Support Board
Date/Time: Sat, 21 Feb 2026 23:07:48 +0000
OS Timers vs SC Timers discussion again... SC Timers chart update limitations
View Count: 1557
| [2026-01-26 04:49:43] |
| blt - Posts: 158 |
|
... I am on Windows 11 pro note: I am in no hurry at all about this Thank you very much |
Private File Attachment Deleted. |
| [2026-01-26 17:00:38] |
| Sébastien - Posts: 123 |
|
V2868, I did some tests with the SC Timer. 0ms Timer allows crazy high refresh rate but CPU utilization is a bit too high for my old Intel 13900H. 1ms seems to be the sweet spot. I get the refresh rate I need for my kind of trading (about 30ms), and CPU utilization seems a bit lower than with OS Timer. Thank you for your work SC team ! |
| [2026-01-26 18:38:24] |
| blt - Posts: 158 |
|
... to: Sierra_Chart Engineering Just tested OpenGL with Ver. 2869 good news is no more violent shut down T&S is working both charts are empty total shutdown did not occur Message Log: "Cannot allocate GPU memory for drawing" SC Timers working perfect 👌 Just FYI Thank you...you doing miracle... |
| [2026-01-27 05:44:49] |
| Sierra_Chart Engineering - Posts: 22936 |
|
See this thread: Critical Update for Sierra Chart Timers 2867 through 2869 | OS vs. SC Timers | 4 MS! Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, use the Teton service: Sierra Chart Teton Futures Order Routing |
| [2026-01-27 15:04:24] |
| blt - Posts: 158 |
|
... to: Sierra_Chart Engineering tested on live chart (SC timers); ver. 2870 Chart Update Interval in Milliseconds....................................................20 (didn't want to test anything lower) Timer Management - Chart Calculation/Draw Time Multiplier for Interval Delay..............1 Minimum Delay Between Chart Updates in Milliseconds.......................................3 RESULTS ARE PHENOMENAL !!!! In my case I have no need for OpenGL at all !!! Thank you Thank you Thank you |
| [2026-01-27 15:28:19] |
| User907968 - Posts: 866 |
|
RESULTS ARE PHENOMENAL !!!!
That is my observation too. I'd been putting off using SC timer based on how it worked when they first released. Now much, much better. |
| [2026-01-27 15:41:44] |
| blt - Posts: 158 |
|
... to: User907968 With this snappiness jitteriness speed looking at my 7 monitors-my only fear comes from developing schizophrenia SIERRA CHART ENGINEERING is doing miracles Thinking that I encountered SIERRA CHART almost by an accident makes me so happy and nervous damn Wish you all the best |
| [2026-01-27 16:21:48] |
| n8trading - Posts: 68 |
|
What do you think about allowing fractional milliseconds (this would be more user-friendly than changing it to microseconds) for the Minimum Delay Between Chart Updates in Milliseconds like .05 or .01? 1ms has good UI responsiveness for me but chart updates are still slower than OS timers, so perhaps there is some more headroom on my computer that I could take advantage of if I could set the milliseconds a little lower than 1ms. This would allow me/us to truly fine tune the delay so that we can get maximum chart update speed while ensuring we have enough UI responsiveness for the particular computer we are on?
