stephenbrooks.orgForumMuon1GeneralBenchmarking with Mpts thread
Username: Password:
Page 1 2 3 4 5 6 7 8 9 10 11 12 13
Search site:
Subscribe to thread via RSS
[TA]z
2006-03-18 16:02:25
2x Dual Core Xeon (Paxville) 2.80GHz (8 threads), 200FSB, 685.16 Kpts/sec

AMD builds smoke this thing on a per core basis O.o
[TA]Assimilator1
2006-03-24 03:58:51
z
Congrats ,your the 1st one to go off the scale!  Big Grin ,Stephen will have to redo the graph Wink
Btw ,mental output from that rig!:Q

Stephen
Just what I was thinking ,& if you look at my XPM (512k L2) vs HaloJones XP1700 it's slightly faster despite having 1/2 the cache & slightly lower clock speed ,seems the FSB or RAM speed counts more their.

Now if only I had RAM that could do 278MHz I could do some more comparisions with my SempronWink

Mezocop
Is you RAM at the same speed as your FSB?
Mezocop
2006-03-24 04:47:21
Assimilator1
Yes.  My Venice doesn`t wish to work stably on freqs higher 2,4 GHz.
Stephen Brooks
2006-03-24 07:44:24
It now goes to 700. Of course this is nothing compared to what will happen if someone runs 4 AMD cores on this.
AttiX
2006-04-24 02:14:41
Intel Core Duo 2400 (2x1.83Ghz).
2x512MB PC3200 DDR2 (3-3-3-8)

AVG: 325 kpts/sec
[OCAU] badger
2006-05-31 04:23:48
now that this thread has reappeared, this P3 E coppermine-T at 995.7 MHz does just over 57 kpts/s. I'll also bench my new pc which it arrives... V4.43a.
[TA]Assimilator1
2006-06-04 22:40:34
lol ,I've just realised I've got the 2nd fastest single core cpu rig in the graph? 
Is my lowly Sempron64 scaring off the high end P4s ,Ath64s & (most)Opterons? 

Mezo
Have you tried with your RAM on a lower speed setting (higher divider)?
[TA]Assimilator1
2006-06-04 22:42:11
(wot no edit function??)

Mezo
It's very likely it's your RAM holding you back not your CPU
Mezocop
2006-06-05 21:58:35
2 Stephen
So strange forum ... No simple functions ...

2 Assimilator1
I`m change my Venice for another and it work at 2700 MHz now.  No influence from RAM. 
danzigrules
2006-06-12 22:44:03
74015,39654.1,237.57
74316,39737.0,237.73
74616,39792.8,237.52
74916,39863.1,237.50
75217,39947.9,237.68
75517,40032.2,237.86
75817,40111.6,237.96
76118,40203.8,238.24
76418,40237.0,237.73
76718,40335.7,238.09
77019,40399.4,237.99
77319,40516.9,238.59
77619,40578.2,238.45
77920,40647.0,238.42
78220,40722.9,238.47
78520,40767.0,238.12
78821,40879.2,238.64
79121,40893.8,237.91
79422,40982.2,238.13
79722,41093.4,238.63

does that seem right for a pentium d-940, with 2 gigs ram?

3.2ghz 200 fsb
[OCAU] badger
2006-06-16 00:57:35
benched my P4 3.0 Ghz (prescott) with 512 Mb Ram at 189kpts/s (2 instances of 4.33c).  This is much slower than my old p4 3Ghz which did 267kpt/s (different version of muon1 though).  I'll have to rebench my barton and xp 2000+ with 4.33c I guess.
[OCAU] badger
2006-06-19 23:45:33
benched my barton 2500+ (@ 186x11.5 = 2140 MHz) overnight with v4.33c, 189 kpts/s. it was 259kpts/s with which ever previous version I was using (also was running a smidge faster), so there is a significant difference in output with the new version.
Stephen Brooks
2006-06-20 00:40:14
I've had a lot of reports saying the new version is slower, which is odd because while writing it I put in a couple of features that I tested and they showed that they made it faster.
[OCAU] badger
2006-06-20 00:44:28
hmm, I recall rewriting a matlab proggie I wrote, removing a bunch of for loops and using arrray operations instead (at the expense of using a lot of memory) to make it faster.  Except it was about 10 times slower...
[OCAU] badger
2006-07-19 00:34:02
Got my new box, from a 3.0Ghz HT P4 with 512Mb ram to a 3.4 with 1Gb ram.  Too bad it's a dell and I can't overclock the thing :|
Anyhoo, v4.43d with 2 single threaded instances, and I got 124 kpts/s and 120 kpts/s - total of 244 kpts/s. This is about 23% more output than the 3.0Ghz machine I had (198kpts/s not the 189 listed above) for only 13% more clock cycles.  This machine has more ram, and a real video card instead of onboard video, otherwise they are the same (same disk image even) however, the version has changed from 4.43a to 4.43d... I'm going to play with the number of threads again to see what has changed in v4.43d

