EPTI dont see it.
Uh sorry i misremembered, here is my own test on latency: https://pastebin.com/raw/7aKxTEFx (which is not accurate by any means)
BW General Discussion - Page 418
| Forum Index > BW General |
|
MeSaber
Sweden1235 Posts
I dont see it. Uh sorry i misremembered, here is my own test on latency: https://pastebin.com/raw/7aKxTEFx (which is not accurate by any means) | ||
|
tec27
United States3702 Posts
On April 28 2022 11:37 MeSaber wrote: @tec27 did you post something after "DTR can be:"? I dont see it. Uh sorry i misremembered, here is my own test on latency: https://pastebin.com/raw/7aKxTEFx (which is not accurate by any means) Yes, it's a link to a Twitter post (which TL turns into an embed). Seems to be working fine for me, maybe something is blocking it for you. As for the latency number, it comes from my table here (which is one-way latencies, double them to get what people usually measure with ping): ![[image loading]](https://i.imgur.com/odSWeYs.png) These values aren't completely exact, as there's an overhead factor that we don't have an exact value for, but they can be seen as the upper bound on max latency for each setting. I'll note also that these are calculated for ShieldBattery's netcode, which manages to reduce the overhead factor a bit, so if you want to be safe on bnet you may want to subtract ~10ms from the numbers. | ||
|
johndoe44
3 Posts
| ||
|
johndoe44
3 Posts
| ||
|
MeSaber
Sweden1235 Posts
On April 29 2022 08:30 tec27 wrote: Yes, it's a link to a Twitter post (which TL turns into an embed). Seems to be working fine for me, maybe something is blocking it for you. As for the latency number, it comes from my table here (which is one-way latencies, double them to get what people usually measure with ping): These values aren't completely exact, as there's an overhead factor that we don't have an exact value for, but they can be seen as the upper bound on max latency for each setting. I'll note also that these are calculated for ShieldBattery's netcode, which manages to reduce the overhead factor a bit, so if you want to be safe on bnet you may want to subtract ~10ms from the numbers. Thats funny. My numbers arent that way off 😂 What causes lag spikes/stutter? Is it when ping goes above the values you posted? Is there a difference of using high TR + xhigh lat vs low TR + low lat? I know Bonyth prefer high TR and high/xhigh. Is this how you should do it? | ||
|
Bonyth
Poland600 Posts
I prefer low latencies. However, because as I mentioned earlier, dynamic turn rate doesn't get the right values, i'm forces to put the latency at 'high' if it was set as 'low' before. | ||
|
MeSaber
Sweden1235 Posts
On May 03 2022 21:50 Bonyth wrote: That's not entirely true what u've said. I prefer low latencies. However, because as I mentioned earlier, dynamic turn rate doesn't get the right values, i'm forces to put the latency at 'high' if it was set as 'low' before. This is what i mean though. If you set high you get higher TR, if you keep low you get lower TR. Which is best? Does it make a difference? | ||
|
Bonyth
Poland600 Posts
| ||
|
Lorch
Germany3692 Posts
| ||
|
tec27
United States3702 Posts
On May 03 2022 18:22 MeSaber wrote: Thats funny. My numbers arent that way off 😂 What causes lag spikes/stutter? Is it when ping goes above the values you posted? Is there a difference of using high TR + xhigh lat vs low TR + low lat? I know Bonyth prefer high TR and high/xhigh. Is this how you should do it? Stutter is caused by not hitting the required timing for turns. When you don't have all of the turns of other players by the time it needs to be executed, the game freezes to wait for them to arrive. That's all a stutter is. Sometimes these are very small (less than 1 turn), sometimes these last for multiple seconds. Generally DTR picks turn rates that cause stutters that are about 1 turn or less, so they aren't super noticeable if you're watching the game, but they tend to feel absolutely terrible for those playing and mess up people's micro, etc. Yes, there is a difference between using low, high, or extra high, versus adjusting turn rates, but people tend to have bad intuition for this (and I think Grant's explanations on the official forums were quite bad and muddied the waters further). In my opinion, the only number you should care about is the average latency between performing an action and having it executed by the game. That is, if I tell a unit to move, how long does it take between my click and the unit actually starting to move? When you increase the in-game latency setting (e.g. low -> high), you're increasing how many turns the game buffers before executing them. The length of each turn is determined by the turn rate. When I perform an action, it may not be right at the end of the turn, it could be anywhere within the turn, that is, it could be anywhere between 0 and 1000 / turn_rate milliseconds before the turn is boxed up and sent to other players. If we add that value to the length of the buffer in milliseconds, we get the minimum and maximum bounds for action latency: 1000 * buffer_size / turn_rate and 1000 * buffer_size / turn_rate + 1000 / turn_rate Therefore if we want the average latency for any given action, the formula is (after simplification): (1000 * buffer_size + 500) / turn_rate Using that formula we find that although TR 24 Extra High and TR 12 Low have the same max latency requirements, TR 24 Extra High has 187ms average