Hooray Signal 10.3!

[Doji3333]

Dloomis, interesting?? But maybe because you did a restore from tape that you replaced everything back to as it was prior to installing 10.3. So, you basically went back to an earlier restore point..

Volume is increasing a bit now and esignal is still running between 10% and 40% of the CPU. I can't imagine adding another 5 or 6 charts will slow it down much more, but I'm going to try.

So far, starting with a new layout has really improved my esignal speed.

[dloomis]

I should also add one other point, I am running my cpu at 100% with a 25% overclock while I run a distributed computing project 24/7.

And even with this I do not have a 'slow' pc on the version mentioned below.

Both cores maxed at 100% 24/7 at 60 degrees C or so.

[Doji3333]

Very interesting about your distributed processing solution?? what apps are you running to accomplish this?? I have 4 PCs here that are pretty fast and did not even know that esignal supported distributed processing solutions??

Please provide more info for interested minds??

[dloomis]

eSignal doesnt support this, distributed processes are stand alone program that use up spare cpu cyles to accomplish group effort computing. Most every one has heard of SETI by this time, the search for extrateresetrial liife.

I run the abc@home on BOINC from Stanford U. I think it is looking for a big prime number. The ABC conjecture is one of the greatest open mathematical questions, one of the holy grails of mathematics. It will teach us something about our very own numbers. abcathome.com

Keeps the cpu busy at 100%, 24/7, giving up its endless need for cpu cycles when some one else comes along.

I never had such a fast machine that could take whatever I threw at, but this one does.

[SteveH]

Doji,

It's been my recent experience that 8.x and 10.x layouts are non-interchangeable. You would have thought that at least the 8.x layouts would be 100% upward compatible.

I can live with that. I only have to redo a layout once to get it right. However, what I can't tolerate for long is flipping back and forth between 2 different time templates each morning with only 1 min difference on the ending time just so my charts will update in real-time for the entire day rather than being completely dead.

I cannot begin to tell you what a rookie bug that is to have in a professional product like this at this level of maturity and 4 minor revisions in to a major release cycle.

[Hey, I bet if these developers were forced to put their money on the line in the markets on a fast moving e-mini on a intraday basis, they'd be a lot more careful in their coding and testing practices.]

[fastflyer]

Volume Spike Stalls

Patience Steve, Patience,

I know what you're saying, I really do, but your tone is too .... These folks already know as well, so try to say something good about them too. We all know we have an excellent platform, and really need to get that response back up to par. These .efs scripts have really helped me " see " the market I look at. To the code writers - we're eagerly waiting your next improvements, VERY EAGERLY

Best Regards and Good Hunting,

Glenn

[JayF]

Originally posted by SteveH
However, what I can't tolerate for long is flipping back and forth between 2 different time templates each morning with only 1 min difference on the ending time just so my charts will update in real-time for the entire day rather than being completely dead.

Steve,

I'm confused by this. Can you clarify what exactly you are seeing?

There should absolutely be no need to flip between time templates just to get real-time data... In fact, that would only force the whole chart to reload, causing more CPU cycles, memory and bandwidth to be consumed. What we should be doing is troubleshooting why your charts are dead in the first place. When did this start and how is it manifesting? A handful of screenshots or a quick movie (with Jing) will be a great start in trying to understand what you are reporting.

[Doji3333]

Steve, I agree... patience and having a backup plan are the key..

Let me state a few things here so everyone will know exactly what esignal provides us...

esignal is an excellent development tool for "Avg. Joe/Jane" to create a complete trading system (if you know what you're doing in EFS). Sure, there are limitations, but I've been developing these types of things for over 7+ years and I can tell you that every time I run into an issue with esignal (efs), I find a workable solution.

Now, with regards to the speed issues... I trust the next release will be able to better address multiple CPUs, x64 and probably some design changes to help us "power users" out.

Trading with real money in the markets every day, I can tell you there are even times when IB is very slow to react. I believe this is a technology issue and the fact that we want everything ASAP.

I built a project for a client on the east coast that had very specific needs for pre-processing large numbers of symbols (like 500+). We started building it with esignal's datafeed (using the API). Everything was working fine, but then the client requested using another datafeed - so we changed it. In the end, it cost them well over $100k to build their robot (and of course there were many changes and alterations to the SCOPE of the project from the very beginning - so this added to the cost too). It traded with IB because IB has a good platform for automation and they wanted to test it first without spending the money to connect it to another broker.

Anyway, I'm very happy with what I've been able to do in esignal. I would really like to be able to use the newest versions, but because of my need to keep my robots running as fast as possible, I'm using 8.0.

My point is.. you can do more with esignal "right out of the box" than you could do with just about any other trading application. When it comes to developing EFS programs, I find EFS to be a feature rich environment for trading system design/deployment.

When it comes to running automated trading systems in esignal, yeah, there are some limitations, but for the Avg. Joe/Jane - it's perfect. If you want to do anything more than what is available in esignal, then you have to build a "stand-alone" solution or find another solution.

I think the developers at esignal will pull from this thread that we're all loyal and want specific "speed" issues addressed ASAP. It is critical to our "business of trading" that we avoid crashes/hangs because of volume or other issues. I'll be here to test the next version when they are ready.

[SteveH]

Jay,

I have six 5 min charts with a 30-day 9:30 - 16:15 timeframe on them.

In the morning, when the market opens, they're completely dead...no updating.

