- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi
I was running a new pc based on z490 board and 10900k, for audio playing, which is require system in low latency mode.
running centos 7.6 and everything tuned in low latency mode(i.e. no c stat and max performance in bios, ht disabled), I running sysjitter app to check my system's latency.
at first it was get really good results then 9900k+z390 board, output as follows:
int_total(ns): | 78830260 | 12534 | 13277 | 13087 | 13557 | 14133 | 13343 | 13692 | 13622 | 9952
int_total(ns): | 78406794 | 13801 | 13344 | 14240 | 14088 | 13578 | 13475 | 14143 | 14621 | 9807
int_total(ns): | 78656008 | 13151 | 13566 | 13668 | 13420 | 13727 | 13740 | 13022 | 14053 | 9957
int_total(ns): | 79154531 | 13406 | 13833 | 14074 | 14687 | 13648 | 13913 | 14215 | 14141 | 9750
int_total(ns): | 78687629 | 13159 | 13752 | 14356 | 13920 | 13985 | 14346 | 14268 | 13940 | 9971
stats means during 10 seconds, the total time which is not running my app. so we can say it is jitter.
but once I open two ssh session, one session running sysjitter, another running nothing, things also running well, if I do anything elso on the second ssh session, even just push Enter button, it will cause huge jitter spike on all of cores, output as follows:
int_total(ns): | 79805825 | 467528 | 358363 | 367667 | 369127 | 369177 | 371507 | 370893 | 386245 | 387845
int_total(ns): | 79327169 | 465557 | 355867 | 365324 | 365759 | 366361 | 365967 | 369585 | 382198 | 379885
int_total(ns): | 80004123 | 464637 | 353287 | 360822 | 363769 | 364829 | 362845 | 365950 | 372751 | 379466
int_total(ns): | 79744417 | 466712 | 355387 | 361916 | 365034 | 367691 | 368237 | 368179 | 384494 | 383931
int_total(ns): | 79575914 | 467936 | 361396 | 361177 | 367285 | 368111 | 368112 | 369193 | 384699 | 380917
int_total(ns): | 79826564 | 465789 | 352476 | 362940 | 364146 | 364948 | 364394 | 366481 | 377708 | 375957
int_total(ns): | 80011866 | 463337 | 356837 | 359012 | 361705 | 362312 | 365301 | 366347 | 375635 | 378728
it's like some thing are taking CPU time from isolated cores (1-9), but I cant find out which device or thread. cat /proc/interrupts and cat /proc/softirqs all looks normal.
So I specific one task on cpu1, ping 192.168.0.1, sysjitter also showing lots of interrupts and taking a lot of cpu time.
int_total(ns): | 243570971 | 1108777 | 1018476 | 1030692 | 1039277 | 1040590 | 1044897 | 1047776 | 1085418 | 1088406
int_total(ns): | 87977770 | 1210447 | 1148266 | 1162262 | 1176262 | 1179898 | 1181904 | 1187575 | 1224952 | 1243542
but when I running 'nohup taskset -c 1 ping 192.168.0.1 &', make the ping program no output on screen. the strangist things happen, everything is in expect. the jitters only on CPU1, like follows
int_total(ns): | 78915692 | 133831 | 14400 | 14229 | 14294 | 14442 | 13883 | 13828 | 13657 | 10009
int_total(ns): | 78943889 | 137861 | 13740 | 14504 | 14698 | 13944 | 14472 | 14125 | 13931 | 9839
int_total(ns): | 78547289 | 132860 | 13719 | 13472 | 13999 | 14165 | 14314 | 13719 | 13889 | 9586
int_total(ns): | 79390416 | 134703 | 13879 | 13332 | 14216 | 13848 | 14281 | 14239 | 14082 | 9940
I suspect the problem is the linux STDOUT on Z490 chipset are unexpectly causing system jitter and it will effect all cpus even the cores are isolated in kernel.
and this is never happened on Z390 chipset.
anyone have info could share with me ?
Thanks, really appreciate!
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello Jiangv,
Thank you for posting on the Intel* Community.
Based on the information you provided the issue seems to be related to the Linux OS* you're using. Our best recommendation is to test the processor, if possible, in a different system using Windows* 10, and test the processor with our tool Intel* Processor Diagnostic Tool:
Intel® Processor Diagnostic Tool (IPDT)
- Intel® Processor Diagnostic Tool download.
- The diagnostic tool checks for brand identification, verifies the processor operating frequency, tests specific processor features and performs a stress test on the processor.
- To save the report, once the test is done, click on "File >> View Results File" and attach the .txt file to your reply.
If you are unable to follow the above recommendation, the best course of action will be to address this to the Linux* community where the distro owner may help you to identify and fix the behavior.
Best regards,
Maria R.
Intel Customer Support Technician
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello Jiangv,
Were you able to check the previous post?
Let me know if you need more assistance.
Best regards,
Maria R.
Intel Customer Support Technician
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello Jiangv,
We have not heard back from you, so we will close this thread. If you need any additional information, please submit a new question as this thread will no longer being monitored.
Best regards,
Maria R.
Intel Customer Support Technician
![](/skins/images/7FC17B7B85029576C25F1E43CE255B51/responsive_peak/images/icon_anonymous_message.png)
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page