All right. Let’s try to find out commonalities.
It’s crazy … often, it’s like you and I are the only users who get problems?? 🙂
Or let others speak up, please ?
I will start with a few environmental “facts”.
I usually run more than one instance of PRT. This does not count for PRT-IG – I have only one instance/account for IG. But PRT-IG and PRT-IB run next to each other. Most won’t be using PRT-IB, so we can skip that all together.
Lately (3 weeks or so) I see that the one Jave instance is influencing the other. This never was so. I don’t apply explicit Java updates (I actually don’t even know it is running). Windows updates, sadly, occur regularly (last one this very morning).
OS is W10.
Processor is some Xeon. I am 100% confident that the processor matters. I feel that I have all sorts of “anomalies” with PRT since I use this processor (which is about a year). Before that I used an i7.
Hyperthreading of this 10-core processor is off.
I don’t recognize at all that my platforms are slow anywhere – the contrary (like startup in 40 seconds with 40+ charts). However, something changed since the same 3-week-ish time : charts are populated differently. They stay black (black is my chosen background) at first and some (types of windows) receive a white background at first. However, backtesting is painful.
Half of my charts are of the 1 or 2 second timeframe. The other half is 5 minutes. One is 1 day.
I use 3 monitors, all connected via HDMI. Taskbar is on one monitor only.
Video card is a GTX 1650.
Mouse is wired. I say this because I seem to have too many weird “click did not come through” or “clicked without knowing” issues that PRT can’t solve.
I have shares to this PC (main C: drive) and also towards another PC.
The account I use most often, is shared with a PC elsewhere. Notice that this relates to the icon you see in the attachment.
Internet speed is 1000Mb. At some stage I went from 80Kb to 400Mb and at that time I was sure this incurred for issues I never had before. Losing packets or so which indeed could be detected at the provider.
Round trip latency is ~4ms (was 20+ with the slower speed).
I can see or feel that data can pile up and choke the platform. Thus, say once per hour for 10 seconds or so nothing moves in one of the instances (the other happily continuing). (nothing moves : the refreshing in the lists with green/red price changes)
I deem this fairly normal, but I notice that this did not happen with the slower internet speed.
When things stall, clicks are nicely queued (read : I can’t loose clicks because of it).
Telephone is over the same internet connection, spread over DECT and IP (2x). TV is usually not over internet (only very occasionally). Using the telephone is not related, it “being there” could.
Larger downloads may happen (mostly via torrents) and could be on-going throughout (this is not something I do myself).
Nobody needs to share his/her environment like I just did. But seeing commonalities and recognising similar issues, could trigger some elaboration.
Careful please, because I don’t recognise performance issues or issues related to my environment as such, at all. So it is merely my idea that everybody has similar issues but diminishes them as a “so what !” ? So … it could be the other way around : people with specific issues could post here and we could try to see what can be done to solve it.
Java issues seem to be on top of matters, while we never see how things are solved. *If* they are solved, because it merely seems that new people just vanish again because of unworkable circumstances.
The issues I do have (daily) are always of a kind you can see below from last night. Mind you, this was during opening times of these futures, but right before closing (usually incurring for a high volatility). In the 3rd attachment – right after the opening again – you see what mysteriously came from it. I did nothing in between ! The top chart solved itself by filling whatever order it was, and the bottom one changed to a very different limit which I did not even notice when I took the screenshot. But it is there all right, and this morning I found the order filled.
I registered really 100s of such issues which always come down to the broker (IB in this case) doing something wrong, which I believe (but with a lot PRT could do to solve things better, but this topic is not about that).
Btw, it is not important for me at all to work out this very issue (I stopped reporting them) – it is only an example of the issues I really have and which seem to have grown or incurred for with more internet speed or the other processor. This is a hunch; a feeling.
People reading this should be aware that I may do some more trades than average. So I incur for more issues as well.
For the backtesting … For me it has become impossible to use the normal Editor. Typing is just impossible in there. Previously (say a month back and beyond) this was only about typing and memory leaking, which very obviously shows if you know where to look. Today it has become impossible to use the Editor all together, which is similar to moving windows in my PRT backtest environment (which is IG). Typing in any field (like Search) has become undoable. I think this currently is the main issue for many. I can see by other means/matters that PRT is changing things, but I am not 100% confident it is PRT doing this to me/us. It can be Windows updates. It can be Java updates without me knowing (I never do anything with Java).
My code is not small and this relates. However, the fact that I use external editor means for that now, did not solve problems for all being sluggish lately. It is a bit tough to judge the effects of the Editor not causing issues any more, because the whole behaviour is different now (like backtesting and growing memory consumption may recover at doing nothing – say giving back 2MB (to 4MB) from 6MB usage). With using the Editior that will never happen and the lot will collapse first. N.b.: I can now backtest for the larger part of the day, especially if I don’t use Graph commands. Another n.b.: using graphs with more than 50K or so bars, has become completely impossible (since mentioned 3 weeks, I’d say). This will be the combination of number of Graphs and the size of the code. But important : this worsened.
Peter
.
