Looking to upgrade motherboard, keep processor

[Yousuf Khan]

I got a pretty good AMD processor right now, a Phenom II 1100T, 6-core,
Socket AM3. It's doing a pretty good job for me right now, I don't feel
it's going to need to replaced anytime soon. However, I'm finding some
features of its current motherboard are starting to feel a little
draggy. It's an ASUS M4A785-M, which as you can tell from the model
number uses an AMD 785 chipset, which is about 2 generations old now.
It's got all of the basic connectors, such as SATA (3Gbps), USB (2.0),
but they are all a bit slower than the most modern ones. So what I'm
looking to upgrade to is something that'll take my existing processor,
but upgrade the speeds of the peripherals. Specifically, looking for
six+ SATA 6Gbps ports, 8 ports is preferred.

Also looking for the USB 3.0 ports, at least 2 of them, maybe all of
them, if possible. However, it seems many motherboards implement USB 3.0
through external chips, rather than through the chipset. I'd prefer that
it be implemented through the chipset.

I've found a motherboard, which seems to fill the bill, except it's a
little subpar on the nice-to-have department. It's got only six SATA
6Gbps ports rather than eight. It's also got USB 3.0 through an external
chip rather than through it's chipset. Also, I assume motherboards
designed for AM3+ will work with AM3 processors, right?

Amazon.com: GIGABYTE GA-970A-D3 AM3+ AMD 970 SATA 6Gb/s USB 3.0 ATX AMD
Motherboard: Electronics
http://www.amazon.com/GIGABYTE-GA-970A-D3-AM3-SATA-Motherboard/dp/B0056G10WK

Any critiques or other recommendations for other mobos?

Yousuf Khan

[Mr. Man-wai Chang]

On 26/05/2013 2:35 AM, Yousuf Khan wrote:
> Amazon.com: GIGABYTE GA-970A-D3 AM3+ AMD 970 SATA 6Gb/s USB 3.0 ATX AMD
> Motherboard: Electronics
> http://www.amazon.com/GIGABYTE-GA-970A-D3-AM3-SATA-Motherboard/dp/B0056G10WK

Make sure that you got the most recent revision of the board! Even Asus
is now playing this Rev. game... :)

--
@~@ Remain silent. Nothing from soldiers and magicians is real!
/ v \ Simplicity is Beauty! May the Force and farces be with you!
/( _ )\ (Fedora 18 i686) Linux 3.9.2-200.fc18.i686
^ ^ 02:54:01 up 6:33 0 users load average: 0.03 0.04 0.05
不借貸! 不詐騙! 不援交! 不打交! 不打劫! 不自殺! 請考慮綜援 (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa

[SC Tom]

"Yousuf Khan" <bbb...@spammenot.yahoo.com> wrote in message
news:51a10468$1...@news.bnb-lp.com...

According to
<http://www.gigabyte.us/support-downloads/cpu-support-popup.aspx?pid=3908>
your CPU is supported.

The specs <http://www.gigabyte.us/products/product-page.aspx?pid=3908#sp>
are pretty good. I could get used to it, and the price is decent :-) Plus,
you can have up to 14 USB2.0 ports available. That's a bunch! Wonder how the
power rail would hold up if they were all used?

Depending on what you plan on doing with it, I would think you'd be pleased
with it.
--
SC Tom

[Paul]

Do a search by CPU here.

http://support.asus.com/CPU.aspx?SLanguage=en

The purpose of doing this, is to establish an "epoch"
for your processor. Now, I run over to Newegg, and see
if any of this is still around.

Phenom II X6 1100T...

