Re: Intel D945GSE vs Zotac ION ITX (was: Support for Zotac MB with nVidia ION chipset)
Jeremie Le Hen
--->> Cc: me when replying, as I'm not subscribed. <<---
I cross-post this reply to freebsd-hardware@ since the result of my
little study around Atom-based motherboard may be of interest for
readers of this ML too.
On Sun, Apr 04, 2010 at 03:00:09PM -0600, Tim Judd wrote:
> On 4/4/10, Jeremie Le Hen <jer...@le-hen.org> wrote:
> > Hi,
> >
> > --->> Please Cc: me when replying, I'm not subscribed. <<---
> >
> > I plan to purchase a Zotac motherboard with a embedded ATOM processor.
> > It uses an NVidia chipset.
> >
> > http://www.zotacusa.com/zotac-ionitx-f-e-atom-n330-1-6ghz-dual-core-mini-itx-intel-motherboard.html
> >
> > My intent is to build a small NAS with ZFS and NFS/CIFS. I'd like to
> > know if anyone successfully ran FreeBSD on this motherboard and what
> > performance could be achieved, especially if ZFS is used. I checked the
> > archives without luck.
>
> A NAS w/ ZFS, NFS and CIFS/SMB, doesn't need any feature of the ion
> chipset. Why are you electing for this board if you're not running
> any graphical environment?
>
> And ZFS is memory hungry, the Atom is a i386-like chip, so you'd have
> too much overhead with ZFS.
>
> I think you've elected the wrong board for your purposes. Will
> FreeBSD run on it? yes. I have freebsd on another atom N-series ASUS
> box.
Yeah, you are right. I should have mentionned that I do not want
necessarily a high-performance NAS, it's for home use so my premary
concern is the low power consumption. This is why I want an Atom-based
motherboard. By the way, I found an post on OpenSolaris forums where
the author achieves something like 35MB/s on a ZFS filesystem through
CIFS using an Intel Atom-based motherboard [1]. This is enough for the
use I intend to have.
Zotac mobo is better than Intel D954GSE because it provides a wireless
interface - although I couldn't figure out which chipset yet, so I don't
know if it's corretly supported on FreeBSD - and three S-ATA connectors.
I'm still not sure about which motherboard to buy actually. After some
additional reading, my leaning seems to go towards Intel's one as it is
less expensive and consumes half the power of the Zotac's one (13W with
a HDD [2] vs. 25W [3]). I can live with two S-ATA connectors and I can
plug a wireless interface on the available PCI connector if I ever need
it.
Regards,
[1] http://mail.opensolaris.org/pipermail/opensolaris-discuss/2009-June/048214.html
[2] http://www.homeserverhacks.com/2009/06/hands-on-whs-build-with-intel-d945gsejt.html
[3] http://www.anandtech.com/show/2765/12
--
Jeremie Le Hen
Humans are born free and equal. But some are more equal than others.
Coluche
Vincent Hoffman
> Hi,
>
> --->> Cc: me when replying, as I'm not subscribed. <<---
>
> I cross-post this reply to freebsd-hardware@ since the result of my
> little study around Atom-based motherboard may be of interest for
> readers of this ML too.
>
> On Sun, Apr 04, 2010 at 03:00:09PM -0600, Tim Judd wrote:
>
>> On 4/4/10, Jeremie Le Hen <jer...@le-hen.org> wrote:
>>
>>> Hi,
>>>
>>> --->> Please Cc: me when replying, I'm not subscribed. <<---
>>>
>>> I plan to purchase a Zotac motherboard with a embedded ATOM processor.
>>> It uses an NVidia chipset.
>>>
>>> http://www.zotacusa.com/zotac-ionitx-f-e-atom-n330-1-6ghz-dual-core-mini-itx-intel-motherboard.html
>>>
>>> My intent is to build a small NAS with ZFS and NFS/CIFS. I'd like to
>>> know if anyone successfully ran FreeBSD on this motherboard and what
>>> performance could be achieved, especially if ZFS is used. I checked the
>>> archives without luck.
