|
On August 18 2017 19:12 Qikz wrote:
Based on your data I can't disagree, but honestly turn rate 8 feels smoother to me and more responsive than iCCUP ever did.
What turn rate did iCCUP use?
Prior to 1.18, these values don't really exist in that sense. All network timings are calculated from the latency value in 1.16 (e.g. 2).
On August 19 2017 00:18 Cryoc wrote:Show nested quote +On August 18 2017 16:21 tec27 wrote:On August 18 2017 04:11 sabas123 wrote:
What are the LL equivalency in SC:R turn rate? I finally ran the tests for this tonight: + Show Spoiler [Data] +All frame counts are from 60fps videos (~16.67ms per frame) of the time it takes
for a probe to begin moving after the click indicator becomes visible.
Blizz Rate 8:
15 frames
14 frames
15 frames
17 frames
16 frames
Min: 14 * 16.67 = ~233ms delay
Average: 15.4 * 16.67ms = ~256ms delay
Max: 17 * 16.67 = ~283ms delay
Blizz Rate 12:
11 frames
13 frames
10 frames
14 frames
Min: 10 * 16.67 = ~166ms delay
Average: 12 * 16.67ms = ~200ms delay
Max: 14 * 16.67 = ~233ms delay
Blizz Rate 12 - High latency:
11 frames
15 frames
12 frames
11 frames
11 frames
11 frames
Min: 11 * 16.67 = ~183ms delay
Average: 11.83 * 16.67 = ~197ms delay
Max: 15 * 16.67 = ~250ms delay
Blizz Rate 12 - Extra High latency:
14 frames
13 frames
12 frames
16 frames
15 frames
17 frames
14 frames
Min: 12 * 16.67 = ~200ms delay
Average: 14.42 * 16.67 = ~240ms delay
Max: 17 * 16.67 = ~283ms delay
Blizz Rate 16:
8 frames
8 frames
11 frames
8 frames
Min: 8 * 16.67 = ~133ms delay
Average = 8.75 * 16.67ms = ~145ms delay
Max: 11 * 16.67 = ~183ms delay
ShieldBattery:
16 frames
13 frames
13 frames
15 frames
15 frames
Min: 13 * 16.67 = ~216ms delay
Average = 14.4 * 16.67 = ~240ms delay
Max: 16 * 16.67 = ~266ms delay
ICCup L2:
15 frames
13 frames
14 frames
16 frames
13 frames
Min: 13 * 16.67 = ~216ms delay
Average = 14.2 * 16.67 = ~236ms delay
Max: 16 * 16.67 = ~266ms delay
Summary of that is that, contrary to a belief which I've seen espoused quite often lately, a rate of 8 is roughly equivalent to 1.16 lan latency (maybe a frame slower, hard to say super precisely without a lot more data). Increasing the latency on turn rate 12 to High seems to increase the variance without meaningfully changing the average latency (but is still faster than what lan latency would have been). Thanks for the test, but I would not come to the same conclusion. A ~15% longer delay time is significant. Basically your data proves, that L2 is indeed right inbetween turn rate 8 and 12, so providing an additional turn rate 10 option would probably be useful to get the L2 setting back.
That assumes the data is far more accurate than it actually is. Measuring by frames here can never give you a truly accurate value as the rendering/recording will basically never line up exactly with network activity and game turn advancement. That's [part of] the reason why these values change between tests. There's no real reason to believe that turn rate 8 is actually different than L2 here. With a lot more samples, it's extremely likely they'd converge to the same thing.
|
I guess this is more of a complaint than anything that really adds to discussion, but vs. R is horseshit. As Z, it upsets me, and I'm sure it's equally shitty for T and P. Maybe instead of playing vs R, it can just pick and show a race? Is that something everyone can get behind?
|
I'm having so much fun with matchmaking ATM. It's such a great addition to BW.
Also, now that the servers are alive, I can host 1v1 games with observers so my friends can see me play. And ramdom people come to observe too ^_^ This is great! I had forgotten how great it felt when people just flood into the games being hosted.