Crosshair III Formula
Crosshair IV Extreme
Crosshair IV Formula
Crosshair V Formula
Crosshair V Formula/ThunderBolt
Crosshair V Formula-Z
M3A32-MVP DELUXE
M3A32-MVP DELUXE/WIFI-AP
M3A78 Pro
M3A78-T
M3A79-T Deluxe
M3A-H/HDMI
M3N78 SE
M3N-H HDMI
M3N-HD HDMI
M3N-HT DELUXE
M3N-HT Deluxe/Mempipe
M4A77T
M4A77T/USB3
M4A77TD
M4A785G HTPC
M4A785G HTPC/RC
M4A785-M
M4A785TD-V EVO
M4A78-E
M4A78-E SE
M4A78LT LE
M4A78LT-M
M4A78T-E
M4A79 Deluxe
M4A79T Deluxe
M4A79T Deluxe/U3S6
M4A79XTD EVO
M4A87T
M4A87T PLUS
M4A87TD
M4A87TD EVO
M4A87TD/USB3
M4A88TD-M EVO/USB3
M4A88TD-V EVO
M4A88TD-V EVO/USB3
M4A88T-M
M4A88T-M LE
M4A88T-M/USB3
M4A88T-V EVO
M4A88T-V EVO/USB3
M4A89GTD PRO
M4A89GTD PRO/USB3
M4A89TD PRO
M4A89TD PRO/USB3
M4N68T LE V2
M4N68T V2
M4N68T-M
M4N68T-M LE V2
M4N68T-M V2
M4N72-E
M4N75TD
M4N78 Pro
M4N82 Deluxe
M4N98TD EVO
M5A78L
M5A78L LE
M5A78L/USB3
M5A78L-M LX PLUS
M5A78L-M LX V2
M5A78L-M/USB3
M5A87
M5A88-M
M5A88-M EVO
M5A88-V EVO
M5A97
M5A97 EVO
M5A97 EVO R2.0
M5A97 LE R2.0
M5A97 PRO
M5A97 R2.0
M5A99FX PRO R2.0
M5A99X EVO
M5A99X EVO R2.0
SABERTOOTH 990FX
SABERTOOTH 990FX R2.0
SABERTOOTH 990FX/GEN3 R2.0

Yup, got some hits right away.

http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=100007625%2050001315&IsNodeId=1&name=ASUS

Picking one at random.

"ASUS SABERTOOTH 990FX R2.0 AM3+ AMD 990FX"
http://www.newegg.com/Product/Product.aspx?Item=N82E16813131877

North Bridge
AMD 990FX

South Bridge
AMD SB950

Storage Devices

SATA 6Gb/s
8 x SATA 6Gb/s

SATA RAID
6 x SATA 6Gb/s port(s) (brown) Support Raid 0, 1, 5, 10

USB 1.1/2.0
8 x USB 2.0

USB 3.0
4 x USB 3.0

( Onboard USB 2 x USB 3.0 + 4 x USB 2.0 )

Now, I can use the Wikipedia article, to determine what is native
in all of that.

http://en.wikipedia.org/wiki/Comparison_of_AMD_chipsets

SB950 May 30, 2011 6 SATA3, 14+2 USB2

That tells me, the board must have a separate storage chip for
the additional two SATA ports. And since the SB950 has no USB3,
then one or two separate USB3 chips are being used.

*******

http://en.wikipedia.org/wiki/Comparison_of_AMD_chipsets

USB3 can be seen in the Fusion FCH section of that
Wikipedia web page. For example, the A75 chipset
has four internal USB3, which would give best integration
of USB3. But there's no board combining A75 with your
processor.

I'd just stick a PCI Express x1 USB3 card in your board,
as that would be less work.

( M4A785-M , your board has an x1 slot )
http://images10.newegg.com/NeweggImage/productimage/13-131-595-16.jpg

This is an example of a card. PEXUSB3S4V . You don't have to
get a quad port, this is just an example. A dual port would be
cheaper. (Duals with Etron chip on them, have had driver issues
in the past.)

http://www.newegg.com/Product/Product.aspx?Item=N82E16815158354

I picked that one, to show that there are quad port USB3 chips
available.

http://images17.newegg.com/is/image/newegg/15-158-354-Z03?$S640W$

Now, being designed to cover low-profile applications, that
Startech design has the connectors too close together. So that's
where it loses. The chip used, appears to be a NEC chip of some sort.

http://ca.startech.com/Cards-Adapters/USB-3.0/Cards/4-Port-PCI-Express-PCIe-SuperSpeed-USB-3-Controller-Card~PEXUSB3S4V

Chipset ID NEC uPD720201

And indeed, the '201 is a quad port. I think Asmedia might have a
quad port now as well.

http://documentation.renesas.com/doc/products/soc/r19pf0043ej0200_usb.pdf

*******

The time to go searching for SATA III ports, is when you own an SSD.
For other usage, SATA II is fine. Only an SSD, really benefits from
an un-constricted pipe. On hard drives, the platter prevents really
great performance (the current record being around 180MB/sec). And
180MB/sec can still be handled with SATA II.

