Esper

I don't check this account often.

[Esper @bertuuuu]

At a GP, small talk before the match: "WOW, you are going to team with @LuisSalvatto ?! It must be awesome to know a Pro Tour Champion". Still my favorite mtg story.

[Esper @bertuuuu]

@legacyggbr Bem-vindo Loading!

https://t.co/0d3Ucyt5AT

[Esper @bertuuuu]

@swayuCS @fREQUENCYCS Newer versions of RTSS have a prebuilt CS2 profile (its on the Profiles folder). Load that and change the fps limit value to your preference, stutters will be gone. -allow_third_party_software in the launch options too.

[Esper @bertuuuu]

@Smdster Warnings and game losses for slow play should be destigmatized and be considered part of the game, like a shot clock violation in basketball. The overall standard of play in mtg should be pushed to be way faster than it is.

[Esper @bertuuuu]

@TheZvi It's simple, in cases like this what you need to do is

[Esper @bertuuuu]

@MaximsTweets @iamcs2kitchen Look for Piwielle thread on reddit with better comparison ldat data at different fps points. The higher the fps value the smaller the difference in latency between driver cap and in-game limiter (which reflex uses). But framepacing always better with drivercap.

[Esper @bertuuuu]

@iamcs2kitchen I think good beats perfect here as a benchmark test, you'll see that the latency measurement differences between runs with the same cap method don't deviate much. It's relevant info still, need to have some data point from actual live game conditions for sanity checks.

[Esper @bertuuuu]

@iamcs2kitchen if you don't want to do live game/conditions test for latency, you should at least show the latency comparison at 400, 500 fps etc using your method, that gives the viewer more data to making a decision. Leave to the viewers to figure out how much fps they can maintain in game.

[Esper @bertuuuu]

@iamcs2kitchen no reason to turn off reflex, you can have it on and prevent it from acting as a dynamic frame limiter as long as the cap you use is far away enough to prevent you becoming gpu bound. Framepacing issues with reflex happen when it activates to limit frames, not just for being on

[Esper @bertuuuu]

@iamcs2kitchen leave reflex on, it shouldnt matter for comparing fps_max and nvcp cap, as long as you are far from being gpu bound

[Esper @bertuuuu]

@iamcs2kitchen Or even compare latency at valve dm that you picked, even though that is less representative of a real life scenario

[Esper @bertuuuu]

@iamcs2kitchen Play 5v5 with Frameview or Presentmon to monitor system latency, wait for a big fight (ie., ancient mid, inferno banana) and compare latency values between nvcp cap and fps_max at 400+ fps when not GPU bound.

[Esper @bertuuuu]

@iamcs2kitchen Why are you testing in valve DM servers instead of 5v5 match or communith DM?

[Esper @bertuuuu]

@iamcs2kitchen The user (not me) used arduino to test end to end latency while using -noreflex launch option. GPU usage was below 80%, so reflex impact on latency is moot. Preveting reflex to kick in as a frame limiter has framepacing benefits, which you ignored as well.

[Esper @bertuuuu]

@iamcs2kitchen Bizzare decision to present comparison only at 110 fps and 225 fps, unless you were trying to prove a point before even looking at all the relevant data.

[Esper @bertuuuu]

@iamcs2kitchen Who cares about 1000 fps, it is at 400 fps that latency becomes the same between methods. Very weird to skip this data point. Also a common fps target range for 5700x3d + systems. You could use frameview or presentmon to test system latency in heavy gunfights as a proxy.

[Esper @bertuuuu]

@edelgenius @Dblinfinityteam Voa bruxão

[Esper @bertuuuu]

@8BitBanana__ @NartOutHere the video mentioned in the post above is exactly that.

[Ondrej Strasky @OndrejStrasky]

https://t.co/RqJxgVw0JO