4930k/4960x Steppings and overclockability ?

[Lanek]

Not tested much, ( not really got the time with my work ), I have not test min Vcore and i use the stock intel air cooler ( my WB is not yet here ): set auto for vcore > XMP 1 ( 2133mhz C9 ) > 45x > boot: let the board do the work. ( batch is L3337C092 E4 ) this have give 1.28vcore ( but i have test with 1.20V not have seen much problem even if i have not test it with Lynx ).

[flesheatinvirus]

Nice, that looks good. Thanks for the batch info.

[Evantaur]

batch: 3331b509

struggling to get this stable at 4.5GHz @ 1.4v
currently sits at 4200MHz @ 1.328v

not sure if i'm missing something

[Doug2507]

Needs more voltage.

[Evantaur]

Needs more voltage.

i'm afraid to go pass 1.4v, you think it would be safe to go 1.5v with custom water?

+ it would be 22/7/386 at 100% load :P

[Doug2507]

it would be 22/7/386 at 100% load :P

?...24/7 365 100% load? I wouldn't even run it at 1.4v!

I think you'd be better off dropping a multi 1st and see where that get's you. I'd be surprised if you could get x44 stable relative to your needs if x42 needs 1.328v. It's just not that great a chip by the looks of things unfortunately...

[Evantaur]

?...24/7 365 100% load? I wouldn't even run it at 1.4v!

I think you'd be better off dropping a multi 1st and see where that get's you. I'd be surprised if you could get x44 stable relative to your needs if x42 needs 1.328v. It's just not that great a chip by the looks of things unfortunately...

i'll have to try to overclock 1 core at the time to see if there's a weak core or 3 somewhere, to see if i can salvage any performance out of this chip until i get more money to buy another one

[NEOAethyr]

I got my replacement cpu, a 4930k 3332B054.

And I checked my pump while my setup was down, it's fine.
I killed my cpu with a bad bios mod.
I just used mmtool to replace my raid rom that was it, seemed to work but killed my cpu.
The csmcore module differed from the modded bios on the rog forum, used the same exact raid rom though and supposedly the same method and tool... (apparently not the same tool or something...).

Anyways no bios modding for me for a while, not at least until I can figure out how to not screw it up.

My old cpu got 4.2ghz stock voltages, without enabling pll overvoltage.
Same settings, new cpu, 4.4ghz is stable.
Tested:
mafia 2 4hrs stable
fc3 1hr stable
batman 2hrs stable
codbo stable
codmw2 stable
sr4 stable

I now have my pump connected directly to the 12v on the psu it's self because I cannot trust the cpu fan header on the board.
It doesn't turn on the pump duriign post, I don't think it even turns it on after post, I gotta hook up a fan to it and find out but regardless the pump is not going back to being powered by that.
Using pk-1 tim on the cpu, I should of replaced that ancient as2 that was on my old cpu as sson as I found my pk-1 but I didn't, I don't think it would of mattered anyways.

I did see a slight burn mark, hardly noticeable on the backside on one of the pins of my old cpu before sending it in, when I was cleaning the tim off it with 99% alco.

Still bums me out, can't believe that happened.
I'm lucky as heck intel decided it was ok to rma it, they told me warranty void and my heart dropped.
Anyways there service was top notch, extremely fast shipping.

I wonder if it's possible to get the overclockers warranty on this replacement cpu, that would be nice.
I don't ever plan on going past stock volts anyways, never did...
Stuff can happened that's not planned for though :\ (Like a bios mod going nuts after booting windows...).
But I doubt they would warranty this.
I know I gotta take it easy, hence I'm not even gonna fool with bios mods, not even from other people.
I can't risk it.

I haven't messed with bclk yet, obviously not going to be using straps because there 100% useless.
I'm also still on the 0403 bios.
Haven't even tried the new bios yet.


Oh at 4.5ghz same settings, mafia2 bench is stable, batman bench is not stable.
So it's close to running 4.5 at stock but it's not 100% for sure.
Again no pll overvoltage (May help).

