My wish is that it be released!!

Does anyone have an idea when ??

I'm going to vote again for a more sophisticated DOM with MP involved.

The ability to keep a certain amount of bars on a chart.So far,when i try to select a certain amount of bars it,for whatever reason,doesn`t keep up.Maybe it`s already there and i just unaware of it.If there is,any suggestions are appreciated.

I wish the optimizer has a more efficient optimizer algo on multi processor/core/thread cpu.

Lets say you have written an indicator, MyCustomIndicator, that contains references to the system indicators SMA and EMA.

If you select MyCustomIndicator and try to export that on its own, then the export fails with the message "Error compiling sources. Please check your logs".

If you also add one system indicator used by the custom indicator (say @SMA) and attempt to export the indicator, then a dialogue box lists all of the other referenced system indicators and there is a prompt to include them automatically.

This prompt to automatically include referenced indicators should be what happened in the first example.

Completely open up and document the custom bar API... Its already there, and many hacks have been done, but it undocumented and unsupported...

Remove restriction barring add(timeframe) only in initialize... custom bar type would allow me to do what I want here, because I could do it in bar creation (I think)

Too late for NT8, but here is my early request for NT9. I've been asking for it for about 4 years already

Support CUDA/OpenCL GPU optimization since everyone will be consumed with automation by then, might as well do it ten times faster than CPU's can. Currently only Matlab and R can use OpenCL/CUDA, but by the time NT9 comes out I imagine many more trading platforms will support it.

25-[AUTOLINK]GPU[/AUTOLINK] cluster cracks every standard Windows password in <6 hours | Ars Technica

opencl - [AUTOLINK]GPU[/AUTOLINK] vs CPU performance for common algorithms - Stack Overflow

Stackoverflow

GPUs are highly specialized hardware designed to do a small set of tasks very well and highly parallelized. This is basically arithmetic (particularly single precision floating point math although newer GPUs do quite well with double precision). As such they're only suited to particular algorithms. I'm not sure if sorting fits that category (in the general case at least). More common examples are pricing of financial instruments, large amounts of matrix maths and even defeating encryption (by brute force). That being said, I did find Fast parallel GPU-sorting using a hybrid algorithm. Another commonly quoted example is running SETI@HOME on an Nvidia GPU but it's comparing apples to oranges. The units of work for GPUs are different (and highly limited) compared to what CPUs ordinarily do.

Mike

Big Mike 

Too late for NT8, but here is my early request for NT9. I've been asking for it for about 4 years already Support CUDA/OpenCL GPU optimization since everyone will be consumed with automation by then, might as well do it ten times faster than CPU's can. Currently only Matlab and R can use OpenCL/CUDA, but by the time NT9 comes out I imagine many more trading platforms will support it. 25-[AUTOLINK]GPU[/AUTOLINK] cluster cracks every standard Windows password in <6 hours | Ars Technica opencl - [AUTOLINK]GPU[/AUTOLINK] vs CPU performance for common algorithms - Stack Overflow Mike

CUDA would be great for training Machine Learning systems.

When adding indicators to a chart it would be nice to sort by most recently added / changed. Trying to find a new indicator when it is named something totally different than the zip file downloaded is a pain.

Robert

NJAMC 

CUDA would be great for training Machine Learning systems.

There will plenty of 2xGPU core's on a single slot, and probably 4xGPU cores on a single slot in a couple years. Scale to four PCIe slots with 2x or 4x per, and you have a beast. And I would be one of the ones that builds such a beast for backtesting and simulation environments. And of course to play

I've been following your ML/SVM's, since you are using an external library, is it possible to use Matlab or R to leverage the existing GPU acceleration in Parallel Computing Toolbox or R's GPU library?

I don't want to go off-topic so I will re-ask my question in your main Support Vector Machine thread:

Support Vector Machines (SVM) To Keep on the Machine Learning and Artificial Intelligence theme, I would like to start a new thread to discuss another approach for machine learning known as Support Vector Machines. There is a very good library called "LIBSVM" which can be …

Mike

Big Mike 

Too late for NT8, but here is my early request for NT9. I've been asking for it for about 4 years already Support CUDA/OpenCL GPU optimization since everyone will be consumed with automation by then, might as well do it ten times faster than CPU's can.

Why too late? I actually kinda assumed that if NT could benefit from GPGPU processing then that would be a really high priority considering it would basically blow away any other retail trader platform and maybe some institutional platforms. And the more granular you get with tick data the more GPGPU processing becomes necessary or at the very least more overclocked cores.

I dream of the day when I can backtest multiple instruments over a decade of data and get the results back in a couple minutes or less.