Login Page - Create Account

Support Board


Date/Time: Fri, 29 Mar 2024 10:12:35 +0000



Disable Hyper Threading Recommendation. Feedback Requested.

View Count: 2108

[2019-08-09 18:05:05]
Sierra Chart Engineering - Posts: 104368
Our recommendation when using Sierra Chart is that you disable hyperthreading if your CPU supports hyperthreading. This can be done in the system BIOS on your computer.

We would like feedback if this helps with Sierra Chart performance assuming that you have had any performance issues currently.

Having as many CPU cores as you can is always going to be best. Not just for Sierra Chart but for your system in general. This is especially true when using multiple instances of Sierra Chart with each instance taking advantage of additional cores.

All communication between Sierra Chart instances occurs on a background thread for the best performance.

For information about the use of threads in Sierra Chart, refer to:
Performance

And we also recommend installing a dedicated solid-state drive with Sierra Chart located on that drive and no other programs. These drives are very inexpensive these days. 250 GB is more than sufficient. It is very easy to transfer Sierra Chart over to a newly installed drive. Refer to:
Transferring Sierra Chart Settings and Data From Existing Installation to New Installation
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, *change* to the Teton service:
Sierra Chart Teton Futures Order Routing
Date Time Of Last Edit: 2019-08-09 18:06:57
[2019-08-11 23:24:54]
User136422 - Posts: 33
Tested many times SC with hypertreadhing on and off. Never seen any notable difference, nor had any issue related to that, it's alway on on my platform.
cpu: 8700K sc is running on a dedicated m.2 250gb samsung 970 evo
[2019-08-12 02:47:39]
User735389 - Posts: 188
Did a quick test, didn't see any difference in terms of time it takes the software to do a round trip, or the "feel" of the system. If you have the ram for it (its super cheap right now), you can probably run it on a ramdrive for even faster performance. Currently using a sqllite db for some chartbook and system communication.
[2019-08-12 03:07:59]
Sierra Chart Engineering - Posts: 104368
Thank you for the feedback. We were not sure whether disabling hyperthreading would help or not. We do disable it on our servers, and have observed a benefit.
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, *change* to the Teton service:
Sierra Chart Teton Futures Order Routing
Date Time Of Last Edit: 2019-08-13 04:40:00
[2019-08-12 11:54:27]
User136422 - Posts: 33
It will be incredible helpful having a sort of benchmark/stress mode to test SC performance with different hardware, especially considering the incredible amount of difference between current cpus, ssd and other components.
I haave no idea if this is technically doable and how, so take this as a simple suggestion.
[2019-08-12 12:25:10]
User735389 - Posts: 188
I have a very simple test study that just reports the amount of time elapsed since the last time the study was called. Just this simple data point gives me a very good idea of how things are doing, if there are any unexpected performance issues after updating studies, the relative "speed" between different computers, weird cpu spikes, etc... For example i can see my system averages about .5s per round trip on win7, .7s on win10. My desktop is closer to .4s average in win 7. You can extend this idea and just see how long your system takes to "replay" a specific timeframe. Something to consider if you need a quick benchmark.
[2019-09-12 04:00:28]
User379468 - Posts: 508
We do disable it on our servers, and have observed a benefit.

Could you share the details observed?
[2019-09-12 04:22:45]
User106180 - Posts: 88
I have a very simple test study that just reports the amount of time elapsed since the last time the study was called.

User735389 - Good idea. It would be useful to have this output in the chart header. Did you build this using the inbuilt studies, or did you code it yourself?
[2019-09-12 06:44:08]
User379468 - Posts: 508
Good idea. It would be useful to have this output in the chart header.

Agree, SC really needs some better tools to measure SC performance such as the benchmark mentioned above for performance testing, easily viewable realtime chart update times, lag vs the desired update interval, CPU efficiency for optimizing instances, etc.

Some interesting info regarding dedicated CPU core allocating, windows scheduler, hyper threading:

https://www.overclock.net/forum/297-general-processor-discussions/1729504-review-explanation-multithread-performance-benchmarks-impact-windows-scheduler-programmed-cpu-core-allocation.html

https://www.agner.org/optimize/blog/read.php?i=838

It does make sense SC may work better with HT disabled to fully make use of the single CPU core it mainly relies on, considering the primary load/chart updating, can only use one CPU thread.
[2019-11-01 00:47:14]
User379468 - Posts: 508
We do disable it on our servers, and have observed a benefit.

Could you share the details observed?

SC Engineering, can you help us on the details here? I am building a new system and would like to optimize it for SC performance.
[2019-12-06 00:14:34]
j4ytr4der_ - Posts: 910
I would also like to know more about hyperthreading or any other potential performance enhancers for SC.

To post a message in this thread, you need to log in with your Sierra Chart account:

Login

Login Page - Create Account