1 |
> Hi, |
2 |
> |
3 |
> I'd like to know if anyone has any experience with either the |
4 |
> Tyan Thunder n3600M (S2932) or the |
5 |
> Tyan Thunder n2000M (S3992) motherboards and, of course, gentoo amd64. |
6 |
> |
7 |
> The CPUs would be 2 2xxx opterons. |
8 |
|
9 |
I've finally received a new S3992 based server yesterday. |
10 |
I've been hit by 4 issues and here they are: |
11 |
|
12 |
1. |
13 |
The one thing that was clearly lacking from Tyan and ThermalTake was that |
14 |
there are actually 2 different Heatsink sizes on the market for Socket F |
15 |
systems. Some heatsinks at 3.5" and others at 4.1". |
16 |
|
17 |
ThermalTake's Socket F heatsinks use 4.1", while the Tyan board uses 3.5". |
18 |
|
19 |
In the end I got 2 Socket 940 heatsinks using the special brackets from |
20 |
Tyan. |
21 |
|
22 |
After rereading the specifications, the only line that was a clue to is |
23 |
was "Uses the standard 3 1/2"/89mm socket connector" which is really not |
24 |
telling much. |
25 |
|
26 |
2. |
27 |
The second issue was that when I disable the "Hide XIOAPIC PCI functions" |
28 |
the linux kernel will only see 1 core of 1 cpu. My only clue to that being |
29 |
the problem was a post to the Linux Kernel List from someone that had the |
30 |
same problem but mentionned he couldn't remember what setting he used. |
31 |
|
32 |
3. |
33 |
If the fan fail led is enabled, then not connected fan are marked as |
34 |
failed and there is no way to tell it to ignore them. The bios will also |
35 |
give a message like "FAN5 RPM doesn't detected". |
36 |
|
37 |
|
38 |
4: |
39 |
lm_sensors require a special program to "switch" which set of fan and cpu |
40 |
is being monitored. I really wish that there is a way to avoid that. I |
41 |
like being able to parse the output of lm_sensors to hobbit (aka big |
42 |
brother) for monitoring purposes. |
43 |
-- |
44 |
gentoo-amd64@g.o mailing list |