And let the PC build begin ...

[Cruachan 705]

Hi Rob,

 

The build is now finished and, very gratifyingly, booted first time without any issues So, I must have done something, indeed several somethings, right! In the end, I decided on using a fan controller rather than plugging straight into the board headers.

 

Now that the honeymoon period has ended I have encountered my first wobble and am looking to the master, once again, for guidance and also, hopefully, some reassurance that all is well thus far. The main problem is, of course, my inexperience with such matters and despite all the research I am having difficulty understanding exactly what is going on at this point.

 

I have tried to be methodical in my approach:

 

ASUS Rampage V Extreme

i7 5960X

UEFI BIOS Version 1601 (updated from 802)

LN2 Mode: Disabled

CPU Cooling: Corsair Hydro Series™ H110i GT 280mm

 

Samsung 850 EVO 1TB SSD x3 and LG BH16NS40 16x SATA Internal BDR: connected and recognised.

G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin DDR4 SDRAM DDR4 3000 (PC4 24000) Desktop Memory Model F4-3000C15Q-16GRR: in place, but not yet tested individually in Slot A1

 

Operating System: Not installed

 

First Cold Boot into UEFI BIOS okay

Q-Code: CE

 

1. Load Optimized Defaults -> Save Changes and Reset

ReBoot into BIOS okay. Ditto with several Resets and a further Cold Boot.

Q-Code: CE (I believe this code confirms successful entry into the UEFI BIOS)

 

2. Set Ai Overclock Tuner: Auto -> XMP (no other changes made)

Screen shows:

XMP: XMP DDR4-3000 15-15-15-35-1.35

CPU Strap: 125MHz

BCLK Frequency: 125.0

 

Save Changes and Reset -> Message: "Overclock failed! Please enter Setup to reconfigure your System

Press F1 to Run SETUP"

Q-Code: A2 (IDE Detect)

 

4. Press Retry Button - System powers on and off twice before continuing to POST -> Message: "Overclock failed! Please enter Setup to reconfigure your System

Press F1 to Run SETUP"

Q-Code: A2

 

6. Load Optimized Defaults -> Save Changes and Reset

POST and Boot into BIOS okay

Q-Code: CE

 

7. Set Ai Overclock Tuner: Auto -> XMP

Changed CPU Strap: 125MHz -> 100MHz (BCLK changes automatically to 100.0)

Save Changes and Reset -> POST and Reboot into BIOS okay

Ditto with several Resets and a further Cold Boot

Q-Code: CE

 

Hardware Monitor:

CPU: Frequency 2800MHz, Temperature 32 C, Core Voltage 1.152V (last 2 digit values fluctuate slightly)

Memory: Frequency 2400MHz, Vol_CHAB 1.356V, Vol_CHCD 1.356V (last 2 digit values fluctuate slightly)

Voltage: 12.192 - 12.288, 5.080, 3.312

 

I notice sometimes the Boot_Device_LED on the MoBo is lit (not unexpected) but at other times it is not.

 

Does this all seem normal to you Rob? I'm guessing that the "Overclock failed" message is being caused by the XMP profile increasing the CPU Strap/BCLK setting without the associated required increase in voltages. Any overclocking guide I have read simply advise applying the Memory XMP profile without any qualification, yet I have this issue which resolves when I reduce the CPU Strap back to 100MHz.

 

At what point should I be running Memtest86 on the memory modules? Should it be done before applying their XMP profile or, as in this case, afterwards with the CPU Strap at 100MHz?

 

Cheers!

Mike

[Guest]

 

 

but not yet tested individually in Slot A1

 

Not sure what you mean here, you have all 4 DDR4 RAM cards in their appropriate slots for Quad channel (manual will show you want slot pairs to use) but should look like my pics in this thread.

 

Layout RAM as in this pic: 

 

 

You didn't mention what your multiplier is?  125Mhz X ??  ... did you set to 35?  Assume you tried the same EFI settings I listed in this thread?

 

Is yes, try increasing the CPU core voltage in small increments.

 

Check the Cache ratio (you'll not be able to run that at 35 ... I seem to recall Auto or 27 value).