[Doug2507]

There's two of these chips over on OCN, both are good clockers.

[flesheatinvirus]

I just got in a 3326B652 yesterday for a fresh build this weekend, will post results later.

[Shiranui Gen-An]

I just got a RIVBE to pair with the 4930k I'd been running on my old RIVE and put it in my bench setup. I never did much experimenting before since it was in my 24/7 system, just 4.5GHz with about 1.35v or so (and I'm not sure if it actually needed that much). Initial stock test shows it using 1.136~1.152v when stressing, batch is 3325A911. Any initial appraisal for OC based on stock VID?

[NEOAethyr]

Just been messing with my vga cads haven't had the time to play with the computer for a while.

Just wanted to give a heads up.
Using auto on my r4be for my cpu voltage, at 4.4ghz, it sets 1.475v...
It only needs 1.2v...
Anyone with this board might wanna stay far away from using auto cpu voltage.

[flesheatinvirus]

I am running tests at 4800 with a remote liquid cooling system, so far I am at 4850 at 1.435 on a r4be.

[NEOAethyr]

The prob I have with mine is that I used to be stable at 2600mhz+ for mem and now I'm not, the mem will doit but the board won't let it be stable anymore.
All I did is swap 1 of my sticks around, chan's d1 and c1.
So I gave up for the time being, just trying to work on my vga bios'es.

I was able to soft strap to a 680mx from a grid k2 but the driver still knows it's a grid card.
Hence not allowing me to use dl-dvi on drivers past 319.xx (the 319's being the last I can use).
I have the bios for the k2 but it has the wrong mem size and no display outputs, so those have to be modded in if I wanted to go that route.
Otherwise I have to hard mod back into a 680 card, I saved the 2 resistors but I can't find them.
I haven't thought about pencil'ing them in but I'm not sure how long that would last.

Right now I'm using 1 vga card instead of the 2 total that I have for this system.

My 1st idea was to use a hybrid sli of sorts with one 680 and one k2, using the 680 for the display output.
But the hybrid sli patches don't work for mixing pro with consumer cards.
Next idea was to soft strap to the mobile 680 card, that worked but the dl-dvi is still disabled because of the hard straps.
Kinda sucks.

Plus I'm not to happy with the motherboard, it killed my 1st cpu.
They're not even fixing the bugs in the bios either.
I did figure out somewhat why my mem bandwith was 25% low in write speeds, but I can't make use of it right now when it's not stable anymore.
So I'm sitting here at stock speeds, at it's slow as heck, so much freaking lag on this board.
And audio is a joke, a gimmick, it distorts alot.
And that thump on power up is annoying.

[Dumo]

Old batch from local micro center

h2o

[Ace123]

I have a 4.85 xtu stable chip here..... still dissapointed

[NEOAethyr]

@ Dumo
What kind of ram is that?, that's quite the low voltage setup...


@ Ace123
Mine is 4.4ghz stable at stokc voltages it seems like, well at 100mhz bclk it is stable.
At 110mhz bclk as soon as I up it past 4.2ghz to 4.4, I lose 25% of my bandwith (aida64 mem writes).
And it stays that way until I load my 100mhz cmos profile and back to my 110mhz one.
It's obviously some voltage is to low but I haven't gotten around to try to figure that out yet.

I haven't clocked mine past stock voltages yet but I decent cpu.
It all around clocks better then my previous one, in bclk and cpu speed.

Ok that does it, I'm gonna swap my sticks back around and try 2hrs of batman lol, to see if I can get stable again like it was on the older bios.

Edit:
Spent 3 and a half hours working on 2600mhz mem stability (2646 actually).
Got batman stable, but after that I made some screens of mem tweakit then aida64's benchmark.
It hardlocked on aida64 and I lost all my usb.
It was close, odd that screwed up on something simple after all that.
I had to swap c1 and d1 ram sticks around to were they used to be, and set vtt and vtt2 to 1.1v, set vccda or whatever it's called to +0.1v.
Not sure which one it actually needed.
But if I were to guess, vtt2.

I may try it again tomarrow night.
I gotta find better tests, I'm sick of playing batman, I beat it like 3-4 times now.

[flesheatinvirus]

Nice one Dumo, same chip I got from Micro and mine will do 5G but needs da juice. My cheap set of Gskill though runs nice just above 2500 at 9-11-11-31.

[Ace123]

Sell me that chip lol. I gotta have it haha

[flesheatinvirus]

I hear ya, I got lucky I guess though it is running in my torture rack which keeps things rather chilly..

[NEOAethyr]

I think I got 2600mhz mem stable again finally.
It's gonna take another day or 2 to finish testing and to finalize a cmos profile for it though.

I went to figure out what I needed for higher cpu multi's at this bclk, normally I lose 20k in mem write bandwith.
So I up'ed my voltages and cpu speed and tested it, had the full bandwith cool, so I'm thihnking it had to be one of those voltages.
In the end it wasn't anything it just did it without any voltage increase.
Weird.
So after that I set my mem speed up, restarted and set my timings to manual from what they had been set to on auto.
Then went on to testing.

The above is what I got.
Read speed fluctuate from 62.5k to 65k.
The above is around 62.5k.
Normally writes are 65-66k.
But oddly enough after setting the timings to manual (without any tweaking), my mem bandwith went up to 68k in writes.
Cool I guess lol.

Anyways I'm gonna go test some more.
Hopefully I can get the side of my case on sooner or later so I can lower the timings on my ram, I got 4 delta 120mm's just waiting to cool off the board and mem.
The ram benchs 2600mhz cas9-11-11 at 1.775v.
I only running 1.65v right now though, for 10-12-12.

Lol @ my typos...


Edit:


The 2nd pic above is a screenshot of the bandwith issue.
Same exact timings, every single one of them, compared to the 1st screenshot.
Same voltages and everything, I saved the profile...

I got 114mhz bclk to bench, 115mhz wouldn't even post.
After that I chnaged my mem div down a notch, loaded up my 100mhz cmos profile.
Then loaded my 110mhz profile back up, the one mentioned above.
Did not change my voltages from the get go because I figured I wouldn't come across that prob but I did anyways.
So I did one at a time, vtt, vtt2, vccsa, and cpu voltage.
Each time after benching I changed my mem div down, save and reset, loaded 100mhz profile, save and reset, and then the 110mhz profile...
Yeah it has to be in that order to give it another go correctly (otherwise the bandwith will never get better)...

The mem div has to go down otherwise it fails to post when switching profiles.
Even though both profiles use the exact same mem div...

Also just noticed another prob, not sure about this one, maybe it's just the overclock that makes it better.
My normal dpc latency is 100-125ms.
A little while ago is was consistently 30-35ms (even with the bad bandwith screen...).

The mem bandwith thing I don't understand yet, I'll give it another go tomarrow night.


I couldn't believe I got 114mhz bclk benching in windows...
Odd how I can do that but no matter what I tried (apart from using slow boot), I couldn't get 115 to post.

[NEOAethyr]

Can you guys take a look at my screens please and tell me wth is going on?
http://www.fileswap.com/dl/DH6N5sAang/

This thing has been bothering me the whole time so far.
It's really tough to get around.

[aussie-revhead]

Old batch from local micro center

h2o

46 degrees on the hottest core at 4500? Is the radiator outside in the snow? :P

What is your ambient there? That's tested on a BE?

[Dumo]

46 degrees on the hottest core at 4500? Is the radiator outside in the snow? :P

What is your ambient there? That's tested on a BE?

It was comfy enough in my benching room

Well, lo volts ambient chip not necessarily good on cold even witout CB/CBB...

Max wall LN2

[flesheatinvirus]

Nice one, that is the highest 4930k R15 score I have seen yet.