Diablo® III

SNS CD PING and you

Posts: 13,644
View profile
Btw I still totally have no idea how CD increases multiplier, it just does not make sense to me. ChC and APS increasing the multiplier makes much more sense.

Anyone kind enough to explain the correlation between ChD and eDPS to me? Not by data but by some simple theorycrafting.


No idea but it does. Those with socket weapons can test ez. Just take out gem. Multiplier will decrease. When I made spec and only had 84 CD I was only running a 7.25x mutli now 8.66x with 249%.

But I think ping is king still.
Edited by Aimless#1700 on 1/16/2013 11:53 AM PST
Reply Quote
Posts: 13,644
View profile
At ~500 ms my multiplier is 3.4.

=]


serious?
Reply Quote
Maybe because the sheet DPS is only calculated based on 2.74 aps? Like, the game only assumes you do 2.74 attacks per sec and calculate the DPS based on that. But in fact because we do so many attacks in 1 sec, the number of attacks > 2.74, so more CD = more off-sheet damage, so higher multiplier. Idk just a guess, maybe dumb lol.

@Aimless: and yes serious, I have to wait like 1 second to see my command executed, you can't imagine how bad it can be for me these days. Well, at least I think I can predict enemies' movement a lot better lol, like I have to run away before the Molten mobs actually die, even if only 1-2 sec after that the explosion animation starts.
Edited by novice#1750 on 1/16/2013 11:58 AM PST
Reply Quote
Posts: 13,644
View profile
Well I have enough data to conclude multiplier have more to do with geography than anything else. So testing is only valid on a individual basis. I will still be doing about 20 CM mutilpliers based on our favorite variants but results should only be taken serious in a relative sense.
Reply Quote
Yeah when testing we should all use the standard SNS with Bone Chill and Worm Hole.
Reply Quote
Actually thus far with the four data points in this thread, we can already see there's a VERY inverse linear relationship with lag:

latency = [500 300 129 24]; %in ms
DPSMultiplier = [3.4 4.95 6.6 8.8];
Correlation Coefficient: r = -0.9762, p=0.02
Best Fit for Predicted DPSMultiplier: DPSMultiplier = latency *-0.0108 + 8.5199 (with normR = 0.8689)

I'll be curious if this relationship holds up with more sampling across CM/WW wizards.
Edited by TekkZero#1963 on 1/16/2013 12:12 PM PST
Reply Quote
http://youtu.be/FQvEtbRL_Ho

I totally forgot my stats at the point when i made that vid. Latency is at 280 to 340 (half of the fight there the ping bar went to yellow)

But I can't get any better result now regardless the same latency. Something is wrong....
Reply Quote
Actually thus far with the four data points in this thread, we can already see there's a VERY inverse linear relationship with lag:

latency = [500 300 129 24]; %in ms
DPSMultiplier = [3.4 4.95 6.6 8.8];
Correlation Coefficient: r = -0.9762, p=0.02
Best Fit for Predicted DPSMultiplier: DPSMultiplier = latency *-0.0108 + 8.5199 (with normR = 0.8689)

I'll be curious if this relationship holds up with more sampling across CM/WW wizards.


I want to cry, I did more DPS back in December despite 20k less sheet DPS, CC APoC aps the same. LOL...
Edited by novice#1750 on 1/16/2013 12:14 PM PST
Reply Quote
Adding in ximae's stats (just realized I missed them), we now have 5 data points: Aimless, worx, ximae, novice, and me. And the relationship seemingly appears even more significant:

latency = [500 300 129 55 24]; %in ms
DPSMultiplier = [3.4 4.95 6.6 8.1 8.8];
Correlation Coefficient: r = -0.9802, p=0.0033
Best Fit for Predicted DPSMultiplier: DPSMultiplier = latency *-0.0110 + 8.5883 (with normR = 0.88)
Edited by TekkZero#1963 on 1/16/2013 12:20 PM PST
Reply Quote
@novice: I'm really sorry man. That's rough to have such bad latency. :(
Reply Quote
Posts: 13,644
View profile
Thanks for your maths Tekk! More results requested! Please use standard SNS with bonechill.
Reply Quote
Ur welcome! I'll keep updating the numbers as more people contribute. I confess I'm actually VERY shocked at how linear it is right now; however, it's worth noting the sample size is small, so we'll have to see if it holds up as more people contribute.
Reply Quote
I really recommend normalizing the results so they're taken over the same elapsed time, because duration of fight also has an impact on effective dps. I'd say something like 50-60s. That would also make it easier to compare multiplier with the same gear with different CD, by removing gem.

