Theck's MATLAB TPS analysis - WotLK/3.x

WATERBOYsh wrote:Hey theckhd, I was just wondering something. How hard would it be to convert your matlab stuff to an exel spreadsheet? I don't know any of the math stuff, but you have all the formulas in place already. It could be made so that you plug in your stats and it spits out graphs showing you which stats you should be stacking (could maybe even throw what avoidance stat you should stack for an optimal ratio of defense:dodge:parry). I might can look into this in a few weeks. I'm a programmer, but my company is trying to push out a beta build by the middle of August so I've been really busy lately.

Also, I don't know much about matlab, but can it spit out an executable? Is it an interpreted language? If we want to run your code, what do we need. I was looking at freemat and it seems pretty good, but I don't really want to mess with the code; just run it.

Don't say me, that I've wasted a lot of my last 4 years: viewtopic.php?f=6&t=3117

Rhî wrote:

WATERBOYsh wrote:Hey theckhd, I was just wondering something. How hard would it be to convert your matlab stuff to an exel spreadsheet? I don't know any of the math stuff, but you have all the formulas in place already. It could be made so that you plug in your stats and it spits out graphs showing you which stats you should be stacking (could maybe even throw what avoidance stat you should stack for an optimal ratio of defense:dodge:parry). I might can look into this in a few weeks. I'm a programmer, but my company is trying to push out a beta build by the middle of August so I've been really busy lately.

Also, I don't know much about matlab, but can it spit out an executable? Is it an interpreted language? If we want to run your code, what do we need. I was looking at freemat and it seems pretty good, but I don't really want to mess with the code; just run it.

Don't say me, that I've wasted a lot of my last 4 years: viewtopic.php?f=6&t=3117

I guess being stickied with the title "Tankadin Spreadsheet" is too discrete... maybe you should change the title to "Rhi's Tankadin Spreadsheet (I was here before Theck's MATLAB TPS analysis (a Jonsey derivative work))".

For what it's worth, I find the spreadsheet useful...

WATERBOYsh wrote:Hey theckhd, I was just wondering something. How hard would it be to convert your matlab stuff to an exel spreadsheet? I don't know any of the math stuff, but you have all the formulas in place already. It could be made so that you plug in your stats and it spits out graphs showing you which stats you should be stacking (could maybe even throw what avoidance stat you should stack for an optimal ratio of defense:dodge:parry). I might can look into this in a few weeks. I'm a programmer, but my company is trying to push out a beta build by the middle of August so I've been really busy lately.

Also, I don't know much about matlab, but can it spit out an executable? Is it an interpreted language? If we want to run your code, what do we need. I was looking at freemat and it seems pretty good, but I don't really want to mess with the code; just run it.

This is kind of ironic, since the matlab code was originally extracted from Jonesy's spreadsheet (which is discontinued, I think). However, Rhî has a sticky thread in this forum for his spreadsheet, which is up-to date.

MATLAB is an interpreted language. It can spit out executables, but I don't have a lot of experience with making it do that.

As for running it yourself, it depends on what you want to do. Running the simulations for yourself wouldn't be hard, you could just download the m-files and run them in FreeMAT. However, to customize them, it would take a little familiarity with how I've structured things. I used to have a "how-to" section in the table of contents, but it was painfully outdated so I removed it a few months ago. If all you want to do is run them for yourself with stats specific to your character, you can just edit the gear_theck.m and base_stats.m files to match your own gear set and stats. Running any of the calculations should then accurately reflect your character. Both of those files are input modules that are coded for easy readability, so they shouldn't be hard to edit.

Ya know... I don't think I even noticed the spreadsheet in the stickies (or if I did, I guess I just forgot about it)... lol. Thanks for pointing that out. I'm playing with it now.

The problem with stickies is that after not changing for so long, I tend to completely overlook them when looking at a forum index.

*edit*
Though I seem to have some problems with the things that are supposedly excel only (the pages with buttons). I have office 2007 and the buttons do absolutely nothing. Is there something special I have to do?

Jasari wrote:For what it's worth, I find the spreadsheet useful...

I do as well, for certain things. I only wrote the MATLAB translation because I was starting to ask questions that spreadsheets don't handle well. A spreadsheet is great if you want to input your character stats and get information about your current state. But if you want to ask "What happens to this as stat X increases dramatically," it's just not the proper tool.

Sure, I could sit there and plug in each value of X, and write down the results, and then plot them. That could take hours though, and it's really not how most spreadsheets are designed to be used. On the other hand, that's exactly what MATLAB is designed to do. Large-dimensional arrays representing many parameters? No Problem!

