• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

SETI TA benchmarking table

Assimilator1

Assimilator1

Elite Member
For us all to be able to compare our rigs in SETI quickly & easily simply involves finding finished WUs for your chosen PC(s) & noting the WU time of *46.xx credit WUs.

This isn't quite as accurate as us all crunching the same WU but if we each give at least 3 46 cred WU scores for each rig then the inaccuracy is kept down to about 5% or less, which is good enough.

To find WU results go to your BOINC SETI stats>Computers>tasks(of chosen PC), note the WU (CPU) time & the precise credit given to at least 3 WUs/PC.
Try not to pick WU times that may of been lengthened by playing games etc alongside it 😉.

When posting results, at the minimium also post your CPU speed,type & SETI name, & ideally chipset & RAM type. There are also columns for RAM timings & speed etc, client type & comments for those who can/want to add that info for better comparisions.
Posting ARs isn't necassary for the table but is useful info, optional extra 😉.

See the prototype TA Top 100 SETI chart here (based on 73.xx cred WUs which were common at the time). Premise changed from Top 100 to general benchmarking table.
I'll fix the error with the missing rank 6 when I update the table.

(Btw anyone know how I could convert my table from an excel file to a table like Roelof's TLC Top 200?)

After investigating 2 methods of benchmarking individual rigs in SETI (for which RAC is mostly useless as it can take 2-4 weeks to level out) as discussed in this thread, I can currently only use the credit specific WU method, as opposed to having a benchmark WU. A benchmarkable WU is the most accurate method but I don't have space to host the needed files for starters. Does anyone else?

Incase you're asking why 46.xx credit WUs? it's because they are WUs with a common AR of ~0.43-44 which I found after examining 200 WUs (see link above) to be one of the most common WUs.

*It now seems 46 cred WUs are rare with 44 being the most common, can people LMK what they are seeing in their results please?
If it turns out that the most common credited WU is constantly changing then this may not be possible, unless there are some common WUs which regularly re-occur.
 
Nearly 30 views (excl mine) & no-one's posted a score? or even any comments?🙁

No one interested?? or was my explanation poor?
I've edited & highlighted a few points.
 
OK here is the first set of scores:

CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 6400+, 2 cores.
Speed: 3175 MHz
Cache: L1= 64KBytes, L2: 1024KBytes
FSB: 200.9MHz
RAM: 2048 MBytes, 401 MHz, 4-4-4-12-21, Dual Channel
Chipset: ATI RD780 rev 00
BIOS: Award
Op-system: Win XP SP3 all updates.
Client: AK_v8 SSE3

WU1: 354839301
CPU-time: 5614.547
True Angle: 0.438763
Pulse count: 4, Gaussian count: 1 other counts: 0
Submitted: Oct 29, 2008, 04:26 UTC

WU2: 355094400
CPU-time: 5038.734
True Angle: 0.436572
Triplet count: 3, all other counts: 0
Submitted: Oct 29, 2008, 08:00 UTC

WU3: 355161359
CPU-time: 5045.109
True angle: 0.438969
Spike count: 3, all other counts: 0
Submitted: Oct 30, 2008, 15:03 UTC

WU4: 355532982
CPU-time: 5084.219
True angle: 0.437273
Spike count: 0, Pulse count: 1, Triplet count 1, Gaussian count 1.
Submitted: Oct 29, 2008, 20:56 UTC

WU5: 355567669
CPU-time: 5043.641
True angle: 0.438723
Spike count: 1, all other counts: 0
Submitted: Oct 30, 2008, 15:03 UTC

WU6: 355653780
CPU: 5207.406
True angle: 0.438895
All counts: 0
Submitted: Oct 30, 2008, 20:30 UTC

I hope this helps ...
Thanks for the job you are doing, Mark! 🙂
 
No probs 🙂, btw shouldn't your CPU speed be 3.2GHz?😕
The average of your WUs is 5172s = 1hr 26.2 mins. I will add your score to the chart when I have some more scores from other posters.

If anyone's interested, some scores from my rigs.

Q6600 @3.35GHz, P45 chipset, DDR2 800 @744MHz 4-4-4-15
Client AK_v8_SSE3, Win XP

Credit _ AR __ Time s _ mins

46.07 _ 0.4389 _ 2578 _ 43
46.01 _ 0.4401 _ 2564 _ 42.7
46.08 _ 0.4391 _ 2834 _ 47.2
46.08 _ 0.4391 _ 2763 _ 46
46.08 _ 0.4391 _ 2582 _ 43
46.20 _ 0.4377 _ 2585 _ 43.1
46.12 _ 0.4387 _ 2574 _ 42.9
46.12 _ 0.4387 _ 2588 _ 43.1

Averaged time in mins (rounded) - 43.9 mins 🙂

Sempron 3100 @2.5GHz (256 KB L2), NF3 chipset, DDR 400 @227MHz 3-3-3, Win XP

Credit _ AR ___ secs
46.51 _ 0.433 _ 8447 _ 2hrs 21mins
(need more scores for these 2)

E4500 (2.2 GHz), P35 chipset, DDR2 800 RAM, Win XP
45.97 _ 0.440 _ 4275 _ 1hr 11.25mins (OK not quite a 46 cred WU but its the closest I have & it is very close!)

(need more scores for these 2)

Lots of views but still not many posters.......
 
Originally posted by: Assimilator1

No probs 🙂, btw shouldn't your CPU speed be 3.2GHz?😕

Lots of views but still not many posters.......
Click to expand...

