Ever since I saw this command on Valve's List of Variables and Console Commands a few months ago, I've been using this at 0.2 since, and I have no clue why this isn't in everybody's autoexecs. The effect of this command is very dramatic, its like changing your cl_interp from 1 to 0 all over again. The default value is 1, which causes a sh** ton of delay. I'm pretty sure that this value is in seconds, considering that there is still a noticeable difference between 0 and 0.2.
Now before you put the value to 0 in your autoexecs, I don't recommend that. Sometimes it can jitter or skip frames if you set it too low. You should set it to the lowest amount without any noticeable jittering. I personally searched using increments of 0.05.
This command also doesn't work well with toasters.
Ever since I saw this command on Valve's List of Variables and Console Commands a few months ago, I've been using this at 0.2 since, and I have no clue why this isn't in everybody's autoexecs. The effect of this command is very dramatic, its like changing your cl_interp from 1 to 0 all over again. The default value is 1, which causes a sh** ton of delay. I'm pretty sure that this value is in seconds, considering that there is still a noticeable difference between 0 and 0.2.
Now before you put the value to 0 in your autoexecs, I don't recommend that. Sometimes it can jitter or skip frames if you set it too low. You should set it to the lowest amount without any noticeable jittering. I personally searched using increments of 0.05.
This command also doesn't work well with toasters.
Does anyone know what it does? Is it a network command?
Does anyone know what it does? Is it a network command?
What is frame latency?
When a frame is rendered, it takes a certain amount of time to do so. That time is measured as the latency, from 0, to render and display the frame.
Doesn't seem to be a network command though I suppose this would control the max amount time before a frame gets shot out. (so more of an engine command in the form of some buffer limit? Somewhat like snd_mixahead? )
http://www.neogaf.com/forum/showthread.php?t=512976
Reading up on this now myself.
Last time I saw this command was in an old version of chris's configs along with mat_framebuffercopyoverlaysize.
He preferred them on defaults I think
Edit: Found stuff!
OctochrisI wouldn’t trust anyone who says that those two settings are better, they’re the sort of people that think setting something to 0 always results in better performance. There’s a reason for having a cache, and there’s a reason for allowing frame latency. Both can result in degraded performance if set to 0.
That seems to be 5 years ago on an etf2l thread. Haven't found much else just yet
[quote]What is frame latency?
When a frame is rendered, it takes a certain amount of time to do so. That time is measured as the latency, from 0, to render and display the frame.[/quote]
Doesn't seem to be a network command though I suppose this would control the max amount time before a frame gets shot out. (so more of an engine command in the form of some buffer limit? Somewhat like snd_mixahead? )
http://www.neogaf.com/forum/showthread.php?t=512976
Reading up on this now myself.
Last time I saw this command was in an old version of chris's configs along with mat_framebuffercopyoverlaysize.
He preferred them on defaults I think
Edit: Found stuff!
[quote=Octochris]
I wouldn’t trust anyone who says that those two settings are better, they’re the sort of people that think setting something to 0 always results in better performance. There’s a reason for having a cache, and there’s a reason for allowing frame latency. Both can result in degraded performance if set to 0.[/quote]
That seems to be 5 years ago on an [url=http://etf2l.org/forum/customise/topic-10435/page-55/]etf2l thread[/url]. Haven't found much else just yet
can you change this command on-the-fly? (as in, without leaving a game.)
it's such a drag if you have to leave a server and re-enter just to see the effect of changing the value of a command :E
can you change this command on-the-fly? (as in, without leaving a game.)
it's such a drag if you have to leave a server and re-enter just to see the effect of changing the value of a command :E
Bucakecan you change this command on-the-fly? (as in, without leaving a game.)
it's such a drag if you have to leave a server and re-enter just to see the effect of changing the value of a command :E
Looks like you can. Brief test in offline mode showed significant decrease in frames when I changed it in large amounts but I think it helped my frame skipping/micro stuttering on hitting a target a little?
Not sure.
Paging comanglia, wareya, panzer or setsul please.
[quote=Bucake]can you change this command on-the-fly? (as in, without leaving a game.)
it's such a drag if you have to leave a server and re-enter just to see the effect of changing the value of a command :E[/quote]
Looks like you can. Brief test in offline mode showed significant decrease in frames when I changed it in large amounts but I think it helped my frame skipping/micro stuttering on hitting a target a little?
Not sure.
Paging comanglia, wareya, panzer or setsul please.
SakiIsaBucakecan you change this command on-the-fly? (as in, without leaving a game.)
it's such a drag if you have to leave a server and re-enter just to see the effect of changing the value of a command :E
Looks like you can. Brief test in offline mode showed significant decrease in frames when I changed it in large amounts but I think it helped my frame skipping/micro stuttering on hitting a target a little?
