Come one mods you must have nominations for this.
Pretty much any Threadspotting thread will get a large number of views, sometimes in the thousands. And threads discussing sex are pretty popular, too. Out of those, which has the most, though? I couldn’t say.
Threads extolling the virtues of Arnold Winkelried usually rank very highly in post/view counts.
In addition those threads have a very good chance of becoming a sticky thread, remaining forever at the top of the forum list.
Click on “Views” and then set “from:” at the bottom of the page to ‘the beginning’. This should show you the thread in that forum with the most views ever. In ATMB, for example, it’s a thread on a fortune-telling function that had almost 8000 views.
GQ: What happens on May 33rd but only on leap year? 18198 views
GD: Ask the Gay Guy IV! 11223
Cafe: Books you read when you were young that absolutely shocked/disturbed you 12177
IMHO: Very vaguely creepy. 26396
MPSIMS: Funny things said during love making/intimate moments… 36343
Pit: Get off my plane, mind-melded Fundie zealots!. 17085
You can do it with replies, too (did not know this, thanks Sublight):
Pit: THIS IS WAR DAMMIT (begun by RosieWolf 10/20/2000; locked; 427 replies)
MPSIMS: Guy Stuff (begun by Uncle Beer; 1,113 replies)
IMHO: Very vaguely creepy (begun by Homer; locked; 504 replies)
Cafe Society: US Big Brother 3 (begun by handy; 457 replies)
Great Debates: For David B, the good Doctor and others who are knowledgable about Creation debate (begun by Satan; locked; 749 replies)
General Questions: What happens on May 33rd but only on leap year? (begin by pkbites; locked; 313 replies)
Comments on Staff Reports: Mormon Temple (begun by hergrace; locked; 133 replies)
Comments on Cecil’s Columns: homeopathy (begun by JillGat; 219 replies)
ATMB: Master List (begun by Chrome Toaster, 447 replies)
Of course, those are the kind of searches that really slow down the server.
Yeah, so don’t you guys do it!
But then again, I thought I read that doing a search in one forum isn’t nearly as bad as searching all forums simultaneously. Isn’t that true? These are only for one forum at a time.
Appalling.
I did some testing on my server, and on that sample case, looking at views seems to be fairly harmless. The following script:
SELECT SUM(views) AS totalviews FROM thread
which sums ALL views of ALL threads on the entire Board runs in less than a tenth of a second on my very pokey server with 51,000 posts.
Let’s see…the SDMB server is about 10 times as powerful, but has about 400 times the posts…so factor of 40…if all things are equal, it would take about 4 seconds. However, it would also depend on your cache settings too…well, it likely would slow the server down a little, but about as much as an ordinary search, I am guesstimating.
This is one of the many reasons I love Anthracite. 
Sorry about the search, anyway. Won’t do it again.
Well, yeah…when you’re comparing one to the other. But how do we measure 0/1 with all the other threads out there?
Anthracite - your simple sum is different than what a search has to do. A search would have to retrieve thread starter, thread title, last post date, etc… (the information shown on the search screen), which would probably involve a join to several other tables, and then sort all the data before presenting it to the user. A much different operation than simply doing a full table scan and only reading one column.
Ah…good point, Arnold. But I figured it still had to tally all of the views, and could do a simple comparison while doing that to find out what is the larger one.
That is, in each case, it must read each thread view, and then either 1) sum it, or 2) compare it. Since the sum and compare seem to be the same effort in SQL, I tried to equate those in terms of CPU time and database access. So I still think they may be equivalent in work.
But let me point out…regardless of if I’m even close to correct, I am not advocating in any way, shape, or form that people argue with or continue to do things that the Staff here would rather not have us do. Sometimes I focus too much on the tech things, and forget that it really doesn’t matter for the overall point of keeping the Board running smoothly.
No Anthracite, they are not equivalent work. What the users do through the search screen is ask to see all the threads, for all dates, in descending view order. This is not the same as saying “I only want to retrieve one thread, the one with the most views.” To accomplish what the search screen is doing, the software would have to read all the threads and do a sort of all that data.
But Sublight/dantheman didn’t use the search screen to get their data, they used the “sorted by…”-option of the forum page.
If it works the way I think it does, the board keeps a separate index for each field that you can choose the forum display page to be sorted by. This way it always has to go only through one pageful worth of consecutive references to collect the corresponding thread data. It shouldn’t drastically change the work to display the forum page if you use a different search order than the default by date.
It still has to look up all the values in the “views” or “replies” fields, and sort out the ones with the highest values, which is a search, though perhaps an efficient one. By contrast, most of the data for the regular forum view (by date, newest first, default page breaks) is probably kept cached, since it’s accessed so heavily. Note: I know nothing at all about the way VBulletin actually works - if you want actual technical details, there are a couple of other posters in this thread who know better than I.
The actual ‘searching’ work would have been already done, by keeping the indexes sorted at the time the data is updated. To find the highest values of a field, you just start reading at the top of its index and follow the references. Compared to looking for the latest post dates it may be less efficient to use another field whose data is less likely to be cached, but certainly not by the magnitude of a full-blown search request scanning through an entire table.
Arnold - I have looked again at the source and asked the vBulletin people about it. I respectfully do not agree that the code is doing what you feel it is, but I’m not going to debate here anymore. Regardless of who is correct, the only real correct thing is to follow the SDMB Staff instructions on use of Board functions.