 

Make sure you powered up the PCIe with that extra power connector on the opposite side from the VRMs.

 

The Q-codes are listed in the back I believe ... there are also other amber diagnositic lights near memory channel (manual goes over what those mean).

 

Cheers, Rob.

[Cruachan 705]

Hi Rob,

Yes the memory sticks are currently in their correct slots: A1,B1,D1,C1 (the red slots). I am intending to follow the advice given whereby each memory stick is tested in turn using only the first slot, Slot A1. As each stick is marked as passed (hopefully) I then move on to install and test the next module in slot A1....and so on. If ALL sticks fail I would then try the same series of tests using either C1 or D1 (not sure which) to clarify whether or not the A1 slot is faulty.

The multiplier, at the moment, appears to be 28X. The BCLK is 100.0

When I Cold Booted to verify the multiplier the system again failed to boot with the familiar message: "Overclock failed! Please enter Setup to reconfigure your System
Press F1 to Run SETUP" However, it succeeded after pressing the MoBo 'Retry' button.

As I said, I have just started this particular phase of the adventure so no BIOS settings have been modified apart from the Ai Overclock Tuner which is now set to XMP.

Should I have reduced the CPU Strap at this stage? I only did it to achieve a successful POST and BOOT into the UEFI BIOS.

Then there is the matter of when I should run Memtest86 on each of the memory sticks. I have been assuming that each stick should be tested while using its XMP profile. Perhaps that is not the case.

I see your updated cooling solution is now in place. Looks very nice! How is it working out?

 

A few images of my new build:

 

 

 

 

Cheers,
Mike

[Guest]

 

 

I am intending to follow the advice given whereby each memory stick is tested in turn using only the first slot, Slot A1.

 

Not a process I would do on a quad channel motherboard ... leave all 4 sticks in, then run RealBench (comes with Asus software suite).

 

Uses the EFI settings screenshots I posted earlier in this thread ... XMP profile should default you to 125Mhz BCLK if the RAM is 3000Mhz rated at 15.

 

Start with 33 X 125 with CPU core at v1.31.

 

My setup is working out very well, the VRM can get VERY hot that's why I went with dual loop and monoblock to cover CPU, VRM, PCH and other loop for GPUs.  Can fly all day and night at 4.5 Ghz @ 3000Mhz with 125 X 36 and rarely see more than about 113F ... I'm running SGSS AA now so I OC'd the GPUs also which helps considerably with SGSS AA (especially SLI) ... in worse case they rarely get above 103F.  

 

I do lock my CPU frequency, not using Intel speedstep ... IMHO consistent temps (frequency) are less likely to cause failures/errors than temps that continually changing from high to low (Visa versa) - material fatigue (repeated heating and cooling of materials can change their structure).

 

Cheers, Rob.

[Cruachan 705]

Not a process I would do on a quad channel motherboard ... leave all 4 sticks in, then run RealBench (comes with Asus software suite).

 

Uses the EFI settings screenshots I posted earlier in this thread ... XMP profile should default you to 125Mhz BCLK if the RAM is 3000Mhz rated at 15.

 

Start with 33 X 125 with CPU core at v1.31.

 

Cheers, Rob.

Leaving all 4 sticks in and running RealBench implies the installation of Windows before knowing the RAM is good. That's why It seemed reasonable to check out the RAM first before going any further.

 

The XMP profile does default the system to BCLK frequency 125.0 and it won't Boot to the BIOS - see item 2 in post #331. However, this may have been down to insufficient CPU core voltage being applied. I'll do as you suggest and report back.

 

Any thoughts about the Cold Boot failure using conservative settings (post #333)? Anything to do with LN2 being disabled, perhaps?

 

Cheers,

Mike

[Guest]

Hi Mike,

 

Suggestions:

1.  Get a working default EFI profile going (all Auto, no XMP) and save that profile, then install Win7 and all the Asus drivers/software

2.  I'm still not clear what your CPU ratio is at ... that's a KEY setting?

3.  Have you tried my EFI settings I listed earlier for an OC profile?

4.  I can't tell from your picture if you have the power female connector (4  pin) connected to the PCIe male connector ... required for a good boot and dual GPUs

 

LN2 doesn't need to be enabled for water cooling but I have tried it ON/OFF didn't notice any difference.

 

Cold boot issue could be any number things (from CPU Cache ratio to PLL Filter to not enough CPU voltage or DRAM voltage, but suggest you answer/try the above first.

 

Cheers, Rob.

[westman 257]

If it not boot at 125strap 3000mhz XMP mem with mult 33 4125mhz.

Probaly mem controller on CPU raise system agent voltage

[Cruachan 705]

Hi Mike,

 

Suggestions:

1. Get a working default EFI profile going (all Auto, no XMP) and save that profile, then install Win7 and all the Asus drivers/software

2. I'm still not clear what your CPU ratio is at ... that's a KEY setting?

3. Have you tried my EFI settings I listed earlier for an OC profile?

4. I can't tell from your picture if you have the power female connector (4 pin) connected to the PCIe male connector ... required for a good boot and dual GPUs

 

LN2 doesn't need to be enabled for water cooling but I have tried it ON/OFF didn't notice any difference.

 

Cold boot issue could be any number things (from CPU Cache ratio to PLL Filter to not enough CPU voltage or DRAM voltage, but suggest you answer/try the above first.

 

Cheers, Rob.

Hi Rob,

 

1. Will do. When you say 'save the profile' do you mean via the Exit (F10) menu or is there some other way of saving and recovering a previously configured profile?

2. At the moment it is sitting at 28x but this will soon change as I implement 1.

3. Not as yet.

4. If you are referring to item (D) on page 1.42 of the Motherboard manual then, yes, the EZ_PLUG is connected.

 

I asked about the LN2 Mode Jumper as it states on page 1.34 of the Motherboard manual:

"With LN2 Mode activated the ROG motherboard is optimized to remedy the cold-boot bug during POST and help the system boot successfully"

However, I can see that this may not be relevant right now until other possibilities have been eliminated.

 

My setup is working out very well, the VRM can get VERY hot that's why I went with dual loop and monoblock to cover CPU, VRM, PCH and other loop for GPUs. Can fly all day and night at 4.5 Ghz @ 3000Mhz with 125 X 36 and rarely see more than about 113F ... I'm running SGSS AA now so I OC'd the GPUs also which helps considerably with SGSS AA (especially SLI) ... in worse case they rarely get above 103F.

Cheers, Rob.

113F = 45C. Wow! That is indeed very impressive. You must be feeling very pleased that all your efforts are reaping these rewards. Looks like you are definitely sorted

 

Many thanks!

 

Cheers,

Mike

 

If it not boot at 125strap 3000mhz XMP mem with mult 33 4125mhz.

Probaly mem controller on CPU raise system agent voltage

Hi Westman,

 

Thank you. I'll certainly keep your helpful suggestion in mind.

 

Regards,

Mike

[HamSammich 17]

Also try this. It bedeviled me, too with a Rampage V--so much so I sold it and stuck with my z97-WS.

 

Naturally, I learned about this only AFTER I sold the board. Seems a number of Asus X99 boards get stuck when scanning your USB devices, The typical symptom is a VERY long pause on the Windows splash screen, then the old Overclock Failed message.

 

It baffles people because 1) it's intermittant and 2) once they get the system booted, it passes all the stress tests at a good overclock.

 

Some gamer ran across it and documented it in a rambling and hard-to-watch Youtube video. Asus was also stumped.

 

Turns out the problem goes away if you set the UEFI to Fast Boot and to NOT initially touch any USB devices on start-up. Which, of course, is counter-intuitive. You'd expect that when you're have trouble booting, slowing everything down and going Old School, as it were, would help.

 

Instead it's basically telling the machine to skip everything and go straight into the boot-loader.

 

This may be ENTIRELY extraneous to your problem. If so, apologies.