Not sure.
Paging comanglia, wareya, panzer or setsul please.
thanks for the reply :-)
i'll try it out tonight
[quote=SakiIsa][quote=Bucake]can you change this command on-the-fly? (as in, without leaving a game.)
it's such a drag if you have to leave a server and re-enter just to see the effect of changing the value of a command :E[/quote]
Looks like you can. Brief test in offline mode showed significant decrease in frames when I changed it in large amounts but I think it helped my frame skipping/micro stuttering on hitting a target a little?
Not sure.
Paging comanglia, wareya, panzer or setsul please.[/quote]
thanks for the reply :-)
i'll try it out tonight
SakiIsa Doesn't seem to be a network command though I suppose this would control the max amount time before a frame gets shot out.
OctochrisI wouldn’t trust anyone who says that those two settings are better, they’re the sort of people that think setting something to 0 always results in better performance. There’s a reason for having a cache, and there’s a reason for allowing frame latency. Both can result in degraded performance if set to 0.
That seems to be 5 years ago on an etf2l thread. Haven't found much else just yet
The number 0 for this command does cause a lot of stuttering, but a low but sustainable number will do lots of good. Also, your description is accurate to what it does.
[quote=SakiIsa] Doesn't seem to be a network command though I suppose this would control the max amount time before a frame gets shot out.
[quote=Octochris]
I wouldn’t trust anyone who says that those two settings are better, they’re the sort of people that think setting something to 0 always results in better performance. There’s a reason for having a cache, and there’s a reason for allowing frame latency. Both can result in degraded performance if set to 0.[/quote]
That seems to be 5 years ago on an [url=http://etf2l.org/forum/customise/topic-10435/page-55/]etf2l thread[/url]. Haven't found much else just yet[/quote]
The number 0 for this command does cause a lot of stuttering, but a low but sustainable number will do lots of good. Also, your description is accurate to what it does.
i'm not perceiving anything differently on both local and public servers, and i'm fluctuating between different extremes
i even set it to 0 without any repercussion
what exactly should i be noticing?
i'm not perceiving anything differently on both local and public servers, and i'm fluctuating between different extremes
i even set it to 0 without any repercussion
what exactly should i be noticing?
mat_maxframelatency and a few other commands are included M0re's fpsconfig but not any of the chris ones.
I was speaking to some american cfg collector a while back who said he had passed the command onto cowmanglia or rhapsody, according to him/them after testing there was negligible fps difference. Leaving fps aside, M0re's cfg has always felt smoother to me than any of the versions of the chriscfg, I'd be interested to see if others have noticed the same and whether they can pin down the commands that make this difference.
mat_maxframelatency and a few other commands are included M0re's fpsconfig but not any of the chris ones.
I was speaking to some american cfg collector a while back who said he had passed the command onto cowmanglia or rhapsody, according to him/them after testing there was negligible fps difference. Leaving fps aside, M0re's cfg has always felt smoother to me than any of the versions of the chriscfg, I'd be interested to see if others have noticed the same and whether they can pin down the commands that make this difference.
Buttnosemat_maxframelatency and a few other commands are included M0re's fpsconfig but not any of the chris ones.
I was speaking to some american cfg collector a while back who said he had passed the command onto cowmanglia or rhapsody, according to him/them after testing there was negligible fps difference. Leaving fps aside, M0re's cfg has always felt smoother to me than any of the versions of the chriscfg, I'd be interested to see if others have noticed the same and whether they can pin down the commands that make this difference.
With this find and this: http://www.teamfortress.tv/post/441037/dx8-graphics-on-dx9 I feel like somebody like Comanglia should reread M0re's configs 'cause they might be hiding something more.
[quote=Buttnose]mat_maxframelatency and a few other commands are included M0re's fpsconfig but not any of the chris ones.
I was speaking to some american cfg collector a while back who said he had passed the command onto cowmanglia or rhapsody, according to him/them after testing there was negligible fps difference. Leaving fps aside, M0re's cfg has always felt smoother to me than any of the versions of the chriscfg, I'd be interested to see if others have noticed the same and whether they can pin down the commands that make this difference.[/quote]
With this find and this: http://www.teamfortress.tv/post/441037/dx8-graphics-on-dx9 I feel like somebody like Comanglia should reread M0re's configs 'cause they might be hiding something more.
i dont see/feel a difference at all
i dont see/feel a difference at all
from what ive seen on steam forums it stands for pre rendered frames. not sure though
from what ive seen on steam forums it stands for pre rendered frames. not sure though
tl;dr: mat_maxframelatency does nothing, and the engine is already tuned optimally for lowest input latency.
