|
Oh My GOD, this is amazing. So I can Stream on 1920x1080, 60FPS, 10 Quality and Starcraft on Low (looks better :D) (Starcraft is constatly on 60FPS, It can drop to around 45FPS in Superlategamebattles (tested).
With an FX-4100 @ 4GHz (HD6970, DDR1600)
When with Xsplit I was barely able to Stream @ 720p (25FPS, 10 Quality) and FFsplit actually didn´t work at all (bug? campatibility?)
This surely is amazing!
|
Friend of mine is having an issue where he specifies a bandwidth limit of 1900 but it will sometimes pull down way more (like 3200).
Anything we can do?
|
Medrea - That's almost always just a miscalculation of the bandwidth on the part of the bandwidth meter. Unless you're settings your buffer size super high or something.
|
FFSplit recently implemented stream statistics to the program. I wonder if it's possible with obs too? + Show Spoiler +Current Framerate Current Bitrate Dropped Frames Session Length Bandwidth Used Lifetime Statistics
|
On October 10 2012 21:42 obs.Jim wrote: Medrea - That's almost always just a miscalculation of the bandwidth on the part of the bandwidth meter. Unless you're settings your buffer size super high or something.
Thats odd considering it will lag him out of LoL when it happens.
|
On October 10 2012 05:24 Avean wrote: I7-960 @ 4ghz. Game was maxed out.
twoscomp: Why are you using DXtory with OBS ? OBS window capture is alot faster than DXtory.
The game doesn't have a fullscreen windowed mode. *shrug*
|
On October 10 2012 05:32 Cyro wrote:Show nested quote +On October 10 2012 05:24 Avean wrote: I7-960 @ 4ghz. Game was maxed out.
twoscomp: Why are you using DXtory with OBS ? OBS window capture is alot faster than DXtory. Preset?
Default: veryfast
|
|
United Kingdom20157 Posts
I have also disabled hyperthreading which you MUST do with your CPU. Newer cpus dont use hyperthreading but older ones like i7-9xxx get a solid boost from it when DISABLED. In DayZ (Arma2) i went from 30fps to 115 fps when disabling hyperthreading.
You gave a source in your other post (http://forums.bistudio.com/showthread.php?80223-Arma-2-amp-Core-i7-Hyperthreading-Tested) which shows:
+ Show Spoiler + All benchmarks are conducted using ArmaII-Mark V1.0.
Now let's get started!
My rig: i7 920 @ 4.00ghz 6GB DDR3 PC-12800 ATI HD 4870 512mb Asus P6T Deluxe Win7 64-bit
Test Settings: Video Memory - High Post Processing - Low Anti-Aliasing - Off All other settings - Normal Resolution - 1680x 1050
RESULTS:
Test 1: Hyperthreading Enabled in BIOS -cpucount=8 1nd Run - 4440 2nd Run - 5129
Test 2: Hyperthreading Disabled in BIOS (max 4 cores) 1st Run - 4626 2nd Run - 5353
First run shows a 4.189% performance increase with HT disabled, second run shows a 4.367% performance increase with HT disabled.
He ends the post with "So what does this mean to i7 users? The game definitely has issues with hyperthreading, and there is definitely room for improvement for i7 processor performance with this game. Bohemia Interactive just has to get around to optimising the game for HT." as well, taking blame onto that specific game and developers, as results that big are uncommon
By your account i went from 30fps to 115 fps when disabling hyperthreading. you got a 283.3% performance gain, over 66x the gap shown in the benchmark with hyperthreading on and off, so you have either major individual problems with the game or your system that are causing horrible performance that is somehow fixed by disabling Hyperthreading. Maybe you messed up something when overclocking to 4ghz?
And also, pulling up a stream isnt going to tell you if you have a solid 60fps or not. It could drop to 40-45 frequently or even float around there or lower and you could not notice. You would need the file and a FPS monitoring tool, or to count frames encoded vs time, etc.
You also have to remember that hyperthreading exists for a reason. People pay $100 to get the 2600k over the 2500k for a reason, and there is pretty much nothing else aside from Hyperthreading added. It will give you a ~20-25% boost in x264 encoding performance and quite a few other tasks that you will sorely need trying to live encode at stressful resolution/fps/preset options.
So I can Stream on 1920x1080, 60FPS, 10 Quality and Starcraft on Low (looks better :D) (Starcraft is constatly on 60FPS, It can drop to around 45FPS in Superlategamebattles (tested).
With an FX-4100 @ 4GHz (HD6970, DDR1600)
Can i see an mp4 file?
|
How do you add a game to the stream? I'm able to monitor capture but it doesn't capture the game I'm playing so I assume some where you would have to add the game? I'm new this streaming thing, please help! Thanks!
|
On October 11 2012 04:25 ktang wrote: How do you add a game to the stream? I'm able to monitor capture but it doesn't capture the game I'm playing so I assume some where you would have to add the game? I'm new this streaming thing, please help! Thanks!