latency vs TR 12 Low's 208ms, so TR 24 Extra High should be preferred. As you can see by Bonyth's post above this one, many people have the opposite intuition. Which is why on ShieldBattery we stopped showing the turn rate + user latency setting at all, preferring the latency number that actually matters. There are still reasons to prefer Low or High latency settings if you're picking a static turn rate however, namely that if the connection quality worsens and you're on Extra High already, there's no setting you can change to improve things without leaving the game. For that reason we try to avoid Extra High settings on SB unless it's the only value that works, but in the future (when we have our own dynamic netcode implementation), we plan to allow larger buffer sizes and remove the user setting entirely, which will fix this issue. | ||
|
MeSaber
Sweden1235 Posts
What happens is that the game fast-forwards like its on crack. Do you have any good explanation why this happens instead of showing a lag screen? This only happens on high TR with some lag, never lower TR numbers like 16 below. Thank for the explanation btw. | ||
|
Bisu-Fan
Russian Federation3339 Posts
| ||
|
WaterSerpentM
United Kingdom139 Posts
you could ignore my elaborations and fill it out yourself for example there is a dispute that its good to scout with a observer or a scout, reason being, observers can scout undetected while scouts can scout and poke(harass), to be honest, any unit can scout, its just that all units are deferent if you want to clarify yourself what is good, you can fill out your own elaborations: Technique 1: Army Army: Use front units backed up with ranged units. Technique 2: Power Up Power Up: Upgrade Army Stats Technique 3: Scouting Scouting: Looking for expansions being built using fast units to destroy expansions. Technique 4: Fortressing Fortressing: To either block or hide behind or creating map control Technique 5: Support Support: Using support units to back up your army Technique 6: Poking Poking: Using fast air units to attack resources carriers to get away with Technique 7: Rendezvous Simply retreating units for backup Technique 8: Drops Drops: on mineral line | ||
KwarK
United States43989 Posts
On May 06 2022 01:10 WaterSerpentM wrote: regarding the 8 techniques, i know it was too much of elaborated statements i wrote that comes under the list you could ignore my elaborations and fill it out yourself for example there is a dispute that its good to scout with a observer or a scout, reason being, observers can scout undetected while scouts can scout and poke(harass), to be honest, any unit can scout, its just that all units are deferent if you want to clarify yourself what is good, you can fill out your own elaborations: Technique 1: Army Army: Use front units backed up with ranged units. Technique 2: Power Up Power Up: Upgrade Army Stats Technique 3: Scouting Scouting: Looking for expansions being built using fast units to destroy expansions. Technique 4: Fortressing Fortressing: To either block or hide behind or creating map control Technique 5: Support Support: Using support units to back up your army Technique 6: Poking Poking: Using fast air units to attack resources carriers to get away with Technique 7: Rendezvous Simply retreating units for backup Technique 8: Drops Drops: on mineral line I think you’re missing a few. What about the clicking technique to see their upgrades? That’s a good technique in PvT. Then there’s knowing, that’s when you use flyers against tanks because you know that tanks can’t shoot up. Also spelling. That’s actually two techniques. It’s useful for communicating by text but can also make enemy units disappear with psionic storm etc. | ||
|
WaterSerpentM
United Kingdom139 Posts
On May 06 2022 03:56 KwarK wrote: I think you’re missing a few. What about the clicking technique to see their upgrades? That’s a good technique in PvT. Then there’s knowing, that’s when you use flyers against tanks because you know that tanks can’t shoot up. Also spelling. That’s actually two techniques. It’s useful for communicating by text but can also make enemy units disappear with psionic storm etc. you are right these are the techniques(a core action to excel at a skill) knowing what units beats other units comes under attributes everything else comes under Tactics(sub techniques), there are loads of them, I don't know them all Also there is Ability(mouse and keyboard control) =] | ||
|
Timebon3s
Norway764 Posts
Thanks for these tips, this is good stuff  | ||
|
Crypto007
1 Post
| ||
|
DiMiaN
Finland306 Posts
| ||
|
redknights
206 Posts
On May 05 2022 04:50 Bisu-Fan wrote: Anyone been following the BW college tournament? I didn't see any coverage of it anywhere. I realize it's like only in Korean, so it's not getting as many eyes, but the storylines are pretty fun and a lot of good training content as well as lots of new people picking up the game. Probably great for the Starcraft/Afreeca industry ($$$) but May will be a bad month for fans of top-level play The matches are going to be ~85-90% non-pros, and the top pros are spending most of their time teaching and training their students for the college tourney. Proleagues are going to be very rare until the tourney is over Queen wrote that he is taking a lengthy break from streaming due to the college tourney (since he is currently not participating in a college) | ||
|
Lorch
Germany3692 Posts
| ||
| ||