Judging by the name, this controls the maximum number of frames that the cpu is allowed to be ahead of the gpu by. This value is in frames, not seconds, so it's an integer, not a float. There's no difference between 0.2 and 0.9, because they both truncate to 0.
As of 2007, this cvar wasn't hooked up to anything, and there's no reason to believe it is does anything now. I believe it was replaced by mat_frame_sync_enable (cheat, default 1) when multi-threaded rendering was added to Source (mat_queue_mode 2). In short, mat_frame_sync_enable forces the renderer to wait for the previous frame to finish before it starts on the next one.
Either way, frame syncing is designed to reduce input latency at the cost of a few fps. If you had sv_cheats 1 and decided to set mat_frame_sync_enable 0, you *might* have improved fps, but you would instead be getting (potentially large amounts of) input lag.
Edit:
marqueemat_framebuffercopyoverlaysize and mat_maxframelatency set to 0:
2639 frames 19.187 seconds 137.54 fps ( 7.27 ms/f) 5.651 fps variability
mat_framebuffercopyoverlaysize 256 and mat_maxframelatency 1:
2639 frames 19.226 seconds 137.26 fps ( 7.29 ms/f) 5.705 fps variability
no changes whatsoever (this is comanglia's config btw)
There is absolutely no reason to include mat_framebuffercopyoverlaysize in your config or anywhere else for that matter. All that cvar does is control how large the debug overlay for the framebuffer copy texture is. It has no performance impact unless you have sv_cheats 1 and mat_showframebuffertexture 1, and even then the fps difference would be measured in microseconds. Here's mat_framebuffercopyoverlaysize 256 (look in the top left):
http://steamcommunity.com/sharedfiles/filedetails/?id=460659888
mat_framebuffercopyoverlaysize 1024:
http://steamcommunity.com/sharedfiles/filedetails/?id=460659916
[b]tl;dr: mat_maxframelatency does nothing, and the engine is already tuned optimally for lowest input latency.[/b]
Judging by the name, this controls the maximum number of frames that the cpu is allowed to be ahead of the gpu by. This value is in frames, not seconds, so it's an integer, not a float. There's no difference between 0.2 and 0.9, because they both truncate to 0.
As of 2007, this cvar wasn't hooked up to anything, and there's no reason to believe it is does anything now. I believe it was replaced by mat_frame_sync_enable (cheat, default 1) when multi-threaded rendering was added to Source (mat_queue_mode 2). [b]In short, mat_frame_sync_enable forces the renderer to wait for the previous frame to finish before it starts on the next one.[/b]
Either way, frame syncing is designed to reduce input latency at the cost of a few fps. If you had sv_cheats 1 and decided to set mat_frame_sync_enable 0, you *might* have improved fps, but you would instead be getting (potentially large amounts of) input lag.
Edit:
[quote=marquee]mat_framebuffercopyoverlaysize and mat_maxframelatency set to 0:
2639 frames 19.187 seconds 137.54 fps ( 7.27 ms/f) 5.651 fps variability
mat_framebuffercopyoverlaysize 256 and mat_maxframelatency 1:
2639 frames 19.226 seconds 137.26 fps ( 7.29 ms/f) 5.705 fps variability
no changes whatsoever (this is comanglia's config btw)[/quote]
There is absolutely no reason to include mat_framebuffercopyoverlaysize in your config or anywhere else for that matter. All that cvar does is control how large the debug overlay for the framebuffer copy texture is. It has no performance impact unless you have sv_cheats 1 and mat_showframebuffertexture 1, and even then the fps difference would be measured in microseconds. Here's mat_framebuffercopyoverlaysize 256 (look in the top left):
http://steamcommunity.com/sharedfiles/filedetails/?id=460659888
mat_framebuffercopyoverlaysize 1024:
http://steamcommunity.com/sharedfiles/filedetails/?id=460659916
pazertl;dr: mat_maxframelatency does nothing, and the engine is already tuned optimally for lowest input latency.
Judging by the name, this controls the maximum number of frames that the cpu is allowed to be ahead of the gpu by. This value is in frames, not seconds, so it's an integer, not a float. There's no difference between 0.2 and 0.9, because they both truncate to 0.
As of 2007, this cvar wasn't hooked up to anything, and there's no reason to believe it is does anything now. I believe it was replaced by mat_frame_sync_enable (cheat, default 1) when multi-threaded rendering was added to Source (mat_queue_mode 2). In short, mat_frame_sync_enable forces the renderer to wait for the previous frame to finish before it starts on the next one.
