Control response time on G4.5
- Thread starter pgoelz
- Start date
Not sure how to interpret that.jeffpn said:
My original problem (excessive latency) was helped by running full screen and also by reducing the resolution while in full screen. I'm still a bit unclear whether that points to CPU or video card bottlenecks.
Since my CPU seems to be running at about 65% utilization while flying (full screen or windowed), does that tip the scale towards the video card? I don't want to invest in a video card only to find my latency is unchanged. And while I would like lower latency, I find it acceptable on the desktop.
My laptop is somewhat worse, but it only has a 1.6GHz dual core as opposed to the 3GHz dual core on the desktop. And of course, a different video card isn't an option on the laptop.
Paul
While we're at it..... what is a good PCI-e video card these days that is inexpensive and doesn't have a noisy dust buster (r) for a blower? I want something that isn't audible outside the case. The PCI-e card in the FAQ seems to be affordable but discontinued (at least Newegg says it is inactive) and I'm not sure what the best replacement would be. I know some video cards make quite a racket.
I have been contemplating a dual monitor setup for video editing, so maybe I can talk myself into a discrete video card after all
Paul
I have been contemplating a dual monitor setup for video editing, so maybe I can talk myself into a discrete video card after all
Paul
jeffpn
Well-known member
I have a pair of 8800GTS cards in my computer that does real well. I've seen people post that 1 card does well, too. It's a couple years old now, so the price should be right, if they are still available. I still think that the benefit to the USB would be negligible by upgrading a video card.
In my case at least, there are two separate latency issues.
The USB interface introduces some latency. I am still estimating it is around 100mS just by looking at it. I agree that a video card upgrade will not affect this latency at all. It is possible that many pilots would not notice it because it is similar in magnitude to the total latency between stick input and the beginning of servo response in a 72MHz Futaba system.
The processing and / or displaying introduces more latency. This is worse in windowed mode and best in full screen mode. In my machine, the added latency over and above the USB latency is not severe in full screen mode. It is quite noticeable in windowed mode. Since the CPU is running around 65% utilization, I would expect this added latency to be addressed by a video card upgrade. Whether it is worth it depends on how good a deal I can find on a good video card
What do you do with two video cards? Do you use two monitors while flying? I'm mainly looking to improve my heli skills, but I can see you can go nuts with this stuff
Paul
The USB interface introduces some latency. I am still estimating it is around 100mS just by looking at it. I agree that a video card upgrade will not affect this latency at all. It is possible that many pilots would not notice it because it is similar in magnitude to the total latency between stick input and the beginning of servo response in a 72MHz Futaba system.
The processing and / or displaying introduces more latency. This is worse in windowed mode and best in full screen mode. In my machine, the added latency over and above the USB latency is not severe in full screen mode. It is quite noticeable in windowed mode. Since the CPU is running around 65% utilization, I would expect this added latency to be addressed by a video card upgrade. Whether it is worth it depends on how good a deal I can find on a good video card
What do you do with two video cards? Do you use two monitors while flying? I'm mainly looking to improve my heli skills, but I can see you can go nuts with this stuff
Paul
Last edited:
opjose
Well-known member
Because G3/G4/G5 models the actual response time of the servos, what you are reporting - MAY - be normal, but you should check....
The best way to check this out, is to go into the aircraft editor, and with the editor open look at the INPUT channel values as you move the stick.
DO NOT LOOK at the control surface or servo values as these have the servo delays applied.
The INPUT channel values show the "raw" data which G4/G5 is seeing from the controller. So any real delays will be most apparent here.
---
The best way to check this out, is to go into the aircraft editor, and with the editor open look at the INPUT channel values as you move the stick.
DO NOT LOOK at the control surface or servo values as these have the servo delays applied.
The INPUT channel values show the "raw" data which G4/G5 is seeing from the controller. So any real delays will be most apparent here.