Random pickers are still hard to deal with, but the way I solve it is by doing proxy 2 gate everytime someone randoms. I mean, 2/3 times the build will be viable. And I consider randoming to be a type of cheese so I don't feel bad about cheesing back. Not that I feel bad for cheesing in any other situation though,
I really hope this lasts.
|
Is anyone else having a lot of people dc when they fall behind on ladder?
About half my games the other guy just drops the second they fall behind or they scout that their cheese is countered.
At least it's stopped giving me losses when the opponent dc's i guess :s
|
Are there any good sites for uploading brood war replays? everything I find is tied to Starcraft 2.
|
On August 19 2017 13:11 dsschnau wrote:
Are there any good sites for uploading brood war replays? everything I find is tied to Starcraft 2.
This thread might be interesting.
www.teamliquid.net
|
Delete please, should have edited.
|
On August 19 2017 02:16 tec27 wrote:Show nested quote +On August 18 2017 19:12 Qikz wrote:
Based on your data I can't disagree, but honestly turn rate 8 feels smoother to me and more responsive than iCCUP ever did.
What turn rate did iCCUP use? Prior to 1.18, these values don't really exist in that sense. All network timings are calculated from the latency value in 1.16 (e.g. 2). Show nested quote +On August 19 2017 00:18 Cryoc wrote:On August 18 2017 16:21 tec27 wrote:On August 18 2017 04:11 sabas123 wrote:
What are the LL equivalency in SC:R turn rate? I finally ran the tests for this tonight: + Show Spoiler [Data] +All frame counts are from 60fps videos (~16.67ms per frame) of the time it takes
for a probe to begin moving after the click indicator becomes visible.
Blizz Rate 8:
15 frames
14 frames
15 frames
17 frames
16 frames
Min: 14 * 16.67 = ~233ms delay
Average: 15.4 * 16.67ms = ~256ms delay
Max: 17 * 16.67 = ~283ms delay
Blizz Rate 12:
11 frames
13 frames
10 frames
14 frames
Min: 10 * 16.67 = ~166ms delay
Average: 12 * 16.67ms = ~200ms delay
Max: 14 * 16.67 = ~233ms delay
Blizz Rate 12 - High latency:
11 frames
15 frames
12 frames
11 frames
11 frames
11 frames
Min: 11 * 16.67 = ~183ms delay
Average: 11.83 * 16.67 = ~197ms delay
Max: 15 * 16.67 = ~250ms delay
Blizz Rate 12 - Extra High latency:
14 frames
13 frames
12 frames
16 frames
15 frames
17 frames
14 frames
Min: 12 * 16.67 = ~200ms delay
Average: 14.42 * 16.67 = ~240ms delay
Max: 17 * 16.67 = ~283ms delay
Blizz Rate 16:
8 frames
8 frames
11 frames
8 frames
Min: 8 * 16.67 = ~133ms delay
Average = 8.75 * 16.67ms = ~145ms delay
Max: 11 * 16.67 = ~183ms delay
ShieldBattery:
16 frames
13 frames
13 frames
15 frames
15 frames
Min: 13 * 16.67 = ~216ms delay
Average = 14.4 * 16.67 = ~240ms delay
Max: 16 * 16.67 = ~266ms delay
ICCup L2:
15 frames
13 frames
14 frames
16 frames
13 frames
Min: 13 * 16.67 = ~216ms delay
Average = 14.2 * 16.67 = ~236ms delay
Max: 16 * 16.67 = ~266ms delay
Summary of that is that, contrary to a belief which I've seen espoused quite often lately, a rate of 8 is roughly equivalent to 1.16 lan latency (maybe a frame slower, hard to say super precisely without a lot more data). Increasing the latency on turn rate 12 to High seems to increase the variance without meaningfully changing the average latency (but is still faster than what lan latency would have been). Thanks for the test, but I would not come to the same conclusion. A ~15% longer delay time is significant. Basically your data proves, that L2 is indeed right inbetween turn rate 8 and 12, so providing an additional turn rate 10 option would probably be useful to get the L2 setting back. That assumes the data is far more accurate than it actually is. Measuring by frames here can never give you a truly accurate value as the rendering/recording will basically never line up exactly with network activity and game turn advancement. That's [part of] the reason why these values change between tests. There's no real reason to believe that turn rate 8 is actually different than L2 here. With a lot more samples, it's extremely likely they'd converge to the same thing.