It's not really a question of which is better or more useful. They're just completely different tools. I wouldn't try to drive in a nail with a paint brush, but trying to paint the fence with a hammer wouldn't be the brightest idea either.

Cylan wrote:No rush on it. I'm just curious how it does compared to threat talents at this point.

By the way, in case you missed it, it's in the post right above yours.

WATERBOYsh wrote:Ya know... I don't think I even noticed the spreadsheet in the stickies (or if I did, I guess I just forgot about it)... lol. Thanks for pointing that out. I'm playing with it now.

The problem with stickies is that after not changing for so long, I tend to completely overlook them when looking at a forum index.

*edit*
Though I seem to have some problems with the things that are supposedly excel only (the pages with buttons). I have office 2007 and the buttons do absolutely nothing. Is there something special I have to do?

Activate your macros (or lower the security level for macros) or use the PTR-version, which isn't running with excel, but with Open office (with macros acivated). I changed the supported program from excel to open office, because everyone could have Open office (Freeware and Open Source).

Jasari wrote:I'm probably ready to throw a "beta" version online somewhere, but don't have any of my own webspace available - I assume the company I work for wouldn't like to host a Paladin threat calculator. So if anyone knows of a cheap (or free) place that works with ASP, please let me know.

I got you covered, give me a few mins (i'm at work ) and I'll pm you the ftp info.

You have some kind of title for it?

I came to this thread because I was searching the forums for information on libram swapping. About 10 pages back, or one month ago, there was talk of analyzing TPS for and against libram swapping. I have spent the past few hours reading this thread and continuing my search elsewhere.

In my search, I found a mod, BANG! Combat Item Swapper, that handles libram swapping somewhat seamlessly, though I question if there's a quicker and more efficient way to handle it with this automation, that's for another thread. What it seems to do, amongst a lot of other fluff, is watch the COMBAT_LOG_EVENT_UNFILTERED and if the spell you indicate casts, then it equips the item you indicate. When I was reading the code, this took me by surprise. I didn't realize the equip functions didn't require a keypress and can be fired by events.

Since there were no results posted, I decided to obtain some data of my own on live. I have three librams I switch to on my 9's - Libram of Resurgence for Consecrate, Libram of Obstruction for Judgement, and Libram of the Sacred Shield for Holy Shield. I ran the tests for one minute. I started off with a Divine Plea and started the clock as soon as I hit the "9" to start the rotation. I ran with Blessing of Wisdom and Seal of Wisdom to give me enough mana to reach the one minute mark. I only have DPS and damage results. I did it a few times both with and without the mod running. Results were similar for each attempt in a given setting, so only posting data from one test. LotSS was used exclusively during non-swap.

Results

With BIS Running
SoR - 6 hits and 2 miss for 22795 damage (3799 avg)
HotR - 8 hits and 0 miss for 13214 damage (1652 avg)
Consecration - 40 hits for 11342 damage (284 avg)
Melee - 15 hits, 4 glancing, 4 crit, 4 miss, and 3 parry for 10289 damage (394, 303, 791, 0, 0 avg)
Judgement - 4 hits and 1 crit for 4406 damage (734 and 1469 avg)
DPS - 920.6

Without BIS Running
SoR - 9 hits and 1 miss for 32282 damage (3587 avg)
HotR - 6 hits, 2 crits, and 1 miss for 15502 damage (1562 and 3067 avg)
Melee - 15 glancing, 11 hit, 4 crit, 3 miss, 3 parry, and 2 dodge for 12113 damage (296, 410, 792, 0, 0 avg)
Consecration - 42 hits and 2 misses for 11328 damage (270 avg)
Judgement - 6 hits and 1 miss for 4571 damage (762 avg)
DPS - 1145.8

My results are hardly conclusive given that my sample size is small, but I think it does hint at the lost value in libram swapping, at least from a DPS perspective. However, we are tanks and we do have that other little issue of surviving. In that regard, libram swapping is pretty impressive. I gained 842 block value with LoO and LotSS up at the same time. That is a nice chunk of damage that I wouldn't take and with the upcoming changes to those librams, we can guarantee close to 100% uptime.

Future testing might consider removal of the Libram of Resurgence. I'm more interested in seeing analysis results with TPS numbers rather than DPS numbers though. I'd be willing to trade away a few hundred TPS for 800+ BV in my full avoidance set. Hoping this post will respark interest in this area now that the talent builds seem to be thoroughly covered.

There's already a thread about this, but I don't have time this evening to sift through 18 pages to find the one bit of info I need.