I should note, my posts 3rd and 4th last above state v4.33a, but I was actually using v4.43a, and I got 198 not 189 kpts/s.
[OCAU] badger
2006-07-19 23:42:12
hmm, benched my barton 2500+ last night with v4.43d, 211kpts/s. V4.43a only gave me 189kpts/s - so obviously something interesting is happening.  Still not up at the old 259 kpts/s though
Stephen Brooks
2006-07-20 11:19:19
It could be the effect of the improved optimiser - the pre-d versions of 4.43 would have done an excessive number of low-yield simulations, which tend to have fewer Mpts per unit time.
[TP]Skatoony
2006-08-21 15:50:48
I'll leave the benchmark program running on my AMD FX-55 ClawHammer and P4 Northwood Extreme Edition (running at 3.2GHZ) for 24 hours then I'll post the results
[TP]Skatoony
2006-08-22 11:30:14
It seems both of my PCs are running this project slower than everyone else.

My results:

AMD FX-55 ClawHammer @ 2.6GHZ: 250 kpts/s
P4 Northwood Extreme Edition @ 3.2GHZ: 227 kpts/s

I think the graph needs to be re-done for v4.4, as no way would my FX-55 be beaten by a Sempron @ 2.5GHZ.
[TA]Assimilator1
2006-09-12 19:06:27
I can only estimate by my posting dates & old DPAD clients I still have stored that I'd benched my XPM & Semp on v442c.
I will re-bench them with v443d (I think that's still the latest) & see what I get.
[TA]Assimilator1
2006-09-12 19:10:49
Btw Stephen the forum link in the 'How to benchmark guide' is out of date
K`Tetch
2006-09-12 22:46:24
Of course, something to bear in mind is that not all simulations are the same.

A batch of 10 simulations, all around 30mpts is going to always take longer than one 300mpts simulation, and it'll also vary by lattice.  In order to really make a comparative baseline, there needs to be a set simulation, or simulations, that stays the same for all computers, so everyone is timing the same event.  Heckm i'd say use an old 300mpts solonoidsto15cm design, and run a timer to time it, and have that as the benchmark.  Of course, I'd code it myself, buyt its rare I have internet access, I've got over 40mb of results for the two current designs waiting to go as soon as I bcan connect it online (and with my P3 based system, thats a LOT of results)
[TA]Assimilator1
2006-09-13 19:10:51
But the benchmark is measuring Kpts/s ,so doesn't that get around the total length problem?
I've no idea about the lattice thing ,Stephen? 

The set simulation sounds a good idea ,in the days of SETI classic TLC had (still has actually) a WU which was used for benchmarking (it was an 'average' WU).

Anyway ,here's my latest benchmarks from my XPm @2.5GHz

24332,521860.7,217.55
24933,521955.2,215.69
25234,522034.8,216.43
25535,522114.5,217.16
26136,522239.0,216.87
26737,522366.7,216.74
27038,522490.3,219.44
27639,522616.0,219.16
27940,522696.6,219.81
28240,522760.6,219.72
28842,522887.1,219.48
29142,522968.0,220.11
30946,523370.7,220.33
31547,523490.3,219.84
31848,523571.1,220.39
32149,523683.6,222.12
32750,523811.5,221.91
33351,523937.9,221.66
36057,524579.6,223.03
37861,524982.4,223.04
38462,525106.7,222.75
38763,525233.4,224.54
40567,525636.2,224.47
40868,525716.5,224.83
41168,525793.8,225.11
41770,525872.2,223.54
43273,526275.0,225.30
43573,526355.4,225.63
44175,526457.7,224.77
44776,526622.8,225.53
46580,527025.7,225.44
47181,527151.7,225.21
47783,527274.1,224.90
48083,527394.9,226.15

