sil3114 sata card

On Fri, 05 Jul 2019 15:10:43 +0000, T. Ment
wrote:

I moved the network card to one of the trouble slots, and now the
network card fails too. Looks like this P3V4X motherboard is shot. I
hate to lose an old board. P3V4X is expensive now on Ebay.

Maybe I'll pull all the capacitors and check them before trashing it.
But that's tedious and not much fun. We'll see.

Picked up from Newegg, long ages ago, an ASUS K8N-E+ -- specifications
being: 5 PCI, an AGP, 4 SATA and 2 PATA;- At an half-off NewEgg
sideline offer, either someone returned or wasn't up to how to build.
Things more apt to happen, back before NewEgg became cutthroat about
RMA-ing on customer money, no less indiscriminately over putting
obvious crap merchandise, in order to dump it. So I ran that Asus
forever, which is long enough to became worse for the wear, distinctly
plagued with issues, and then some for an added measure. I wore that
MB into the ground, then to begin going through a regular routine and
series of power supply replacements, which helped for a spell longer,
which broke when I plugged into it the only one server-grade power
supply (Fortron/Sparkle), I've owned, whereupon the ASUS it ate it in
a poof of particularly evil-smelling smoke.

Not to mention expensive: A value that, to me, exceeded mere special
PS units. That PS was unlike anything before or since. It was
literally server grade all the way, smaller and very much like a
brick, (for low profile cases), and not an ounce less than as heavy.

I don't trust going back, as a rule, risking dated OEM stock and
rejects. Not that I haven't had good luck with limited purchases from
white-boxed MBs, DELL or HP sells unbranded for nickel-dime, from
special production runs lacking any characteristic nomenclature off
the Pacific Rim.

It's still at risk over what averages are, on a market, with the most
value, for DollarCostAverging all sales, has to offer. That's
industry bread and butter: The mean on recent production fabs and at a
volume significant enough for decent representative Quality Control.

Which is secondary to why I then switched to Gigabyte. I'd avoided
them since being told by those I respected of their value and quality.
And I'd even run into an odd instance of an Asus employee, who was
fired, who advised me of Asus deteriorating core values. Some day,
perhaps, I'm due up and I'll now have to try an ARock.

Looking at my Gigabyte in a UK link, the first thing I saw was a
negative comment on some outlandish support-chipset temperatures I
hadn't read to notice prior. I mean I've since run into it: recorded
my own at 140F extremities (one of note out of 4 Gigabyte MB
thermistors) on the MB chipsets. But it was satisfactorily cheap
enough of a MB for me -- for 8 SATA ports and more AMD overlap
support, among generations of CPUs, than is reasonably sane to
expect;- Probably, something that's addressable for enduser
modifications in cooling.

All in all, at one PCI slot, which is more along contemporary
standards (inasmuch a sixfold hard focus as adapted to solidstate
transfers on memory storage considerations, than CPU efficacy refined
for thread/core count on non-gaming measures), which, of course,
places your six PCI slots as outlandish, even if only one were out of
the loop for nothing much else spectacularly off kilter.

Indeed, the augmented computer of ISA and PCI MBs is increasingly one
falling behind, the shrinking micro-platform of advancing form, where
MBs increasingly are defined for self-containment within admissible
limits of their own specifications. A skew, to be su 3 PCI slots,
no less an rich estate, and just as likely to be a $300+ concession,
in today's pricing and, not to discount ASUS, more and above any
$50US, give or take a little, on value-oriented MB ranges.

T. Ment wrote:
On Fri, 05 Jul 2019 05:52:22 +0000, T. Ment wrote:

Four slots can't all be bad

I moved the network card to one of the trouble slots, and now the
network card fails too. Looks like this P3V4X motherboard is shot. I
hate to lose an old board. P3V4X is expensive now on Ebay.

Maybe I'll pull all the capacitors and check them before trashing it.
But that's tedious and not much fun. We'll see.



The PCI bus is a shared bus. Each slot has a chip select
and a copy of a clock signal (which would be unique).

If the bus was shot, then chances are, moving a card to
any slot would not work.

If the motherboard is not positioned properly in the
tray, a card can be inserted crooked and cause a problem.
You have to ensure when placing a motherboard, that the slots
at the extreme ends are equally smooth for (unpowered)
test card insertion.

Paul

On Fri, 05 Jul 2019 13:28:17 -0400, Flasherly wrote:

Which is secondary to why I then switched to Gigabyte. I'd avoided
them since being told by those I respected of their value and quality.
And I'd even run into an odd instance of an Asus employee, who was
fired, who advised me of Asus deteriorating core values. Some day,
perhaps, I'm due up and I'll now have to try an ARock.

I don't trust name brand reputation. I buy cheap junk and try to fix it
up. My K7S5A was an Ebay "parts or not working" special for $6, shipping
included. I replaced a couple of bad caps. It works.