If you do stick an SSD on a SATA II, it will still feel fast. It's
just if you "benchmark against your friends", it'll look slow. It's
still a great benefit, even if the ports are SATA II.

Paul

[Yousuf Khan]

On 25/05/2013 4:38 PM, SC Tom wrote:
> According to
> <http://www.gigabyte.us/support-downloads/cpu-support-popup.aspx?pid=3908>
> your CPU is supported.
>
> The specs
> <http://www.gigabyte.us/products/product-page.aspx?pid=3908#sp> are
> pretty good. I could get used to it, and the price is decent :-) Plus,
> you can have up to 14 USB2.0 ports available. That's a bunch! Wonder how
> the power rail would hold up if they were all used?

I gotta 750W PSU right now, so I'm not worried about that one.

> Depending on what you plan on doing with it, I would think you'd be
> pleased with it.

It's pretty equivalent in the number of ports as my existing
motherboard, except each of those ports are faster. USB 3.0 is
implemented through an Etron chip rather than through its chipset. That
one has me a little worried.

Yousuf Khan

[larrymo...@my-deja.com]

Do you want a UEFI BIOS instead of a regular BIOS?

[SC Tom]

"Yousuf Khan" <bbb...@spammenot.yahoo.com> wrote in message

news:51a42a21$1...@news.bnb-lp.com...

> On 25/05/2013 4:38 PM, SC Tom wrote:
>> According to
>> <http://www.gigabyte.us/support-downloads/cpu-support-popup.aspx?pid=3908>
>> your CPU is supported.
>>
>> The specs
>> <http://www.gigabyte.us/products/product-page.aspx?pid=3908#sp> are
>> pretty good. I could get used to it, and the price is decent :-) Plus,
>> you can have up to 14 USB2.0 ports available. That's a bunch! Wonder how
>> the power rail would hold up if they were all used?
>
> I gotta 750W PSU right now, so I'm not worried about that one.

I meant the power rail on the MB. That's where the major stress would be,
not on the PSU.

[Paul]

SC Tom wrote:
>
>
> "Yousuf Khan" <bbb...@spammenot.yahoo.com> wrote in message
> news:51a42a21$1...@news.bnb-lp.com...
>> On 25/05/2013 4:38 PM, SC Tom wrote:
>>> According to
>>> <http://www.gigabyte.us/support-downloads/cpu-support-popup.aspx?pid=3908>
>>>
>>> your CPU is supported.
>>>
>>> The specs
>>> <http://www.gigabyte.us/products/product-page.aspx?pid=3908#sp> are
>>> pretty good. I could get used to it, and the price is decent :-) Plus,
>>> you can have up to 14 USB2.0 ports available. That's a bunch! Wonder how
>>> the power rail would hold up if they were all used?
>>
>> I gotta 750W PSU right now, so I'm not worried about that one.
>
> I meant the power rail on the MB. That's where the major stress would
> be, not on the PSU.
>

<<snip>>

It could be stressed two ways. Motherboards now, tend to run the USB
ports off +5VSB. Which is a poor choice from the perspective of getting
enough amps. If you had 14 ports, and actually supported all of them,
that would be 14*500mA or 7 amps (this assumes every port draws as much
current as the standard would suggest is possible). And power supplies
tend to top out at around 5VSB @ 3A perhaps. So if there really were
14 ports, all in heavy usage, the supply would shut off on overcurrent.
(But you could easily run 14 optical mice, and the supply would stay up. :-) )

The motherboard itself, has ampacity limits on conductors. The average
board, over time, was probably four layers thick. A few manufacturers use
2 ounce copper, versus say 0.5 ounce that was more typical for signal
layers. And the heavy copper, means the tracks don't have to be as wide,
for a given max current flow. Tracks are sized according to target
delta_T (the motherboard heats up, over top of the track). There are
tables you can consult, for picking copper track thickness, for a
desired delta_T.

