Support Board
Date/Time: Wed, 10 Dec 2025 01:20:13 +0000
Post From: Replay slows progressively the longer I run it
| [2025-10-17 14:29:52] |
| User98657 - Posts: 391 |
|
I'm running extended periods of replay (a year) and experiencing a progressive slowing of replay speed. I'm not at all suggesting this is a Sierr Chart fault, cause, or issue. I'm just asking for suggestions, or ideas. I'm running replay with an ACSIL Custom Study for long-ish periods (as I said, a year) on tick data and Renko bars. I'm running this on a separate instance of SC. I'm setting the chart to use specific dates, and condensing my evening session time to only include the 15 minutes prior to the start of the US RTH open at 9:30e. "Skip Empty Periods" is enabled on replay. I'm using Accurate Trading System Back Test Mode for consistency. I run these tests Disconnected from the data feed. I've read as much documentationn and help/trouble tickets as I can find related to the issue, and combed through all the suggested chart settings to help replay run efficiently, such as Update Intervals. When I start the back test, things run fine. There's a noticeable difference between speed selections, and I can see a fairly smooth data feed until I start pushing above 240x, which is when I begin to see that start-stop / "data gathering" pause, data dump on to the screen, pause while it gathers more data...... I don't particularly care for the start, pause, dump, but I understand it, and the testing has been VERY consistent (amazingly consistent). I generally run it at 960x. Beyond 960, there's no noticeable difference in actual speed. HOWEVER, It seems to run the first few months of data consistently, then begins to slow. And it continues to slow the longer it runs. Last night, it only processed two months of data in the 7 hours I was sleeping. This is obviously painfully slow, and not at all what I see when I begin the run. I'm running an I9-13900 with 32GB of Ram and SC is housed on a 2TB SSD. I'm using two Nvidia 8GB Display Adapters for 5 24" monitors, but not using Open GL due to the driver conflicts we began experiencing with Open GL. I realize there are many factors you can't speak to. I realize you can't know how many ticks are being processed on any given day and busy days will be very data intensive. But this isn't slowing down for one, two, or even a few days. It's slowing down overall. I'm not using very many studies on the chart, and none I would consider to be extremely data intensive other than perhaps the ZigZag study, but I realize you can't possibly know if the Auto-Trading Custom Study I'm using is as efficient as it could possibly be. But again, things seem to start out fine, then become progressively slower. There is of course, no way to accurately quantify any of this, which is as frustrating to me as it must be to you when I can't provide any quantifiable measures. I'm not at all suggesting this is a Sierr Chart fault, cause, or issue. I'm merely reaching out to ask if there's anything you may be aware of that I can change or look at to help avoid this slowing. Is this perhaps a memory caching issue where, due to the nature of how SC data is processed during replay, my machine's memory cache is loading up and leaving less overhead room for data transfer? Though my memory is only loaded to 19GB at the peak of everything running, would I benefit by having 64GB of RAM? Any suggestions would be appreciated, and as always, thanks for the support. |