Star-fst-l mailing list

["Visser, Gerard" <gvisser AT indiana.edu>]

Yes, 700 Hz will be 2.2% deadtime at minimum. (In all this discussion, "at minimum" means the frontend's fixed deadtime with 300 RS cycles as it is now set, which seems to be a good setting. There is little prospect to reduce that significantly for 3 timebin readout, though we could​ go to 1 timebin readout if that worked for analysis needs. (It's plausible...) I am not really sure whether 2 timebins would be much lower than 3, also could be tried.

The deadtime due to data transfer and processing bottlenecks should be almost always ~zero until we hit the various rate ceilings and then they will turn on very sharply.

       Gerard

---

From: Ye, Zhenyu
Sent: Wednesday, December 22, 2021 10:22 AM
To: Visser, Gerard
Cc: Tonko Ljubicic; Star-fst L
Subject: Re: [Star-fst-l] Final FST deadtime/rates with 3 timebins

Hi Gerard,

This is a good question. I think it needs to be discussed with the trigger board, and the decision will depend on the overall forward trigger rate that is still under study, while the the mid-rapidity triggers are more or less settled. For the latter, HF PWG is interested in including forward detectors in dimuon and HT triggers which have a total rate of about 600-700 Hz. Including FST in such triggers will add 2% dead time if I am not mistaken. 

Zhenyu

On Dec 22, 2021, at 9:09 AM, Visser, Gerard <gvisser AT indiana.edu> wrote:

hi Zhenyu,

   The minimum deadtime will be 15% at 4.6 kHz, 21% at 6.5 kHz. It would be good to know if that will be acceptable, i.e. losing 21% of rare triggers maybe will be objected to?

      Gerard

---

From: Star-fst-l <star-fst-l-bounces AT lists.bnl.gov> on behalf of Ye, Zhenyu <yezhenyu AT uic.edu>
Sent: Wednesday, December 22, 2021 10:05 AM
To: Tonko Ljubicic <tonko AT bnl.gov>
Cc: Star-fst L <star-fst-l AT lists.bnl.gov>
Subject: Re: [Star-fst-l] Final FST deadtime/rates with 3 timebins

 

Hi Tonko and Gerard

Thanks for the effort to reduce the dead time. 

As you have heard from the trigger board meeting, there is an interest to including forward trackers in mid-rapidity program.
From what Tonko wrote, there seems to be large room for improvement on FST rate to go up from 4600 to 6500 Hz. If we could do that, it will provide an possibility to include FST in such triggers. I don’t know how much effort on your side is needed to do the code optimization and I would strongly encourage you to do that.

Thanks and Best,
Zhenyu

> On Dec 22, 2021, at 4:04 AM, Tonko Ljubicic <tonko AT bnl.gov> wrote:
> 
> Hello Gerard, all,
> 
> I played a bit with the intrinsic deadtime and:
> 
> 1) 290 still crashes
> 2) 300 is OK -- I will set it to 300
> 
> Running full blast I get 100% dead at:
> 
> 1) 4600 Hz
> 2) 140 MB/s per fiber
> 3) 94% CPU usage
> 
> Which means that at this point the CPU is the limiting factor.
> We'll see how it will go in reality, with actual required rates, etc
> but if we still have a problem I can mess with the code
> and likely optimize the CPU code more to reach the final
> limiter which is 200 MB/s per fiber, at ~6500 Hz.
> 
>        -- Tonko
> _______________________________________________
> Star-fst-l mailing list
> Star-fst-l AT lists.bnl.gov
> https://nam04.safelinks.protection.outlook.com/?url="https%3A%2F%2Flists.bnl.gov%2Fmailman%2Flistinfo%2Fstar-fst-l&amp;data=04%7C01%7Cyezhenyu%40uic.edu%7C1b0ab303f5db43f8619508d9c5327d09%7Ce202cd477a564baa99e3e3b71a7c77dd%7C0%7C0%7C637757642922188300%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&amp;sdata=LvVJUpOx%2FwbCR9iuKJEqtmJuMrK4IBsycmbUqaXnDMs%3D&amp;reserved=0

_______________________________________________
Star-fst-l mailing list
Star-fst-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/star-fst-l