Overclock.net › Forums › Intel › Intel Motherboards › ASUS X99 Sabertooth Owners Thread
New Posts  All Forums:Forum Nav:

ASUS X99 Sabertooth Owners Thread - Page 67

post #661 of 1958
Originally Posted by Lykorian View Post

Those are nice speeds xkm, with decent voltage too, is it stable?

What did you use for stability and for how long?

ASUS Realbench 12hr stable. So pretty good.

I wish I can say the same for my FuryX OC. Ever since I switched to 16.1 I have been having all kinds of problem during OC.
post #662 of 1958
Oh...ASUS Realbench.

Thank you.
post #663 of 1958
May have killed my ram slots. While working on mods for builds took the armor off i used the mobi no problems for a while. Had to test fit the mobo back in case with armor. When i was done i put it back in my old case. Well the first boot i realized i plugged the wrong drive in the wrong sata spot so was attempting to boot with no os on the drive. I shut down switched now nothing. The red dram light comes on, ive gotten 00b1 for post code and 0053. I tried removing ram from slots a1, b1, and c1 nothing, tried removing one by one nothing. Any help is appreciated.
Edited by Dave6531 - 2/1/16 at 3:29am
post #664 of 1958
Had a couple of days hands on woth this board and I am impressed how it can cope with the RVE on Core and Cache OC.

Can't wait for the other 5 to be set-up.

Had all 7 5930Ks tested on the RVE for max clock allowed. All chips have no issue running Real Bench X264 Encode with 4.5GHz at 1.25 Vcore. With the exception of the 2 that do 4.6 at same mentioned Voltage.

Same OC and Voltages apply to the Kitty. VRMs have no issues.
Edited by mus1mus - 2/1/16 at 6:38pm
post #665 of 1958
Is it also Prime95 Blend, OCCT and Intel Burn Test stable?
post #666 of 1958
No way I'm killin 7 chips. doh.gif
post #667 of 1958
No way those 7 chips are stable at your rated speed and voltage.

post #668 of 1958
Originally Posted by Lykorian View Post

No way those 7 chips are stable at your rated speed and voltage.


you have no idea. buttkick.gif

Edited by mus1mus - 2/1/16 at 6:49pm
post #669 of 1958
Haha, ASUS RealBench FTW!
post #670 of 1958
If you think Prime95, OCCT or IBT or AIDA64 (first 4 tests +GPU test) are killing your CPUs, we shouldn't be having a conversation to begin with.

But I will indulge for the sake of others that are looking for an honest observation of Haswell-E (5820k in this case) when it comes to overclocking.

The reason I will write the following is because over the past months I've seen alot of misleading information on the internet about the overclockability potential of the Haswell & Haswell-E CPUs.

And it's all because of this article:


While that article is great for starters and a general guideline...

IT DOESN'T reflect the truth for 24/7 stable machines, VERY FAR from it actually.

- unless you have a CPU from that very rare batch - under 10% - AND that still wouldn't qualify as 24/7 stable by testing with ASUS Realbench alone and overclocking via software - applaud.gif

It may reflect some truth for benchmarking purposes aka for those people who want to reach a certain frequency in order to obtain a certain score in a certain program.
Obtaining stability for a few seconds or minutes in order to get a certain highscore and qualify in a top 10-20 etc is ANYTHING BUT 24/7 stability.
Yes, I know you said 12h+ ASUS Realbench. ROFLMAO.

How do I know it's less likely that your CPUs are stable, besides your attitude?
Because everyone is posting similar results everywhere, with surreal voltages, temperatures, noise levels and whatnot (many even on air cooling).
All this independent of cooling, mainboard, PSU, case, voltages and all the other parameters that make up for a truly 24/7 stable machine.
Which means everyone on the internet forums must've have had divine intervention upon them and thus receiving CPUs out of that 10% quality margin.
They're so damn rare that everyone has them. Right.

Also, I happen to have the same "stable" config as you and tons of other enthusiasts.
Guess what, I'm not a gratuitous believer in ASUS Realbench's prowess (or any other ONE-ONLY-stress-test only solution).

But I digress.

I've been toying with many different Haswell-E CPUs for the past months and while I initially & reluctantly jumped the ASUS Realbench wagon for stability testing, I obviously found out it counts for some but mostly for nothing.
I could run ASUS Realbench on 1.25-1.3 volts on many if not most CPUs (both adaptive and direct - didn't try offset mode yet) and obtain 4.5-4.6Ghz for 24h+ hours but fail OCCT, Prime95, AIDA64 and IBT (usually Prime95 and IBT).
Why? Because opening multiple Photoshop and Adobe Premiere "rendering" instances means poop.

Since I've almost filled a A4 paper size notebook with settings, results and various stuff from the past 5-6 months, I will only post some screenshots of the latest CPU (5820k) which has been in stress-testing for the past 3-4 weeks.This CPU didn't differ much from the others, quite the contrary.

Some of my reasonable results of the very, very, very many:

CPU & MEM: 4.25Ghz (125x32) - XMP (2666-13-14-14-39) - UNCORE/CACHE 3Ghz (NO CACHE OC) (if anyone should care, this CPU's VID=0.8250)

Input Voltage=1.970
Core Voltage (adaptive): +0.222
VSSA (adaptive)=+0.300
UNCORE/CACHE (adaptive)=+0.400

VRM Spread Spectrum=Enabled ---> CPU Power Phase=Optimized
LLC=Level 9
CPU Integrated VR Fault Management=Disabled
CPU SVID Support=Disabled
VRM Spread Spectrum=Disabled
C States and EIST Enabled (I like the CPU to rest when not under load, hence the adaptive on voltages too)

- room temperature for all the following 3 tests varying from 25.5 Celsius to 28.5 Celsius, no open windows, doors to outside or AC running
(you bet those details matter!)

PRIME95 - BLEND.25h+ (same stability goes for Small and Large FTTs)

OCCT (same as above, same as above, yes I know, forgot to print HWiNFO64 information, voltages vary from PRIME to OCCT to AIDA etc...)

AIDA (same as above, yes I know, forgot to print HWiNFO64 information, voltages vary from PRIME to OCCT to AIDA etc...)

Since everyone is posting their successes, it's about time someone posted a failure.
Same settings as above, except what's underlined in red:

CPU & MEM: 4.25Ghz (125x32) - XMP (2666-13-14-14-39) - UNCORE/CACHE 4Ghz (1GHZ CACHE OC)

Everything looks in order, right? All workers stoppped after 21h:40min with 0 errors.
Wrong, take a look in HWiNFO64 at Core #2-Thread#0 usage ---> one worker stopped working in between 21h and 21h:30min.

Therefore test FAILED.
You think this would pass ASUS Realbench 12h+ ? thinking.gif
It passes an eternity of that + everything else but I still wouldn't call this STABLE.

All CPUs have been subjected (and some still are, running Prime95 on one of them as we speak) to the same kind of tests (alongside HCI Memtest 1000%, Memtest86 all tests + torture and some others) and I'm not seeing anywhere near your and others' performance levels for 24/7 STABLE (emphasis on 24/7).

Yes, 3 of the past CPUs are now dead and I'm writing this stuff because I'm in mourning and feel like sharing the pain is going to alleviate some of it. upsidedwnsmiley.gif
One of the PSUs also exploded because of Prime95 with AVX instructions. graduated.gif

(is joke bo$$)

Anyway, hopefully the future "summer" profile for one of them (Prime95 BLEND running right now).
Why "summer profile", because I like stable & silent computers even when it's 40 degrees Celsius outside with AC running in the other side of the house.
Computers with less or equal to 7 fans, as I'm not a housemaid to be dusting furniture everyday.

CPU & MEM: 4Ghz (125x32) - XMP (2666-13-14-14-39) - UNCORE/CACHE 4Ghz (125x32)

Input Voltage=1.8
Core Voltage (adaptive): +0.125
VSSA (adaptive)=+0.300
UNCORE/CACHE (adaptive)=+0.300

VRM Spread Spectrum=Enabled ---> CPU Power Phase=Optimized
LLC=Level 9
CPU Integrated VR Fault Management=Disabled
CPU SVID Support=Disabled
VRM Spread Spectrum=Disabled
C States and EIST Enabled (I like the CPU to rest when not under load, hence the adaptive on voltages too)

Room temp varying from 25.5 to 26.8 Celsius (no windows opened, no AC running)

Running Prime95.BLEND (timer is in RealTemp window, 19h+)

I'm not here to prove you wrong, you know you can't run Prime or Intel Burn Test and that matters to nobody but you in fact.
What matters is what other people might draw from your statements, which are very misleading.

In other words, I would like to think I am here to contribuite with a correct interpretation to one of the most popular articles about Haswell-E regarding its overclocking
- and I'm repeating myself - namely this:


... but also would like to add 2 more realistic, independent, less towards-ASUS-biased, more down-to-earth articles on the same topic, that will, in the end,
lead users to a better understanding of what they're doing and most importantly, what they're trying to obtain out of their machine:




+ (a bit more on the general side)


Don't get me wrong, I can understand the psychological comfort one gets from running a certain stress test and passing (be it for 1, 2, 3, 6, 12, 24h etc..).
Wishfully calling that ONE result as a stable computer is well...wishful thinking (LESS emphasis on thinking).
I truly wish it would be that easy but things are far more complex than that, reason why one needs different test, for different purposes for as many hours one has the time for.

Also I can sympathize with bragging rights about a certain overclock result.
(I get to change my signature in some forum, I ***** slap some noob who runs stock etc..mature stuff)
What I can't sympathize with is arrogance and ignorance when it comes to a 24/7 stable overclock result.

This ain't religion mate, you have to have proof.gif
Yes, I know, ASUS Realbench told you so, it's wrriten there in the results log.popesmiley.png

I will come back in another post with some observations about BIOS 2101 and some of its peculiarities (some may call them bugs and rightfully so).
Also I might have found out some some workarounds, for whom it may concern.
I see some people here have been troubled with some of those issues but in my experience, it's ASUS standard (yes, it counts for RVE too but in a far lesser extent).


Did you by any chance overclock via ASUS AI Suite instead of BIOS?

The only one you could brag your ASUS Realbench results to and also be impressed is someone who thinks Haswell-E is the latest boy band.
Edited by Lykorian - 2/1/16 at 10:36pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel Motherboards
Overclock.net › Forums › Intel › Intel Motherboards › ASUS X99 Sabertooth Owners Thread