That's not from 24hrs as I didn't have the PC on that long ,plus I was watching SGAt so I chopped that bit out ,this list represents about an 8hr period.
If that's not long enough then LMK & I'll rerun it.
But in the meantime it looks like my output has dropped about 15% (assuming an average of 220 Kpts/s)
Stephen Brooks
2006-09-14 12:02:25
--[Btw Stephen the forum link in the 'How to benchmark guide' is out of date]--

OK, fixed that.
[TA]Assimilator1
2006-09-14 22:55:01
Cool

Update on the benchmark ,seems to be higher this time.
This time its about a 14hr period

Uptime (secs),Mpts in file,Estimate kpts/sec
4936,529000.3,0.00
5537,529130.1,215.87
7341,529532.2,221.14
9145,529934.9,222.04
9446,530063.1,235.66
10047,530187.6,232.30
10348,530253.3,231.53
12151,530656.1,229.47
12753,530783.2,228.08
13354,530909.3,226.77
14858,531310.1,232.80
15158,531390.7,233.84
16962,531793.6,232.26
17563,531920.7,231.27
17864,532000.9,232.09
18165,532097.2,234.10
18465,532160.8,233.60
20269,532563.3,232.36
20871,532690.1,231.55
21171,532770.9,232.24
22975,533174.0,231.36
23276,533298.8,234.38
23577,533359.7,233.86
26283,533994.2,233.94
26884,534121.5,233.33
27185,534201.3,233.77
27485,534281.9,234.22
27786,534350.1,234.13
28387,534478.2,233.59
28988,534606.9,233.10
29590,534732.0,232.49
30191,534914.1,234.16
30792,535010.4,232.44
31394,535147.9,232.35
31694,535277.1,234.57
31995,535296.2,232.67
32296,535421.8,234.70
32897,535549.8,234.23
33198,535629.6,234.57
33799,535758.0,234.13
34100,535817.4,233.75
35904,536220.2,233.14
36204,536299.5,233.44
36505,536380.4,233.77
36806,536461.3,234.11
37407,536625.0,234.81
38008,536751.9,234.38
39812,537154.7,233.81
40113,537209.9,233.38
40714,537338.2,233.04
41015,537419.1,233.34
41316,537528.0,234.41
41917,537653.9,234.00
43721,538056.7,233.50
44022,538184.4,234.97
45826,538587.5,234.46
46126,538668.2,234.71
46427,538710.4,234.03
47028,538836.8,233.69
47329,538959.2,234.92
49133,539361.4,234.43
49433,539442.3,234.66
50035,539564.7,234.25
50335,539645.6,234.48
50937,539772.7,234.18

Average seems a little over 233 Kpts/s ,which is about 11% down on the old score
HuubNWW
2006-10-30 12:32:20
FYI
Intel Core 2 Duo X6800 2,93Ghz

V4.43d

After close to 90 hours my graph levels out at 495kpts/s
tomaz
2006-11-21 14:01:13
Better later than never...Here are my benchmarks made in May 2006.

Athlon 1200 (1.2 Ghz).......127
Athlon MP 2400+ (2x2.0 Ghz)........366

Since first machine is about 8 years old and second about 4 y.o. I find it interesting. 
It would be nice if overclocked machines would be marked (and oc frequencies given) in Stephen's graph.
[TA]Assimilator1
2006-11-23 07:26:08
Athlon skt A's didn't come out until about mid 2000 ,so its about 6yrs old at most
Btw CPU clock frequencies are shown in the graph
Xanathorn
2007-01-09 08:06:49
Time to put my findings in as well.  I use 1 client per core as its slightly faster than 1 client using 2 cores.  Wonder if I can do any better (probably with better ram)...

