Asus P9X79 four short beeps

[Bill Anderson]

It began happening a few weeks ago -- I'd try to boot but before post
I'd hear four short beeps followed by the post chirp (different sound)
and then nothing. But it was an intermittent problem and I let it go
when rebooting seemed to solve things.

Then yesterday the problem stuck around through several attempted
reboots and I figured I'd better do something. So I looked up the error
beep table on the Asus website and found nothing about four short beeps.
Thanks, Asus. Then I branched out and learned Asus uses American
Megatrends' AMI BIOS, and that four beeps mean System Timer Failure,
which further means something's wrong with memory.

That was when I actually looked at BIOS to see how much memory I had and
I learned my four 8-gig memory sticks were producing not 32 gigs of
memory, but just 8 and change. In other words, only one stick was working.

So just now I removed all the memory and began replacing sticks one at a
time. At first I couldn't get past the four beeps, but I kept replacing
sticks in the D1 slot until one worked. (The manual says to put a
single stick in D1.) Then I put a stick in B1 as shown in the manual and
went right back to the four error beeps.

But I persevered, removing and re-inserting, and eventually I had two
working, then four, and now I'm booting nicely and BIOS shows total
memory at 32 gigs and change.

Funny things was -- the AIDA 64 Extreme system monitor software always
showed four slots filled with 8-gig sticks. It saw them when BIOS
didn't. Weird.

But things are fine now. I think. Maybe I oughta run Memtest just to be
sure...

--
Bill Anderson

I am the Mighty Favog

[Shadow]

Either one or more of the sticks were badly seated or you have
bad RAM.
Since it's usually an intermittent problem, I'd run Memtest
ASAP. Let it run overnight.
[]'s
--
Don't be evil - Google 2004
We have a new policy - Google 2012

[Paul]

Memory is detected two ways at BIOS level.

1) Read config info from SPD PROM on each DIMM.

2) Once the BIOS knows a DIMM is present, it uses "peek & poke"
testing to prove "a RAM is a RAM" and that it can actually
store stuff.

One of the "proof cases" for this, was the day when
some brand of DIMMs, had the wrong SPD chip soldered
to them. The DIMM may have declared it had 256MB on it,
when the physical chips were 128MB. The BIOS (correctly)
did "peek & poke" and measured 128MB, and the system
started just fine and ran with the reduced amount
of memory. Because in fact, that's all the physical memory
that was present on the stick.

This issue seemed to first show up on triple channel memory.
Maybe a 12GB system would be detected as an 8GB system. At
first, people might have tried blaming a "socket contact"
issue for the problem.

Then later, people started randomly adjusting the IMC or
VNorthbridge. And then there were claims that this
was "fixing it".

I don't know if I've ever seen a company web site
(Asus or Intel) making claims as to why this happens.

When the OS is running, hardware identification software
continues to have access to the SPD and can then
claim that 32GB are "installed", even if the BIOS
has tested and chosen to only use a subset, because
"peek & poke" is failing.

"Peek & poke" is a quick check of memory presence and
does not represent a full memory test. That may come
later in the POST, if the user has enabled it.

*******

And yes, doing some memtest is a good idea.
Between that, and something like Prime95, you'll get
a better idea whether it's really working properly or not.

Paul

[Bill Anderson]

I'm thinking three of the sticks, or maybe just one, affecting the
others, somehow became unseated. Maybe fluctuations in temperature over
time caused it, I dunno. I certainly hadn't been in there bumping
around. I learned there's a technique to seating the sticks on the
P9X79. Most memory slots I've seen over the years have locking levers on
both ends, but the P9X79 has a lock on only one end, with a passive
slotted block on the other. To get things seated, I had to press the
stick firmly in the slotted block end and then press hard on the locking
end to snap the stick in place. Pressing evenly across the stick seemed
not to work as well.

I did run one full pass of Memtest and saw no errors. Maybe I'll let it
run all night as you suggest, but I've found if there are going to be
Memtest errors, they'll pop up long before one full pass is completed.

[Bill Anderson]

I think re-seating the memory sticks was what I needed to do. BIOS is
showing 32 gigs of RAM now and so is Win10. I think you must be right
about how software might see the RAM sticks when BIOS couldn't. It's
always something. Thanks, Paul.

