Intel D525MW BIOS Configuration, Jitter Results
30 Jul 2015 03:27 #61024
by AggieMEEN
Intel D525MW BIOS Configuration, Jitter Results was created by AggieMEEN
I thought I'd put this information out there for anyone who has an Intel D525MW and is working to tune it for LinuxCNC 2.6+ on the Hybrid ISO (Debian Wheezy). Also in case something happens and I need a reference for later.
Beginning with BIOS. This is every setting I've selected, based on reading forum posts here and elsewhere.
MAIN PAGE:
Intel Hyper-Threading Technology - <DISABLE>
BOOT CONFIGURATION:
Numlock <ON>
System Fan Control <DISABLE>
System Fan Speed <100%>
Max CPUID Value Limit <DISABLE>
Display Setup Prompt <ON>
Display POST Codes <ENABLE>
PERIPHERAL CONFIGURATION:
Serial Port <DISABLE>
Parallel Port <ENABLE>
Mode <EPP> (Note, I read EPP should be used for those people who run a Gecko G540, like me)
Audio <DISABLE>
On-Board LAN <ENABLE>
INTERNAL CHIPSET DRIVE CONFIGURATION:
Configure SATA as <IDE>
S.M.A.R.T. <ENABLE>
Hard Disk Pre-Delay <0>
EVENT LOG:
Clear Event Log <DISABLE>
Event Logging <DISABLE>
VIDEO CONFIGURATION:
DVMT Mode <DVMT>
IGD DVMT Memory <256 MB>
IGD Aperture Size: 256MB
Primary Video Adapter <Int Graphics (IGD)>
CHIPSET CONFIGURATION:
PCI Latency Timer <32>
HPET <ENABLE>
USB CONFIGURATION:
USB Ports <ENABLE ALL>
USB LEGACY <DISABLE>
SECURITY:
Expansion Card Text <DISABLE>
Execute Disable Bit <ENABLE>
POWER:
After Power Failure <STAY OFF>
Wake on LAN from S5 <STAY OFF>
ACPI Suspent State <S1 STATE>
S1 State Indicator <ON>
Wake System From S5 <DISABLE>
Wake on PS/2 Keyboard from S5 <STAY OFF>
BOOT:
Boot Menu Type <NORMAL>
Hard Drive Order <my HDD identified here>
Boot to Optical Devices <DISABLE>
Boot to Removable Devices <DISABLE>
Boot to Network <DISABLE>
USB Boot <DISABLE>
ZIP Emulation Type <FLOPPY>
Boot USB Devices First <DISABLE>
USB Mass Storage Emulation Type <AUTO>
UEFI Boot <DISABLE>
I also employ the 07_RTAI method to run off of a single core for LinuxCNC.
So. With all of that being said, I run a latency test with one instance of GLXGEARS, a terminal window and the Iceweasel browser running. The result is a Max Jitter of 10273 when I'm watching. I let things run while I get some lunch, then come back in about 30 minutes to find the Max Jitter is 20537. And if I leave it for longer, the Max Jitter value bumps up higher still (to around 50,000+). If I run more than one instance of GLXGEARS (let's say four) and let the latency test run for the evening, I will get values well above 100,000.
Is this consistent with what other people are seeing with these boards, running the Hybrid ISO flavor of LinuxCNC? Does the Latency Test score gradually increase over time?
Am I, for lack of better words, "properly abusing" the system with the Latency Test conditions I use?
Does anyone see any BIOS entries that I should change to improve the consistency of my latency scores?
My apologies for the lengthy posting.
Thanks,
Beginning with BIOS. This is every setting I've selected, based on reading forum posts here and elsewhere.
MAIN PAGE:
Intel Hyper-Threading Technology - <DISABLE>
BOOT CONFIGURATION:
Numlock <ON>
System Fan Control <DISABLE>
System Fan Speed <100%>
Max CPUID Value Limit <DISABLE>
Display Setup Prompt <ON>
Display POST Codes <ENABLE>
PERIPHERAL CONFIGURATION:
Serial Port <DISABLE>
Parallel Port <ENABLE>
Mode <EPP> (Note, I read EPP should be used for those people who run a Gecko G540, like me)
Audio <DISABLE>
On-Board LAN <ENABLE>
INTERNAL CHIPSET DRIVE CONFIGURATION:
Configure SATA as <IDE>
S.M.A.R.T. <ENABLE>
Hard Disk Pre-Delay <0>
EVENT LOG:
Clear Event Log <DISABLE>
Event Logging <DISABLE>
VIDEO CONFIGURATION:
DVMT Mode <DVMT>
IGD DVMT Memory <256 MB>
IGD Aperture Size: 256MB
Primary Video Adapter <Int Graphics (IGD)>
CHIPSET CONFIGURATION:
PCI Latency Timer <32>
HPET <ENABLE>
USB CONFIGURATION:
USB Ports <ENABLE ALL>
USB LEGACY <DISABLE>
SECURITY:
Expansion Card Text <DISABLE>
Execute Disable Bit <ENABLE>
POWER:
After Power Failure <STAY OFF>
Wake on LAN from S5 <STAY OFF>
ACPI Suspent State <S1 STATE>
S1 State Indicator <ON>
Wake System From S5 <DISABLE>
Wake on PS/2 Keyboard from S5 <STAY OFF>
BOOT:
Boot Menu Type <NORMAL>
Hard Drive Order <my HDD identified here>
Boot to Optical Devices <DISABLE>
Boot to Removable Devices <DISABLE>
Boot to Network <DISABLE>
USB Boot <DISABLE>
ZIP Emulation Type <FLOPPY>
Boot USB Devices First <DISABLE>
USB Mass Storage Emulation Type <AUTO>
UEFI Boot <DISABLE>
I also employ the 07_RTAI method to run off of a single core for LinuxCNC.
So. With all of that being said, I run a latency test with one instance of GLXGEARS, a terminal window and the Iceweasel browser running. The result is a Max Jitter of 10273 when I'm watching. I let things run while I get some lunch, then come back in about 30 minutes to find the Max Jitter is 20537. And if I leave it for longer, the Max Jitter value bumps up higher still (to around 50,000+). If I run more than one instance of GLXGEARS (let's say four) and let the latency test run for the evening, I will get values well above 100,000.
Is this consistent with what other people are seeing with these boards, running the Hybrid ISO flavor of LinuxCNC? Does the Latency Test score gradually increase over time?
Am I, for lack of better words, "properly abusing" the system with the Latency Test conditions I use?
Does anyone see any BIOS entries that I should change to improve the consistency of my latency scores?
My apologies for the lengthy posting.
Thanks,
Please Log in or Create an account to join the conversation.
31 Jul 2015 01:47 #61054
by andypugh
The test shows the highest latency seen so far, so the value will only ever increase.
I would be tempted to try running "xset -dpms" in the terminal to disable display power saving. Does your display sleep when you leave the machine untouched?
Replied by andypugh on topic Intel D525MW BIOS Configuration, Jitter Results
Is this consistent with what other people are seeing with these boards, running the Hybrid ISO flavor of LinuxCNC? Does the Latency Test score gradually increase over time?
The test shows the highest latency seen so far, so the value will only ever increase.
I would be tempted to try running "xset -dpms" in the terminal to disable display power saving. Does your display sleep when you leave the machine untouched?
Please Log in or Create an account to join the conversation.
01 Aug 2015 09:42 #61087
by AggieMEEN
OK, that makes sense about the value only increasing. So, to get a feel for how this machine handles latency, would a set time period make more sense for the test? Fire it up and let it run for 15 minutes?
As for the display, I set the power settings under xFce Power Manager's monitor tab to <NEVER> put display to sleep, switch off display or reduce screen brightness. That seems to do the job; I turn the monitor off whenever I am not using the CNC computer. Or does the terminal command do something different than what I've selected in Power Manager?
Replied by AggieMEEN on topic Intel D525MW BIOS Configuration, Jitter Results
The test shows the highest latency seen so far, so the value will only ever increase.
I would be tempted to try running "xset -dpms" in the terminal to disable display power saving. Does your display sleep when you leave the machine untouched?
OK, that makes sense about the value only increasing. So, to get a feel for how this machine handles latency, would a set time period make more sense for the test? Fire it up and let it run for 15 minutes?
As for the display, I set the power settings under xFce Power Manager's monitor tab to <NEVER> put display to sleep, switch off display or reduce screen brightness. That seems to do the job; I turn the monitor off whenever I am not using the CNC computer. Or does the terminal command do something different than what I've selected in Power Manager?
Please Log in or Create an account to join the conversation.
01 Aug 2015 09:46 #61088
by andypugh
My recent experience is that the display will sleep no matter what you do in power manager.
Replied by andypugh on topic Intel D525MW BIOS Configuration, Jitter Results
Or does the terminal command do something different than what I've selected in Power Manager?
My recent experience is that the display will sleep no matter what you do in power manager.
Please Log in or Create an account to join the conversation.
03 Aug 2015 04:19 #61130
by AggieMEEN
I executed xset -dpms and ran an instance of GLXGears while leaving the browser window open. Left it for 15 minutes and came back to a Max Jitter (ns) on the Base thread of 22996. Just before that it was 18567.
Is there anything else I can try beyond what I've listed, and the xset -dpms command, or is a score of ~23000 to be expected from a computer like mine under GLXGears and browsing?
Replied by AggieMEEN on topic Intel D525MW BIOS Configuration, Jitter Results
My recent experience is that the display will sleep no matter what you do in power manager.
I executed xset -dpms and ran an instance of GLXGears while leaving the browser window open. Left it for 15 minutes and came back to a Max Jitter (ns) on the Base thread of 22996. Just before that it was 18567.
Is there anything else I can try beyond what I've listed, and the xset -dpms command, or is a score of ~23000 to be expected from a computer like mine under GLXGears and browsing?
Please Log in or Create an account to join the conversation.
03 Aug 2015 04:51 #61131
by andypugh
If 23,000 is as bad as it ever gets then I would call say that's good enough and move on to making parts.
Replied by andypugh on topic Intel D525MW BIOS Configuration, Jitter Results
[Is there anything else I can try beyond what I've listed, and the xset -dpms command, or is a score of ~23000 to be expected from a computer like mine under GLXGears and browsing?
If 23,000 is as bad as it ever gets then I would call say that's good enough and move on to making parts.
Please Log in or Create an account to join the conversation.
04 Aug 2015 06:52 #61151
by SrLange61
Replied by SrLange61 on topic Intel D525MW BIOS Configuration, Jitter Results
I've been testing my D525mw as a TVHeadend server recently and its just not powerful enough.
I tried my LinuxCNC ssd in it from my Q1900B-itx CNC Box and got the same high latency results as you.
I think its time to bin my D525. and move on.
The Q1900 board tops out at 23000 latency running a full screen Youtube and it does it well.
srl
I tried my LinuxCNC ssd in it from my Q1900B-itx CNC Box and got the same high latency results as you.
I think its time to bin my D525. and move on.
The Q1900 board tops out at 23000 latency running a full screen Youtube and it does it well.
srl
Please Log in or Create an account to join the conversation.
Time to create page: 0.092 seconds