• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Specs + 7GHZ question

L

light65536

n00b
Joined
Jan 19, 2006
Messages
29
My specs:


i7 3770k
Gigabyte UD5 MB
Phanteks super air cooler

I am able to run 4.8ghz at what appears to be really stable (run linx as many passes as desired) at 1.41V.

I am running this configuration 1 Core at 5Ghz, 2 at 4.9, 3 @ 4.8, and 4 @ 4.8

This seems pretty solid. I was able to run around 8 iterations of burn test before temps spiked to 87c and caused an error.


however... when I run Sci Sandara with the .NET test my processor speed spikes to 7ghz at reported by the Intel turbo boost monitor and the program errors.

I do not see this behavior from any other program. Any ideas why that is happening?

I use Intel Turbo Boost Tech monitor 2.5 for monitoring the turbo boost technology

-- Quick update

I also ran Prime 95 at the 4.8ghz and it does error running the 4 cores after a few minutes and when the processor pegs 83C. Stability does seem to decrease when this processor gets much over 80C. I usually don't run P95 because I've found it not very good for testing stability compared to LINX. I'd say its 99% stable but I'll probably have to drop it to 4.7GHZ for 99.999% stability. Dropping to 4.7 @ 4, 4.8 @ 3, 4.9 @ 2, and 5.0 at might get me 1 notch higher stability.


My observations

* This chip becomes unstable as temps increase into the 80C + range. You might do 81C,82C, 83C and be stable but somewhere in the 80C the chip starts to struggle.
* It is much harder to get a stable 4.8 then 4.7. Its like a huge jump
* Using the turbo boost technology it is possible to get more "real world speed"
* I am now testing 1.39V @ 4.7ghz x 4, 4.8x3, 4.9x, and 5.0x1. This is the configuration that I will probably try to run it. P95 is now stable.

*** Update 2

Interestingly.. the processor won't pull to 5ghz when I set it to 1.39V but it didn't error either. I'm back to 1.41V and trying the 4.7,4.8,4.9,5.0 ramp combination.
 
Last edited:
Just a final update...

I was able to pass several iteration of Prime95 torture test, LINX, and SiSoftware .NET Arithmetic module running 4.7x4, 4.8x3, 4.9x2, 5.0x1. The SiSoft .NET Arithmetic module somehow causes the turboboost to boost beyond the setting. I boosted to 5.38GHZ for a few moments.

I'm running at 1.41V and have read that 1.45 is max safe voltage.

I used Intel Turbo Boost to monitor turbo boost speeds

http://downloadcenter.intel.com/Detail_Desc.aspx?DwnldID=19105

This is very nice because I can see what I'm doing in my actual programs that I'm using.

Overall I'm very happy with this upgrade and my results.
 
..... new update

Unfortunately, even though I had LINX and Prime95 operating 100% stable. I ran a real world application and found instability.

I increased voltage to 1.44V but did not resolve issue. Temps were excellent with max temps probably below 70C.

I will not go over 1.44V and have dropped back down to 1.42ishV. I have decreased PLL to see if that helps.

Will update when I find a stable setting... I don't think I've seen a chip that did as well in testing as this one has but not be stable in normal apps.

In monitoring the real world apps, I've found that many apps don't use over 2 cores and even 1 core. I also monitor rapid flux in the turbo boosting. This makes me wonder if VTT Load Line Caliberation might help. I believe that increases the stock voltage to the chip.
 
New update...

Okay, these are probably my latest specs because I can't do any more...

4.9ghz x 1 core
4.8ghz x 2 core
4.7ghz x 3 core
4.7ghz x 4 core

Voltage: 1.42V
Final Modification: Only VCORE, turbo multipliers, and set the the PWM to extreme and Fast.

Experiences:
I've never seen a processor run system stability tests as well as this one and then fail when opening an average program. My recommendation for those trying to overclock this thing is don't run a stress test forever but try a lot of different programs.

The current settings allow LINX, Prime95, and the "real world" application that was failing on me to run stable. However, there may be a problem with running the Windows Experience test so I can't say that its 100% stable. If I drop the voltage any then the system becomes unstable very quickly.

While real world programs don't generate the temps, if I were to increase the voltage any more then stress tests would start to fail.

In general, the chip doesn't seem to like to operate above around 85C but its a combination of the voltage+temps+speed. I've ran tests on it up to 89C with no problems. Real world apps don't hit over 65C (heavy duty calculating apps) and seem to benefit from the increased voltage.

It might be possible to do a better in real world apps if one was willing to forgo the stress tests because the stress tests tend to generate a lot more heat then the real world apps but don't guarantee stability.
 
update...

I still didn't get 100% stability. I would say 4.5ghz is the only speed I'd say could be rock solid. So, this is a difficult one to get to run. Temps can be good and increasing voltages helps but doesn't guarantee.

i am currently running 45,46,47,48 at 1.38v and I think i'm getting closer to a 99.999999% stability at these levels.
 
You must log in or register to reply here.
Back
Top