[Bill Anderson]

On 5/1/2018 12:59 PM, Paul wrote:

I wrote a long post explaining I still have the problem, then I had a
bright idea about how to test, and now I still have the problem only my
testing has shown it isn't exactly what I thought it was. Got that?

So I've deleted the draft of my earlier post, and now I'll try to keep
this as short as possible:

1) I could boot and Windows was seeing all 32 gigabytes of RAM, four
8-gig sticks in four slots.
2) If I shut down and immediately rebooted (not a restart - a full
shutdown and reboot) I'd get the four beeps before POST error. A simple
restart was no problem. Booting from a full shutdown was the problem.
3) If I waited like 10-15 minutes after a full shutdown I could reboot
and everything would be back to normal -- Windows seeing all 32 gigs of RAM.
4) I decided to test each memory stick, so I pulled all but the one in
slot D1 (the slot for single sticks) and booted and shut down and
rebooted and no problem
5) Tested another stick the same way -- boot, shut down, reboot. No problem.
6) Tested a third stick -- oops! Four beeps on reboot. Wait a while
and boot easily. Reboot and -- four beeps. So I set this stick aside.
7) Tested the fourth stick -- no problem.
8) Tested the suspect stick again -- boot once, OK. Second time -- four
beep error.
9) Aha! I've found the culprit. Right?
10) Put one good stick in slot D1. Boots fine.
11) Put a second stick in slot B1 as shown in the manual. Boots fine.
But...what's this? BIOS is showing only 8 gigs of RAM.
12) Reset stick in slot B1. No go. And again. And again. And let's
try the other good stick. No go. Still only 8 gigs of RAM. I should
be seeing 16 gigabytes of RAM but I'm getting nothing out of slot B1.
Remember -- when I started this test I was seeing 32 gigs, so B1 must
have been working then. But not now.

So here I am running my system (nicely, I suppose) with one 8-gig stick
of RAM installed. And I'm thinking my Asus P9X79 MBO is going bad. Is
that what you're thinking?

[Bill Anderson]

Forgot to mention I ran Memtest on all 32 gigs for six hours last night
and there were no errors.

[Paul]

The memory controller is inside the processor itself.

The motherboard contributes power regulators.

VDimm
Vtt (terminator)
VNorthbridge (powers IMC, uncore)

The motherboard also contributes copper tracks.

The sockets all have to make contact.

It's very hard to determine a root cause in this particular
case, by testing one combination after another of DIMMs. This
is how the first wave of testing concluded it was the CPU socket
was at fault. And you can tell from your testing, that's not
likely to be the case. If a contact was truly flaky, you'd
expect to have the CPU "crash" in the middle of a session.
And that's not happening. Notice how it's always an
orderly BIOS issue.

While I might look at the VNorthbridge, and adjust it one
notch, it's up to you to decide what to test next, due
to the large number of variables, and the possibility of
disturbing something while going from one test setup
to the next.

When I had a stick on a previous set of RAM start to throw
errors, one or two notches on VNorthbridge fixed it (on
a X48 Northbridge). The regulators have a fairly fine
gradation, so we're not talking about a very large
voltage change here.

It's a good thing that previous motherboards here were
rather lacking in voltage adjustments, but at the same
time, they didn't have RAM problems like these >2channel
boards do. Both 3 channel and 4 channel processors have
exhibited this problem.

There have been older boards, where somebody at the factory
didn't test the memory map when "full". One board, if you
installed 4GB (its max), the USB ports would indicate "overcurrent",
which means that the process of setting up the memory,
was splattering some control bits in a USB register. Which
means the memory map wasn't planned properly by the BIOS.

You could also consider examining the voltage controls
on your board, just to make sure there wasn't a BIOS
settings corruption at the heart of this. And that the
voltage regulators are *undervolted*. Usually the boards
have a color scheme, with green (normal), amber (slight boost),
red (might be getting close to a hardware limit). But undervolt
would be below green in a sense. If the BIOS is well designed,
the "default" value or factory value should be shown in
a different color, so you know what the normal value is.

Paul

[Paul]

Bill Anderson wrote:
> Forgot to mention I ran Memtest on all 32 gigs for six hours last night
> and there were no errors.