|
| [2026-01-27 17:04:20] |
| blt - Posts: 158 |
|
... to: n8trading (hope you are addressing me) Allow me to explain how I see all these phenomenal improvements SIERRA CHART is doing for us🤔 (feel free please to correct my thinking) SIERRA CHART is enabling us to have as miniscule as possible time overhead to usual retail inability to have pipe directly into the wall of data center (although I am bloody close to that) Meaning 10-30-80 ms plus time used by our hardware/software In my opinion SIERRA CHART went above phenomenal (it could easily be that SIERRA CHART is better architecture than many professional machines/software because SC Wizards🧙♂️ are addicted to bring the impossible from C++) Did you notice that I set my interval to 20 ms (I am lucky one having my internet node positioned to data center in Aurora and CME to unbelievable 10-20 ms; jitter 1 ms ; ping 7-8 ms) because I am a bit afraid if I goes lower I will turn psycho Another thing is my trading machine-no other retail trader has better one-handling everything coming my way to perfection (trading is my only way making living) I am so happy 😊 with SIERRA CHART beyond describable with ONE (1) PROBLEM-I can't "force" them to take my $$$ as an appreciation for their work; Engineering; John-SUPPORT I NEVER encounter anything similar to SIERRA CHART Sorry to bother you with all this;in short-I was perfect with 100 ms then 50 ms; 30 and now 20 ms and will probably go down to 10 ms Build the strongest trading machine you can (if you are on Win with a lot of RAM take paging out) surround yourself with monitors start trading 💹 make money 💵 Wish you all the best p.s. With this new improvement I really have no need for OpenGL |
| [2026-01-27 17:12:26] |
| n8trading - Posts: 68 |
|
I was asking SC engineers about making fraction milliseconds an option. But thank you blt for that info, good points.
|
| [2026-01-28 13:41:34] |
| @sstfrederik - Posts: 412 |
|
Great work on the timers. I am trying to see proof of custom study execution at the set chartupdate interval when UpdateAlways = 1. running SC version 2870. One chart, one custom study, code posted below. Chartupdate interval set at: 20ms. Minimum chartupdate interval in ms for ACSIL updatewalways: 20ms. Using SC timers. Timer management enabled. max delays 20ms. min delay 10ms. The custom study simply prints a line at every calculation. Not sure why the log isn't showing more output based on custom study calculations. Output is shown at ~200ms with regular occasions of ~500ms calculation intervals. What am I missing? Seems like the study calculations are not following the chartupdate and update always settings. With some of my work this will be a limitation for real-time applications. A custom study can't pick up any real-time data from other custom threads within 20ms if there is a 200ms gap. I see the same ~200-500ms gap on that custom work. Thanks for your time to look into this. SCSFExport scsf_SST_TestFunction(SCStudyGraphRef sc) { if (sc.SetDefaults) { // Set the study configuration and defaults. sc.StudyDescription = "Test"; sc.GraphName = "Test"; sc.AutoLoop = 0; sc.UpdateAlways = 1; return; } int& counter = sc.GetPersistentInt(1); // Data processing if(sc.IsFullRecalculation){ counter = 0; return; } //push message to log to see update interval SCString updateIntervalMsg; updateIntervalMsg.Format("Update interval counter is %d", counter); sc.AddMessageToLog(updateIntervalMsg, 1); counter++; } |
| [2026-01-28 18:00:50] |
| User719512 - Posts: 419 |
|
Frederik, you can't write to the log every iteration and expect to measure perf at levels like 20ms; well at least I would not. You could track/accumulate over say 10 seconds and get avg min max or something and log that, reset, and accumulate again. For the most accurate timings on Windows, you'd want to incorporate Performance Counters. Instead of logging, using GDI/Graphics is very low overhead for more real-time visualizations. I get a good steady state after ~5 seconds. And at 20ms settings, my chart/machine will measure ~33ms for a single chart/DOM test chart setup. The study was called 211 times in 7 seconds for about 30 times a second or 33ms interval (see attached). There are parts in play not measured like Sierra's own processing time, OS overhead for network and file I/O etc. Measuring a single study perf for "how often can my study get called" seems close enough to gauge responsiveness so I never expect to "see" 20ms with that setting. During replay, I have seen sub ~10ms updates with the same setup. I don't think my CPU/GPU would get me much past that. I don't think Update Always will go faster than Chart Update Interval, so might check/verify that. Sierra is blazingly fast. |