On Fri, 05 Jul 2019 15:50:29 -0400, Paul wrote:

The PCI bus is a shared bus. Each slot has a chip select
and a copy of a clock signal (which would be unique).

If the bus was shot, then chances are, moving a card to
any slot would not work.

I don't know much about PCI design.


If the motherboard is not positioned properly in the
tray, a card can be inserted crooked and cause a problem.
You have to ensure when placing a motherboard, that the slots
at the extreme ends are equally smooth for (unpowered)
test card insertion.

But I make sure the motherboard and cards are fully seated and making
good contact.

There are some capacitors near and between the PCI slots. I'll desolder
and check them, to see if I can find any bad ones. Something is messing
up INT-C and INT-D. INT-A and INT-B work.

T. Ment wrote:

There are some capacitors near and between the PCI slots. I'll desolder
and check them, to see if I can find any bad ones. Something is messing
up INT-C and INT-D. INT-A and INT-B work.

Interrupts are swizzled on the motherboard.

https://www.coreboot.org/images/8/84/Routing_full.png

Most cards think they are accessing INTA. (A card doesn't
necessary need to use more than one signal, so INTA on the
card is the one that is used. The table in the motherboard
manual is likely how "INTA" on all slots is wired. The swizzle
diagram in the previous picture, is more important on expensive
PCI cards with say, multiple chips or functions.)

But the IRQ from the motherboard could be one
of many such signals. And the old manuals
had a table with the details.

Chipsets began to acquire more signals, making
a difference to how onboard chips were treated,
versus the "empty slot" wiring. The more modern
systems had "a few good choices" for tricky cards.

https://i.postimg.cc/2jWmqhzB/INTA-slot-wiring.gif

PCI Express don't need this, because they can
use "Interrupt packets", and then the wiring
doesn't matter. Everything is carried over
a physical star connection (hub in center inside
chipset, each slot has a "private" bus and does not
share resources). This makes PCIe virtually
bulletproof to the issues that existed on PCI.
Unless someone designs a bad motherboard with
bad controlled impedance values, the signal
quality on PCIe approaches "perfection". Nice
wide eye opening. No wiggly-jiggly **** like
on PCI.

Intel did 1000 hours of analog simulation (something
not common at the time of introduction), to ensure
that sufficient PCI slot combinations were tested so
that there would be no slot dependencies. I've done
some of those simulations, for projects at work,
the difference being, the stuff I worked on,
the chips were "permanently affixed" to their
slots, so I only needed to run one sim topology
to prove it worked. The bus could seemingly be
quite long, before it ran into trouble (longer
than the wiring we needed at the time).

The P2B-S would not allow a SATA card to operate.
The theory was, that the BIOS PNP code refuses to
map cards where the "type" field is not recognized.
The P2B was an IDE ribbon cable system, and nobody
knew of SATA at the time. Selecting PNP_OS=yes might
change the behavior, as the OS is then tasked with
recognizing cards. Normally, PNP_OS=no is the
recommended value. The P2B-S cannot boot from a DVD
drive either. Once it determines the drive isn't
exhibiting CD behavior, it won't even send out
any probes. The light won't blink.

It's hard to say what kind of table manners a P3V4X would have.

Paul

On Fri, 05 Jul 2019 19:53:53 +0000, T. Ment
wrote:

I don't trust name brand reputation. I buy cheap junk and try to fix it
up. My K7S5A was an Ebay "parts or not working" special for $6, shipping
included. I replaced a couple of bad caps. It works.

I saw the same on my MB, about dozen instances of the Gigagbyte on
Ebay, two or three similarly listed for parts. $20, or $6 and a raise
the postal services gave themselves since then. But it's also on the
"brand reputation" circuit, or should be. I only got as far as UK's
Amazon listing while conveniently echoing off a virtual TOR network.
New and in a box for $60 and still boots DOS or a UNIX variant.

I trust Gigabyte. I don't know why the HDD controller balks, on my
particular model, over fragmentation routines and is too slow, nor how
a support chip can reach 140F. And I don't want, not especially, to
solder on it to find out more to fix it.

(OK, I will go so far to cut up old heatsink material, custom mount
and fit a small fan. And if this "new socket AM3+ MB" actually did
had more than one PCI slot, I might be tempted to put back in a SLI
controller, have 10 SATA ports, and kludged it up further to get a
smooth defrag routine.)

Besides Gigabyte was using solid-state capacitors ahead of the
industry curve, notably for a MB advertised as adverse to operating
under rugged conditions. Solid-state capacitors is "hot stuff" as I
understand.

There had been an issue, sometime prior to SS capacitors, with some
motherboard manufacturers buying "bad-batched" and substandard
electrolytic capacitors - ASUS being named pre-eminent in the
deceptive practice. I'm not sure to what extent subsequent
distribution may have been affected or how long it went on. As
mentioned the ASUS K8N-E+ is old enough as it, a 3rd generation prior
AMD socket, yet I was already raising my eyebrows at ASUS pricing
schemes, or I may as well instead have bought it new rather than at
half off from Newegg.