So it's probably not the RAM itself.

Something about the processor and its built-in memory
controller. Or the voltage regulators that power the
various parts there.

Paul

[Bill Anderson]

I think you put me on the right path in your previous message, Paul, and
once again I thank you. I've adjusted things in BIOS and reinstalled
all four RAM sticks and I'm showing 32 gigs of healthy memory and the
system has been booting flawlessly from shut-down state. All I did was
go to the memory adjustments in BIOS and try to make everything as
untweaked as possible. Here's how things look now:

https://goo.gl/D3x3L7

All this overclocking tweaking stuff is too much for me to even want to
figure out. I do remember a few years back trying to tweak timings and
so forth, based on advice I'd either read on the internet or that you'd
given me. I just don't remember. Whatever, It had all been working
fine up until recently and I'd given it no more thought. Something must
have goosed BIOS, I dunno. It certainly wasn't me. Maybe now I can
stop thinking about it again.

Of course things may all go south any time, but right now I'm feeling
pretty good about my current setup and I've stopped looking at the
latest and greatest Asus motherboards and Intel i7 processors. Boy, you
can spend a lot of money on that stuff...

Thanks again.

[Bill Anderson]

And within 20 minutes of posting that I was back to the four beeps
again. I'm tired of this. My current build will be six years old in
September. It's time.

First step is an inventory of everything I have hanging off the P9X79.
I'll want to use as much of it as possible. But maybe my new MBO could
have more SATA III ports than the P9X79? I've had to get a PCIE card to
give me more SATA III ports. And I'd like USB3 internal blocks for USB
ports on the front of my case.

And I'd like to keep using my NZXT Kraken X61 RL-KRX61-01 liquid CPU
cooler, but apparently the only Intel processors it supports are 2011
/1366 /1156 /1155 /1150. Maybe there's an adapter. If not, I'll have to
get a new cooler, and I like the Kraken so much -- it's quiet and solved
my overheating problem -- I'll probably want to get one that fits
whatever new processor I come up with.

Let's see what I have in each of the P9X79's slots:

PCIEX16_1 slot
Silverstone PCI Express Card with USB 3.0 Internal Connector
Shouldn't need this if new MBO has internal USB 3.0 connection for front
panel

PCIEX1_1 slot
Hauppauge Colossus video capture card
I'll need this in the new build

PCIEX1_2 slot
ATI Theater 650 Video Card
It's a TV tuner card I'm no longer using. Guess I oughta pull it out.

PCIEX16_2 slot
GEFORCE GTX970 Video Card

PCI1 slot
Unoccupied as the video card covers it up

PCIEX16_3
StarTech 2 Port SATA 6 Gbps PCI Express SATA Controller Card
Shouldn't need this in new build if the MBO has plenty of SATA6 ports.


So what to get? This looks good, but expensive, ASUS PRIME X299-DELUXE
https://goo.gl/iWZhQY
It's got seven SATA 6 ports and it has one USB 3.1 Gen 2 front panel
connector port. Plus lots more USB.

It'll work with this processor, i7 Core X
https://goo.gl/kagSsV

And of course there's RAM. I'm thinking my 32 gigs of DDR3 was
overkill. Maybe 16 gigs for this one? A couple of these Corsair
Vengeance sticks?
https://goo.gl/NmfLRT

I want power for video editing, not so much for gaming, though the
Oculus Prime VR device looked interesting and I was disappointed when I
ran the routine to check my P9X79 and I learned I didn't have a
processor powerful enough. It's an i7 for heaven's sake!

I confess I'm at a loss about what I should do. I wouldn't even be
thinking about a new build if my P9X79 were dependable, but it's going
flaky and I don't want to mess around with it. I want a computer that
works.

Other ideas?

[Paul]

Scroll down a bit further and check the other values ?

https://s18.postimg.cc/d2udxby2h/my_machine_p9x79.gif

Paul

[Paul]

My first thought, would be to double check the Oculus requirements
and see if it was the video card that was the issue, and not the CPU.