Near the main power connector, there is a lot of need to move amps of
current away from the connector. And not a lot of layers to do it with.
That's the puzzling part of motherboard design. How do they manage to
do that ?

So that's the motherboard stress part. Figuring out how to route
enough wide copper, so there isn't too much delta_T on any
individual conductor.

The motherboard has a Tg or glass transition temperature, which defines
how much heat the motherboard can take. And the Tg provides the
budget for allowing the motherboard to "get hot". The price of the
bare motherboard varies with the Tg of the material selected, and
with motherboards, you'd assume not the best material got used
(for pricing reasons).

If you scroll half way down this page, you can see such a curve for
sizing conductors in a PCB.

http://wiki.xtronics.com/index.php/Wire-Gauge_Ampacity

If you actually were to carry 7 amps in the motherboard (which the
power supply would not allow on +5VSB), and were willing to accept
a delta_T of 20C, then the track is about 0.060" wide. Which sounds
quite easy to do, assuming there are no other heavy loads on the
motherboard.

http://wiki.xtronics.com/images/7/74/PCB_current.png

All those ports would have to be fused as well, and there is a
Polyfuse per pair of ports, typically. When picking the distribution
track size, you'd assume one port goes shorted at a time, not all of
them :-) So you'd engineer for 12*500ma + approx 2A on the shorted
port pair (to open the Polyfuse), or about 8 amps. Which again,
is stopped by the power supply.

Given how stupid this design idea is (USB +5VSB-only operation), the
track never has to be sized for more than about 3 amps. Just enough,
to hit the supply limit, on the "best" power supply. So now the
track is down to around 0.030" wide, to carry +5VSB without significant
heating.

Now, if they'd allowed it to be powered from +5V, the end-users
would like that, but the track to distribute USB power ends up
being wider. And then, you can actually use all the ports, without
worrying about it (run 14 iPhone chargers or whatever).

Paul

[Flasherly]

GA-970A-UD3 (classic design)
Surprisingly seem over it, even for HDs, at roughly 1:2 for
dissatisfaction.

GA-970A-UD3 (VRM pwr design @1/5 more costlier)
...follows closer to 1:4

GA-990FXA-UD3
black ops same 4a gameboard top fashion garner

GA-990FXA-UD5
does 1:3

GA-78LMT-S2P
1:5

Closer to what I'm running (USB2) & probably whatever this superceded
GA-78LMT-USB3
1:7

(discounting possibly
GA-78LMT-S2
for 1:3)

Initial low price high value point. GA-78LMT-USB3

For acceptable (if not outdated) features first, then ratings which
all, mostly, have some degree of strike-outs over technilogical
challenges. Return policy non-factored - Newegg may be screwing you
or getting more particular these days about using your dollar to
return it, I do indeedy believe - though, I'd personally want no lower
than 1: (out of) 4 dissatisfied. And you can discount that, too, say
by Gigabyte over an ASUS. (From my biasis. Haven't done any MSI
recently, though may be interesting things that way.

I'd also, as it were if me, want to know more about the non-classic
version you've submitted. It's drawing down twice yours at bulk
interests and may have subsequently have attained some special status
ratings;- apart from any horseshit, lower classic satisfaction is
sufficient reinforcement stimuli to being well-informed before even
hitting the hardware sites for bitting into it for wires dripping
between your teeth. Reason enough to get by lazily -eh- with a
proven/settled GA-78LMT-USB3 for half-priced hassles. Maybe. Whatever
stokes up the fire for you, my man.)

[Yousuf Khan]

On 28/05/2013 1:33 AM, larrymo...@my-deja.com wrote:
> Do you want a UEFI BIOS instead of a regular BIOS?

Doesn't matter to me, probably it would be better to go with UEFI to
stay with modern trends.