>>>
>> A NAS w/ ZFS, NFS and CIFS/SMB, doesn't need any feature of the ion
>> chipset. Why are you electing for this board if you're not running
>> any graphical environment?
>>
>> And ZFS is memory hungry, the Atom is a i386-like chip, so you'd have
>> too much overhead with ZFS.
>>
Actually I run a zotac atom board, forget the model just of hand but can
look it up tomorrow if you like, Its amd64 so zfs should be viable, I
dont run it though.
CPU: Intel(R) Atom(TM) CPU 330 @ 1.60GHz (1601.61-MHz K8-class CPU)
Origin = "GenuineIntel" Id = 0x106c2 Stepping = 2
Features=0xbfe9fbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,CLFLUSH,DTS,ACPI,MMX,FXSR,SSE,SSE2,SS,HTT,TM,PBE>
Features2=0x40e31d<SSE3,DTES64,MON,DS_CPL,TM2,SSSE3,CX16,xTPR,PDCM,MOVBE>
AMD Features=0x20100800<SYSCALL,NX,LM>
AMD Features2=0x1<LAHF>
TSC: P-state invariant
real memory = 4294967296 (4096 MB)
avail memory = 3356155904 (3200 MB)
ACPI APIC Table: <050809 APIC1106>
FreeBSD/SMP: Multiprocessor System Detected: 4 CPUs
FreeBSD/SMP: 1 package(s) x 2 core(s) x 2 HTT threads
cpu0 (BSP): APIC ID: 0
cpu1 (AP/HT): APIC ID: 1
cpu2 (AP): APIC ID: 2
cpu3 (AP/HT): APIC ID: 3
>> I think you've elected the wrong board for your purposes. Will
>> FreeBSD run on it? yes. I have freebsd on another atom N-series ASUS
>> box.
>>
> Yeah, you are right. I should have mentionned that I do not want
> necessarily a high-performance NAS, it's for home use so my premary
> concern is the low power consumption. This is why I want an Atom-based
> motherboard. By the way, I found an post on OpenSolaris forums where
> the author achieves something like 35MB/s on a ZFS filesystem through
> CIFS using an Intel Atom-based motherboard [1]. This is enough for the
> use I intend to have.
>
> Zotac mobo is better than Intel D954GSE because it provides a wireless
> interface - although I couldn't figure out which chipset yet, so I don't
> know if it's corretly supported on FreeBSD - and three S-ATA connectors.
>
>
The zotac board I have is atheros
ath0: <Atheros 9280> mem 0xfebf0000-0xfebfffff irq 16 at device 0.0 on pci4
ath0: [ITHREAD]
ath0: AR9280 mac 128.2 RF5133 phy 13.0
which is working fine for me (8.0-RELEASE)
Hope that helps a little.
Vince
Peter Jeremy
>Yeah, you are right. I should have mentionned that I do not want
>necessarily a high-performance NAS, it's for home use so my premary
>concern is the low power consumption. This is why I want an Atom-based
>motherboard.
Unfortunately, FreeBSD has some issues with low memory handling that
make FreeBSD/i386 a bad choice for ZFS. You would be far better off
running amd64 with as much RAM as you can fit onto the board. (And
this is one case where you want amd64 even if you don't have >4GB RAM).
Note that this doesn't mean you can't use an Atom - some Atoms include
EM64T - you just need to check.
IMO, the biggest disadvantage of using an Atom in a ZFS NAS is the
lack of ECC support on the Atom. ZFS can detect bitflips in the
I/O sustem but you can still get screwed by a bitflip in RAM.
>I'm still not sure about which motherboard to buy actually. After some
>additional reading, my leaning seems to go towards Intel's one as it is
>less expensive and consumes half the power of the Zotac's one (13W with
>a HDD [2] vs. 25W [3]).
I'd recommend against buying anything with the Atom combined with a 945.
