or Connect
Head-Fi.org › Forums › Misc.-Category Forums › Members' Lounge (General Discussion) › PC Enthusiast-Fi (PC Gaming/Hardware/Software/Overclocking)
New Posts  All Forums:Forum Nav:

PC Enthusiast-Fi (PC Gaming/Hardware/Software/Overclocking) - Page 216  

post #3226 of 9059
 
Originally Posted by NamelessPFG View Post
 
  • 4.2 GHz (42x100) core, 1.25Vcore, 4.0 GHz (40x100) cache
  • 4.4 GHz (44x100) core, 1.365 1.37 1.38Vcore, 4.0 GHz (40x100) cache (STILL TESTING...crap, a thread failed on Prime95 small FFTs despite doing well on blend. Seems like that mode really does put stability to the test. Argh, these 0x124 BSoD errors...)

 

 

Oof, looks like you hit a wall there at around 4.2ghz.

 

Are you allowed to return the CPU for another one? I'd imagine most shops will refuse an exchange, because the chip isn't faulty at all, it works perfectly fine.

post #3227 of 9059
Quote:
Originally Posted by deadlylover View Post
Oof, looks like you hit a wall there at around 4.2ghz.

 

Are you allowed to return the CPU for another one? I'd imagine most shops will refuse an exchange, because the chip isn't faulty at all, it works perfectly fine.

 

I'm pretty sure Micro Center will allow one return, maybe two...but the third would definitely make them suspicious.

 

I'll back off and run at 4.2 GHz in the meantime, which I know is stable. Trying to just hit 4.4 GHz is taking all the fun out of this new build.

 

Meanwhile...I just noticed that some of my GTX 480's screws were not tightened all the way, so I corrected that. For some reason, that was enough to get the card to work like it's supposed to. I suspect a somewhat faulty BGA connection somewhere that's being held together by the HSF heatplate being clamped against the PCB pretty hard by those screws.

 

It seems reliable enough for now, but this does mean I can't use my DD-GTX480 waterblock, which sucks. One thing's for sure, though: whatever my next graphics card is, I'm going to seriously consider a pre-waterblocked one. Or make sure to get a waterblock with a bloody backplate this time. I don't know what Danger Den was thinking, hanging a heavy mass of copper off a PCB with no supporting backplate...

 

Anyway, let's see what the 4670K can do over the ol' Q6600, especially with the GTX 480 operational again to mitigate any GPU bottlenecks.

post #3228 of 9059
Quote:
Originally Posted by linglingjr View Post
 

Does any one here have experience "un bending" cpu pins? A friend might be sending me an 8350 that came with bent pins and AMD nor Newegg will accept a return.  I expect it to be a PITA but as long as I'm patient and careful...... worth it for a free cpu upgrade :biggrin:

 

Not sure if anyone is familiar with popping the anti-reversal latch on airsoft gear boxes w/o opening them but I expect it to be on the same level.

... I'd advise not using it at all as worst case scenario something might short out on your Mobo >.> unless you've got a spare mother board lying around to tinker around with the CPU in 

 

Quote:
Originally Posted by NamelessPFG View Post
 

 

I'm pretty sure Micro Center will allow one return, maybe two...but the third would definitely make them suspicious.

 

I'll back off and run at 4.2 GHz in the meantime, which I know is stable. Trying to just hit 4.4 GHz is taking all the fun out of this new build.

 

Meanwhile...I just noticed that some of my GTX 480's screws were not tightened all the way, so I corrected that. For some reason, that was enough to get the card to work like it's supposed to. I suspect a somewhat faulty BGA connection somewhere that's being held together by the HSF heatplate being clamped against the PCB pretty hard by those screws.

 

It seems reliable enough for now, but this does mean I can't use my DD-GTX480 waterblock, which sucks. One thing's for sure, though: whatever my next graphics card is, I'm going to seriously consider a pre-waterblocked one. Or make sure to get a waterblock with a bloody backplate this time. I don't know what Danger Den was thinking, hanging a heavy mass of copper off a PCB with no supporting backplate...