Yousuf Khan

[Yousuf Khan]

Anybody know what the quality of an Etron USB 3.0 chip is? Any known
issues? Most motherboards I've looked at so far don't seem to have
native USB3 support, they need it through some secondary chip.

Yousuf Khan

[Paul]

Do a Google check for driver issues.

I haven't traced down, whether current drivers fix all issues,
or whether there is a hardware component to the problems.

But some Googling should clarify.

Paul

[Yousuf Khan]

On 29/05/2013 1:43 PM, Paul wrote:
> Do a Google check for driver issues.
>
> I haven't traced down, whether current drivers fix all issues,
> or whether there is a hardware component to the problems.
>
> But some Googling should clarify.

I did look it up before, but my only problem is that it didn't look like
there was much details on what the problem was. Here's an example from
the Gigabyte forums:

new ETRON USB 3.0 drivers
http://forum.giga-byte.co.uk/index.php?topic=8243.0

Yousuf Khan

[Yousuf Khan]

On 29/05/2013 1:43 PM, Paul wrote:

> Do a Google check for driver issues.
>
> I haven't traced down, whether current drivers fix all issues,
> or whether there is a hardware component to the problems.

Looks like with AMD, AMD only included native USB 3.0 support with their
A75 chipset for Llano (FM1) and possibly Trinity and later (FM2), but
not for any of their AM3+ platforms. I don't think Intel still has
native USB 3.0 support either.

http://www.tomshardware.com/reviews/usb-3.0-controller-performance,3024-2.html

The external chips available are from Etron, ASMedia, and Renesas/NEC.
They seem to get connected via a single PCI-e/x1 lane. Is an x1 lane
good enough to supply two USB3.0 ports?

I'm really leaning towards this motherboard now:

GIGABYTE - Motherboard - Socket AM3+ - GA-990FXA-UD3 (rev. 3.0)
http://www.gigabyte.com/products/product-page.aspx?pid=4397#ov

It has my required number of SATA 6.0G ports (8 of them, 6 via internal
chipset, 2 via Jmicron chip, and even 2 more with eSATA). Unlike its
bigger brothers the UD5 and the UD7, this one doesn't go crazy on
SLI/Crossfire support, as I'll never use that feature, but I do need the
USB and SATA ports. It keeps to a simple two-way SLI/Xfire support, thus
keeping the cost down. The UD5 has 3-way support, while the UD7 has
4-way support! Nuts.

It has quite large number of USB 2.0 ports, all of which are powered on
standby. The only thing that's concerning me now is the USB 3.0 ports.
I've had major problems with unknown chipsets from unknown companies
before, and their lack of driver support.

Because of the Etron chips, I was just about set to go and look at the
Asus Sabertooth again, but that one has another unknown chipset,
ASMedia. It's like rolling the dice now, on which unknown chipset I'm
going to be going with.

Yousuf Khan

[Paul]

Only with the integrated chipsets, do you stand a chance of not
having a bottleneck in the path. But for all practical purposes,
for most jobs, the run of the mill add-on is good enough. Whether
using a 250MB/sec x1 lane or a 500MB/sec x1 lane, a 200MB/sec USB
disk enclosure is not going to be too badly treated. If you expect
to put a SATA III SSD in a USB3, then you'd want to select
that slot with more care.

On a two port USB3, the assumption would be that one device
is reading and the other writing. Rather than both reading
simultaneously, or both writing simultaneously. Since PCI Express
has full duplex wiring (x1 read, x1 write, separate wires), that
helps the two port pattern be less restrictive. (See, I just
waved my hands like a marketeer :-) )

One solution to this I've seen, is an x4 switching chip and four
separate USB3 chips. On each chip, only one USB3 port is used.
This is the "Cadillac" approach to USB3. The card has been
discontinued. This will still be a few megabytes per second
behind an integrated chipset. Still, I'm impressed with
someone going to this much trouble. (It seems we just cannot
expect any company making chips, to put an x4 interface right
on the USB3 chip. That's too much to ask. It always has to
suck a little bit.)