I don't get your chain of arguments. What is your reason to believe that the delays are not different, good faith?
I feel ingame that TR 8 has a slightly higher delay than L2 and I am glad you took the time to do some tests because I wasn't 100% sure because neither TR 8 nor TR 12 felt like L2. Your sample size might not representative measurement but it is much more likely, that a significant difference really exists.
|
What does the 'environmental effects' setting do. Maybe I'm crazy but when I toggle it on and off I can't see a difference.
|
On August 20 2017 00:02 CobaltBlu wrote:
What does the 'environmental effects' setting do. Maybe I'm crazy but when I toggle it on and off I can't see a difference.
I think one of the things is waves on water.
|
On August 20 2017 00:02 CobaltBlu wrote:
What does the 'environmental effects' setting do. Maybe I'm crazy but when I toggle it on and off I can't see a difference.
Since SC often has very similar tiles next to each other, I think it just increases the variety in the tiles. I agree it doesn't do an amazing job though.
|
Ranked seems down atm. I haven't been able to get a match, have tried restarting and whatnot. Played quite a few games just before now though.
|
does anyone have any guesses on how big the korean scene will grow ? Id love for it to get big again
|
On August 20 2017 02:00 LifeIsGood wrote:
does anyone have any guesses on how big the korean scene will grow ? Id love for it to get big again
Anyone's guess really. It's looking pretty darn good as it is. We are already in the second golden age.
The question is if we will have a third golden age when the current elite gets too old or begin their military service. Will there be Korean teenagers devoting their lives to BW once again? Not yet, but it might happen.
|
On August 19 2017 20:11 Cryoc wrote:Show nested quote +On August 19 2017 02:16 tec27 wrote:On August 18 2017 19:12 Qikz wrote:
Based on your data I can't disagree, but honestly turn rate 8 feels smoother to me and more responsive than iCCUP ever did.
What turn rate did iCCUP use? Prior to 1.18, these values don't really exist in that sense. All network timings are calculated from the latency value in 1.16 (e.g. 2). On August 19 2017 00:18 Cryoc wrote:On August 18 2017 16:21 tec27 wrote:On August 18 2017 04:11 sabas123 wrote:
What are the LL equivalency in SC:R turn rate? I finally ran the tests for this tonight: + Show Spoiler [Data] +All frame counts are from 60fps videos (~16.67ms per frame) of the time it takes
for a probe to begin moving after the click indicator becomes visible.
Blizz Rate 8:
15 frames
14 frames
15 frames
17 frames
16 frames
Min: 14 * 16.67 = ~233ms delay
Average: 15.4 * 16.67ms = ~256ms delay
Max: 17 * 16.67 = ~283ms delay
Blizz Rate 12:
11 frames
13 frames
10 frames
14 frames
Min: 10 * 16.67 = ~166ms delay
Average: 12 * 16.67ms = ~200ms delay
Max: 14 * 16.67 = ~233ms delay
Blizz Rate 12 - High latency:
11 frames
15 frames
12 frames
11 frames
11 frames
11 frames
Min: 11 * 16.67 = ~183ms delay
Average: 11.83 * 16.67 = ~197ms delay
Max: 15 * 16.67 = ~250ms delay
Blizz Rate 12 - Extra High latency:
14 frames
13 frames
12 frames
16 frames
15 frames
17 frames
14 frames
Min: 12 * 16.67 = ~200ms delay
Average: 14.42 * 16.67 = ~240ms delay
Max: 17 * 16.67 = ~283ms delay
Blizz Rate 16:
8 frames
8 frames
11 frames
8 frames
Min: 8 * 16.67 = ~133ms delay
Average = 8.75 * 16.67ms = ~145ms delay
Max: 11 * 16.67 = ~183ms delay
ShieldBattery:
16 frames
13 frames
13 frames
15 frames
15 frames
Min: 13 * 16.67 = ~216ms delay
Average = 14.4 * 16.67 = ~240ms delay
Max: 16 * 16.67 = ~266ms delay
ICCup L2:
15 frames
13 frames
14 frames
16 frames
13 frames
Min: 13 * 16.67 = ~216ms delay
Average = 14.2 * 16.67 = ~236ms delay
Max: 16 * 16.67 = ~266ms delay
Summary of that is that, contrary to a belief which I've seen espoused quite often lately, a rate of 8 is roughly equivalent to 1.16 lan latency (maybe a frame slower, hard to say super precisely without a lot more data). Increasing the latency on turn rate 12 to High seems to increase the variance without meaningfully changing the average latency (but is still faster than what lan latency would have been). Thanks for the test, but I would not come to the same conclusion. A ~15% longer delay time is significant. Basically your data proves, that L2 is indeed right inbetween turn rate 8 and 12, so providing an additional turn rate 10 option would probably be useful to get the L2 setting back. That assumes the data is far more accurate than it actually is. Measuring by frames here can never give you a truly accurate value as the rendering/recording will basically never line up exactly with network activity and game turn advancement. That's [part of] the reason why these values change between tests. There's no real reason to believe that turn rate 8 is actually different than L2 here. With a lot more samples, it's extremely likely they'd converge to the same thing. I don't get your chain of arguments. What is your reason to believe that the delays are not different, good faith? I feel ingame that TR 8 has a slightly higher delay than L2 and I am glad you took the time to do some tests because I wasn't 100% sure because neither TR 8 nor TR 12 felt like L2. Your sample size might not representative measurement but it is much more likely, that a significant difference really exists.
Because adding a single frame of delay over what existed previously would be nonsensical. Stilwell (or some other Blizzard person) stated that 8 was equivalent around when they originally added the option. Many people (such as yourself) didn't accept that at face value, so I attempted to give you some data so that it can be accepted. If you really want to believe something otherwise, I guess you will, but I actually talked to Stilwell (like, on the phone) when they were making 1.17, and know for a fact that targeting L2 was their priority. It'd be pretty damn strange for them to suddenly decide that wasn't the priority (and not only that, but for such a small amount of added latency) once they made turn rate configurable.
It is far more likely that any delay you perceive is related to input and/or graphical changes between 1.16 and now, and not due to the rate at which network turns are sent or processed.
|
Norway28673 Posts
On August 19 2017 02:25 gumbum8 wrote:
I guess this is more of a complaint than anything that really adds to discussion, but vs. R is horseshit. As Z, it upsets me, and I'm sure it's equally shitty for T and P. Maybe instead of playing vs R, it can just pick and show a race? Is that something everyone can get behind?
The main problem is that they're not allowing match picking at the moment. I think once they do, random isn't gonna be a big thing anymore. Also, the disadvantage from having to learn every matchup is genuinely bigger than the advantage you get from randoming.
|
Seems to me skill in SC2 doesn't transfer well over to BW.
Watching Dragon get owned at 15-1600mmr got me thinking that.
Same goes for me really. I can't control jackshit in terms of army when I get beyond 100 supply. So I just end my games as fast as I can. Literally played last 15 games like that.
|
51461 Posts
lol, if dragon is getting owned at 1600 mmr i really doubt he was ever a+ on iccup
edit: could be possible that the skill floor on korea is just super duper higher than in the west, but still.
|
Korea is way harder than EU/NA, way way way harder. ~1900 on KR = B rank players on fish (well during the beta at least, i'm sure the ranks have gotten inflated after the release.. could be around 2K MMR now).
That said Dragon wasn't A+ on ICCup, he was an ok player tho. His id used to be Yongma[ArnC], he played on EU a lot.
|
How do the gateways work? If you play on Korea, do you get matched versus koreans more often? I thought it was a global ladder and didn't matter what gateway you selected.
|
|
|
|
|
|
EPT