 

Anyway, let's see what the 4670K can do over the ol' Q6600, especially with the GTX 480 operational again to mitigate any GPU bottlenecks.

let us know how it goes :D 

post #3229 of 9059
Quote:
Originally Posted by HybridCore View Post
 

I think they use metal ones on the Define. I can't remember.

Plastic on the R3, dunno about the new R4.

 

Quote:
Originally Posted by NamelessPFG View Post
 

I doubt I'm doing much to other bus speeds; I should also point out that I haven't even touched the base clock, just the multiplier/core ratio. Even though Intel brought back BCLK straps to Haswell, most people still leave BCLK at 100 MHz and work with the multiplier. It's much simpler that way.

 

 

It's so far below the apparent average that I'm honestly going to return it within a week. Might even step up to the i7-4770K if Hyper-Threading proves valuable for future games as well as video encoding and whatnot. I'd probably make out better on the chip lottery with just about any other Haswell CPU right now.

 

Next time...they say batch numbers don't matter much, but I'm making damn sure they hand me a Malaysian one.

 

Keep in mind I'm using manual voltage because I'm still stress-testing for stability, and adaptive voltage means a nasty case of overvolting while stress-testing. Once everything's finalized, I'm switching to adaptive.

 

Here's what I've managed so far:

  • 4.2 GHz (42x100) core, 1.25Vcore, 4.0 GHz (40x100) cache
  • 4.4 GHz (44x100) core, 1.365 1.37 1.38Vcore, 4.0 GHz (40x100) cache (STILL TESTING...crap, a thread failed on Prime95 small FFTs despite doing well on blend. Seems like that mode really does put stability to the test, and after 1.365V, it seems to be getting LESS stable. Argh, these 0x124 BSoD errors...well, if this is anything like Sandy Bridge, I guess I gotta do something about the low-power states.)

 

Vdroop?

VCCIO?

System agent Voltage?

All power scaling options off?

 

Those things said, it does sound like your chip sucks.

post #3230 of 9059
Quote:
Originally Posted by evilhippie View Post
 

 

How bent exactly exactly are we talking?  I've ran into the issue a very long time ago, but I can't for the life of me remember which CPU it happened to.  if its not bent too badly (like completely clumped together at a 90 degree angle), it may very well be salvagable with minimal effort.  If its just a few pins then you'll most likely be good to go, just have to be very careful, obviously.

Not bent enough that he can tell where the bad pins are.  That being said he never even got it in the socket and probably bent even more trying to jam it in.  As for it shorting your motherboard is that being just as paranoid as the people that religiously wear antistatic bracelets?

 

Tried playing Bf4 last night.  So far all the maps look like remakes of CoD maps, Rush is ruined, they felt the need to make it timed and takaway getting more tickets with each M comm (Don't fix something if it's not broken) Twice I went up to somebody for a knife take down, both times I got disconnected a second before I went into animation.  All that being said I still look forward to playing it.

post #3231 of 9059
Quote:
Originally Posted by linglingjr View Post
 

Not bent enough that he can tell where the bad pins are.  That being said he never even got it in the socket and probably bent even more trying to jam it in.  As for it shorting your motherboard is that being just as paranoid as the people that religiously wear antistatic bracelets?

 

I think this youtube video might be helpful, this guy has a pretty good method for bending the pins back into place. 

 

http://www.youtube.com/watch?v=IWAvwpANZ7c

 

Interestingly, he mentions that a solution to fixing completely broken pins is to stick the pin in the corresponding socket slot, but yeah, theres nothing to be paranoid about as far as shorting a motherboard out.  I doubt any modern board will even turn on if the pin configuration is incorrect or a critical pin is missing, much less "explode" per se.  I wouldn't be so quick to throw away that CPU regardless, as it sounds like its probably salvagable.  If he can't even locate where the bent pins are chances are it a very insignificant bend.  You might not even have to bend anything, just sort of "wiggle" the CPU into the socket, if that makes any sense.

post #3232 of 9059
Quote:
Originally Posted by Tangster View Post

Vdroop?

VCCIO?

System agent Voltage?

All power scaling options off?

 

Those things said, it does sound like your chip sucks.

 

I'm still uncertain of the exact amount of Vdroop right now, and I haven't quite figured out the other two. The way Asus names all these voltage settings in their BIOS doesn't help. I just leave all that extra crap on Auto for the time being until I know exactly what to tweak, but maybe it wouldn't hurt to fiddle with the Load Line Calibration setting for Vdroop compensation.

 

This should give you an idea of what I'm working with. I'm using a Maximus VI Hero, not an Extreme (that's way too expensive), but the BIOS options are mostly the same.

 

Also, I was under the impression that Manual voltage adjustment mode (not Offset or Adaptive) turns off power scaling by default, since it's always running at the set Vcore.

post #3233 of 9059

Sorry i missed that post about you using JDEC memory speeds and multiplier only clocking.  Both good.


In my experience (Asus Sabertooth x79) the bios settings on asus boards are confusing and take some trial and error. I got the best results (i.e. most consistent) with a static CPU voltage, none of the vdroop offset stuff really worked IMO.  Find a voltage that is stable with stress testing (something like prime95), then enable speedstep when your done and call it a day.  Running static voltage 24/7 dosent matter if the clock speed can dynamically adjust which = really low idle heat and power consumption.  But for initial OC and testing leave it disabled.

 

LLC i bumped up to 120% for the cpu and vccsa, with optimal profile setting.

 

To be honest i wouldn't get too caught up chasing that last few hundred mhz, certainly not at the expense of that much extra voltage.  There is a point of diminishing returns, when i could play BF3 with 25-30% cpu usage going from 4.6ghz to 5ghz was a purely academic pursuit.

post #3234 of 9059
I'm running a x79 Sabertooth as well. The one thing that helped me stabilize my 4.6 OC that you don't read mentinoed enough online is the voltage of the memory controller needs a fairly substantial boost if you have populated more than 2 memory slots on the board.

So if Namless here has all 4 slots populated, he might need to up the VCCSA (I believe). You can see Vdroop just by opening CPU-Z or the AI suite (if you installed it) and watch the voltage drop a few tenths when you run Prime95 or whatever. If you have the Load Line Calib. (LLC) high it might not droop much at all, but then your VRM temps can go crazy. It's all a balancing act.
Edited by Muinarc - 10/29/13 at 4:03pm
post #3235 of 9059
Quote:
Originally Posted by Muinarc View Post

but then your VRM temps can go crazy.

Better waterblock them then :P

post #3236 of 9059
Quote:
Originally Posted by Muinarc View Post
I'm running a x79 Sabertooth as well. The one thing that helped me stabilize my 4.6 OC that you don't read mentioned enough online is the voltage of the memory controller needs a fairly substantial boost if you have populated more than 2 memory slots on the board.

So if Namless here has all 4 slots populated, he might need to up the VCCSA (I believe). You can see Vdroop just by opening CPU-Z or the AI suite (if you installed it) and watch the voltage drop a few tenths when you run Prime95 or whatever. If you have the Load Line Calib. (LLC) high it might not droop much at all, but them your VRM temps can go crazy. It's all a balancing act.

 

It's a tip I'll keep in mind...when I'm running 32 GB later on.

 

I currently only use two slots, 8 GB modules in each.

 

I haven't installed Asus' Ai Suite yet. I usually don't hear good things about motherboard software of this sort since everyone favors BIOS/UEFI-level overclocking, but maybe Asus has far better software developers for this sort of thing than Gigabyte did. (EasyTune 5 on my old GA-P35-DS3P was pretty terrible. Then again, this is also a board whose driver/software disc was printed with a malware infection. Not even joking about that...)

 

My monitoring's currently done with the latest HWMonitor version. Not sure how accurate that is right now, but it seems to get the VID and temps right, at least. Meanwhile, the current CPU-Z actually shows a voltage a bit ABOVE the set VID, to my recollection. For instance, if it's supposed to be 1.25V, it'll show 1.264V. Maybe it's showing VRIN and not Vcore?

 

I'll look into all of this in a bit more detail when I get home.

post #3237 of 9059
Quote:
Originally Posted by Tangster View Post

Better waterblock them then :P

 



I would have if I had planned that into my water loop, and I bought this Day one of the x79 launch....and only the ROG board got vrm waterblocks that soon frown.gif

I just ghetto rigged a spare 140mm fan to blow on em. Been doing protein folding ever since!
post #3238 of 9059
Quote:
Originally Posted by NamelessPFG View Post

 

It's a tip I'll keep in mind...when I'm running 32 GB later on.

 

I currently only use two slots, 8 GB modules in each.

 

I haven't installed Asus' Ai Suite yet. I usually don't hear good things about motherboard software of this sort since everyone favors BIOS/UEFI-level overclocking, but maybe Asus has far better software developers for this sort of thing than Gigabyte did. (EasyTune 5 on my old GA-P35-DS3P was pretty terrible. Then again, this is also a board whose driver/software disc was printed with a malware infection. Not even joking about that...)

 

My monitoring's currently done with the latest HWMonitor version. Not sure how accurate that is right now, but it seems to get the VID and temps right, at least. Meanwhile, the current CPU-Z actually shows a voltage a bit ABOVE the set VID, to my recollection. For instance, if it's supposed to be 1.25V, it'll show 1.264V. Maybe it's showing VRIN and not Vcore?

 

I'll look into all of this in a bit more detail when I get home.

 



AI suite has decent monitoring tools and if you change one thing at a time, you can play with some OC setting in it, but I would only use that for testing purposes myself. It is kind of a resource hog though.

I think VID within 0.01V is "close enough". especially at idle. My 3930K droops by like 0.08V under full load and it's amazingly stable. "You have it set too high then" you may say.. but If I lower it at all in the BIOS all hell breaks loose lol. I think it's set at 1.4V, but I haven't touched my BIOS settings in over a year.
post #3239 of 9059
Quote:
Originally Posted by NamelessPFG View Post
 

My monitoring's currently done with the latest HWMonitor version. Not sure how accurate that is right now, but it seems to get the VID and temps right, at least. Meanwhile, the current CPU-Z actually shows a voltage a bit ABOVE the set VID, to my recollection. For instance, if it's supposed to be 1.25V, it'll show 1.264V. Maybe it's showing VRIN and not Vcore?

Multimeter or bust.

post #3240 of 9059
Quote:
Originally Posted by Muinarc View Post

I'm running a x79 Sabertooth as well. The one thing that helped me stabilize my 4.6 OC that you don't read mentinoed enough online is the voltage of the memory controller needs a fairly substantial boost if you have populated more than 2 memory slots on the board.

So if Namless here has all 4 slots populated, he might need to up the VCCSA (I believe). You can see Vdroop just by opening CPU-Z or the AI suite (if you installed it) and watch the voltage drop a few tenths when you run Prime95 or whatever. If you have the Load Line Calib. (LLC) high it might not droop much at all, but then your VRM temps can go crazy. It's all a balancing act.

 

Now i think about it mine is @1.6v as dictated by the XMP profile, thats with 4x4gb. But good point..

 

The load calibration if i'm not mistaken controls maximum current capability, not voltage.

 

As you say the VCCSA (voltage regs) run hotter than the CPU, seems perfectly normal for an x79 from what i have read.

 

@NamelessPFG: Install the AiSuite and have a play with it, the sensor recorder is a fantastic app for monitoring temps.

New Posts  All Forums:Forum Nav:
  Return Home
This thread is locked  
Head-Fi.org › Forums › Misc.-Category Forums › Members' Lounge (General Discussion) › PC Enthusiast-Fi (PC Gaming/Hardware/Software/Overclocking)