On Fri, 05 Jul 2019 15:50:29 -0400, Paul wrote:

If the motherboard is not positioned properly in the
tray, a card can be inserted crooked and cause a problem.

I desoldered one capacitor. My cheapo capacitor checker said it's good.
I soldered it back on the board. I didn't check any more. They all look
good, no bulges or leaks.

I put the motherboard back in the case, and the network card in one of
the trouble slots. Now it's working.

I temporarily used three screws to hold the motherboard in the case.
Maybe it will fail again, when I put the other three screws in. Maybe
that warps it just enough, to expose a defect not visible to the eye.

On Fri, 05 Jul 2019 18:55:37 -0400, Flasherly wrote:

I trust Gigabyte. I don't know why the HDD controller balks, on my
particular model, over fragmentation routines and is too slow, nor how
a support chip can reach 140F. And I don't want, not especially, to
solder on it to find out more to fix it.

I killed more than one motherboard learning how to solder.

It's not hard to get a capacitor off the board. You just add a little
blob of solder to the pins on the reverse side of the board, heat one
pin and wiggle with your hand on the opposite side of the board, heat
the other pin and wiggle, and repeat that cycle until you have it out.
That's the easy part.

But then you're left with two holes still plugged with solder. This is
what's hard. You have to clear the holes without damaging them, and they
are delicate.

The factory uses lead free high temp solder. With a common soldering
iron, it's not easy getting enough heat on the holes, to wick out the
factory solder without damaging the holes. It's doable, but tedious.
Once cleared, you put the capacitor back in, and solder it with leaded
solder, which is easier to work with.

Motherboards are not designed for easy repair.

On Sat, 06 Jul 2019 00:38:25 +0000, T. Ment
wrote:

I killed more than one motherboard learning how to solder.

It's not hard to get a capacitor off the board. You just add a little
blob of solder to the pins on the reverse side of the board, heat one
pin and wiggle with your hand on the opposite side of the board, heat
the other pin and wiggle, and repeat that cycle until you have it out.
That's the easy part.

But then you're left with two holes still plugged with solder. This is
what's hard. You have to clear the holes without damaging them, and they
are delicate.

The factory uses lead free high temp solder. With a common soldering
iron, it's not easy getting enough heat on the holes, to wick out the
factory solder without damaging the holes. It's doable, but tedious.
Once cleared, you put the capacitor back in, and solder it with leaded
solder, which is easier to work with.

Motherboards are not designed for easy repair.

My impression of my first Gigabyte was, after I've went through so
many others, why can't I kill this one? (Seems it may have seen 8,
going on 10 years use;- I run them 24/7.)

Someone was listening and heard me, because a lightning storm emerged,
struck the transformer pig on a pole in my back yard, leaving me
looking at one fried modem and MB.

Maybe there's some truth to 3-year capacitor (liquid) average
lifespan. Five years is what I'd expect of a MB prior to Gigabyte. SS
caps are also now to be expected, more widely adapted for an industry
standard. MSI, a year or two ago, was trying on a comeback for
reputability to foist SS caps for MBs built for high-quality standards
at Military Specs. I left MSI to ASUS for reliability after a couple
bad MBs, long before SS caps.

https://www.gigabyte.com/webpage/8/a..._all_solid.htm

On Wed, 03 Jul 2019 23:06:34 +0000, T. Ment wrote:

I need it for a DOS computer

The Ebay seller told me it does not support DOS.
Looks like my plan for DOS won't fly.

So back to Ebay for a Promise SATA/150 TX4. It can see the SATA drive,
but wont' use it, until you enter its BIOS and create an array -- which
kills any data on the drive.

I had already cloned the drive using linux DD on another computer which
has onboard SATA, and I didn't want to kill the data and start over. But
at this point I had no other option.

With nothing more to lose, I put the sil3114 back in the P3V4X, booted
from IDE, and ran updflash with BIOS version 5.5.00, the latest one. But
as before, updflash failed, and when rebooting, the computer hung again.
That problem was gone for a while, but it came back. Not sure why.

So back to the USB programmer to reflash. Then the computer booted OK.

Next I booted into the sil3114 BIOS and ran the low level quick format,
hoping to make the sil3114 happy.

I disabled the onboard IDE, then installed DOS 6.22 from floppy to the
SATA drive on the sil3114. Then I enabled IDE, and copied a 40 meg file
from IDE to SATA, all in DOS. That worked too. No errors as before.

Disabled IDE again, booted DOS from SATA. Copied the big file again,
from one directory to another. Very fast. about 1.5 seconds, compared to
10 seconds with the IDE drive and P3V4X bios.

So my conclusion is, you can't clone the drive on some other computer
and expect it to work in DOS with the sil3114. You must make the sil3114
happy by using its low level format first. I don't know what INT13 magic
is going on there, but it seems to work.

I will do more testing after reinstalling the other partitions, but DOS
was the only problem before, so hopefully it will all work.