 

Marshall 

[Cruachan 705]

Turns out the problem goes away if you set the UEFI to Fast Boot and to NOT initially touch any USB devices on start-up. Which, of course, is counter-intuitive. You'd expect that when you're have trouble booting, slowing everything down and going Old School, as it were, would help.

 

Instead it's basically telling the machine to skip everything and go straight into the boot-loader.

 

Marshall

 

Hi Marshall,

 

This is very useful info - thank you! I think the UEFI BIOS defaults to FAST BOOT when the optimised defaults are loaded. However, I'm not sure I understand what you mean by 'not initially touching any USB devices during start-up'? Are you referring to, for example, moving the mouse?

 

In fact, apart from a little dabbling to familiarise myself with the UEFI BIOS I haven't, as yet, attempted any overclocks with voltage tweaking. That part of the adventure is yet to come

 

Right from the start I decided to take my time while carefully considering each and every step with this build. So far, this seems to have been the right approach as everything has worked without the need for any modification or returns to the drawing board to work out what I've done wrong.

 

I have a feeling that the cold boot issue (never a problem while running with optimised default settings) is related to insufficient voltages - that I have yet to confirm.

 

Memory checks with MemTest86 V6.1.0 have completed and all 4 sticks have passed with flying colours! MemTest86 was run from a bootable USB drive. I didn't have to change boot priorities - it ran automatically and started the test run on each installed stick. It runs 4 full passes and completes after roughly 4 hours for each stick, 16 hours total. I decided to test one stick at a time as it would make it easy to identify which stick was faulty in the event I needed to RMA for replacements. My understanding is that you do have to replace all sticks in a matched set.

 

As an aside, I noticed a wee problem during testing: I thought I would switch off the monitor rather than have it display a static image for several hours at a time. Switching back on after a few minutes was fine and the MemTest86 display returned. However, trying the same after, say, an hour I was presented with the 'No Signal' message!? So, the monitor was left on thereafter and that was okay - I thought perhaps it might drop into a power saving mode (not likely outside Windows but, as yet, I am not fully familiar with the workings of the Asus ROG Swift PG278Q) so I quickly learned the MemTest86 'ESC > 3 > X' key combo to get me out of trouble.

 

Fortunately all is well and now I can proceed to the next stage which is installing Windows 7 Pro 64bit. I will do so while keeping the UEFI BIOS at optimised defaults.

 

Regards,

Mike

[HamSammich 17]

Mike,

 

The setting in the UEFI is found in Boot|USB Support. Choose Disabled. This means USB devices don't get initialized until the OS loads and does it. The upside is it seems to cure that hanging-on-boot syndrome I mentioned. The downside is you can't get into the UEFI on POST by hitting the F1 or DEL key. You need to either used the Boot Setting Utility from Asus in Windows.. or hit the BIOS Direct button on your MB.

 

Marshall

[Cruachan 705]

Hi Marshall,

 

Okay, understood. Thanks.

 

Is the 'BIOS Direct Button' the same as the 'SAFE_BOOT' button in the Overclock area of the R5E MoBo? My understanding is that this gets you back into the UEFI BIOS while preserving your latest settings. One of the many features which, together, make the R5E a truly impressive board!

 

Regards,

Mike

[Guest]

Interesting, I've never experience a USB boot issue with my Asus Rampage V board and I have 28 USB devices connected with a mix of 3.0 and 2.0 and even 1.1 devices.

 

But ran a quick test with FAST BOOT enabled and then disabled, no difference with cold or warm boot with my setup.

 

I am running an additional USB 2.0 5 port PCIe card because USB 3.0 ports are limited to 96 devices where as USB 2.0 has 255 devices (note device count is per port and not necessarily how many actual USB devices you have attached).  As you can see from the image below LOTS of USB 3.0 ports but not many USB 2.0 ports ... even though USB 3.0 is backward compatible it's limited to 96 devices and you can hit this limit quickly if you connect a few hubs which use up many device counts ... I hit the limit at only 12 USB devices physically connected going thru 2 HUBs.  Hence why I got the USB 2.0 PCIe card.

 

I don't use the USB/ROG port at all ... don't know if you are, might want to try booting by removing any USB device you may have plugged into the USB/ROG port.

 

 

Cheers, Rob.

[HamSammich 17]

Rob:

 

It is indeed odd-- I do know it happened to me with a Rampage and an X-99 Deluxe. A Thrustmaster Warthog and, at the time, two Saitek Quads and rudder pedals.

 

What egregious, noisy, wobbly, crap Saitek makes, IMO.

 

In any case, the problem manifested itself for me and vanished when I went back to the Z97-WS, which has been bulletproof. I stumbled across the USB boot solution by accident. If I had to guess, Asus went through some kind of funky batch of controller hardware/firmware or something similarly subtle from a part supplier. Whatever it was certainly didn't affect everyone. In fact, if affected so few people I'd imagine after-sales QC measures didn't capture it--it was lost in the noise..

 

If I had to guess, it was in the USB controller/controllers, because the problem appeared even with no devices but a keyboard attached. Of course it could have been the keyboard itself--or any keyboard, because I tried five.

 

Anyhow--just something to explore that only takes a minute and costs nothing.

 

Unless you do what I did and sell your board--after buyting and selling a 5820K to swap in, in case it was a bad CPU.

 

Grrrrrrrrr.

 

I really need my 5960X back--All those cores made texture-loading SOOOOOO seamless.

 

Marshall

[Cruachan 705]

Hi Rob,

 

Happy New Year!

 

Thought you might be interested to hear that I have finally taken the plunge and overclocked to 4.4GHz (4399.0) stable with Vcore=1.310 (also stable) using x44 Multiplier.

 

However, and this is the main reason for my follow-up post, I confess I wimped out by electing to follow the automatic route using the ASUS AI Suite 3. Prior to making the decision I carried out a fair bit of research and, as part of that, came across this interesting article:

 

http://m.hardocp.com/article/2014/12/08/asus_x99a_lga_2011v3_motherboard_review/7#.VofFzq_fWK3

 

In particular, you might like to look at paragraph 11 which starts with the word 'Unfortunately'. I was a tad wary about increasing the CPU Strap to 125MHz and it was this revelation that persuaded me down the automatic route while leaving the Base Clock at 100MHz.

 

I have flown Prepar3D v3.1 under changing conditions (at relatively high in-sim settings and using NI 4xSGSS) for quite long periods while monitoring the 5960X and both GTX980s. All is well. CPU temp rarely rises above 40 degreesC (104F) and both GPUs never exceed 68-70 degreesC (158F), while their usage can vary between 30-40% all the way up to 99% depending on what they are being asked to do by the sim. No anomalies or screen display artefacts have been encountered. As an aside, yesterday my 12 year old grandson showed me how to play 'Project Cars' over a 2 hour period and the performance and visuals were fantastic throughout with frame rates often exceeding 100fps. Again, the above noted temps were never exceeded.

 

By adopting the dynamic approach I note that my performance plan has been changed to 'Balanced'. This has the added advantage of throttling back the CPU speed at or near idle to 1197.7 MHz, Vcore=0.818, the multiplier drops to x12 and the system is virtually silent.

 

This overclock solution seems to be providing me with all the performance I am likely to need while providing a welcome level of hardware protection over the longer term. It has not allowed an increase in G-Skill memory clock to match yours, Rob, nor does it offer an option to clock the CPU higher unless I opt to change parameters manually. However, I do wonder how much more real performance I can expect for a further boost in DRAM frequency and/or a modest increase of 100MHz in CPU frequency. My intention was never to exceed 4.5GHz with my current Corsair H110i GT water-cooled 8 core CPU (HT Disabled) and I was concerned that an unvarying higher o/c would lead to distracting noise levels. This dynamic solution seems to me to be the ideal compromise for my situation.

 

Oh, and Prepar3D v3.1 runs great under SLI while remaining smooth as the proverbial. Actually it ran very well at stock speeds, but the overclock has provided some welcome additional overhead. Also, I've finally decided on sticking with G-Sync and unlimited frame rates rather than mucking around with VSync and frame rate locking, or not as the case may be. That may change, who knows?!

 

Cheers!

Mike