 Screenshot 2026-01-28 093344.png / V - Attached On 2026-01-28 17:52:15 UTC - Size: 194.95 KB - 31 views Screenshot 2026-01-28 094133.png / V - Attached On 2026-01-28 17:52:20 UTC - Size: 202.74 KB - 30 views |
| [2026-01-28 19:02:14] |
| @sstfrederik - Posts: 412 |
|
thanks. Sure sending to msg log at every call might not be the best case. But even sending a counter to the log every 10 seconds only shows about 33 calls. If study gets called more often that counter should be way higher. Update: on a fresh SC installation it seems to be fine (420 calls in 10 seconds). Unsure why the other installation is not responding the same way ... SCSFExport scsf_SST_TestFunction(SCStudyGraphRef sc)
{ int& PriorArraySize = sc.GetPersistentInt(1); // Set the configuration variables if (sc.SetDefaults) { // Set the study configuration and defaults. sc.StudyDescription = "Test"; sc.GraphName = "Test"; sc.AutoLoop = 0; sc.UpdateAlways = 1; return; } int& counter = sc.GetPersistentInt(1); SCDateTime& Now = sc.GetPersistentSCDateTime(1); SCDateTime& LastCall = sc.GetPersistentSCDateTime(2); // Data processing if(sc.IsFullRecalculation){ counter = 0; return; } //push message to log to see update interval Now = sc.CurrentSystemDateTime; if(Now > LastCall + SCDateTime::SECONDS(10)){ LastCall = Now; int calls = counter; SCString callsMsg; callsMsg.Format("Calls in the last 10 seconds: %d", calls); sc.AddMessageToLog(callsMsg,1); counter = 0; } counter++; } Date Time Of Last Edit: 2026-01-28 19:26:38
|
| [2026-01-28 20:04:38] |
| blt - Posts: 158 |
|
... to: SIERRA_CHART ENGINEERING Being discretionary trader would like to report to WIZARDS of SIERRA CHART ENGINEERING that right now (FOMC press conf.) your marvel is performing to perfection on all of my 7 monitors (SC TIMERS engaged with moderate settings) My issue with OpenGL could be totally dismissed; I do NOT need that at all; feel free to entertain yourself with that at your discretion and timing Phenomenal work you did Thank you very much |
| [2026-02-01 20:57:21] |
| n8trading - Posts: 68 |
|
SC Engineering, From my testing so far in 2873, I cannot find the sweet spot yet of getting at least as good of chart update speed and UI responsiveness as OS timers using different combinations of chart update interval and minimum delay between chart updates. It's looking like I might be able to find that sweet spot if I could slightly increase (or decrease) the minimum delay between chart updates such as to 1.1ms or 1.2ms. If I increase it to 2ms, that seems to be huge in terms of effective delay as it significantly slows down my chart updates. And note that I am also increasing the chart update interval itself to say 50ms. Right now with 50ms chart update interval and 1ms minimum delay, UI responsiveness is pretty bad but effective chart update speed is just as fast as I get when using OS timers. If I go to 100ms chart update speed with 1ms minimum delay, there is a noticeable degradation in effective chart update speed but UI responsiveness ok (not great but useable). Thank you, |
| [2026-02-02 23:16:10] |
| Sierra_Chart Engineering - Posts: 22936 |
|
We have looked into this and it is very difficult to support a fraction of a millisecond. There is no standard support for this in the operating system. It would require a special implementation and probably would cause a performance issue.
Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, use the Teton service: Sierra Chart Teton Futures Order Routing |
| [2026-02-03 21:38:45] |
| Sierra_Chart Engineering - Posts: 22936 |
|
Regarding this: OS overhead for network and file I/O etc.
Network I/O is on different threads. Not the primary thread.File I/O is also on other threads (both reading and writing). When an Intraday chart is updating, it is only reading from a memory cache. Not from a file. Although as historical data is being downloaded, the chart data is read from the file. When the download finishes, and the file reading is complete there is no more reading from the file. Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, use the Teton service: Sierra Chart Teton Futures Order Routing Date Time Of Last Edit: 2026-02-03 21:39:06
|
To post a message in this thread, you need to log in with your Sierra Chart account: