HostID 1001562 - Richard Haselgrove's Q6600 Quad Core

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109403264653
RAC: 35523820

I've uploaded the latest data

Message 78897 in response to message 78893

I've uploaded the latest data for the 8 hosts shown below. The smoked trout host is still inactive.

1001562 belonging to RH (#1).
1001564 belonging to RH (#2).
997488 belonging to peanut.
1084360 belonging to nivekfalk.
1101980 belonging to RAMA.
1106990 belonging to Herr Datenrat.
989926 belonging to Elphidieus.
1109009 belonging to th3.

I haven't had time to look at any of the above at all. The main reason for collecting this data is to make sure there is a goodly collection so that Mike can continue fine tuning his RR tool if he wishes. It's also there for Richard to plot if he chooses. I would hasten to add that I would fully understand if he doesn't have the time to continue doing that.

I'd like to thank him most sincerely. The information in his plots has added a whole new dimension to our understanding of the crunch times variations. With Bikeman's explanation of the wiggles, and the prediction (now being borne out) of decreasing runtime variance with the new apps, the whole exercise has been most worthwhile from an understanding point of view.

If we hadn't had the original plots from Richard and the cycle model proposed by Peter to explain it all, and then Mike's RR tool to really start analysing the variation, we would not have been in a position to so quickly investigate the whole flock of new apps that have been recently released. There would have been a lot more confusion out there as to whether or not there was any significant benefit for a particular transition.

So my personal thanks go to all those who have contributed to the various discussions. It has been most enjoyable and most worthwhile.

Cheers,
Gary.

Richard Haselgrove
Richard Haselgrove
Joined: 10 Dec 05
Posts: 2139
Credit: 2752853842
RAC: 1405366

With all the fun and games,

With all the fun and games, I've been neglecting my own hosts - in danger of being called 'off topic' in my own thread. So:

4.32

(direct link)
This is pretty close to a full cycle - peaks should be 338 and 508.

4.36 - first show

(direct link)
The left-hand axis is almost peaking - last point 303, theoretical peak 302. Those two lonely little pink points in the bottom right are nowhere near the trough - that doesn't come until 378. Just look at those times. (For anyone comparing with Gary's datafile, I've left out the first four 'pure' 4.36 datapoints - I was still flushing SETI tasks from the system while these were running, and they were significantly slower). From Bernd's post:

Quote:
The Hough part is largely limited by memory bandwidth, which is somewhat orthogonal to FP speed.


I wonder if that explains Datenrat's Peak as well?

th3
th3
Joined: 24 Aug 06
Posts: 208
Credit: 2208434
RAC: 0

RE: 1109009 belonging to

Message 78899 in response to message 78897

Quote:


1109009 belonging to th3.


Most of my 4.27 results are pretty worthless, had motherboard issues and was using a mATX board for a while, the overclock varied greatly during a period, only the last of my 4.27 resuslts were done with the same board and speed as the 4.35 results.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109403264653
RAC: 35523820

I've uploaded the latest data

Message 78900 in response to message 78897

I've uploaded the latest data for the 8 hosts shown below.

1001562 belonging to RH (#1).
1001564 belonging to RH (#2).
997488 belonging to peanut.
1084360 belonging to nivekfalk.
1101980 belonging to RAMA.
1106990 belonging to Herr Datenrat.
989926 belonging to Elphidieus.
1109009 belonging to th3.

We should be starting to accumulate some sizeable result files now!

Cheers,
Gary.

Richard Haselgrove
Richard Haselgrove
Joined: 10 Dec 05
Posts: 2139
Credit: 2752853842
RAC: 1405366

Sorry, I was busy yesterday.

Sorry, I was busy yesterday. From Gary's data -

Peanut:

(direct link)
A good portion of a cycle, with some wiggles higher up the peak.

Nivekfalk:

(direct link)
Another good portion of a cycle, though I think we can probably discount even more of 864.35 than we already had.

Herr Datenrat:

(direct link)
Well, the peak wasn't a fluke after all. I think he just has a very slow memory controller on that machine (16 cores, 48 GB RAM - it ain't a standard desktop). Bernd said the Hough part was memory-intensive, and varied with declination.

Elphidieus:

(direct link)
Make of it what you will!

RH 4.32:

(direct link)
I'm going to retire this one now - I think we've got all we need from 4.32, and should concentrate on 4.36: and CPDN will be wondering what's happened to my trickles. My 4.36 monitor has been getting rather scattered datapoints recently, but seems to be starting on a cleaner run - I'll plot it next week.

Brian Silvers
Brian Silvers
Joined: 26 Aug 05
Posts: 772
Credit: 282700
RAC: 0

RE: I'm going to retire

Message 78902 in response to message 78901

Quote:

I'm going to retire this one now - I think we've got all we need from 4.32

Awww... To think, I was just beginning to enjoy the double-helix action going on in the trough...

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109403264653
RAC: 35523820

I've uploaded the latest data

Message 78903 in response to message 78900

I've uploaded the latest data for the 8 hosts shown below.

1001562 belonging to RH (#1).
1001564 belonging to RH (#2).
997488 belonging to peanut.
1084360 belonging to nivekfalk.
1101980 belonging to RAMA.
1106990 belonging to Herr Datenrat.
989926 belonging to Elphidieus.
1109009 belonging to th3.

I'll stop collecting for 1001562 (RH #1) once its cache of work is exhausted.

Thanks very much for the plots, Richard. Very nice job there.

Cheers,
Gary.

Richard Haselgrove
Richard Haselgrove
Joined: 10 Dec 05
Posts: 2139
Credit: 2752853842
RAC: 1405366

RE: ..... I'll stop

Message 78904 in response to message 78903

Quote:
.....
I'll stop collecting for 1001562 (RH #1) once its cache of work is exhausted.


You'll be waiting a long time! I just meant I would retire it from the intensive analysis of 4.32 - it'll still be contributing to the cause, but just as an ordinary 4.36 worker, alongside other projects.

@ Brian - yes, that intrigued me as well. I even got as far as plotting alternate points as two separate trend lines - it worked well for a while, but ascended into chaos as they left the trough. Still, I don't think we've left the wiggles behind entirely with 4.36 - I'll see if I can plot you a nice double helix from one of the others.

Edit - just thought it would be nice to put my sig. on this one - this research run has taken me over 1M for Einstein, and 4M for BOINC.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109403264653
RAC: 35523820

I've uploaded the latest data

Message 78905 in response to message 78903

I've uploaded the latest data for the 8 hosts shown below.

1001562 belonging to RH (#1).
1001564 belonging to RH (#2).
997488 belonging to peanut.
1084360 belonging to nivekfalk.
1101980 belonging to RAMA.
1106990 belonging to Herr Datenrat.
989926 belonging to Elphidieus.
1109009 belonging to th3.

All the above are continuing to add completed results to their respective lists.

Cheers,
Gary.

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109403264653
RAC: 35523820

Once again, another update

Message 78906 in response to message 78905

Once again, another update for the 8 hosts shown below.

1001562 belonging to RH (#1).
1001564 belonging to RH (#2).
997488 belonging to peanut.
1084360 belonging to nivekfalk.
1101980 belonging to RAMA.
1106990 belonging to Herr Datenrat.
989926 belonging to Elphidieus.
1109009 belonging to th3.

All the above, except for RH's #1 host, are continuing to add completed results to their respective lists.

Cheers,
Gary.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.