I have another time template that is 9:30 - 16:14. I change templates on all of the dead charts to this one and the charts come to life as though there is nothing wrong.

Next morning, ALL the 9:30 - 16:14 time template charts are dead when the market opens. I change the time templates on those charts to the original 9:30 - 16:15 time template (that didn't work in the first go-around), and, voila, charts come back to life for the rest of the day.

Jay, I'm playing musical time template switches at the start of each day between 2 time templates, varying only by 1 min on the end to bring them back from the grave each morning.

Jay, run this kind of test with 10.3 RC:

1. Create several 5 min ES #F and TF #F charts with a time template of 9:30 - 16:15 for 30 days.

2. Create a script that puts like 8 different emas on the chart...doesn't matter, start at like 20 and scale up to like a 900 ema (and don't insert them with the Basic menu option, actually create the EFS with the "new Array(...)" returning the values for each price bar. I do call setComputeOnClose() in the preMain().

3. Bring up the chart before the U.S. market opens at 9:30 EST.

Now, when the market opens are all of your 5 min charts working or are all or some of them dead?

If you find a dead one (i.e., not updating in real-time), then create a separate Time Template where it starts at 9:30 and ends at 16:14 and make it for 30 days.

You will find that when you switch to the other TT, things will work just as you expect them to with the chart updating.

Another curious thing is that I could change those charts to any other timeframe like a 1, 2, 3, 4, 6, 7, etc. etc. min chart and they would update in real-time without even bothering with changing over to a 9:30 - 16:14 time template.

Jay, hope this helps! I think the key to exposing this bug is for you to have your charts open before the 9:30 U.S. start on that 9:30 - 16:15 30 day template and see what happens when the real-time data hits them at 9:30.

What I'm hoping you'll find is..."holy s--t!...SteveH is right! This is a major bug we must find the answer to immediately!"

[Note to Glen (aka fast flyer): there's no way you could know this but I have 2 computer science degrees with 16+ years of professional software experience (before trading full-time in mid-2003) and I have developed in complex real-time systems before so I'm not an "outsider" to the biz of developing quality software products which other people depend on to be as perfect as possible in order to make money at something else or ensure their safety with using high quality simulation systems. With my credentials, it's not a cheap shot to say what I said. It comes from being on their side of the fence and knowing when quality is beginning to degrade with each ensuing release.

Glen, there's another saying in the software business called "eating your own dog food" which means that you're actually using the software you're developing. I develop my own trading software (C# interacting with EFS) for personal use to work with eSignal and help me make the $$$ trading. I guarantee you that software wouldn't be near as good if I hadn't been depending upon it to help make/save me money in the markets!

Glen, I simply love working with the Javascript (EFS) / C# code combo. eSignal's real-time model it presents to EFS users is simple yet elegant and I wouldn't think of changing to another product for my trading. But, I call them as I see them, and I see bugs in a product which should NOT be there at this level of maturity if the quality assurance levels area weren't becoming more lax. You have to have a passion for software development / testing to create a top flight product. Someone in that group is losing their passion and needs a sit-down pow-wow pronto!]

[SteveH]

Today (1/28/09), out of six 5 min charts, 3 of the 6 were dead on the 30 day 9:30 - 16:15 Time Template for the "ES #F" and "TF #F" symbols I had up. The only way to bring the other 3 back to "life" was to switch their Time Templates to the 9:30 - 16:14 one I have setup.

Sometimes it's 3, sometimes it's 6, etc. What I seem to be guaranteed of is that not all 6 are going to open up in real-time for the new day and function properly if I don't change the ending time by at least 1 min on the dead ones.

This is with 10.3 RC and was a fresh install on a new Windows XP SP3 installation with all of the patches applied first before installing it.

Tomorrow, I'm going to put NO STUDIES on like 6 5 min ES/TF charts and see what happens. That will at least determine whether eSignal has a problem in the EFS area of their source code "goo".

[ACM]

Steve
FWIW enclosed below is a video capture [approximately 2 minutes in duration] of the test I ran this morning using a Page and efs set up as per your description [or at least my understanding of it]. The efs used in the charts is attached and the Page [Test SteveH.pg] is available here in the Pages and Charts folder
The Page was loaded shortly before 9:30 EST and as you can see all charts started updating at 9:30. FYI I ran the same test also yesterday with the same results.
Alex

[SteveH]

Thanks for testing. I'll try your test script with a few mods and see how that goes tomorrow morning.

[ACM]

Steve
You are welcome
It may be helpful if you post the test script after you have modified it so that eSignal can test this issue under the same conditions
Alex

Originally posted by SteveH
Thanks for testing. I'll try your test script with a few mods and see how that goes tomorrow morning.

[JayF]

We have also been testing this in our QA department using a similar setup. We had 5 different PC's setup over the last two days and have not been able to duplicate the issue described in SteveH's post.

Steve,

Does your description match up with your actual usage or is it different in a slight way? If so, could you describe the differences in more detail?

Also, please provide your system specifications.

[SteveH]

Now I'm completely confused. Since last Friday, none of my charts are showing this "dead chart syndrome" and it appears that eSignal did nothing deliberately on their side to make a correction.

*Something* has changed. The only thing I can figure is that I either got assigned to a different data server or a data server that was suspect with problems on a completely different issue got "fixed" and a side-effect was clearing up the dead charts.

I'll let you know if it happens again.