Until native capture mode is supported, you'll have to run in some sort of windowed mode (fullscreen window) or use DXtory to capture.
|
On October 09 2012 08:03 obs.Jim wrote: RiSkyToss - The actual capturing FPS can now be seen in the latest daily builds I have posted on my site. The captured FPS is the FPS that you stream at. The FPS set in video settings is your maximum FPS, not necessarily your sustained FPS, but I do not artificially or intentionally lower the FPS of the stream or something, and I don't see why ffsplit would do that either. The rate you capture at is the rate you get, at least for my app. I'm not sure if I'm blind but I'm using v0.411a and I don't see captured FPS anywhere.
|
Disabling hyperthreading does seem to often improve performance.
Its all over Xsplit forums. And I see it in other places as well. I have no idea why. And it really sucks since streaming is the biggest reason to get an i7 in the first place.
|
Capture FPS and dropped frames is going to be in 0.42a - hopefully I'll be uploading it today. Been pretty busy lately tracking down and getting rid of some rather nasty bugs
|
I don't know if this has been reported but when switching scenes and I have my webcam on one of the scenes the sources don't fade out and in nicely, it kinda lags over to the other scene. But when the webcam is off there are no issues with the fade animations.
|
United Kingdom20157 Posts
On October 11 2012 07:51 Medrea wrote: Disabling hyperthreading does seem to often improve performance.
Its all over Xsplit forums. And I see it in other places as well. I have no idea why. And it really sucks since streaming is the biggest reason to get an i7 in the first place.
Improve performance marginally, in cases, yes. I have yet to see a practical case with a >5% difference, and i personally get much better performance with xsplit when pushing stressful resolution+fps+preset combo's with HT on. I had a bunch of issues with it that seem to be gone after doing a double check on disabling core parking and messing with a few other things
|
On October 11 2012 07:51 Medrea wrote: Disabling hyperthreading does seem to often improve performance.
Its all over Xsplit forums. And I see it in other places as well. I have no idea why. And it really sucks since streaming is the biggest reason to get an i7 in the first place. If all you were doing was encoding, hyperthreading would probably be an improvement. However, streaming is usually a compromise between stream quality and game performance. Using hyperthreading effectively halves the cache sizes available to the encoding threads, which results in an overall increase in memory accesses for the same work. This increases the memory latency for the game threads, which has a substantial impact on game performance.
On top of that, stream encoding operates in bursts because it's a finite task, so by using more threads, you're increasing the jitter on the game threads.
This is true even if you keep the encoding threads away from the cores that are running the game.
|
On October 11 2012 09:26 jaj22 wrote:Show nested quote +On October 11 2012 07:51 Medrea wrote: Disabling hyperthreading does seem to often improve performance.
Its all over Xsplit forums. And I see it in other places as well. I have no idea why. And it really sucks since streaming is the biggest reason to get an i7 in the first place. If all you were doing was encoding, hyperthreading would probably be an improvement. However, streaming is usually a compromise between stream quality and game performance. Using hyperthreading effectively halves the cache sizes available to the encoding threads, which results in an overall increase in memory accesses for the same work. This increases the memory latency for the game threads, which has a substantial impact on game performance. On top of that, stream encoding operates in bursts because it's a finite task, so by using more threads, you're increasing the jitter on the game threads. This is true even if you keep the encoding threads away from the cores that are running the game.
I guess I can buy that.
Why is it true even if I segregate the two processes to run on completely different cores? Or is it?
My example is this. I run a game on two cores, I run the stream on two cores and TWO hyperthreads. (Stream does NOT run on the hyperthreads associated with the physical cores that are running the game because that would be impossible, running a game on a core blocks out the hyperthread from being used as it is already being used by the game).
|
I don't know if I'm doing something wrong or not but I'm getting constant crashes while trying to stream in fullscreen with dxtory. Whenever I tab out of the game there is like a 50/50 chance with the scene switcher R1ch released for the 32bit version that it stops working and reconnects and crashes. (I have not tried with any other version of the program) But I'm going to assume its similar for the 64bit version or if your manually change scenes aswell.
Is this a known issue that your working on or is there a workaround already?
|
United Kingdom20157 Posts
On October 11 2012 09:26 jaj22 wrote:Show nested quote +On October 11 2012 07:51 Medrea wrote: Disabling hyperthreading does seem to often improve performance.
Its all over Xsplit forums. And I see it in other places as well. I have no idea why. And it really sucks since streaming is the biggest reason to get an i7 in the first place. If all you were doing was encoding, hyperthreading would probably be an improvement. However, streaming is usually a compromise between stream quality and game performance. Using hyperthreading effectively halves the cache sizes available to the encoding threads, which results in an overall increase in memory accesses for the same work. This increases the memory latency for the game threads, which has a substantial impact on game performance. On top of that, stream encoding operates in bursts because it's a finite task, so by using more threads, you're increasing the jitter on the game threads. This is true even if you keep the encoding threads away from the cores that are running the game.
Makes sense, but again two different variables, game performance vs encoding performance, your game might not be running as well as you want or at the framerates you want, but being unable to encode the stream at the resolution, framerate and preset you want because your cpu is maxing out and you are dropping frames is a big deal too.
You wanna encode a 1920x1080 stream at 60fps with veryfast on a first gen i7 (quad)? You are going to need the boost from Hyperthreading, and even then its probably completely out of reach for a lot of scenes
|
|
|
|