My video card cost $50 or so. And this was my test result.

https://s18.postimg.cc/51b7cm67t/compat-check.gif

*******

You're not really using a lot of PCIe lanes in your list.

https://www.cpubenchmark.net/high_end_cpus.html

Intel Core i7-7820X @ 3.60GHz 18,609 $ 589 (no cooler) 28 lanes
Intel Core i7-8700K @ 3.70GHz 16,043 $ 359 95W 16 lanes(CPU)
AMD Ryzen 7 2700X 17,099 $ 369 105W Apr19,2018 20 lanes

https://www.cpubenchmark.net/singleThread.html

Single thread of execution (typical desktop apps). CPU is
probably turboed on one core for these, and the clock was
actually higher than this.

Intel Core i7-7820X @ 3.60GHz 2,422 quad channel RAM (room for more sticks)
Intel Core i7-8700K @ 3.70GHz 2,709 dual channel RAM
AMD Ryzen 7 2700X 2,209 dual channel RAM

The 8700K is only two channels of RAM and the usual
low-end lane count. But right now, your 970 video
card is the only big consumer of lanes. Buying any
more machine constitutes "future proofing". And
you "wore out" your current machine in only six years :-)

**********************************************************

The following are some tables I made up earlier this year,
when a few of the regulars caught "upgrade fever". There's
nothing of great consequence here, except to note that in
your proposed purchase, you're thinking of buying a 28 lane
processor, and so the bottom x8 slot on the motherboard
I made an example of, it wouldn't be working with a 7820X
plugged in. The 7740X, the worlds worst processor, is
even worse in this regard, of being "lane-starved".

The table was for someone who proposed running older OSes, so
I threw in older gear for comparison purposes.

X79 LGA2011 V1
Lane
2011 Core i7 i7-4960X(S1) Ivy Bridge-E 3.6GHz 15MB 130W P3.60
2011 Core i7 i7-4930K(S1) Ivy Bridge-E 3.4GHz 12MB 130W P3.60 40?
2011 Core i7 i7-4820K(S1) Ivy Bridge-E 3.7GHz 10MB 130W P3.60 40?
2011 Core i7 i7-3970X(C2) Sandy Bridge-E 3.5GHz 15MB 150W P2.30
2011 Core i7 i7-3960X(C2) Sandy Bridge-E 3.3GHz 15MB 130W P1.90
2011 Core i7 i7-3960X(C1) Sandy Bridge-E 3.3GHz 15MB 130W P1.90
2011 Core i7 i7-3930K(C2) Sandy Bridge-E 3.2GHz 12MB 130W P1.90
2011 Core i7 i7-3930K(C1) Sandy Bridge-E 3.2GHz 12MB 130W P1.70 40?

X99 LGA2011 V3 Lane
2011-3 Core i7 i7-6950X(R0) Broadwell-E 3.0GHz 25MB 140W P3.00
2011-3 Core i7 i7-6900K(R0) Broadwell-E 3.2GHz 20MB 140W P3.00
2011-3 Core i7 i7-6850K(R0) Broadwell-E 3.6GHz 15MB 140W P3.00
2011-3 Core i7 i7-6800K(R0) Broadwell-E 3.4GHz 15MB 140W P3.00
2011-3 Core i7 i7-5960X(R2) Haswell-E 3.0GHz 20MB 140W All
2011-3 Core i7 i7-5930K(R2) Haswell-E 3.5GHz 15MB 140W All 40?
2011-3 Core i7 i7-5820K(R2) Haswell-E 3.3GHz 15MB 140W All 28?

LGA2066 4 channel

X299 LGA2066 V1 via DMI V3 Lane
2066 Core i9 I9-7980XE(M0) Skylake-X 2.6GHz 24.75MB 165W
2066 Core i9 I9-7960X(M0) Skylake-X 2.8GHz 22MB 165W
2066 Core i9 I9-7940X(M0) Skylake-X 3.1GHz 19.25MB 165W
2066 Core i9 I9-7920X(M0) Skylake-X 2.9GHz 16.5MB 140W
2066 Core i9 i9-7900X(U0) Skylake-X 3.3GHz 13.75MB 140W 44
2066 Core i7 i7-7820X(U0) Skylake-X 3.6GHz 11MB 140W 28 <===
2066 Core i7 i7-7800X(U0) Skylake-X 3.5GHz 8.25MB 140W
2066 Core i7 i7-7740X(B0) Kaby Lake-X 4.3GHz 8MB 112W 16
2066 Core i5 i5-7640X(B0) Kaby Lake-X 4.0GHz 6MB 112W