AMD Athlon 64x2 Toledo 4400+ @2.42 Ghz
2x1GB Twinmos PC3200 DDR (2.5-3-3-8-1T)

Average: 502 kps/sec (206 kpts/sec per core)
Client: 4.43d
Xanathorn
2007-01-09 08:08:20
Hmm that must be 251 kpts/sec per core, oh well its early in the morning
[OcUK]diogenese
2007-01-09 20:46:52
Here's another AMD X2 4400, running as standard with 1Gb ram and No.  of cores in setup set to auto, I presume it's running two as I get 50% usage when it's set to 1.
462.8 kpts/sec
bzm
2007-01-20 12:45:26
Intel Core 2 Duo QX6700 @2.93GHz
Average: 928 kpts/sec (232 kpts/sec per core)
Client: 4.43d on Windows XP
bzm
2007-01-21 10:52:56
As above, but with 4 clients: 263 kpts/sec per core = 1052 kpts/sec
[TA]Assimilator1
2007-01-23 20:50:20
Awesome CPU!  ,& nice score too ,though the score/core seems a bit dissapointing.

Your CPU is roughly clocked 17% faster than mine & scores roughly 13% more (for 1 core) ,I would of expected a Core 2 to absolutely destroy an Athlon XPM!  ,maybe DPAD needs optimising for them?
[us-distributed]Merlin
2007-01-25 04:25:46
Stupid question.

I tried to run the program, but for some reason it takes 70% processor when Moun1 if run.

Please explain how to use the program.

Is it stand alone?
I start it after starting Moun1 or from Moun1 command line version?

I would like to bench mark this 64-3800 I have against a 64-3000

Thanks
[TA]Assimilator1
2007-01-27 07:19:13
From DPADS FAQ

How do I use the Muon1Bench.exe program?  First, make sure sample files downloading is switched off in config.txt.  Then put the program in your Muon1 directory (same place as muon1.exe) and simply start and leave it running while Muon1 is working.  It will produce a file BenchCSV.log which has three columns: one logging the time in seconds, the second showing the number of new Mpts Muon1 has produced and the third showing a running estimate of the speed in units of kpts/sec (1000 kpts = 1 Mpts).  Once you've run it for a day or so (enough for the estimate to be stable), post your results to the benchmarking thread in the forum.
Mezocop
2007-02-02 04:38:23
I can offer you the simple program written by our member of a command (Dhara): http://www.berk.ru/muon/muonstats.rar
Just put the muonstats.exe in your Muon folder, run it and wait some results

P.S. Beta version, mistakes are possible
iNSaNiTi
2007-02-17 02:37:36
Amd Venice 3000 + @ 2872 Mhz

Average 351 ... mpts , got different results on each period maybe cauze changing lattices ?
havent run muonbench for long time, but soon i will give it a try

[url]http://img440.imageshack.us/img440/9158/muon351kptsaveragete9.png[/url]
K`Tetch
2007-02-26 20:23:59
eachperiod is different because it only checks every 6mins (default).  If you had a result come in soon after the previous check, it'll be a lower value than if it finished just before the check.  Thats why you need an a fairly lengthy run to get the average out right.

Personally, I think it'd be best if a decent size (say 70MPTS) simulation was set as a standardised one, and timed.  a much better comparison
K`Tetch
2007-02-26 20:38:41
that way would also give comparisons with different client versions (forgot to add)
iNSaNiTi
2007-03-11 17:31:22
Here my results again :

Uptime (secs),Mpts in file,Estimate kpts/sec
8359,4161798.3,0.00
9571,4162246.5,369.89
11091,4162681.9,323.46
12303,4163113.2,333.41
13515,4163546.0,338.96
14727,4163978.1,342.29
15939,4164408.2,344.30
17151,4164807.1,342.22
18363,4165238.6,343.89
19575,4165670.0,345.19
20484,4165983.6,345.19
21696,4166414.5,346.13
22908,4166843.9,346.81
24120,4167274.9,347.48
25332,4167707.4,348.15
26544,4168138.3,348.64
27756,4168550.7,348.11
28968,4168983.0,348.62
30181,4169417.3,349.15
36241,4171580.3,350.83
37454,4172012.8,351.08
38666,4172443.1,351.23
39878,4172875.4,351.44
41090,4173311.0,351.73
42303,4173747.0,352.01
43817,4174178.6,349.15
45029,4174612.9,349.45
46242,4175044.9,349.67
47454,4175475.3,349.84
48667,4175906.4,350.01
49879,4176377.9,351.14
51092,4176810.8,351.31
52304,4177244.3,351.49
53516,4177672.5,351.53

* DFI Lanparty nF4 Ultra-D rev A3 (s939) *
* MP: 317 x9 *
* AMD Athlon 64 3000+ 1800MHz @ 2853MHz *
* 2x 512Mb G.Skill @ 285Mhz 2,5-4-4-7 1T *

Average is even higher with less Mhz than before

Screen :

Free Image Hosting at www.ImageShack.us
HaloJones
2007-03-13 23:53:23
Nice!  I think that's the new fastest single-cpu/core benchmark!  AMD 64's seem to be very good at this stuff.  My Opteron at 2700 is three times faster than a P4 at 2400!  In fact it takes 6xP4-2400s to keep up with one Opty at 2700 and an Athlon XP at 2400.
TurtleBlue
2007-03-15 11:31:41
Here is my first time results post for my Tower box...

Uptime (secs),Mpts in file,Estimate kpts/sec
31392,67189.5,0.00
32293,67300.0,122.69
36796,67999.1,149.81
37697,68134.9,149.95
38298,68194.8,145.59
41000,68598.3,146.64
41300,68676.5,150.09
42201,68812.8,150.19
45203,69216.3,146.76
45503,69297.1,149.36
46104,69377.9,148.76
46704,69446.5,147.40
47605,69581.3,147.53
50324,69984.8,147.65
50625,70049.0,148.68
53327,70452.0,148.74
54228,70586.7,148.77
54829,70667.7,148.41
55429,70749.8,148.12
56361,70884.0,147.97
57305,71020.4,147.84
57695,71101.0,148.71
58595,71211.1,147.84
61297,71633.0,148.59
62198,71766.7,148.58
63154,71904.7,148.46
63754,71982.9,148.12
64355,72115.3,149.44
65256,72249.6,149.43
66156,72384.5,149.44
67057,72516.4,149.36
67958,72652.5,149.40
68859,72788.1,149.43
69159,72811.0,148.85
71862,73214.5,148.88
72463,73295.4,148.67
73063,73376.4,148.47
73364,73427.7,148.63
74265,73557.6,148.54
76967,73960.9,148.58
77868,74092.6,148.53
82372,74760.1,148.50
83273,74893.2,148.49
83874,74974.6,148.34
86576,75378.2,148.39
87177,75458.9,148.24
87777,75549.0,148.26
88678,75682.3,148.25

CPU Family: Intel(R) Pentium(R) 4 CPU 3.06GHz Model 2, Stepping 7
CPU Speed: 3066 MHz
No Overclocking
2 - 512Megs Corsair DDR333 (2700) Memory Chips - Value Select Latency » 2.5-3-3-7-2T
Mainboard : Intel D845GERG2 FSB Frequency : 533.3 MHz (QDR)
Chipset : Intel i845GE
Processor : Intel Pentium 4 @ 3066 MHz Northwood
Video Card : ATI Technologies Inc Radeon X800 Series
Hard Disk : HDS722516VLAT80 (165 G
Hard Disk : HDS722525VLAT80 (250 G
Hard Disk : HDS72251 (165 G
Hard Disk : IC35L040 (41 G
Hard Disk : IC35L090 (82 G
Monitor Type : NEC Technologies NEC LCD1530V - 15 inchs
Operating System : Microsoft Windows XP Professional 5.01.2600 Service Pack 2
DirectX : Version 9.0c

Have 2 other PC's working on dpad as well...will post their muon bench result soon...

Quite depressed about the 3.06ghz northwood dpad results compaired to the rest of the pack

I have read all of the posts in this group and thanks to [TA]z, Assimilator 1, Stephen and the rest of the gang I have come up
with the ideal to put together another box (or mini crack-rack) using an old AMD Thunderbird 1ghz setup I built for my nephew which no
longer works (just as well...afraid that if it did it could out-perform my 3.06ghz northwood!) and put together a AMD 3600 or 3800 Winsor
system.  From the post I understand that the speed of the memory is very important so if you guys (and dolls) have any recommendations on
this and on any other parts, I am all ears.  I have some bits & bobs around (agp vid cards, antec power supplies, etc.) so I may not have to
buy a complete set of goodies and keep the cost DOWN!  I would devote this setup (or mini-crack rack) to dpad and put the 3 other INTEL pc's back to working for F@H where there is a desperate need of help for Team Anandtech!

