1 |
Nuitari <nuitari@××××××××××××××××.net> posted |
2 |
Pine.LNX.4.64.0705020412370.22431@××××××××××××××××.net, excerpted below, |
3 |
on Wed, 02 May 2007 04:14:11 -0400: |
4 |
|
5 |
> Hi, |
6 |
> |
7 |
> I'd like to know if anyone has any experience with either the Tyan |
8 |
> Thunder n3600M (S2932) or the |
9 |
> Tyan Thunder n2000M (S3992) motherboards and, of course, gentoo amd64. |
10 |
> |
11 |
> The CPUs would be 2 2xxx opterons. |
12 |
|
13 |
No experience with those models, but Tyan is generally one of the better |
14 |
mobo manufacturers in terms of Linux support. I'm running a now older |
15 |
s2885 (dual Opteron 2xx, currently 242s, soon to be upgraded to 290s -- |
16 |
Tyan was very good about upgrading their BIOS to support the dual-cores |
17 |
when they came out as well), and have been VERY well pleased with their |
18 |
Linux support. |
19 |
|
20 |
Among other things, they had a pre-configured lm_sensors.conf file for my |
21 |
board, and it's certified for several Linux distributions. (These |
22 |
appeared in the shipped manual, not in the PDF on the Tyan site, so while |
23 |
I knew it worked with Linux from my research, the actual certifications |
24 |
were a surprise to me. I'm guessing that's what the blank page in the |
25 |
manual on the site will have in the shipped version, if they've gotten |
26 |
them.) Tyan also has Linux FAQ sheets both in general and for specific |
27 |
boards. Sometimes they have Linux drivers too, but those are usually the |
28 |
proprietaryware versions of stuff like RAID drivers, and I'd recommend |
29 |
using the native Freedomware Linux drivers instead, as it's generally |
30 |
possible to do so. (I'm running kernel software RAID here, rather than |
31 |
their BIOS/software solution, and the kernel solution is more flexible, |
32 |
more widely tested, and more portable should the hardware I'm on fail, |
33 |
all three.) |
34 |
|
35 |
One of the the reasons I decided to go with Tyan was that at the time I |
36 |
bought the board (back in late 2003), there were only a very few makers |
37 |
of dual Opteron boards, four I think listed on pricewatch.com. The two |
38 |
that ended up on my short list were MSI and Tyan. Where the Tyan site |
39 |
used standard (and therefore Linux viewable) PDFs for its manuals and |
40 |
some other documentation, and standard zip files to ship its BIOS and |
41 |
other utilities, all I could find on the MSI site were *.exe files |
42 |
(probably self-extracting zips, but there was no way to tell for sure |
43 |
without further investigation and it was enough to get me to drop them |
44 |
from consideration). I emailed MSI too, telling them exactly why I |
45 |
dropped them from consideration. Apparently, it and perhaps the emails |
46 |
of others had an effect, as people have reported since that MSI's |
47 |
downloads are now standard format as well, no more *.exe. Anyway, I made |
48 |
the right decision, as Gentoo/amd64 MSI users have seen more problems (as |
49 |
reported here) than Tyan users, which have seemed pretty happy with their |
50 |
purchases, from what has been reported here anyway. |
51 |
|
52 |
At the time I got mine, as I said in late 2003, Tyan's BIOS flash |
53 |
instructions still required MSDOS, but I wrote to them inquiring about |
54 |
using FreeDOS (since I didn't have an MSDOS around by that point, having |
55 |
dumped everything MS). They were generally helpful and supportive but |
56 |
couldn't verify whether FreeDOS would work for flashing or not. It did, |
57 |
and I've used it the couple times since when I needed/wanted to do a |
58 |
BIOS flash. I wrote back verifying that FreeDOS did indeed work, and |
59 |
suggested that they do what I think it's ASUS does, actually ship |
60 |
complete FreeDOS boot images for the purpose of flashing, so folks didn't |
61 |
have to create their own FLASH images, possibly screwing up the image by |
62 |
loading memory managers and the like. I've not checked to see if they |
63 |
have or not, but it'd be nice to see them at least mention in their Linux |
64 |
FAQ that some users have reported that FreeDOS works for flashing, as an |
65 |
alternative to using MSDOS. They could do that without testing and |
66 |
without liability. Anyway, if it's not in their instructions, yes, |
67 |
FreeDOS for BIOS flashing seems to work quite well. =8^) |
68 |
|
69 |
The single problem I did have wasn't specifically Linux related at all. |
70 |
Their original BIOS didn't support the memory speed adjustments that I |
71 |
had grown accustomed to (and apparently erroneously thought were more or |
72 |
less standard in moderm BIOSs), and for some time I was running generic |
73 |
memory that really wasn't stable at its rated PC3200 (200 MHz DDR to |
74 |
400). A later BIOS update DID include memory speed limiting, and I |
75 |
lowered my memory the speed a single notch, to 183 MHz (DDR to 366, |
76 |
PC3000). My system was rock-stable after that, as a Linux-system should |
77 |
be. At the slower nominal speed I was in fact able to tweak the |
78 |
individual memory latency settings beyond factory settings to bring speed |
79 |
back up somewhat. It just couldn't handle the full 200/400/PC3200 |
80 |
nominal speeds. Later I upgraded the memory (now running 8 gig) , and |
81 |
the new memory didn't have the issue at all; I was able to reset the BIOS |
82 |
to full speed. It was thus in fact a problem with the generic memory I |
83 |
had been running, not of the board, but having the additional memory |
84 |
speed tweaking options in the earlier BIOSs would have saved me a LOT of |
85 |
grief. They eventually got them, but I sure could have used them earlier! |
86 |
|
87 |
As I said, however, that wasn't Linux related at all. On the Linux |
88 |
front, as I've seen the various issues folks have reported here, I've |
89 |
been VERY glad I bought Tyan, and in fact I'm likely to stay with them in |
90 |
the future, precisely because they DO support Linux, and quite well, at |
91 |
that. |
92 |
|
93 |
One thing you may wish to research a bit is the integrated video. You |
94 |
don't mention your intended usage. If your purpose is a no-X server, the |
95 |
integrated PCI-only ATI es1000 should be fine. If you will be running X, |
96 |
it appears the xorg-native radeon driver supports it, based on the output |
97 |
in my xorg log (I've a Radeon 9200 AGP, my board didn't have integrated |
98 |
video). I've no idea what the proprietary ATI fglx driver supports as I |
99 |
won't run proprietary drivers. However, it may be that you'll simply |
100 |
disable it and install a PCI-E card if you want 3D accelerated X. |
101 |
Anyway, it's likely you are simply running it as a no-X (or even |
102 |
headless, serial console redirect) server, in which case the integrated |
103 |
video should be fine. |
104 |
|
105 |
-- |
106 |
Duncan - List replies preferred. No HTML msgs. |
107 |
"Every nonfree program has a lord, a master -- |
108 |
and if you use the program, he is your master." Richard Stallman |
109 |
|
110 |
-- |
111 |
gentoo-amd64@g.o mailing list |