http://www.newegg.com/Product/Product.aspx?Item=N82E16816115104

The U3S6 came before that card, and used a similar approach.
The switch chip gives a 500MB/sec lane to the USB3 chip,
even when the motherboard doesn't have one of those lanes.
Four 250MB/sec lanes, leaves room to make two 500MB/sec x1 lanes
for the chips. These were a bargain at $25 each, when they came out.
Still not entirely complaint free, as a product.

http://www.newegg.com/Product/Product.aspx?Item=N82E16813995004

Intel has some chipsets with integrated USB3.
Starting with Z77, from Apr 2012.

http://en.wikipedia.org/wiki/Intel_chipsets

On AMD, it is Fusion chipsets like A75, that have it. There
was some news item, about AMD planning on buying an IP logic
block for USB3, rather than designing it themselves. Which
may account for the limited distribution to date.

http://en.wikipedia.org/wiki/AMD_chipsets#Fusion_controller_hubs_.28FCH.29

As for the "no-name" nature of the add-on USB chips, you
should be able to find out if they had driver problems.
I could find references to Etron having a few problems, but
didn't notice the others. Some equipment combinations may have
the odd problem. NEC was the first to market with a USB3 host,
just like NEC was first to market with USB2. I gather that
was a point of pride with them, beating the other idiots who
had a head start. Intels strategy, leaves me scratching my head.
They held back on purpose, on releasing USB3. Maybe it had
something to do with writing the USB3 spec, and not being
seen to profit from the advantage that gives.

HTH,
Paul

[Yousuf Khan]

On 30/05/2013 9:16 PM, Paul wrote:
> Only with the integrated chipsets, do you stand a chance of not
> having a bottleneck in the path. But for all practical purposes, for
> most jobs, the run of the mill add-on is good enough. Whether using a
> 250MB/sec x1 lane or a 500MB/sec x1 lane, a 200MB/sec USB disk
> enclosure is not going to be too badly treated. If you expect to put
> a SATA III SSD in a USB3, then you'd want to select that slot with
> more care.

I expect to put external enclosure USB3 hard drives on that thing, not
SSD's. If I have an SSD (and I do), then I'm putting it on a SATA
connection (and I am). Actually, I currently only have SATA 3.0G
connectors, so SSD is not fully taxed yet, hence also the need for many
SATA 6.0G connectors.

> On AMD, it is Fusion chipsets like A75, that have it. There was some
> news item, about AMD planning on buying an IP logic block for USB3,
> rather than designing it themselves. Which may account for the
> limited distribution to date.
>
> http://en.wikipedia.org/wiki/AMD_chipsets#Fusion_controller_hubs_.28FCH.29

Yes, one of the articles I posted previously mentioned that AMD bought
their IP from NEC/Renesas. Renesas is the modern name for NEC, either
they bought or NEC or renamed themselves to Renesas, I don't know which
one though.

> As for the "no-name" nature of the add-on USB chips, you should be
> able to find out if they had driver problems. I could find references
> to Etron having a few problems, but didn't notice the others. Some
> equipment combinations may have the odd problem. NEC was the first to
> market with a USB3 host, just like NEC was first to market with USB2.
> I gather that was a point of pride with them, beating the other
> idiots who had a head start.

I think they were the first to market with it, and it looks like they
are also the gold-standard for USB3 chipsets now, since AMD took their
design rather than try to build their own.

> Intels strategy, leaves me scratching my head. They held back on
> purpose, on releasing USB3. Maybe it had something to do with writing
> the USB3 spec, and not being seen to profit from the advantage that
> gives.

I can answer that. Remember at that time Intel was pushing that
completely unnecessary Thunderbolt standard, hoping it would replace USB
completely. But only Apple took them up on it. When they found their
manipulation was going nowhere they waved their little white flag of
surrender, by finally adopting the standard everyone else had adopted.

It's not the first time Intel has tried to get away with something like
this and utterly failed. If you go back in their history, you'll see
such hall of shame actions like pushing Rambus RAM over the original
DDR1 standard, they eventually adopted DDR1 like everyone else. And
their most shameful example was of course, trying to foist the whole
Itanium and the entire IA64 marchitecture over the AMD64/x64 standard
which they eventually had to adopt themselves.

Yousuf Khan