Whilst the Atom is low-power, the 945 isn't. That is also an older
motherboard using an older, superseded Atom. I suggest you look for
motherboards built around the new Pinetrail Atoms (which _do_ support
EM46T and hence can run amd64).
Supermicro make a number of potentially suitable boards:
http://www.supermicro.com/products/motherboard/ATOM/ICH9/X7SPA.cfm
this is pricier but supports remote management - other options at:
http://www.supermicro.com/products/motherboard/Atom/
If you want an Intel MB, search for BOXD510MO
Note that I'm not sure how well FreeBSD's X.org supports the Pinetrail
yet. There have been some commits but I don't know if support is
complete.
--
Peter Jeremy
Dan Naumov
free to completely ignore people saying that ZFS overhead is too much
for an Atom to handle efficiently, they have no idea what they are
talking about. I am using a Supermicro X7SPA-H board (Atom D510) and I
an easily achieving ~85mb/s transfers over Samba to and from the
machine. 85mb/s is also the best these drives will do and my CPU is
nowhere near maxed during these transfers, so with better disks I
would be easily saturating gigabit, while still having plenty of
available CPU time. What you want is a good disk controller and fast
and reliable disks, 2gb RAM is enough, but with 4gb ram you can
basically safely enable prefetch for a very noticable boost in
sequential pattern reads. Below are some numbers from my personal Atom
NAS system:
===================================================================
bonnie -s 8192
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
8192 29065 68.9 52027 39.8 39636 33.3 54057 95.4
105335 34.6 174.1 7.9
dd if=/dev/zero of=test1 bs=1M count=8192
8589934592 bytes transferred in 111.300481 secs (77177875 bytes/sec) (73,6mb/s)
dd if=/dev/urandom of=test2 bs=1M count=8192
dd if=test2 of=/dev/zero bs=1M
8589934592 bytes transferred in 76.031399 secs (112978779 bytes/sec)
(107,74mb/s)
===================================================================
This is a ZFS mirror of 2 x 2tb WD Green drives with 32mb cache with
the automatic headparking disabled via WDIDLE3. The drives are very
cheap and hence, are the bottleneck in my case.
- Sincerely,
Dan Naumov
Jeremie Le Hen
Thank you for your enlightening replies.
I think I will go for the Supermicro X7SPA-H. It seems a better
hardware although it consumes more power.
Nonetheless I'm a little worried by what you said about the lack of ECC.
Computers has been used for years before ECC came out and obviously they
worked :). Do you really think it might happen to be a problem? Would
an Intel board would compensate for this? Dan, have you ever
experienced weird problems that could be explained by bitflips?
For the records, I've found an interesting and very recent post about
someone running OpenSolaris on this Supermicro motherboard [1]. He uses
a thumbdrive for the operating system and with four drives connected
onto it, the whole system sucks 41 watts when idle (27 without any HDD,
which is twice as the Intel D945GSE, but I guess this is the price for
better performance). Now, it seems that OpenSolaris unfortunately has
some problems with this hardware, but according to Dan, FreeBSD runs
correctly on it.
Thanks again for your help.
Regards,
[1] http://sorenragsdale.livejournal.com/19875.html
Dan Naumov
> Nonetheless I'm a little worried by what you said about the lack of ECC.
> Computers has been used for years before ECC came out and obviously they
> worked :). �Do you really think it might happen to be a problem? �Would
> an Intel board would compensate for this? �Dan, have you ever
> experienced weird problems that could be explained by bitflips?
Personally, I haven't had any issues, but then again on the ZFS scale
of things, both my current pool size (2 TB) and projected pool size
when I add more disks (6 TB) is pretty small. If this was a heavily
used machine with a 10 TB pool or bigger, I would definately give
strong consideration to ECC.
> For the records, I've found an interesting and very recent post about
> someone running OpenSolaris on this Supermicro motherboard [1]. �He uses
> a thumbdrive for the operating system and with four drives connected
> onto it, the whole system sucks 41 watts when idle (27 without any HDD,
> which is twice as the Intel D945GSE
The power draw (from the wall) for the Supermicro X7SPA-H without any
disks attached is as following:
26W - During boot.
24W - IDLE at console
28W - Full load
This is with a 80+ rated Corsair 400CX PSU. Sadly, I did not have the
opportunity to measure the power draw with powerd enabled. The D945GSE
is unsuitable for use as a ZFS NAS due to it's severe feature
limitations when compared against the X7SPA-H, of biggest concern
would be the limitation of RAM, followed by the amount of native SATA
ports, followed by the fact that you only get a PCI-E x1 (both
physical formfactor and speed-wise) slot for expansion, while most
controller cards are either 4x or 8x, meaning they simply wouldn't
physically fit into the slot.
Singlecore 1,6Ghz Diamondville Atom VS Dualcore 1,66Ghz Pineview Atom
1 RAM socket supporting a max of 1GB VS 2 RAM sockets supporting a max
of 4GB (note that X7SPA-H uses SO-DIMMs, not regular DIMMs)
2 SATA ports vs 6 SATA ports
1 Realtec NIC vs 2 x Intel NIC
PCI-E x1 Slot VS PCI-E x4 Slot (in x16 form factor) for expansion
- Sincerely,
Dan Naumov
Peter Jeremy
>Nonetheless I'm a little worried by what you said about the lack of ECC.
>Computers has been used for years before ECC came out and obviously they
>worked :).
Not really. Most early computers had fairly extensive error detecting
hardware. Early microprocessors didn't because the novelty of getting
an entire on a CPU on a chip was enough. Most 486 based PCs supported
parity RAM but maufacturers and end-users found they could save pennies
by leaving the parity bits off.
ECC support was a requirement for building servers with microprocessors
and some support has trickled down to the desktop. It hasn't been
really popular because wider memory costs more and most people want the
fastest, cheapest system possible to make their games render faster.
Occasional glitches don't matter.
With the current generation of CPUs, Intel appear to have made a
marketing decision to not support ECC on their desktop CPUs - if you
want ECC, you need to user a server-grade CPU (with a much greater
profit margin). AMD have gone the other way and have have ECC support
in all their x64 chips except mobile ones. You are still at the mercy
of motherboard manufacturers who decide to not include the tracks
between the DIMM sockets and the CPU.
> Do you really think it might happen to be a problem?
There's no way to know. Definitely, the added error checking in ZFS
have resulted in a number of "ZFS kept reporting errors and I found I
actually had bad hardware even though I've been using it for years"
reports.
> Would an Intel board would compensate for this?
No. The memory controller is embedded in the Atom and doesn't support
ECC. If you decide to go the ECC path, you need to pick a different CPU.
--
Peter Jeremy
Robert Bonomi
> Date: Tue, 6 Apr 2010 07:34:05 +1000
> From: Peter Jeremy <peter...@acm.org>
> To: Jeremie Le Hen <jer...@le-hen.org>
> Cc: freebsd-...@freebsd.org, freebsd-...@freebsd.org
> Subject: Re: Intel D945GSE vs Zotac ION ITX (was: Support for Zotac MB with
> nVidia ION chipset)
>
>
> --n8g4imXOkfNTN/H1
> Content-Type: text/plain; charset=us-ascii
> Content-Disposition: inline
> Content-Transfer-Encoding: quoted-printable
>
> On 2010-Apr-05 12:20:12 +0200, Jeremie Le Hen <jer...@le-hen.org> wrote:
> >Nonetheless I'm a little worried by what you said about the lack of ECC.
> >Computers has been used for years before ECC came out and obviously they
> >worked :).
>
> Not really. Most early computers had fairly extensive error detecting
> hardware.
Depends on what machines you're talking about. One fairly well-known supe-r
computer class architecture from the mid 1960s ran without *any* error checking
in the CPU *or* main memory. Dr. Seymour Cray analyzed things and concluded the
significant extra component count for just doing 'parity' checking, let alone
ECC made for a net _reduction_ in overall system reliability, *IF* the machine
was run under very tightly controlled operating conditions -- the big ones being
extremely stable power and a very limited temperature range. So, he specified
the design to tight tolerances, and ran truely 'naked' hardward. Scary, but true.
