Textual exploitation stats for short NL, comment & discuss

[PeteX]

Ok, importing & housekeeping done.

Some figures in case someone is interested:
- About 30mil short nl hands imported
- Import took about ~100h (started with ~255h/s and went down to ~100h/s at the end)
- Full housekeeping after that took about 62h
- Database size on disk is about 155G
- This all on a brand new 1,5T Seagate Barracude 3.0G/s SATA-drive that had no other content except the newly created database.
- My computer is Quad Core Q6600 with 4G memory and Vista Home Premium.

Just so that you know what you are up against when doing such a huge import...

And yes, PT3 sucked it all in without crashing... Count me as a happy customer...

[kraada]

Thanks for the update, it will definitely be useful to be able to refer to

[APerfect10]

Ok, importing & housekeeping done.

Some figures in case someone is interested:
- About 30mil short nl hands imported
- Import took about ~100h (started with ~255h/s and went down to ~100h/s at the end)
- Full housekeeping after that took about 62h
- Database size on disk is about 155G
- This all on a brand new 1,5T Seagate Barracude 3.0G/s SATA-drive that had no other content except the newly created database.
- My computer is Quad Core Q6600 with 4G memory and Vista Home Premium.

Just so that you know what you are up against when doing such a huge import...

And yes, PT3 sucked it all in without crashing... Count me as a happy customer...

With an updated cache; how is HUD and tracker performance? HUD performance should be good although I suspect some of the tracker reports will take awhile to load...

Best regards,

Derek

[PeteX]

With an updated cache; how is HUD and tracker performance? HUD performance should be good although I suspect some of the tracker reports will take awhile to load...

Best regards,

Derek

Well, my intention is not to use this particular database (it has no hands from me, only datamined hands) for the HUD, hence I didn't update the cache, and that is not included in the housekeeping time reported above. I'm only doing SQL-queries on this database.

I'm currently waiting for the Summary-tab page to refresh. The list of (253,126) players was populated maybe in 1-2 minutes, the Site Summary was populated maybe in 2-3 minutes and now I'm waiting for the Player Summary to populate and that has been going on for a few minutes now, probably will take quite a while.

[kraada]

The reports can use the cache as well and the reports should populate faster when the cache is up to date.

[PeteX]

The reports can use the cache as well and the reports should populate faster when the cache is up to date.

Oh, right! I didn't know that. I thought the cache was for HUD only.

How does the report know when to use the cache and when to query the actual tables?

[kraada]

PT3 uses the cache whenever possible; the only time it's not possible if:
(1) There is no cache
or
(2) There is a new stat used in the report which has not been cached.
or
(3) You're using the HUD in the replayer (since it filters to the HUD stats at the time the hand was played, the cache is not usable).

[PeteX]

PT3 uses the cache whenever possible; the only time it's not possible if:
(1) There is no cache
or
(2) There is a new stat used in the report which has not been cached.
or
(3) You're using the HUD in the replayer (since it filters to the HUD stats at the time the hand was played, the cache is not usable).

Ok, so what if I import 1000 hands and then build the cache -> Reports use the cache, I presume?

But what if I then import another 1000 hands and do NOT build the cache, what happens with the reports? Do they show the information from the first 1000 hands or from all 2000 hands?

Judging from your response (1) is false as there is a cache and (2) is false as there are no new stats and (3) is def false when looking at the reports.

So, if I interpret your response literally, the report shows only stats as calculated from the first 1000 hands but disregards the last 2000 hands...?

[kraada]

I believe that PT3 keeps track of when the caching was last done and will use data since the caching was completed as well, just that data isn't cached.

So in your 2000 hand example it will give you the right data, just the first half of it will be gettable a lot faster (as for the second 1000 it needs to select sum(x) where x is newer than the first 1000 hands, etc.

[PeteX]

Ok, that makes sense. So basically the performance will be highest with a totally up-to-date cache and the more you import new hands without updating the cache, the more the performance will degrade.