• 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 benchmark updates

N

nexus9

Senior member
I've been thinking that Thunder's seti benchmark page is a bit out of date (with seti client versions), and was thinking of putting something together. Any interest?

-Nexus9
 
Yes, this would be valuable. A lot of people asking about times / cpus / work units out there.
 
You guys might want to do like Ars and pick a specific WU (or perhaps WUs from different angle ranges, eg., one from the VLARs, one from the "mids" and one from the VHARs) for people to run so you get a more accurate way to compare times... 🙂

Alternately, a bunch of us have found that WUs with the same angle range will run pretty close to each other time-wise (if you run them on a dedicated cruncher - they'll vary if you actually use the machine), so people could report times for clients, CPU, etc., for any specifically named angle range and then compare that way...
 
Great idea🙂
I have had thoughts along that line myself but couldn't commit time to it🙁

I would suggest that the Ars standard WU be used.
It would give a broader base to an already existing DB and give us a chance to compare our times to theirs. (Are we running better or are they?)

I know you are a php/mysql wizard-maybe a sheet to fill out on your site with its info fed to mysql for an up to the minute DB🙂 < Im not expecting too much am I > 🙂

😎 😎
 
I'd be glad to help. But, I don't see using good cycle time to benchmark a &quot;standard&quot; WU. Poof's altenative: <<...a bunch of us have found that WUs with the same angle range will run pretty close to each other time-wise...>> would be better to me. Those running SetiSpy, SetiDriver, SetiQ and others have lots of logged data already.
 
I agree with JWM. A standard WU might be cool for those willing to run it. But many of us have logs that can provide &quot;real world&quot; averages and such. I would be happy to compute my averages from my logs and add them to a database.

Good idea.
 
is there any good way to extract text from setispy's log and turn it into a spreadsheet?
 
ElFenix, There is not really an easy way. But, I cut and paste the data into Word2000 and do a few find/replace actions then convert it to a table. The table is then pasted into Excel2000. It takes a few minutes to do, but it works great.
 
Setispy log -> Excel the easy way:

copy all of your setispy log to the clipboard.

Paste into excel.

Highlight first column.

Under the data menu, select text to columns

Voila your done takes 5 seconds.

😎
 
Once all your data is in nice little neat columns you'll want to figure out your average time. My column says &quot;proces time-hr&quot;. I went to the very bottem (after most recent result)and clicked in the cell directly below this column. Then, under the insert menu I clicked on function and choose average out of the list. This averaged the entire row.

My result:

7.09 hrs

this is an overclocked tbird (980 mhz 9.5X103)
Win98se
384mb ram

lately it has been faster after some recent tweaks, but all the previous WU raised average.
 
wow thanks artemedes!

106 3.03 WUs complete (there were a couple that weren't 100%, i took those out)
935.643 hours
8.74 hours/wu std dev 0.50

p3 700/100 MHz
win2000 pro (no service pack)
256/512 meg pc100 222 ram
maxtor diamondmax plus 40 on promise ultra66

and...

79 3.03s
995.789 hours
12.6049 hr/wu

p2 400/100MHz
win2k pro
64 megs pc 100
 
Sounds Good to Me!

Why not average the data instead of messing with a standard work unit? Set the minimum number of values (n)for the database at like 100 wu's using the same client.

Thanks Nexus9 and DOUBLE thanks for your stats website!
 
Yah, badthad thats what I was thinking too. 100 Work units would certainly be a nice sample size to produce a good average.
 
well, it will take a few more days for the slower computers to hit 100 WUs, but its a good limit. the t and standard normals are pretty much the same value there.
 
Thanks artemedes! I'd forgotten about that Excel option. It is easier!

One other option we could include would be to add a min and max to the average. That would cover the Low and High Angle-Range WUs.

For example:
Min. Avg. Max.
5.62 7.12 9.37

I use to do that on my sig sheet, but changed to using SetiQ data since it is available and much easier to gather.
 
we should include a sigma... averages are useless without a standard deviation. xl will do that too.
 
SD or RSD (better) would be cool, but a minor PITA for me since I use SETI Watch. I would have to do a good bit of massaging in Excel to get it out using the CSV file, not fun with 20 PC's. LOL

It comes down to this, absolute accuracy is not really needed. This is for general information only, let's use the KISS method here. 😀
 
Yeah Assim1, they are getting carried away. El prob just completed a Stats class! 😉

I think the min and max with the avg would help and be easy to understand. A range is also helpful since there can be a big spread, especially on slower box. So, when someone says &quot;My machine is taking this long...&quot; we can refer to the chart and say that it is within the range.
 
Don't get it Assim1? hehehehehe

SD is Standard Deviation. It's a measure of the statistical variation around the mean. Usually expressed at +/- around the mean, i.e. 10 +/- 0.5 hours.

RSD is Relative Standard Deviation. It the SD divided by the mean times 100 to convert it to percent.
 
How about keeping it real simple:

using CLI fastest slowest all with WinME 128 or 256 Cas2
C667@850 9.36 11.21 with games running in the evening
P667@750 7.21 8.42 with Mon browsing and e-mail
P1G @1100 6.21 7.19 on my server (DirecPC, WinRoute and etc)

I would love to see times for 486, P200, K6-2, PPro, and Xeon

What would the average user expect to see?

We will pass 500 WU's Thursday!
 
Nexus9 has been quite lately about this topic, sure he's busy with his new house. Glad to see this thread brought back because I would like to see this stat. me=stats junkie 🙂

cakin - I can tell you some odd cpu times, since you're interested:

CLI v3.03
PPro 200, W2K AS, 304MB EDO (dually) - 38-39 hr/wu (runs 2 processes similtaneously)
K6/2-500, ME, 256MB - 25-28 hr/wu
P3-500, NT4, 128MB - 10.5-11.5 hr/wu

[edit]correct PPro times!!...doh![/edit]
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top