Slow Redraw Of Crosshair Cursor in NT8

[NinjaTrader_Brett]

This will be resolved in the next beta. Thanks for mentioning!

[rfjoeyb]

Cursor has serious lag issue

A lot has already been said about the lag in the NT8 cursor function, but let me also add my two cents to this discussion. I just loaded the latest version of NT8 today, and compared its performance to the NT7. One of the most obvious / annoying issues to me was cursor 'out of sync' cursor performance. I personally would not update from NT7 for this reason alone. I use 3 full page charts (basic tick charts on different timeframes) with not much else running indicator wise on those charts. The NT8 cursor is so slow that it totally confuses me since my hand movements are not in sync with the cursor. Further, if I want to move a line, or an arrow, it is extremely hard to get the drawing item highlighted in order to move it. Seems like in NT7 I just needed to get pretty close, and the object is selected. In NT8, I have to use an extremely sluggish cursor and get right on the object before the item can be selected. Wastes a lot of time. If this is a hardware problem on my end, then fine, guess it's time for me to update. But, I don't want to buy a new graphics card, if that is not going to fix the issue. The computer setup I'm using is about 5 years old, but was top of the line at the time of purchase. This computer has been running NT7 with no issues. So, in conclusion, I think you sort of have a 'show stopper' here for me. FYI. Otherwise, looks like a lot of nice updates in NT8. Glad to hear the Ninja team is going to fix this, in the next release I hope. I also have issues with the NT8 text feature, but I'll discuss that elsewhere.

[sledge]

Glad to hear the Ninja team is going to fix this, in the next release I hope. I also have issues with the NT8 text feature, but I'll discuss that elsewhere.

I believe this fix mentioned is related to this:

Would be nice if the crosshair would be directly drawn on the new position when switched to it by keyboard shortcut.

Nothing to do with crosshair performance.

[NinjaTrader_Brett]

For those of you posting and willing to help to get a baseline of performance. Could you please respond with the following survey information.

1) What video card you have?
2) What CPU do you have?
3) How many monitors are you running on that card?
4) What resolution is your biggest monitor running at?
5) What delay (slight/severe/none) do you experience on a fresh all defaults 1 minute chart (no customized settings or templates loaded) that is maximized on the monitor in question 3 using the 'local crosshair'?

[NinjaTrader_Brett]

I'll Start...

Computer 1 (laptop):

1) Nvidia Quadro K1100M
2) Intel Core i7-4700MQ @ 2.4 (8 logical cores)
3) 2
4) 1920x1080
5) None

Computer 2 (desktop):

1) Nvidia GTX 1060
2) Intel Core i5 @ 3 ghz (4 logical cores)
3) 2
4) 1920x1080
5) None

[Segwin]

Computer Specs

1) Nvidia GTX 550 Ti

2) Pentium E5300 2.6GH 2 core

3) 2

4) 1680x1050

5) None

[Taddypole]

1. ATI FirePro 2450 - GDDR3 SDRAM - 512 MB - NT8 has problem with this video card
2. I7 - 4790K - 4 Ghz - NT7 works good with fast processor
3. 4 monitors
4. 1920 x 1200
5. Severe

Regards,
taddypole...

[GaryAlbers]

Cross Hair cursor unusable

1) 4 * ATI FirePro 2450. 8 monitors total
2) i7 3.0Ghz. 1st Gen
3) 2 monitors per card
4) 2048 * 1152
5) SEVERE delay. Cross Hair cursor unusable.

Gary

[RonnyKeller]

Not sure, but maybe I found something that might explain the different behavior.

I use a good PC under Windows 7, Windows index 7.7, 16GB Ram and a good NVIDIA-card GeForce GTX-680. My setup has a 30" at 2560x1600 plus three 24"-monitors at 1920x1200 - performance was never an issue with NT7 (or anything else I use).

Now I installed NT8 RC1. I opened two different charts, each in a different workspaces, both charts in fullscreen-mode on the 30"-monitor, both showing approximately 220 candles, only very few very simple indicators (e.g. one standard EMA)

And here we go now: one chart in fullscreen-mode shows quite annoying crosshair-lagging, but the other works just fine. (both Local crosshair only, no global crosshair). I'm still looking what causes the issue.

Chart with lagging crosshair:
Black chart background, red and green candles, crosshair set to DimGray. Style Dot, Width 1

Chart with not-lagging crosshair:
White chart background, black and white candles, crosshair set to DimGray, Solid, Width 1

I couldn't figure out yet what exactly causes the problem. But IF there is different behavior on the SAME PC it's mostly likely neither the graphic card nor the CPU nor the RAM.

Plus: IF this lag is related to some chart settings, it might even increase the problem if the "wrong" setting is used on several or all charts parallel, and probably getting really, really ugly if lag-causing settings are connected through "Global Crosshair"-usage.

I will check and compare further during the weekend, but you may want to point your tests in that direction rather than checking hardware-settings.

Just to make clear: there are more differences between those charts than only a dotted crosshair and different colors, e.g. one was created under Ninja Theme A and the other under Ninja Theme B and so on, that's why I need to test further.

Or is it possible to send those charts, workspaces, settings etc. to a Ninja Supporter to check the details?

[NinjaTrader_Brett]

The cause in your tests below is likely the Dots applied to Crosshair. Dots are a lot harder to render then just lines so you have more chance to run into a performance issue there.

-Brett

[TraderWan]

Lagging Crosshair Issue

So, let me see if I get this straight. NT8 is using dedicated GPU and multiple cores, and now doted crosshairs is "hard to render" and causing a "performance issue?"

Mine is so ridiculous I going back to NT7 for awhile as I look at other charting options. Not the only game in town and I'm not hanging around for ever. I didn't know I was buying the McDonalds of charting software. Can we kick it up a notch, fellas? If you can't or won't fix the crosshair issue, please just let us know so we can move forward somehow.

Even for Beta, this is a huge step backward in software performance and is unacceptable.

[NinjaTrader_Brett]

TraderWan,

Here is where we are at:

• NinjaTrader 7 did not utilize the video card as much for rendering. Mostly CPU driven.
• NinjaTrader 8 heavily leans on the video card for rendering. This in turn means you needed a decent video card else you could fall behind in performance. (CPU might be awesome but video card not so much)
• We are in the information gathering stage right now where we are trying to get information on the systems which are having crosshair problems. Some are having issues and others are not.
• There perhaps programmatically might be more we could do, we're in the middle of reviewing if any such option would be viable and would return with more information.

[sledge]

• NinjaTrader 8 heavily leans on the video card for rendering. This in turn means you needed a decent video card else you could fall behind in performance. (CPU might be awesome but video card not so much)
• We are in the information gathering stage right now where we are trying to get information on the systems which are having crosshair problems. Some are having issues and others are not.

I sent some videos via dropbox to you.

Full Screen does get laggy. I loaded up 1 day of replay data, 1 minute.

[jcrza]

1) What video card you have?

NVidia 780ti

2) What CPU do you have?

Intel i5 4670k

3) How many monitors are you running on that card?

3

4) What resolution is your biggest monitor running at?

1920x1080

5) What delay (slight/severe/none) do you experience on a fresh all defaults 1 minute chart (no customized settings or templates loaded) that is maximized on the monitor in question 3 using the 'local crosshair'?

None. There is only delay when using global crosshairs.

This is a really serious issue. It makes the entire program feel sluggish. I spent 95% of my time interacting with charts. I've had global crosshairs on 20+ open charts on multiple screens on thinkorswim without a single dropped frame.

Here's how you reproduce the issue in NT8. Get three monitors, do a three wide setup, plug them all into the GPU, put one ES chart on each and instrument link them, then enable global crosshairs on the leftmost and rightmost monitors ONLY while leaving local crosshairs enabled on the center one. Wiggle your mouse around with the middle chart selected. It's still responsive, right? Feels good? Now enable global crosshairs on your middle screen as well so that now all three are synced.

Notice the difference in how it feels the second you toggle it? Notice how the framerate on the chart is reduced so that when you stretch the chart the performance is degraded. If you don't notice it perhaps you need more charts so you're consuming the number of threads/cores on your system or something. Or maybe try quickly toggling with the crosshairs hotkeys for better A/B testing on how it 'feels' if you're not using an FPS counter to watch the performance degrade.

How much horsepower can you really ask of the user to draw some charts and synchronize some lines in a 2D environment? Please look into it more seriously. This is something that should have been caught in alpha, not RC1. I won't be renewing my lease if this isn't fixed. It seems incredible to be saying things like "Dots are a lot harder to render then just lines" when it's obvious there's a serious issue here. We're living in the age of rendering games like crysis on laptops and you're telling us rendering lines is taxing for desktop computers? C'mon guys. Embarrassing.

[sledge]

How much horsepower can you really ask of the user to draw some charts and synchronize some lines in a 2D environment? Please look into it more seriously. This is something that should have been caught in alpha, not RC1. I won't be renewing my lease if this isn't fixed. It seems incredible to be saying things like "Dots are a lot harder to render then just lines" when it's obvious there's a serious issue here. We're living in the age of rendering games like crysis on laptops and you're telling us rendering lines is taxing for desktop computers?.

You haven't dealt with .NET. lol

And we've been complaining since early on...

And since NT8B10 - my full market replay of 2.75 years of data is worthless @ MAX speed or any other speed... It's faster in NT7@500x split on 2 computers. Probably will need 3 computers with 3 years of data.