Either way, frame syncing is designed to reduce input latency at the cost of a few fps. If you had sv_cheats 1 and decided to set mat_frame_sync_enable 0, you *might* have improved fps, but you would instead be getting (potentially large amounts of) input lag.
??????????????????????????????????????????????????????????????????????????????????????
I noticed a huge difference. It does something. Also, I call complete bs on this because even 0.2 and 0.5 has a noticeable difference for me. Maybe Comanglia's config makes this command useless or something... Or it might be dependent on hardware. But I know for a fact that it does something.
ButtnoseM0re's cfg has always felt smoother to me than any of the versions of the chriscfg
This is what I think people have wrong with usual Configs. Smoothness > FPS, because about anyone with a computer built in the last 5 Years can play this game with good FPS.
[quote=pazer][b]tl;dr: mat_maxframelatency does nothing, and the engine is already tuned optimally for lowest input latency.[/b]
Judging by the name, this controls the maximum number of frames that the cpu is allowed to be ahead of the gpu by. This value is in frames, not seconds, so it's an integer, not a float. There's no difference between 0.2 and 0.9, because they both truncate to 0.
As of 2007, this cvar wasn't hooked up to anything, and there's no reason to believe it is does anything now. I believe it was replaced by mat_frame_sync_enable (cheat, default 1) when multi-threaded rendering was added to Source (mat_queue_mode 2). [b]In short, mat_frame_sync_enable forces the renderer to wait for the previous frame to finish before it starts on the next one.[/b]
Either way, frame syncing is designed to reduce input latency at the cost of a few fps. If you had sv_cheats 1 and decided to set mat_frame_sync_enable 0, you *might* have improved fps, but you would instead be getting (potentially large amounts of) input lag.[/quote]
??????????????????????????????????????????????????????????????????????????????????????
I noticed a huge difference. It does something. Also, I call complete bs on this because even 0.2 and 0.5 has a noticeable difference for me. Maybe Comanglia's config makes this command useless or something... Or it might be dependent on hardware. But I know for a fact that it does something.
[quote=Buttnose]M0re's cfg has always felt smoother to me than any of the versions of the chriscfg[/quote]
This is what I think people have wrong with usual Configs. Smoothness > FPS, because about anyone with a computer built in the last 5 Years can play this game with good FPS.
Mightypazertl;dr: mat_maxframelatency does nothing, and the engine is already tuned optimally for lowest input latency.
Judging by the name, this controls the maximum number of frames that the cpu is allowed to be ahead of the gpu by. This value is in frames, not seconds, so it's an integer, not a float. There's no difference between 0.2 and 0.9, because they both truncate to 0.
As of 2007, this cvar wasn't hooked up to anything, and there's no reason to believe it is does anything now. I believe it was replaced by mat_frame_sync_enable (cheat, default 1) when multi-threaded rendering was added to Source (mat_queue_mode 2). In short, mat_frame_sync_enable forces the renderer to wait for the previous frame to finish before it starts on the next one.
Either way, frame syncing is designed to reduce input latency at the cost of a few fps. If you had sv_cheats 1 and decided to set mat_frame_sync_enable 0, you *might* have improved fps, but you would instead be getting (potentially large amounts of) input lag.
??????????????????????????????????????????????????????????????????????????????????????
I noticed a huge difference. It does something. Also, I call complete bs on this because even 0.2 and 0.5 has a noticeable difference for me. Maybe Comanglia's config makes this command useless or something... Or it might be dependent on hardware. But I know for a fact that it does something.
ButtnoseM0re's cfg has always felt smoother to me than any of the versions of the chriscfg
This is what I think people have wrong with usual Configs. Smoothness > FPS, because about anyone with a computer built in the last 5 Years can play this game with good FPS.
didnt you use default net settings in that mge video you posted... i think im trusting pazer more on this one.
[quote=Mighty][quote=pazer][b]tl;dr: mat_maxframelatency does nothing, and the engine is already tuned optimally for lowest input latency.[/b]
Judging by the name, this controls the maximum number of frames that the cpu is allowed to be ahead of the gpu by. This value is in frames, not seconds, so it's an integer, not a float. There's no difference between 0.2 and 0.9, because they both truncate to 0.
As of 2007, this cvar wasn't hooked up to anything, and there's no reason to believe it is does anything now. I believe it was replaced by mat_frame_sync_enable (cheat, default 1) when multi-threaded rendering was added to Source (mat_queue_mode 2). [b]In short, mat_frame_sync_enable forces the renderer to wait for the previous frame to finish before it starts on the next one.[/b]
Either way, frame syncing is designed to reduce input latency at the cost of a few fps. If you had sv_cheats 1 and decided to set mat_frame_sync_enable 0, you *might* have improved fps, but you would instead be getting (potentially large amounts of) input lag.[/quote]
??????????????????????????????????????????????????????????????????????????????????????