---
Raylon
Active member
pgoelz said:While we're at it..... what is a good PCI-e video card these days that is inexpensive and doesn't have a noisy dust buster (r) for a blower? I want something that isn't audible outside the case. The PCI-e card in the FAQ seems to be affordable but discontinued (at least Newegg says it is inactive) and I'm not sure what the best replacement would be. I know some video cards make quite a racket.
I have been contemplating a dual monitor setup for video editing, so maybe I can talk myself into a discrete video card after all
Paul
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127423
That will work well. You seem to be dead set on something quiet, but to be honest, power and quietness don't really go together when it comes to computers, unless you start switching to massive low RPM fans or watercooling. That card will be more than enough for G5. Better than an 8800 GTX and quite cheap.
Thanks for the recommendation. And not too bad a price either. My mother board has a PCI-e 16X slot. I assume it will work? Not sure about 2.0 vs. 16X. Way too many "standards" out there these days.
Yes, I am dead set on quiet. My other two applications for the PC are audio recording and editing and video editing. Both are much nicer in a quiet room. My CPU, case fan and power supply are pretty quiet and I don't want to muck that up too badly. It doesn't have to be silent.... just no more noisy than my CPU fan, for example.
I'll have a look in the aircraft editor and see what kind of delay is visible. But truthfully, in full screen mode on my 3GHz machine it isn't too bad. It could be better but it is quite usable. I've been enjoying flying instead of fighting the simulator. And I even discovered another club member put a photofield of our club field on the swap pages. What more can you ask!
Paul
Yes, I am dead set on quiet. My other two applications for the PC are audio recording and editing and video editing. Both are much nicer in a quiet room. My CPU, case fan and power supply are pretty quiet and I don't want to muck that up too badly. It doesn't have to be silent.... just no more noisy than my CPU fan, for example.
I'll have a look in the aircraft editor and see what kind of delay is visible. But truthfully, in full screen mode on my 3GHz machine it isn't too bad. It could be better but it is quite usable. I've been enjoying flying instead of fighting the simulator. And I even discovered another club member put a photofield of our club field on the swap pages. What more can you ask!
Paul
jd_ee_2000
New member
I run with a 9600GT card (low power version) and it is very quiet and works great for RF 4.5. You can easily find it for <$90 and occasionally with a $15 rebate. If you cant find the rebate then the GT250 at $100 is probably a better deal.
Back to the original question for a moment......
I have played around with settings and find that smoke and shadows are my worst enemy. Both cut the frame rate about in half. With minimum smoke and no shadows, I get around 60FPS. If I enable shadows, that drops to about 32FPS give or take. Which raises the question.... seems to me that shadows should be harder to compute than to image. So is the bottleneck more likely the CPU vs. the graphics card? Or does the program offload the shadow calculation to the graphics processor? Last I looked, the CPU was nowhere near 100% utilization....
Paul
I have played around with settings and find that smoke and shadows are my worst enemy. Both cut the frame rate about in half. With minimum smoke and no shadows, I get around 60FPS. If I enable shadows, that drops to about 32FPS give or take. Which raises the question.... seems to me that shadows should be harder to compute than to image. So is the bottleneck more likely the CPU vs. the graphics card? Or does the program offload the shadow calculation to the graphics processor? Last I looked, the CPU was nowhere near 100% utilization....
Paul
Raylon
Active member
pgoelz said:
Both of those are definitely GPU heavy processes. Back when I had a crappy PC, which was a P4 1.4Ghz, 256MB Ram, Nividia 4600, I would kill Particles, smoke, and shadows first, and even that machine would run G3.5 just fine. It's all about give and take. Sounds like you can run yours just fine it you don't want to spend the extra $$$ for the extra bells and whistles. If you do decide to update, don't cheap out. Get something that will get the job done and then some, it's always nice to have that extra space to play with.
Hmmmmm...... although I am quite PC literate, I am stuck in the last century when it comes to understanding graphic cards and what tasks are handled by the graphics processor these days. If I understand you correctly, the graphics card (as opposed to the CPU) handles things like computing and rendering shadows? So if my CPU reports less than 100% utilization while I get 30FPS rendering shadows and 60FPS with shadows disabled, that is an open and shut case for the graphics card being the bottleneck? I can believe it since the "card" in this case is the onboard graphics processor on the motherboard (Intel 82945G). I just don't want to go get an actual card and find that it wasn't the problem.
