Official Haswell-E thread

[AdamK47]

The first Gigabyte X99 Gaming G1 I bought had obviously bad components around the VRM area. There was a very noticeable squealing sound immediately after turning the system on. Fortunately, it didn't kill my 5960X. My new Gaming G1 has been running without issues.

I don't plan to ever go nuts with voltage and clock speeds on the CPU or memory, so I hope nothing happens. Those looking to ramp up the clocks and volts with water cooling may want to watch out.

 

[guskline]

The first Gigabyte X99 Gaming G1 I bought had obviously bad components around the VRM area. There was a very noticeable squealing sound immediately after turning the system on. Fortunately, it didn't kill my 5960X. My new Gaming G1 has been running without issues.

I don't plan to ever go nuts with voltage and clock speeds on the CPU or memory, so I hope nothing happens. Those looking to ramp up the clocks and volts with water cooling may want to watch out.

From what I am reading I think the mbs may need beefed up for OCing. Your 4.125 Ghz OC sounds safe, solid and fast for 8 cores/16 threads.

 

[WhoBeDaPlaya]

pm would probably be the best person to weigh in on this, since he worked on the power delivery for Haswell-E.

 

[biostud]

hmm doesn't sound good.

 

[EvilNodZ]

Thats not good news. Im defo not pushing my motherboard any harder until they get to the bottom of this.

 

[AdamK47]

I just checked the user reviews on Newegg and it looks like I wasn't the only one that got a whiny Gigabyte X99 Gaming G1. My new board is much better though.

 

[Bubbleawsome]

Dave over at TPU said that the ASUS OC socket really does help keep the CPU stable. He seems to be getting higher clocks out of most of the chips he has for reviews.

 

[pm]

pm would probably be the best person to weigh in on this, since he worked on the power delivery for Haswell-E.

What should I weigh in on?

 

[WhoBeDaPlaya]

What should I weigh in on?

Whether there's a problem with the FIVR, or are mobo manufacturers cheaping out on their X99 VRMs.

 

[pm]

Whether there's a problem with the FIVR, or are mobo manufacturers cheaping out on their X99 VRMs.

To be honest, my expertise is below the package level down on the silicon and so board level issues are pretty far above what I worked on - actually the part that I worked on was "levelshifting" signals crossing voltage domains on the die as well as making sure that analog and digital circuity works as the supplies come up and go down... I didn't actually even work on FIVR - my stuff was pretty esoteric.

I will say that I have worked with engineers from Gigabyte in the past and they are a pretty solid group so I wouldn't want to say anything bad about them... but given how FIVR works, I'd be more inclined to blame the VRM (hard to picture how the problem described could be due to FIVR but it's possible, I guess). My guess is that it's a bad VRM - some sort of resonance issue maybe... parts that kind of work can be hard to detect in the manufacturing line - and so I would recommend replacing the board as the user appears to be doing.

* As always, I'm just a random engineer who works for Intel and who posts on Anandtech... my opinions are my own

 

[AtenRa]

Dave over at TPU said that the ASUS OC socket really does help keep the CPU stable. He seems to be getting higher clocks out of most of the chips he has for reviews.

Gigabyte has two sockets as well. They dont use 2011-3 for the OverClocking motherboard models

 

[WhoBeDaPlaya]

...
* As always, I'm just a random engineer who works for Intel and who posts on Anandtech... my opinions are my own

Thanks for the informative post! Multi-power states/domains are a real PITA at the back-end, especially since CPF/UPF handling don't seem to be quite mature/robust just yet. Then again, SDCs (as an example) are ancient and still allow for plenty of PEBCAK issues

Not even going to touch on functional verification in multi-power scenarios...

 

[RussianSensation]

The first Gigabyte X99 Gaming G1 I bought had obviously bad components around the VRM area. There was a very noticeable squealing sound immediately after turning the system on. Fortunately, it didn't kill my 5960X. My new Gaming G1 has been running without issues.

I don't think the quality of the components on the X99 Gigabyte boards is bad, it's just the chokes they use are known to exhibit coil whine under load. The chokes Giga used are similar to the ones from 7950/7970 reference cards -- chokes by Cooper Bussmann. While they have a proven track record of rock solid 24/7 100% operation under strenuous loads given how many thousands of 7950/7970 withstood mining for years and years, they have 1 negative side effect of often squealing/buzzing.

Here is a shot of the reference 7970 PCB:

Just 5 of these are used for the GPU and 2 for the memory which is a cakewalk for a 275W load on this card for 7 chokes. The Gigabyte X99 G1 has 8 of these which means 300W+ load on the CPU for this board is a not a problem for 24/7 100% load operations for many years. However, why Gigabyte went with non-solid/non-alloy choke is puzzling indeed:

 

[Techhog]

Here are some tests with a retail chip. Still looks decent.

https://www.youtube.com/watch?v=o7SSacvmTVo

 

[FiLeZz]

My parts arrived today, However so did the twins my wife has been cooking in her tummy.

This will delay my put together by a day or two. No biggy, I am just happy to finally be a DAD.

 

[Earwax]

I got the ASUS X99 Deluxe, 32GB (4x8GB) GSkill 2400 DDR4 memory and the Nepton140XL closed loop cooler they used here for the review. I can't get this thing to OC well at all. Perhaps I'm just way too out of practice, maybe I should reset the CPU and redo the thermal paste.

Currently at stock when I run OCCT my cores are hitting the low to mid 60s. My idle temp is also around 33C. Am I doing something wrong?

I briefly tried putting the core voltage up to 1.2 and running it around 4GHz, but literally any increase in the voltage results in the cpu going overtemp in under a minute under stress testing.

Any suggestions? It's been a while since I put a system together.

Additional Specs:
EVGa Geforce Titan Black Signature Series
Corsair 1200i PSU
Samsung EVO 500GD SSD
Win 7 x64 Ultimate

EDIT: BIOS update solved all my issues, was able to get a stable 4.1GHz clock at stock voltage. Curiously 4.3GHz caused windows crashes at 1.1, 1.2, and 1.3 volts but that may be due to my RAM. I can live with 4.1 at stock voltage.

 

[AdamK47]

That is odd. Especially so considering the latest OCCT doesn't put nearly as much load on my 5960X as the latest version of LinX. I'd try to reseat the cooler.

 

[Zucker2k]

My parts arrived today, However so did the twins my wife has been cooking in her tummy.

This will delay my put together by a day or two. No biggy, I am just happy to finally be a DAD.

Congratulations, man! I can only imagine the excitement you must surely be feeling recently.

 

[Det0x]

Have anyone seen some information related to about what stepping/batches to look for ?

I have just picked up my new 5960X, and the stepping+batch are the following:

SR20Q 3429A714

Most of the CPU's the reviewers got was QFRA as far as i know.. Does anyone know the difference between SR20Q and QFRA ?

Iam still waiting for my EK-KIT X360 set, so i haven't had the time to test it yet.. So iam wondering if iam sitting on a above or below average CPU

*edit*

I actually managed to find one guy on a Norwegian hardware forum with the exact same batch as me, and he is running Cinebench R15 @ 4.625 ghz with 1.3 voltage..

Seems to be my lucky day.

 

[Earwax]

BIOS Update fixed all of my issues, I can run it at 4.1GHz at stock voltage now and it tops out around 80C after 24 hours of constant load. Good enough for me.

I should have tried the BIOS update first, lesson learned.

If you are purchasing an ASUS X99 Deluxe, be sure to flash to most recent bios before doing any stress testing--the current release updates a few glitches with temperature reporting.

 

[Grooveriding]

My X99 deluxe finally arrived and the top right corner opposite the IO area had a slight bend in the PCB in the area of the mounting hole. Tried to boot it and got a mix of 61 and 53 q codes. Returning it and should have a replacement on Monday.

edit: I tested my ram and CPU in a local shop and same issue with q code 53. They swapped a different CPU in and it posted fine. May not even be the bend in the board I received. Can you actually get a brand new dud CPU ? And could it manifest as a memory issue at post ?

 

[FiLeZz]

CPU: 5960x
S-Spec: SR20Q
Batch#: 3424B585
4.374GHz @ 1.3V
memory @ 3000 with 15-15-15 timings.
Not enough time to go further yet. Temps on the loop stay at 32C.
I am sure there is more room to push the chip as the temps are nice and low.

 

[Burpo]

Nice FiLeZz! Sounds like you got a good one

Can you actually get a brand new dud CPU ? And could it manifest as a memory issue at post ?

Of course..

 

[Grooveriding]

Nice FiLeZz! Sounds like you got a good one



Of course..

I have never even read an anecdotal account of a DOA CPU. :| What a double whammy. Damaged mobo and DOA CPU.

Pic of mobo damage and DOA CPU Stepping.

 

[AdamK47]

I would suspect the board killing the CPU.