[TA] TurtleBlue
[TA]JonB
2007-03-20 00:13:17
I've been using Muonbench to find my optimum overclock.  I have an Opteron 165, currently running at FSB of 240 x 9, so 2160mhz.  I had to drop my memory speed from DDR400 to DDR333 to get that high.  Currently, I get a combined value of 425kpts/sec.  To go higher on the FSB, I'll probably have to drop my memory speed to DDR226. I'm not sure what the impact will be, but I'll know soon because I just rebooted with the same 240 x 9 setting and the lower memory speed.  Any guesses what the new Muonbench value will be?
[TA]JonB
2007-03-20 15:08:35
The experimenting on my Opteron 165 (ASROCK 939Dual-SATA2) continues.  I left the FSB at 240mhz as described above but dropped the memory speed down to DDR266. As expected, it continued to run DPAD smoothly, but the MuonBench numbers didn't drop as much as I anticipated (from 425kpts/sec to just 415kpts/sec).  This may be because the memory timings were automatically improved slightly by the SPD chip. 

So now I've moved the FSB up to 245mhz then 250 then 255 and will continue reporting.  Memory speed will have to stay at DDR266. Would be nice to edit the updates in rather than add new entries.  hint hint
[TA]JonB
2007-03-23 02:21:36
My best current settings for the above Opteron 165.
Motherboard settings are:
FSB 265mhz
HT multiplier is 4x
CPU speed is 2.39Mhz (9x multiplier)
Memory is 265mhz (DDR200 setting is a 1x multiplier, but even though the memory is DDR400, it doesn't like speed ).  I'll try boosting this to DDR266 (1.33x mulitplier) to see if its stable there.
All in all, much peppier than it was at the stock 1.8mhz.  Much peppier.

here are the muonbench numbers and the average:


430.86
446.06
430.01
436.81
423.33
427.77
425.03
420.99
423.68
415.85
418.87
421.53
430.55
419.23
424.54
426.25
425.17
424.95
426.55
420.46
422.74
425.08

425.74 (average, and my best run so far)
Pascal
2007-03-23 16:13:58
The program does not produce a file with values, only something like this:

Uptime (secs),Mpts in file,Estimate kpts/sec
Uptime (secs),Mpts in file,Estimate kpts/sec
Uptime (secs),Mpts in file,Estimate kpts/sec
Uptime (secs),Mpts in file,Estimate kpts/sec

As I did not find a good manual on how to use it, it would be nice to get one if this is possible.
Thanks.
[TA]JonB
2007-03-23 17:52:29
Not much of a manual needed.  Copy the muonbench.exe file into the same folder that your DPAD program stores all of its files.  It works by periodically examining the results.dat file, so it just needs to be in the same place as results.dat.

As it runs, it creates a muonbench log file.  What you pasted above looks like you copied it from the DOS Command window that muonbench runs in, but your lack of any other numbers tells me you just don't have it in the correct folder.
Pascal
2007-03-25 15:28:08
JonB,
I did everything correct, as I have seen in between.  I have to wait for more data before posting details.  At the moment I bench two clients on my dual core Athlon 64 X2 4200+ Windsor (2 x 2200 MHz, FSB 200) with standard rates and 2048 MB PC-667 Memory.  I do not like overclocking because of errors in results.

On my second machine, an Athlon XP 2400+ (2000 MHz, FSB 166) with 256 MB RAM and Windows 2000 Prof., the muon1bench crashes.  I tried it again and it crashes.  I can assure that the system is running well.  With folding@home and 17 or bust this machine is running stable 24/7.
Whats wrong with the benchmarking tool?
[DPC]white_panther
2007-03-25 19:50:52
try to install the program again.....
: contact : - - -
E-mail: sbstrudel characterstephenbrooks.orgTwitter: stephenjbrooksMastodon: strudel charactersjbstrudel charactermstdn.io RSS feed

Site has had 25162802 accesses.