Well, I checked with CPU-z: 3 279 MHz (a small OC ... FSB 205MHz...) 🙂

 
I don't know why I didn't think of this before, but it would be far less restricting if we could use results from any credit rated WUs. This could be done easily just by dividing the WU time (in secs) by the credit number.
I'm not sure if the crunching time is linear though, I'll do some quick numbers on it now.

Q6600 @3.35GHz, v8.0 app

Credit _ Time(s) _ credit/s

79.16 _ 4033 _____ 50.95
69.87 _ 3477 _____ 49.76
65.47 _ 3258 _____ 49.76
56.20 _ 3602 _____ 64.09
56.03 _ 3619 _____ 64.59
46.20 _ 2648 _____ 57.31
44.03 _ 2521 _____ 57.26
40.53 _ 2411 _____ 59.49
38.47 _ 2146 _____ 55.78
35.97 _ 2029 _____ 56.41
33.11 _ 1862 _____ 56.23
16.87 __ 750 _____ 44.46
16.87 __ 761 _____ 45.11
14.39 __ 688 _____ 47.81

Average credit/s - 54.215
Range from average, approx +19% to -18%

Damn, that's no good then 🙁, if we're going benchmark with SETI then we're going to have to stick to a specific credit WU, which means that specific number has to stick around or at least regularly come back.

(On a side note, it seems 44 credit (AR 0.447) WUs are the most common now!, )

I'd appriecate some feed back on which credit scored WU is the most common, TIA.
 
Last edited:
Ok, here's a few of mine. I'll update as more come up. The super slow one was by my home desktop, I updated the optimized app on it so we'll see how much it improves.

Credit_AR_Time
44.20 0.442251 4733.1 SSSE3x (linux) Q6600@2.4
45.35 0.432085 4896.9 SSSE3x (linux) Q6600@2.4
44.18 0.442360 7080.9 setiathome_enhanced 6.02 Q6600@3.28
44.95 0.438569 5020.3 SSSE3x (linux) Q6600@2.4
 
Cancelled due to lack of interest 🙁.

Any late comers feel free to post results, if I get enough requests/results then I will resurrect this little project, until then consider it dead.

If anyone would prefer the benchmarkable WU method then speak up, I could do that instead.
 
Cheers Petrus 🙂, I appriecate the input you & Alyx put in :thumbsup:, I know a 3rd person was going to (still can) once his job allowed time (dejeepster I think) which was fair enough.

But I am disappointed by the lack of response, including, strangely, people who had replied in my previous related thread 😕, also it's not just the lack of results but also near zero feedback, I would understand if a bunch of poeple said something like 'I don't think the results method is accurate enough so I'll hold out for the benchmark WU' or that everyone had a common credit rating which was different making it impossible, but nothing at all!? 🙁.

I guess most SETI folks now just aren't interested in comparing performance between different & like for like rigs.......

Btw in your 1st post you said your CPU was at 3175MHz, but then in your 2nd post you said it was slightly o/ced at 3279Mhz, which was it for the results shown? 😉
 
Originally posted by: Assimilator1
Btw in your 1st post you said your CPU was at 3175MHz, but then in your 2nd post you said it was slightly o/ced at 3279Mhz, which was it for the results shown? 😉
Click to expand...

It was the latter speed - the first one was a wrong reading ... I can sometimes not read my own scrawl - I won't even call it hand writing ... 😉

Anyhow, I am somewhat surprised about the lack of interest. OTOH I assume that the comps are now powerful enough for crunching - and that tweaking out more speed is not as interesting now than before. I mean: now you can oc a processsor (with proper hardware) from 2.7GHz to almost 4GHz - using a cookbook and some patience.
I think that other concerns are now more important than the last % of performance: energy, efficieny, noise, keepig the CPU at low temps ...
Furthermore: if someone wants more power a CPU-upgrade (e.g. to a dual or quad core) is more efficient than tweeking ...

Just my 2 cents ...
 
Yea that's what I did , but you see then you can tweak that quad too 😉.
When I swapped the MSI P6N Nvidia 650 mbrd to the Asus Intel P45 I was able to increase the CPU clock of my quad from 3GHz to 3.35GHz & it also draws 5w less from the mains! 😎 (more power efficient chipset).

No matter what level performance the rig is, it's still interesting to see how it fairs against others, well for some of us anyway, used to be many of us.....
Also it's interesting to see how different CPU architectures fair against each other, particularly in DC.

Talking of which, it's interesting to see that though mine is just clocked ~2% faster than your X2 it's nearly 50% faster!, of course that's a totally unfair comparison as the C2Q design is much newer & it's a more expensive CPU 😉.

Ath64 X2 6400 @3.28 GHz, average 46 cred WU time - 1hr 26mins (5172s)
C2Q Q6600 @3.35 GHz, average 46 cred WU time - 44 mins
 
Yeah, I would not buy <a AMD64 X2 today ... I am looking at C2Q now - I need to replace at least 3 comps within the next month ... OTOH: it would mean going from 3 cores to 12 cores ... I like that. Especially in Seti@Home. 🙂
 
That would give you a huge boost in output!:Q😎

Btw FYI, my Q6600 @3.35GHz @1.34v (when running SETI & DPAD), Asus Intel P45 mbrd, 2GB DDR2, X1950 Pro, 1TB & 300GB HDD etc draws ~220w from the mains (~230v), which I think is quite reasonable 🙂, especially considering my old XPM 2500 @2.5GHz drew 180w! (running DC).
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top