky about PPD/MHz?
hrm,Quote:
Dual X5450 (8 Threads) @ 3,6Ghz; WinXP x64; 3:273:00:01:42 6,422,064 9,326
so what is that?
even at BOINC points it's unreasonable that these processors do so much,
at WCG points, it is extremely low...
**o.k,
so this is the results returned**,
newcomers would have a great deal of confusion, trying to understand it :shrug:..
and:maybe i'm confusing something here, yet it seems quite hard to understand :shrug:...Quote:
For example, this is my HTPC's data blurb from the Device Statistics page:
0:043:07:56:30 226,691
It's an i5 750 at 3500MHz running Win7 x64 and BOINC 6.10.43.
With this info, I can see I'm doing 20926 PPD at 3500MHz. That works out to 6PPD per MHz.
how did you get 20926 a day, from this number?
0:043:07:56:30 226,691?
yeah,Quote:
Take any of the numbers in my chart, multiply by system MHz, and it's a good estimate of what to expect from that machine for PPD.
the issue, is Westmere, can be a lot of processors.., different cache, and gainstown too, and any other family..,
you see..?
Ok, that number translates to:Quote:
Originally Posted by onex
0 years, 43 days, 7 hours, 56 minutes and 30 seconds
and
226,691 points
yeah, of course, this is understandable,Quote:
Ok, that number translates to:
0 years, 43 days, 7 hours, 56 minutes and 30 seconds
and
226,691 points
yet, where does the 20926 number (PPD) is coming from?
if you divide 226,691 by (even) 11 days, it still comes out 20608...
E:
huh :),
got a new title :)
226691 (points) / 43.33 (wcg days (don't forget to convert to decimal)) * 4 (cores or threads) = 20926 PPD (average points per real day)Quote:
Originally Posted by onex
Vapor's i5-750 is at 3500MHz, so 20926 PPD / 3500MHz = 5,98 PPM (average points per megahertz)
From here, if you have an i5-750 you can make an educated guess at what will be your estimated average PPD by multiplying the 6 PPM by the speed of the processor.
pff,Quote:
226691 (points) / 43.33 (wcg days (don't forget to convert to decimal) * 4 (cores or threads) = 20926 PPD
fine, that makes it simple.
it isn't that obvious from the posts, Vapor..,
newbies would have a very hard time figuring that out :yepp:.
after that,
the graph makes much more sense now, (though it would been nice to get a clue about the differences between westmere different L2 cache sizes).
dual Harpertown seems to give only ~27736 for dual E5430, which is only about 18% more then a Bloomfield 920 at ~23408,
2*6MB L2 cache vs 4*256KB L2 and 8MB L3.
so maybe WCG is more L3 cache dependent,
yet why would it be :confused:..,
HT, should not give any CPU such a close productivity to native cores, these are quite odd results :head scratching:.
Why shouldn't it? (I just don't know one way or another, that's why I'm asking.)Quote:
Originally Posted by onex
How much boost you get from HTT varies enormously, I'd guess depending on how much cache each thread is using and how often each thread gets blocked waiting for IO or something. A whole bunch of IO bound threads, with low cache usage, would benefit enormously while CPU bound threads would get less benefit.
In this case, what I mean by "low cache usage" is that the cache doesn't need to be refreshed massively every time the CPU changes the thread it's working on. You'd probably find there's a benefit to running a HTT enabled system on a single project (like HCC or FAAH) rather than several different projects.
it should be project related,
maybe 25%-60?
rough estimate.
All newbies are recommended to read the whole thread :up:Quote:
Originally Posted by onex
Quote:
0:053:07:08:59 197,914
53.29 days / 4 cores = 13.32 days running time
197914 points / 13.32 days = 14858.4 WCG PPD (should be POINTS)
14858.4 POINTS / 3600MHz = 4.13 PPD/MHz
Phenom II x4 @ 3600MHz = 4.13 PPD/MHz
no, it's his fault,Quote:
All newbies are recommended to read the whole thread
i insist :P.
j/k,
totally missed it :).
though it is worth mentioning through the OP as well :yepp:.
I have a Clarkdale I3-540 now and it is doing better than the 3.8 PPD/MHz that is on the chart now.Quote:
Originally Posted by Vapor
Not sure what it is doing exactly, because it is only running for 3 days now. And I'm still playing with the OC.
But the 2 full days that it has been running almost full time it did 4.4 PPD/MHz.
I will leave it like it is now for a few days and post the results here.
4.4 is more inline with what I was expecting from Clarkdale (it's essentially half a Bloomfield). I'll update the chart when you get some more solid numbers :)Quote:
Originally Posted by Rob_B
Changed my Q6600 @ 3200MHz from almost dedicated in Windows 7 x64 to fully dedicated in Ubuntu 10.04 X64.
http://img517.imageshack.us/img517/988/wcgc.jpg
It went from about 4.8 PPD/MHz to 5.7 PPD/MHz.
Looks like that 4.4 was a little high. The first few days when I had this PC running I had been playing a little with OCing and maybe that messed up the numbers a bit.Quote:
Originally Posted by Vapor
So after 2 days running at the same speed I didn't touch the PC and let it do it's job.
This is what I got over 6 full days running all projects.
Clarkdale i3-540 @ 3533 MHz. HT on. Windows 7 x64. 24/7 cruncher.
05/06/10 0:003:12:28:55 12,855
06/06/10 0:004:09:22:58 15,972
07/06/10 0:004:17:53:36 16,259
08/06/10 0:003:06:48:50 11,679
09/06/10 0:004:19:40:40 18,183
10/06/10 0:003:08:00:24 12,418
That is 4.1 PPD/MHz over 6 days crunching. Not to bad for a PC that only uses 85 watt. :)
Now I let it finish all the WU that are downloaded and try it again with Ubuntu.
My average for my i3-540 is about 4.125 PPD/Mhz, and i also do some gaming/browsing/whatch movies.
Rob_B you might overclock it easily to 4000 Mhz, just stay below 1.3V and it will be energy efficient.
I choose that speed after reading this article on cpu overcloking vs Power consumption.
I'm going to speed it up after I get it watercooled. I want it to run cooler and quieter.Quote:
Originally Posted by pfm3136
But it can do 3694 MHz @ 1.15V and run WCG without problems.
Thanks for the link.