I noticed a huge difference. It does something. Also, I call complete bs on this because even 0.2 and 0.5 has a noticeable difference for me. Maybe Comanglia's config makes this command useless or something... Or it might be dependent on hardware. But I know for a fact that it does something.
[quote=Buttnose]M0re's cfg has always felt smoother to me than any of the versions of the chriscfg[/quote]
This is what I think people have wrong with usual Configs. Smoothness > FPS, because about anyone with a computer built in the last 5 Years can play this game with good FPS.[/quote]
didnt you use default net settings in that mge video you posted... i think im trusting pazer more on this one.
I always get really excited in these kind of threads and just end up feeling disappointed :( I was hoping for the next command to bring consistent 420 fps
I always get really excited in these kind of threads and just end up feeling disappointed :( I was hoping for the next command to bring consistent 420 fps
911 placebo cops what's your emergency
911 placebo cops what's your emergency
oktober_Mightypazertl;dr: mat_maxframelatency does nothing, and the engine is already tuned optimally for lowest input latency.
Judging by the name, this controls the maximum number of frames that the cpu is allowed to be ahead of the gpu by. This value is in frames, not seconds, so it's an integer, not a float. There's no difference between 0.2 and 0.9, because they both truncate to 0.
As of 2007, this cvar wasn't hooked up to anything, and there's no reason to believe it is does anything now. I believe it was replaced by mat_frame_sync_enable (cheat, default 1) when multi-threaded rendering was added to Source (mat_queue_mode 2). In short, mat_frame_sync_enable forces the renderer to wait for the previous frame to finish before it starts on the next one.
Either way, frame syncing is designed to reduce input latency at the cost of a few fps. If you had sv_cheats 1 and decided to set mat_frame_sync_enable 0, you *might* have improved fps, but you would instead be getting (potentially large amounts of) input lag.
??????????????????????????????????????????????????????????????????????????????????????
I noticed a huge difference. It does something. Also, I call complete bs on this because even 0.2 and 0.5 has a noticeable difference for me. Maybe Comanglia's config makes this command useless or something... Or it might be dependent on hardware. But I know for a fact that it does something.
ButtnoseM0re's cfg has always felt smoother to me than any of the versions of the chriscfg
This is what I think people have wrong with usual Configs. Smoothness > FPS, because about anyone with a computer built in the last 5 Years can play this game with good FPS.
didnt you use default net settings in that mge video you posted... i think im trusting pazer more on this one.
I feel a really clear difference. The only possible explanation by now is that it helps because my hardware generally 'delays frames' more???
I guess this one is just me-specific
[quote=oktober_][quote=Mighty][quote=pazer][b]tl;dr: mat_maxframelatency does nothing, and the engine is already tuned optimally for lowest input latency.[/b]
Judging by the name, this controls the maximum number of frames that the cpu is allowed to be ahead of the gpu by. This value is in frames, not seconds, so it's an integer, not a float. There's no difference between 0.2 and 0.9, because they both truncate to 0.
As of 2007, this cvar wasn't hooked up to anything, and there's no reason to believe it is does anything now. I believe it was replaced by mat_frame_sync_enable (cheat, default 1) when multi-threaded rendering was added to Source (mat_queue_mode 2). [b]In short, mat_frame_sync_enable forces the renderer to wait for the previous frame to finish before it starts on the next one.[/b]
Either way, frame syncing is designed to reduce input latency at the cost of a few fps. If you had sv_cheats 1 and decided to set mat_frame_sync_enable 0, you *might* have improved fps, but you would instead be getting (potentially large amounts of) input lag.[/quote]
??????????????????????????????????????????????????????????????????????????????????????
I noticed a huge difference. It does something. Also, I call complete bs on this because even 0.2 and 0.5 has a noticeable difference for me. Maybe Comanglia's config makes this command useless or something... Or it might be dependent on hardware. But I know for a fact that it does something.
[quote=Buttnose]M0re's cfg has always felt smoother to me than any of the versions of the chriscfg[/quote]
This is what I think people have wrong with usual Configs. Smoothness > FPS, because about anyone with a computer built in the last 5 Years can play this game with good FPS.[/quote]
didnt you use default net settings in that mge video you posted... i think im trusting pazer more on this one.[/quote]
I feel a really clear difference. The only possible explanation by now is that it helps because my hardware generally 'delays frames' more???
I guess this one is just me-specific