And, it worked.
This was also a machine where, at any given moment, a fair part of the data in
the CPU was 'in the wires' ("in transit" from one part of the CPU to another),
and significant parts of the wiring harness had to be of _just_the_right_length_
(speed-of-light considerations) for the box to work.
Incidentally, this computer COULD NOT ADD two numbers together. Literally!!
It performed addition by 'complement and subtract'. Yeah, it -sounds- silly,
but there were valid architectural reasons for it.
per...@pluto.rain.com
> One fairly well-known super computer class architecture from the
> mid 1960s ran without *any* error checking in the CPU *or* main
> memory. Dr. Seymour Cray analyzed things and concluded the
> significant extra component count for just doing 'parity'
> checking, let alone ECC made for a net _reduction_ in overall
> system reliability, *IF* the machine was run under very tightly
> controlled operating conditions -- the big ones being extremely
> stable power and a very limited temperature range. So, he
> specified the design to tight tolerances, and ran truely 'naked'
> hardward. Scary, but true. And, it worked.
CDC-6600 and/or 7600, I presume?
The flaw in that reasoning is that, while an unchecked machine may
indeed be faster and/or have a somewhat better MTBF, the symptom
of a failure may well be silently incorrect results. If reliable
production results are what's valued, as opposed to time between
detected failures while running diagnostics*, a checked or corrected
design wins hands down.
> This was also a machine where, at any given moment, a fair part
> of the data in the CPU was 'in the wires' ("in transit" from one
> part of the CPU to another), and significant parts of the wiring
> harness had to be of _just_the_right_length_ (speed-of-light
> considerations) for the box to work.
Second- (or third?) hand war story from the manufacturing dept:
Occasionally the instructions would call for pin so-and-so to be
connected to pin thus-and-such with, say, a 6" wire -- when the
pins in question were 8" apart! The source of the story claimed
that the standard practice in such cases was to use the shortest
wire that would reach, and let the QA dept worry about the fallout.
* A diagnostic is a program that runs when the hardware is
malfunctioning -- R. F. Rosin.
Jeremie Le Hen
-->>> Please Cc: me when replyig, as I'm not subscribed. <<<--
I would finally like to buy a SuperMicro X7SPA-H motherboard.
Unfortunately it seems very difficult to find this product in France or
even in Europe. Either it is not in the catalog or not in stock.
I tried to purchase it from a US website, but shipping fees are very
high for Europe, around the half of the price of the product itself.
Does anyone know a website that would ship this product in France for a
reasonable price?
Thanks for your help.
Patrick Proniewski
> I would finally like to buy a SuperMicro X7SPA-H motherboard.
> Unfortunately it seems very difficult to find this product in France or
> even in Europe. Either it is not in the catalog or not in stock.
> I tried to purchase it from a US website, but shipping fees are very
> high for Europe, around the half of the price of the product itself.
>
> Does anyone know a website that would ship this product in France for a
> reasonable price?
did you check http://www.ahead-it.eu/ ?
patpro
Jeremie Le Hen
Thanks for you quick answer. Unfortunately, they have this board in
their catalog but they do not have it in stock as many other websites
I've found. I'm going to contact them nonetheless to know if they can
give me an estimation.
Regards,
Patrick Proniewski
On 06 avr. 2010, at 23:45, Jeremie Le Hen wrote:
>> did you check http://www.ahead-it.eu/ ?
>
> Thanks for you quick answer. Unfortunately, they have this board in
> their catalog but they do not have it in stock as many other websites
> I've found. I'm going to contact them nonetheless to know if they can
> give me an estimation.
Last time I asked for a 1U server they wouldn't have in stock, they
told me it could ship in 5-10 days. Not that bad ;)
patpro