Ghostcrawler wrote:Only white attacks can stack the Sea of Vengeance / Corruption dot. However special abilities should still trigger the bonus weapon damage. I'm not sure the current PTR build you have reflects that last bit.

He was speaking about Ret, so I assume this includes Judgement, DS, and CS. I'm not certain whether that includes ShoR though. Does anyone have any information about that? From my understanding, the build on the PTR doesn't reflect this change, so at the moment unless we have a blue post saying one way or another, all I can do is speculate.

theckhd wrote:

Cylan wrote:No rush on it. I'm just curious how it does compared to threat talents at this point.

By the way, in case you missed it, it's in the post right above yours.

I appreciate it the work. It's interesting to see how much Redoubt contributes on a pretty plot.

Jerey-Darkspear wrote:My results are hardly conclusive given that my sample size is small, but I think it does hint at the lost value in libram swapping, at least from a DPS perspective. However, we are tanks and we do have that other little issue of surviving. In that regard, libram swapping is pretty impressive. I gained 842 block value with LoO and LotSS up at the same time. That is a nice chunk of damage that I wouldn't take and with the upcoming changes to those librams, we can guarantee close to 100% uptime.

Future testing might consider removal of the Libram of Resurgence. I'm more interested in seeing analysis results with TPS numbers rather than DPS numbers though. I'd be willing to trade away a few hundred TPS for 800+ BV in my full avoidance set. Hoping this post will respark interest in this area now that the talent builds seem to be thoroughly covered.

As far as TPS goes, just multiply everything by 1.43 and then everything but melee by 1.9. Given your numbers:

With:
51757 Holy Damage = 140.6k Threat = 2343 TPS
10289 Phys Damage = 14.7k Threat = 245 TPS

Without:
63683 Holy Damage = 173k threat = 2884 TPS
12113 Phys Damage = 17.3k threat = 289 TPS

However, you're right about your sample size being very small. I suspect that the bigger factor in this is the larger number of misses (1 more miss on ShoR, 2 miss 3 parry on melee, though 1 less miss on HotR).

theckhd wrote:

Jerey-Darkspear wrote:My results are hardly conclusive given that my sample size is small, but I think it does hint at the lost value in libram swapping, at least from a DPS perspective. However, we are tanks and we do have that other little issue of surviving. In that regard, libram swapping is pretty impressive. I gained 842 block value with LoO and LotSS up at the same time. That is a nice chunk of damage that I wouldn't take and with the upcoming changes to those librams, we can guarantee close to 100% uptime.

Future testing might consider removal of the Libram of Resurgence. I'm more interested in seeing analysis results with TPS numbers rather than DPS numbers though. I'd be willing to trade away a few hundred TPS for 800+ BV in my full avoidance set. Hoping this post will respark interest in this area now that the talent builds seem to be thoroughly covered.

As far as TPS goes, just multiply everything by 1.43 and then everything but melee by 1.9. Given your numbers:

With:
51757 Holy Damage = 140.6k Threat = 2343 TPS
10289 Phys Damage = 14.7k Threat = 245 TPS

Without:
63683 Holy Damage = 173k threat = 2884 TPS
12113 Phys Damage = 17.3k threat = 289 TPS

However, you're right about your sample size being very small. I suspect that the bigger factor in this is the larger number of misses (1 more miss on ShoR, 2 miss 3 parry on melee, though 1 less miss on HotR).

Yeah. It just wasn't possible to go for longer than a minute on a full rotation. I figure with scaling from raid buffs, that gap will become even larger. For modeling purposes, it seems that equipping extends the GCD by more than just .1 sec as was speculated before. I'd be quite comfortable with saying it extends GCD by .25 to .50 sec. That should help you get a more realistic model at least.

I'm going to modify that add-on and make it more specific to paladins. See if removing all the extra "junk" will cut down on runtime and make it swap librams faster.

Jerey-Darkspear wrote:Future testing might consider removal of the Libram of Resurgence. I'm more interested in seeing analysis results with TPS numbers rather than DPS numbers though. I'd be willing to trade away a few hundred TPS for 800+ BV in my full avoidance set. Hoping this post will respark interest in this area now that the talent builds seem to be thoroughly covered.

That's quite interesting. Your threat is still decent even though you're swapping three librams. For a pure mitigation scenario like holding Steelbreaker or Freya, it'd be pretty ridiculous to gain 200 dodge rating with the new libram and 842 block value at the same time.

The problem with libram swapping, at least to me, is that it seems obviously unintended for you to be able to get the benefit of multiple librams at the same time. If it truly is possible it should be reported as a bug.