C621 Xeon??? via UPI

*******
RAMCH PCIE AVX512
i9-7980XE (24.75M cache, 18 Cores, 36 Threads, 2.60 GHz) $1,979 4 44
i9-7960X (22M cache, 16 Cores, 32 Threads, 2.80 GHz) $1,684 4 44
i9-7940X (19.25M cache, 14 Cores, 28 Threads, 3.10 GHz) $1,387 4 44
i9-7920X (16.5M cache, 12 Cores, 24 Threads, 2.90 GHz) $1,189 4 44
i9-7900X (13.75M cache, 10 Cores, 20 Threads, 3.30 GHz) $ 989 4 44
i7-7820X (11M cache, 8 Cores, 16 Threads, 3.60 GHz) $ 589 4 28 <===
i7-7800X (8.25M cache, 6 Cores, 12 Threads, 3.50 GHz) $ 383 4 28 Yes
i7-7740X (8M cache, 4 Cores, 8 Threads, 4.30 GHz) $ 339 2 16 No
i5-7640X (6M cache, 4 Cores, 4 Threads, 4.00 GHz) $ 242 2 16

Page 11 here, shows you what happens when you don't have enough PCIE lanes on the CPU.

http://download.gigabyte.us/FileList/Manual/mb_manual_ga-x299-ud4_e.pdf

HTH,
Paul

[Bill Anderson]

I'm curious what you think of the "high-end" build on this page.
https://www.custompcreview.com/computer-builds/best-video-editing-pc-builds/

It uses the Asus MBO I like and the i97900X you show above is "only"
$899.99 at Newegg. The killer for me is the price of the RAM. If I go
this route I'll probably start with a quarter, possibly half the RAM
recommended. But I'm thinking seriously about purchasing the MBO and
CPU they recommend, some of the RAM, and whatever else I need from the
rest.

This would be expensive. And what makes it worse is that my current
system is behaving itself most of the time today. Don't know how long
that will last, though.

Thoughts?

[Paul]

Bill Anderson wrote:
> On 5/4/2018 1:23 AM, Paul wrote:

>> LGA2066 4 channel
>>
>> X299 LGA2066 V1 via DMI V3 Lane

>> 2066 Core i9 i9-7900X(U0) Skylake-X 3.3GHz 13.75MB 140W 44

>> *******
>> RAMCH PCIE AVX512

>> i9-7900X (13.75M cache, 10 Cores, 20 Threads, 3.30 GHz) $ 989 4 44
>

> I'm curious what you think of the "high-end" build on this page.
> https://www.custompcreview.com/computer-builds/best-video-editing-pc-builds/
>
> It uses the Asus MBO I like and the i97900X you show above is "only"
> $899.99 at Newegg. The killer for me is the price of the RAM. If I go
> this route I'll probably start with a quarter, possibly half the RAM
> recommended. But I'm thinking seriously about purchasing the MBO and
> CPU they recommend, some of the RAM, and whatever else I need from the
> rest.
>
> This would be expensive. And what makes it worse is that my current
> system is behaving itself most of the time today. Don't know how long
> that will last, though.
>
> Thoughts?

CPU Intel Core i9-7900X ~$989 4 channel CPU, use 4 DIMMs minimum
Motherboard ASUS PRIME X299-Deluxe LGA 2066 as you'd expect, $$$
Memory 64GB Kingston HyperX FURY 2133MHz (4x16GB) Kit <=== could replace with 4x4GB DIMMs
GPU PNY Nvidia Quadro P6000 24GB ~$4500.00 !!! <=== certified OpenGL support $$$
Case Corsair Obsidian 750D
Boot Drive Samsung 850 PRO 1TB SSD <=== probably the last 850 Pro (MLC) still in stock, 256/512 gone
Storage Drive Western Digital Caviar Black 4TB
Optical Drive ASUS Blu-Ray Writer
PSU EVGA SuperNOVA 850 G3
CPU Cooler Corsair H100i v2
Operating System Windows 10 Home 64-Bit <=== Cheapskates! This should use Pro.