Paul
Paul
I tried an experiment last night. I have a GeForce 6200 that I bought for a previous motherboard that didn't have a video card onboard. My current MB does and can handle the native resolution of my monitor so the 6200 went back ino its box.
I installed it last night as an experiment to see if it improved my response time. It is plain PCI, not PCI-e so I was doubly curious. It speaks DX 9 and has 128MB I think.
Well, the graphics looked better for sure.... smoother and more detailed textures. But the frame rate was significantly lower than my Intel onboard card. So that tells me that my bottleneck is indeed the graphics card.
Paul
I installed it last night as an experiment to see if it improved my response time. It is plain PCI, not PCI-e so I was doubly curious. It speaks DX 9 and has 128MB I think.
Well, the graphics looked better for sure.... smoother and more detailed textures. But the frame rate was significantly lower than my Intel onboard card. So that tells me that my bottleneck is indeed the graphics card.
Paul
opjose
Well-known member
Remember because you tried with a PCI card, the on board integrated chip can get video data faster than the 6200.
The 6200 is really a DX7/DX8 compliant card, and not really up to DX9 compliancy.
Now that said, your on board Integrated display chip uses a shared memory scheme that is inherently slower than a card with dedicated vram.
However in your case that is offset by the PCI bus and the old video chipset used on that card.
In some respects then the integrated chipset is a bit faster, but your dedicated card has more rendering pipelines, etc.
What you're looking at, is a situation where the deficiencies of both of the things you tried, tend to balance out the results.
There would be a dramatic improvement if you could go with a PCI-E x600 or x800 video card.
Also remember that the 6200's video card defaulted to settings that improved the overall video quality.
I'll bet if you dialed things down so that the graphics were about equal, you may have found that the 6200 did far better than your integrated chipset video.... but there's that PCI bus issue....
The 6200 is really a DX7/DX8 compliant card, and not really up to DX9 compliancy.
Now that said, your on board Integrated display chip uses a shared memory scheme that is inherently slower than a card with dedicated vram.
However in your case that is offset by the PCI bus and the old video chipset used on that card.
In some respects then the integrated chipset is a bit faster, but your dedicated card has more rendering pipelines, etc.
What you're looking at, is a situation where the deficiencies of both of the things you tried, tend to balance out the results.
There would be a dramatic improvement if you could go with a PCI-E x600 or x800 video card.
Also remember that the 6200's video card defaulted to settings that improved the overall video quality.
I'll bet if you dialed things down so that the graphics were about equal, you may have found that the 6200 did far better than your integrated chipset video.... but there's that PCI bus issue....
Thanks, that's about the way I saw it too. I'm just trying to avoid running out and wasting $50-100 on a video card that I don't really need.... it runs OK on my onboard card after all..... just a tad slow. But that said, I did see some nice cards that would do the job and also accelerate h.264 videos.... that would be a plus
Paul
Paul
OK, report from the trenches.... I looked at a lot of video cards and settled on an nVidia GeForce 210 with 512MB DDR2 memory for $69. Running G4.5 in fullscreen mode with everything set to default and soft shadows and depth of field disabled I get around 130-150 FPS in a photofield airport. Since my LCD monitor only refreshes at 60Hz, that seems pretty good, and way better than the 35FPS I was getting with my onboard Intel chipset. Even at 35FPS, it wasn't all THAT bad. But this card removes the last bit of control lag that was bugging me.
With soft shadows and depth of field enabled, the refresh rate drops to about 25FPS so this card isn't for everyone. But for me it was a great compromise.
Oh, and the fan is inaudible.
Paul
With soft shadows and depth of field enabled, the refresh rate drops to about 25FPS so this card isn't for everyone. But for me it was a great compromise.
Oh, and the fan is inaudible.
Paul
Last edited: