pfSense Support Subscription

Author Topic: Unofficial QOTOM Hardware Topic  (Read 33371 times)

0 Members and 1 Guest are viewing this topic.

nivek1612

  • Guest
Re: Unofficial QOTOM Hardware Topic
« Reply #645 on: January 02, 2018, 05:26:28 pm »
As others have said once you get the Qotom take it apart and remove the thermal paste they used, its poor quality, thin and poorly applied

Then apply a good quality paste and reassemble

Avg Core Temp
- Before 60 deg C
- After 37 deg C


Offline johnkeates

  • Hero Member
  • *****
  • Posts: 739
  • Karma: +57/-1
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #646 on: January 02, 2018, 05:28:08 pm »
As others have said once you get the Qotom take it apart and remove the thermal paste they used, its poor quality, thin and poorly applied

Then apply a good quality paste and reassemble

Avg Core Temp
- Before 60 deg C
- After 37 deg C

That is one hell of a difference. I have seen 47 to 37 etc, but 60 by default, that's real bad in comparison (but the mobile CPU doesn't really care, it rides all the way up to 100+).

Offline Failover

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #647 on: January 03, 2018, 08:25:13 am »
Not sure what temps before but 37oc now


I used MX4 paste

Offline Koenig

  • Jr. Member
  • **
  • Posts: 65
  • Karma: +1/-0
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #648 on: January 03, 2018, 09:07:41 am »
Why is mine stuck at 27.9 C , anyone know?

Offline johnkeates

  • Hero Member
  • *****
  • Posts: 739
  • Karma: +57/-1
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #649 on: January 03, 2018, 10:41:04 am »
Why is mine stuck at 27.9 C , anyone know?

You are using the wrong sensor. Add the temperature widget and also select the sensor in the settings.

Offline noons

  • Newbie
  • *
  • Posts: 23
  • Karma: +0/-0
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #650 on: January 04, 2018, 08:24:23 am »
I ended up picking a Q355G4 - i5 up for another usecase and I decided to load pfsense on it to compare it to my j1900 and SG-3100. It performs about equal to my j1900 when running speed tests (which is unfortunate - I was expecting a pretty nice boost) and it actually performs a little worse then the sg-3100. Downloading uses about 40-50% interrupt @936 Mb/s and 60-70% on upload speed tests @870Mb/s. This means this box is not capable at handling full duplex 1Gb/1Gb connections (something to keep in mind). I had virtually no rulesets loaded (other then defaults) and no plugins. For those looking to potentially handle 1Gb/s I think the SG-3100 is probably the better buy.

Offline johnkeates

  • Hero Member
  • *****
  • Posts: 739
  • Karma: +57/-1
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #651 on: January 04, 2018, 10:04:44 am »
I ended up picking a Q355G4 - i5 up for another usecase and I decided to load pfsense on it to compare it to my j1900 and SG-3100. It performs about equal to my j1900 when running speed tests (which is unfortunate - I was expecting a pretty nice boost) and it actually performs a little worse then the sg-3100. Downloading uses about 40-50% interrupt @936 Mb/s and 60-70% on upload speed tests @870Mb/s. This means this box is not capable at handling full duplex 1Gb/1Gb connections (something to keep in mind). I had virtually no rulesets loaded (other then defaults) and no plugins. For those looking to potentially handle 1Gb/s I think the SG-3100 is probably the better buy.

2x1GB FD here, works fine. Not sure about your settings...

Offline noons

  • Newbie
  • *
  • Posts: 23
  • Karma: +0/-0
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #652 on: January 04, 2018, 10:39:03 am »
I ended up picking a Q355G4 - i5 up for another usecase and I decided to load pfsense on it to compare it to my j1900 and SG-3100. It performs about equal to my j1900 when running speed tests (which is unfortunate - I was expecting a pretty nice boost) and it actually performs a little worse then the sg-3100. Downloading uses about 40-50% interrupt @936 Mb/s and 60-70% on upload speed tests @870Mb/s. This means this box is not capable at handling full duplex 1Gb/1Gb connections (something to keep in mind). I had virtually no rulesets loaded (other then defaults) and no plugins. For those looking to potentially handle 1Gb/s I think the SG-3100 is probably the better buy.

2x1GB FD here, works fine. Not sure about your settings...

Define fine? My primary concern is routing and NAT performance and not having a bottleneck, unlike others I am not using openvpn or ipsec. What does your CPU load look like when running a speed test and what speeds are you getting during that test?

As far as settings on my end nothing crazy. I did a fresh install, disabled eee, and enabled powerD high adaptive. This is a DHCP wan connection so PPPoE isn't even involved..

Offline johnkeates

  • Hero Member
  • *****
  • Posts: 739
  • Karma: +57/-1
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #653 on: January 04, 2018, 11:07:33 am »
I ended up picking a Q355G4 - i5 up for another usecase and I decided to load pfsense on it to compare it to my j1900 and SG-3100. It performs about equal to my j1900 when running speed tests (which is unfortunate - I was expecting a pretty nice boost) and it actually performs a little worse then the sg-3100. Downloading uses about 40-50% interrupt @936 Mb/s and 60-70% on upload speed tests @870Mb/s. This means this box is not capable at handling full duplex 1Gb/1Gb connections (something to keep in mind). I had virtually no rulesets loaded (other then defaults) and no plugins. For those looking to potentially handle 1Gb/s I think the SG-3100 is probably the better buy.

2x1GB FD here, works fine. Not sure about your settings...

Define fine? My primary concern is routing and NAT performance and not having a bottleneck, unlike others I am not using openvpn or ipsec. What does your CPU load look like when running a speed test and what speeds are you getting during that test?

As far as settings on my end nothing crazy. I did a fresh install, disabled eee, and enabled powerD high adaptive. This is a DHCP wan connection so PPPoE isn't even involved..

I have one of those i3 models doing routing between a few VLANs on a LAGG, works fine at 980Mbit/s bi-directional according to iperf. End-users have no lag either, it's mostly used for media uploads and downloads between mobile teams, they get about 90MB/s on SMB transfers. CPU is about 12% during usage.
« Last Edit: January 04, 2018, 11:52:25 am by johnkeates »

Offline Koenig

  • Jr. Member
  • **
  • Posts: 65
  • Karma: +1/-0
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #654 on: January 04, 2018, 12:47:15 pm »
Why is mine stuck at 27.9 C , anyone know?

You are using the wrong sensor. Add the temperature widget and also select the sensor in the settings.

Thing is I've switched back and forth a couple of times, since it seems too work for a while andthen get stuck.
But I havn't had the widget, I've only looked at the "System information"

Right now I switched it from AMD to Intel and added the widget.

Hope works.

I thank you!

Offline johnkeates

  • Hero Member
  • *****
  • Posts: 739
  • Karma: +57/-1
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #655 on: January 04, 2018, 03:55:00 pm »
It should look something like this:

Offline noons

  • Newbie
  • *
  • Posts: 23
  • Karma: +0/-0
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #656 on: January 04, 2018, 04:32:48 pm »
I ended up picking a Q355G4 - i5 up for another usecase and I decided to load pfsense on it to compare it to my j1900 and SG-3100. It performs about equal to my j1900 when running speed tests (which is unfortunate - I was expecting a pretty nice boost) and it actually performs a little worse then the sg-3100. Downloading uses about 40-50% interrupt @936 Mb/s and 60-70% on upload speed tests @870Mb/s. This means this box is not capable at handling full duplex 1Gb/1Gb connections (something to keep in mind). I had virtually no rulesets loaded (other then defaults) and no plugins. For those looking to potentially handle 1Gb/s I think the SG-3100 is probably the better buy.

2x1GB FD here, works fine. Not sure about your settings...

Define fine? My primary concern is routing and NAT performance and not having a bottleneck, unlike others I am not using openvpn or ipsec. What does your CPU load look like when running a speed test and what speeds are you getting during that test?

As far as settings on my end nothing crazy. I did a fresh install, disabled eee, and enabled powerD high adaptive. This is a DHCP wan connection so PPPoE isn't even involved..

I have one of those i3 models doing routing between a few VLANs on a LAGG, works fine at 980Mbit/s bi-directional according to iperf. End-users have no lag either, it's mostly used for media uploads and downloads between mobile teams, they get about 90MB/s on SMB transfers. CPU is about 12% during usage.

I am not seeing any lag or dropped packets or anything that would suggest an issue, just high interrupt to suggest either something performance wise with the 211's (doubtful unless they're fakes) or the cpu just can't keep up as well as I would have liked. Like I said for the supposed improvement from the j1900 I would expect to be < 40% and that is being generous.

was that 12% during your iperf or 90MB/s smb? Mind running an external speedtest from speedtest.net and checking on your cpu?

nivek1612

  • Guest
Re: Unofficial QOTOM Hardware Topic
« Reply #657 on: January 04, 2018, 05:18:48 pm »
It should look something like this:

or this if your a little naughty :-)

Offline drewstreib

  • Newbie
  • *
  • Posts: 4
  • Karma: +0/-0
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #658 on: January 04, 2018, 09:41:50 pm »
It should look something like this:

I *just* received a Q355G4 a couple days ago, and got the 5250U cpu option on the build. I'm idling at ~52-58C, and running stress-ng on all cpus lands me at ~77-80C.

I normally don't think this is horribly out of range for modern cpus, but I'm jealous of numbers in the 30's or low 40's for idle. I'm a bit curious as to what you see at full cpu burn.

Looking at some pics online of a different Qotom box, it looks like the cpu on the backside of the board (on top by the fins as the machine is right side up) is basically just kind of mounted directly to the top/sink along with the board. Is this the case with the Q355 as well? Normally I'm pretty sparing with thermal paste, but the tolerances seem pretty loose for this arrangement, at least compared to a heat sink designed to clip right over the cpu housing. Anything folks have found that needs any particular trick/advice? Just a small pea sized dollop of arctic silver on the cpu, spread around and squished against the case/sink?

Edit: Any Qotom folks watching this thread? Is there a reason it seems to be generally accepted that these machines need to always be fixed with 10c of thermal paste? It seems odd that there would be such an obvious manufacturing flaw.
« Last Edit: January 04, 2018, 10:00:03 pm by drewstreib »

Offline johnkeates

  • Hero Member
  • *****
  • Posts: 739
  • Karma: +57/-1
    • View Profile
Re: Unofficial QOTOM Hardware Topic
« Reply #659 on: January 05, 2018, 09:09:54 am »
It should look something like this:

I *just* received a Q355G4 a couple days ago, and got the 5250U cpu option on the build. I'm idling at ~52-58C, and running stress-ng on all cpus lands me at ~77-80C.

I normally don't think this is horribly out of range for modern cpus, but I'm jealous of numbers in the 30's or low 40's for idle. I'm a bit curious as to what you see at full cpu burn.

Looking at some pics online of a different Qotom box, it looks like the cpu on the backside of the board (on top by the fins as the machine is right side up) is basically just kind of mounted directly to the top/sink along with the board. Is this the case with the Q355 as well? Normally I'm pretty sparing with thermal paste, but the tolerances seem pretty loose for this arrangement, at least compared to a heat sink designed to clip right over the cpu housing. Anything folks have found that needs any particular trick/advice? Just a small pea sized dollop of arctic silver on the cpu, spread around and squished against the case/sink?

Edit: Any Qotom folks watching this thread? Is there a reason it seems to be generally accepted that these machines need to always be fixed with 10c of thermal paste? It seems odd that there would be such an obvious manufacturing flaw.

I don't think there are Asian Qotom people here. Regarding the thermals: it's well within spec, and they aim for 'good but cheap' at Qotom it seems, so it doesn't surprise me the thermals aren't optimal.