That video card choice totally swamps out any notion of "cost savings".

I generally try to track down what the "desktop" equivalent of
the card is. In this case, it would be a Titan X (~$1200 original price).
The difference is, the "certified OpenGL support" is for CAD software, and the
card ends up with "business" pricing to go along with the
"professional" creds of the card. The Titan X also has half
the VRAM, but really "12GB is the same as 24GB" for all except
the most extreme applications. SimCity doesn't need 12GB.

https://wccftech.com/nvidia-pascal-quadro-p6000-gaming-benchmarks/

If you were Bill Gates, that would be a great system.

So when I click the button, it goes to Amazon, and is "sans video card" ???
If so, you'd probably want to do a few edits in the parts list.

The motherboard is $500 class.

https://www.newegg.com/Product/Product.aspx?Item=N82E16813132990R

You could go crazy with the RAM in terms of speed... but read the reviews.

G.SKILL TridentZ Series 16GB (4 x 4GB) 288-Pin DDR4 SDRAM DDR4 3600 $257

https://www.newegg.com/Product/Product.aspx?Item=N82E16820231980

Pros: Looks nice.

Cons: Did not hold stock clocks or timings.

When it would hold them, it required such High VSA and VCCIO
that the CPU was under threat.

Would only load past the windows logo once every 10+ boots
during these moments of testing.

Spreaders conflict with larger, say 135mm, CPU heatsink fans.

Your build is using a water cooler for the CPU, so it doesn't
matter how tall the heat spreaders are. The hose barbs and
tubing have to be routed so they don't conflict with
RAM or something.

These other ones, the five reviews seem to be better for some
reason, perhaps because they were testing with a different
CPU type ? It's hard to find ordinary DIMMs that actually
claim to be tested with X299 (for whatever that is worth -
the machine should really work with any DDR4 you select).
The reason for selecting this, is the hope the XMP profile
works to save on the need for "tuning".

https://www.newegg.com/Product/Product.aspx?Item=N82E16820231907 $238.99

This is easily a $2000 build, just quickly doing a few sums
in my head. And say, reusing your video card. A $500 motherboard,
a $1000 CPU, $250 for RAM, a bit more other junk, and so on.
The CPU liquid cooler probably costs a few bucks. But
is going to be more friendly than putting a huge Noctua
air cooler in there. I had to seat the DIMMs on my system,
using a "pusher stick" made of wood, to snap the DIMMs into place.
A water cooler leaves a lot more room (although it doesn't give
any blowdown cooling or anything, for motherboard components).
You might want to check the temps on the VCore regulator
during your build/test. When Prime95 is running (non-AVX version).

Paul

[Bill Anderson]

Heh, I didn't price the GPU because I planned to use the one I have. And
yes, I figured about $2K, which I can swing, but which I don't want to
do if I don't have to. I hadn't even looked at the price of the GPU.
Holy moly.

I've spent the bulk of today working on (OK, playing around with) my
P9X79 computer. I have a dual boot system and I wanted to upgrade my
secondary OS with the latest version of Win10, a clean install.

A digression: I downloaded the new (this week) Win10 ISO and installed
it clean in a small partition I use for my secondary OS. Pretty soon it
was telling me it needed some "updates." Well, where the initial
installation took about 20 minutes, the "updates" took over an hour, but
now I have an installation of Win10 version 1803, OS Build 17134.1. And
now that I look, I see my main OS installation is Version 1709, Build
16299.371. And I thought I'd successfully updated it to the Spring 2018
version earlier this week. Sometimes Windows puzzles me.

My main point, though, is that throughout today I've heard the four-beep
error message only once, and that was when I'd disconnected my main boot
drive (an SSD that holds the OS and all applications) in order to
prevent Win10 from spotting it while installing my secondary OS on
another drive. (I hate dealing the with boot manager, and this way it
never gets triggered.) But all I had to do was power down and then
power up again and I heard the Post beep and I was in business for the
rest of the day.