Also, if anyone is having ghom move out of their twisters, you need to omit that run. You can keep him 100% in your twisters by moving to the upper left corner of the room at the start. Just wait til he's in melee range to activate SA. The small AP cost should be compensated by the adds he'll likely spawn while waddling toward your location.
Reply Quote
I really recommend normalizing the results so they're taken over the same elapsed time, because duration of fight also has an impact on effective dps. I'd say something like 50-60s. That would also make it easier to compare multiplier with the same gear with different CD, by removing gem.

Also, if anyone is having ghom move out of their twisters, you need to omit that run. You can keep him 100% in your twisters by moving to the upper left corner of the room at the start. Just wait til he's in melee range to activate SA. The small AP cost should be compensated by the adds he'll likely spawn while waddling toward your location.


or if you have any kind of recording software, just record the whole thing and pick out a ~45 second steady state portion and note his hp at the start and end of it.
Reply Quote
aimless why is that i have higher dps then you much higher cd and your **#@*##@!% my dps?
Could it really be the ping?
Reply Quote
So far it seems like it's a strong candidate. I just re-did one of my Ghom tests using both ximae's and Loroese's suggestions, and I did improve my numbers; however, I'm still well below xiame's and Aimless's multipliers (6.92 now, rather than 6.03). :(
Edited by TekkZero#1963 on 1/16/2013 1:30 PM PST
Reply Quote
Posts: 13,644
View profile
@ worX- ping and ur unbuffed dps is lower and u'r not even at acceptable breakpoints. 2.73, so you miss casting opportunities. 2.5 is terrible on single target. Cooldowns wont reset for crap.
Edited by Aimless#1700 on 1/16/2013 1:29 PM PST
Reply Quote
Agreed with Aimless. 2.5aps is horrible for single target unless you're using Cold Snap. :(
Reply Quote
Posts: 3,955
View profile

Oh and killer i did shell too 4u again it's 7.23x...shards more than 8-10%..more like 19% more dam.


I am touched ;) You are going to find my latest results very interesting.

So I recorded two more videos. One with shards and one with shell. My profile probably shows my archon gear so here is a brief summary:

218k dps
351% CD
116 ms ping

ShockNadoShard took 83 seconds. 138,425,408 HP / 83 seconds = 1667776 eDPS. 1667776/ 218438 = 7.635 multiplier.

http://youtu.be/xuVokbNjfJg

Shocknadoshell took 101 seconds. 138,425,408 HP / 101 seconds = 1370548 eDPS. 1370548/218438 = 6.3 multiplier.

http://youtu.be/xuVokbNjfJg

This is a very strange turn of events. I last uploaded a couple of videos a month or so ago showing only a 10% difference between shards and shell. This is what Aimless is referring to. I cannot explain why this time the difference between shell and shards is 21%. The only major difference is that I went through some upgrades and now my dps is 218k from 186k. Thank god I have videos for proof or I would look like some kind of a liar!

The above videos are going to take approx 30 minutes to upload.

**edit**

Here are the old videos

http://www.youtube.com/watch?v=fC_KBBkCI3s

http://www.youtube.com/watch?v=K_nQ0ZDIRy8
Edited by KillerElite#1853 on 1/16/2013 1:58 PM PST
Reply Quote
01/16/2013 12:24 PMPosted by Aimless
Thanks for your maths Tekk! More results requested! Please use standard SNS with bonechill.

I can run it a few times and keep track if someone links the thread that had the info how to do it...
Reply Quote

Please report any Code of Conduct violations, including:

Threats of violence. We take these seriously and will alert the proper authorities.

Posts containing personal information about other players. This includes physical addresses, e-mail addresses, phone numbers, and inappropriate photos and/or videos.

Harassing or discriminatory language. This will not be tolerated.

Forums Code of Conduct

Report Post # written by

Reason
Explain (256 characters max)
Submit Cancel

Reported!

[Close]