As I've mentioned in earlier posts, a simple reboot never triggers the
error beeps. The system has to shut down -- fans off -- before a reboot
triggers the error. And the other day that was happening regularly.
Every time I'd exit BIOS the system would completely power down, pause a
few seconds, and then power up again, triggering the error. Today that
hasn't happened once, I transition smoothly from BIOS into OS boot and
I've been in and out of BIOS a number of times.

So...I'm not going to buy a new system as long as this one keeps
behaving. But if things go south again anytime soon, I really believe
I'll get the board and the processor mentioned above, and some of the
RAM, and as for that $4500 GPU...ha ha.

Thanks for following up. I'm going to take a deep breath now and wait
to see if my problems recur anytime soon. I'm hoping I just might be
able to escape this with few injuries to my wallet.

[Bill Anderson]

Well, pissonit. As time passed I became adept at working around the
problem. Restarts are never a problem -- I can restart over and over
all day long and everything's perfect. But let me shut down -- fans off
-- and getting things going again becomes a hit-or-miss proposition,
with lots of misses before a hit. Press power button -- hear the error
-- power off -- wait -- try again -- no luck -- wait -- try -- wait --
try -- wait...hey, that's the post beep. And then everything's fine
until the next power down.

When all this started, I could always power on successfully first thing
in the morning, but that's no longer the case. Now I expect an error
every time I power up.

I bought a new power supply just in case that was the problem. I
figured if a new PS could solve the problem my wallet would get off
easy. But today I've installed my new Corsair RM750x and on first boot
it was easy sailing, but after a shutdown and power on, I got the error
beeps again. So for a semi-reasonable sum I've ruled out the power
supply. Next up, I guess -- MBO, CPU, memory, M2 card. Good news
about the cooling system: apparently my current Kraken X61 will fit a
socket 2066.

I don't know what else to do. Memtest never finds an error and Windows
and Aida report my 32 gigabytes of DDR3 are behaving normally. I think
there's got to be a MBO problem, but if you have another idea I'd like
to hear it. Otherwise I'm about to do some expensive shopping. At
least I'll be starting with a nice new power supply.

[Bill Anderson]

On 5/25/2018 2:16 PM, Bill Anderson wrote:

> I don't know what else to do.  Memtest never finds an error and Windows
> and Aida report my 32 gigabytes of DDR3 are behaving normally.  I think
> there's got to be a MBO problem, but if you have another idea I'd like
> to hear it.  Otherwise I'm about to do some expensive shopping.  At
> least I'll be starting with a nice new power supply.
>

I'm now convinced the P9X79 is my problem.

I bought a new MBO -- Asus Prime X-299A and outfitted it with a new i7
processor and 16 gigs of ddr4 memory and (my new pride and joy) a 500 GB
Samsung NVMe m.2 socket SSD card. Man that thing is blazing fast -- I'm
convinced even faster than my old Samsung 500 GB SSD. And in case you're
wondering, a Kraken X61 cooler moves from a LGA 2011 socket to an LGA
2066 socket with no fuss at all. Fits perfectly.

In building the new system I came to the embarrassing (but mistaken)
conclusion that my whole problem had a bad Colossus video capture card,
as I couldn't get my old one to work on the new MBO for some reason.
The new system wouldn't start with that Colossus card installed. Well I
just knew that must have been my problem all along, so I ordered a new
Colossus 2 card which is working flawlessly in the new system. I was
confident I'd identified the problem and pretty annoyed at myself for
buying a new MBO and processor and all the rest when all I'd needed to
do was replace that Colossus card. Sure, I wanted a faster system, but
still...

So last night, with the new system up and running great, I gathered some
unused parts and rebuilt my old system. I had a spare case and cooler
and power supply and really everything I needed except an optical drive
and strangely, a little motherboard speaker -- both easily obtainable.
I certainly had enough parts to fire up the rebuilt P9X79, which I did
and it started perfectly, everything working. And then I shut down and
restarted and ... nothing. I head no beeps because I had no MBO
speaker, but it was easy to tell it wasn't booting.

Wait, try, wait, try, wait ... then it booted. Same behavior as before,
and this time with practically no peripherals installed. Removed all
memory but one card ... nothing. Switched out that card ... nothing.
Wait, try, wait, try, wait ...and it boots.

So I want a new motherboard for my Intel LGA 2011 i7 3930 processor and
32 gigabytes of Corsair Vengeance DDR3 memory. And I can't seem to find
anything worth having for less than about $400. I can find micro
motherboards that come with a processor already installed for under
$150, but I want a full ATX with onboard sound and a few slots for
peripherals like my spare nVidia GTX 570 video card. And I don't want
to pay $250 to $400 to build a system I don't really need but would like
to have around just in case...you know.

Anybody have a suggestion for where to find something cheap? And good?
Please help me find a use for all my spare parts....

[Bill Anderson]

Maybe this?

https://goo.gl/wqcfA7

https://www.amazon.com/Procaja-Motherboard-LGA2011-Channels-E5-2670/dp/B07BV8T8R7/ref=sr_1_9?s=electronics&ie=UTF8&qid=1528568092&sr=1-9&keywords=motherboard+lga+2011+socket

[Paul]

The Amazon entry, the dimensions are somewhat strange.
You'll want to convert the measurements to inches and
see whether it's even MicroATX.

I couldn't find any reference to the brand.

But apparently, X79 motherboards from China are a "thing".
The dude here receives one, that doesn't even have a
heatsink on the VRM components.

https://www.youtube.com/watch?v=pg739TdPFeA

Now if I was building that up:

1) Build on kitchen table.
2) Lay motherboard in the open so it can be seen.
3) Run Prime95.
4) Stick finger on VRM. Get burnt or whatever.
5) Plan a cooling strategy.
6) Then... install in a case.

You'd probably check the VRM even at idle in the case of
that motherboard (with no heatsink on VRM). The one in your
case has an extruded aluminum piece bolted to it for
looks. Aluminum fins, it doesn't pay to make them too
tall, as you can't get heat flow up to the top of the
fin all that well. And that's why the shortness of
those fins probably isn't a total disaster. While surface
area is what you want on heatsinks, if you have a
really thin 1" tall fin, the top half inch isn't
doing anything. This is why for CPU coolers, heatpipes
are wound through the fin arrays to shorten the distance
from heatpipe "perfect thermal conductor" to the fin.
If the fins were thicker, then they could be taller,
but with thicker fins there would be fewer of them.

I ended up using impingement cooling on my VRM,
but the heatsink on my board isn't really shaped
for any good cooling strategies. I just "pointed
the fan at it and hoped for the best". I installed
the fan, because I burned my finger on the VRM
heatsink.

Since the dude in the video received a three page
fold-out "manual", chances are you'll be doing
a build with zero support from the manufacturer.
If you're lucky, there might be an all-Chinese
manual as a PDF on the CD. I've worked with
all-Chinese manuals before, and the diagrams
still have value.

The BIOS on that Youtube motherboard is AMI,
but who knows whether AMI knows about that
particular manufacturer.

I would prefer to see a "CPU support table", with
some notion of BIOS version versus supported CPUs.
Just dropping a few CPU model numbers into a
single sentence answer, isn't confidence building.

But it is cheap :-) Er, ah, I mean "good price".
The larger Chinese plants use automation for mfg
(standard pick and place, double IR reflow tunnels),
and only test is done with humans. They cannot
afford to make too much garbage - in a low margin
business, you'd go bankrupt in a year if you
couldn't achieve some level of build quality.
But that doesn't address whether the board layout
and controlled impedance were done well. They have
to steal an experienced designer from one of the
larger motherboard houses, to expect good work.
I don't think we're at the level yet where a
high school student can have "first time success".

Paul

[Bill Anderson]

You make some very good points, as usual. This one at least has reviews
and the cost is about the same:

https://goo.gl/Mhq69H

https://www.amazon.com/Beyang-Intel-x79-ATX-Motherboard/dp/B06WWHJHT4/ref=pd_sbs_147_1?_encoding=UTF8&pd_rd_i=B06WWHJHT4&pd_rd_r=Z7FAMF066A0QEG53VKKC&pd_rd_w=z9Bp3&pd_rd_wg=YhaUI&psc=1&refRID=Z7FAMF066A0QEG53VKKC#customerReviews

I'm going to give it some more thought as I'm headed out of town
tomorrow. But for the price I think this is worth a shot. We shall
see. And thanks for pushing me toward a heat sink.