Those run times are all over the place, tells me your voltage isn't high enough. With IBT testing you are stable when you can pass a full "Extreme" setting test and with almost no variation in run times or GFLOPs produced. For quick general stability tests like the setting you're using, run times should not vary by more than about .1 second from fastest to slowest, and you're getting about half a second. Each 100MHz increment starting at about 4.5GHz requires a voltage increase in the range of .03V to .1V, in my experience. You should use Core Temp instead of Real Temp, it shows the VID, which is the expected voltage for a given workload. Start there, adding small amounts until you find stability.

Are you trying to say that there is a connection between Vcore and cooling/temps? This is for the first time when i hear that and i am very interested, i saw that you have experience with overclocking and fsx systems. Does this mean that i can get lover Vcore with water cooling vs air?

 

Probably not with water vs. air, but definitely with sub-zero cooling. Hasse (Westman) talked about it in a thread he started on his experience with the phase change cooler he runs now. The thread is here. The specific post I'm thinking of says this: "It runs @5.2ghz 1.34v , to run that with Air or a unlapped H100 is @4.6ghz 1.34v some needs less or more voltage"

So with "normal" above zero cooling you can run a 3770k @ 4.6GHz and need something like 1.34V. When you drop the temp below zero you can push the clockspeed much higher while needing the same amount of juice. It has to do with energy being wasted as heat. Lower temps mean lower volts.

 

 

Can anyone confirm that you can run fsx noticeable smoother with 2133MHZ RAM vs 1600MHZ? I belive that 2133 will be more stable than 1600, but i'd like confirmation

 

I don't know if I'd say "noticeably smoother", as that's subjective and depends greatly on the situation. If you're running at say 15 FPS instead of 13 FPS, that might be a noticeable difference but 33 FPS vs. 30 FPS may not be.

 

That's is pretty good Vcore and temps for 4.6GHz, just stay away from auto-overclock features(i saw that you tried this in your first post) and auto Vcore settings in bios. CPUs will degrade over time whatever voltage they are at, but the higher - voltage the faster the CPU degrades.

According to some very experienced overclockers, you'r CPU should be fine for at least 3-5 years near 1.4 Vcore and max temps around 80 when runing stability tests(under 70 for everyday usage).

My 2700K @4.8GHz and 1.375 Vcore(stable at 1.370, but i allways add +0.05Vcore when stable, that's just me - i have some theories about stability B) )reached max 80c with intel burn test, max 70c with Prime95 1h test, and never goes above 58c with FSX. I am happy with that overclock.

 

 

That's actually a pretty good rule.... But with Ivy, every .000000005 of voltage counts

I'm still testing and tweaking, but I'm definitely going to post my final config on here for the gurus to approve of

 

Those run times are all over the place, tells me your voltage isn't high enough. With IBT testing you are stable when you can pass a full "Extreme" setting test and with almost no variation in run times or GFLOPs produced. For quick general stability tests like the setting you're using, run times should not vary by more than about .1 second from fastest to slowest, and you're getting about half a second. Each 100MHz increment starting at about 4.5GHz requires a voltage increase in the range of .03V to .1V, in my experience. You should use Core Temp instead of Real Temp, it shows the VID, which is the expected voltage for a given workload. Start there, adding small amounts until you find stability.

 

Sounds good, I'm going to start now, at 4.6 and 1.25V, and I'll push for (almost) equal timings, see what voltage I get, temps, and then see if I can take 'er any further.

 

Can anyone confirm that you can run fsx noticeable smoother with 2133MHZ RAM vs 1600MHZ? I belive that 2133 will be more stable than 1600, but i'd like confirmation

 

I can't for the life of me figure out how to setup my Trident X 2400 kit, it's running at 1600 right now, that's the only way it's stable

Sounds good, I'm going to start now, at 4.6 and 1.25V, and I'll push for (almost) equal timings, see what voltage I get, temps, and then see if I can take 'er any further.

 

I'd guess you'll need closer to 1.30V

 

I can't for the life of me figure out how to setup my Trident X 2400 kit, it's running at 1600 right now, that's the only way it's stable

 

Make sure you run the XMP profile and 1.65V in the BIOS, but also be sure to set the speed to 2400 manually.

Make sure you run the XMP profile and 1.65V in the BIOS, but also be sure to set the speed to 2400 manually.

 

Okay, just did that, stress testing on IBT now.

 

I'd guess you'll need closer to 1.30V

 

1.35 for 4.7? Whaddya think about that?

Probably not with water vs. air, but definitely with sub-zero cooling. Hasse (Westman) talked about it in a thread he started on his experience with the phase change cooler he runs now. The thread is here. The specific post I'm thinking of says this: "It runs @5.2ghz 1.34v , to run that with Air or a unlapped H100 is @4.6ghz 1.34v some needs less or more voltage"

So with "normal" above zero cooling you can run a 3770k @ 4.6GHz and need something like 1.34V. When you drop the temp below zero you can push the clockspeed much higher while needing the same amount of juice. It has to do with energy being wasted as heat. Lower temps mean lower volts.

Thanks for the link and great explanation, i learned that I can get 5.4 - 5.8GHz if i go for sub-zero cooling, and that sub-zero cooling can be configured for everyday usage, but this is realy extreme cooling for fsx purpose. I reached 4.9 on air with noctua nh-d14.

 

 

I think i can go for 5GHz too, but i don't like to use Vcore above 1.4 for 7/24 usage - becouse of CPU degradation. Friend of mine used 2600k at Vcore 1.425 for about 1 year, and now CPU needs 1.45 Vcore to be stable(i don't remember @ how much GHz). But, in the other hand, who knows what really happen to that CPU.

 

I don't know if I'd say "noticeably smoother", as that's subjective and depends greatly on the situation. If you're running at say 15 FPS instead of 13 FPS, that might be a noticeable difference but 33 FPS vs. 30 FPS may not be.

I understand, there is some fps increase

Okay, so I think I've FINALLY gotten my stable 24/7 OC!

 

4.7GHz @ 1.305V

75% LLC

RAM @ 2400MHz

 

Test Results:

 

 

Okay, so I think I've FINALLY gotten my stable 24/7 OC!

 

4.7GHz @ 1.305V

75% LLC

RAM @ 2400MHz

 

Test Results:

 

 

Temps are ok, but i passed intel burn test many times and after that i get random BSOD when runing fsx or with PC at idle. I think you should try to run Prime95 test for at least 1h, if you don't get test errors(workers errors) or BSOD, it's probably good Vcore.

And i must to correct myself, i add +0.005 Vcore when stable, not 0.05

Temps are ok, but i passed intel burn test many times and after that i get random BSOD when runing fsx or with PC at idle. I think you should try to run Prime95 test for at least 1h, if you don't get test errors(workers errors) or BSOD, it's probably good Vcore.

And i must to correct myself, i add +0.005 Vcore when stable, not 0.05

 

 

Hmmmmm, well, about 12 hours later, I've finished a couple of flights with the NGX, no problems so far.. 2 hour + flights...

 

The only problem was that, because I haven't played FSX in quite a while, so my landings were utter rubbish

<br />Hasse (Westman) talked about it in a thread he started on his experience with the phase change cooler he runs now. The thread is here. The specific post I'm thinking of says this: "It runs @5.2ghz 1.34v , to run that with Air or a unlapped H100 is @4.6ghz 1.34v some needs less or more voltage"<br />

 

Hi Max

I done futher tuninig of the Cap tube , tune it for 210 watt have it for 300watt on begining.

it gain a 3-4 C , have -37c full load on the 3770k now.

raised the Vcore to 1,44v and run @5.4 ghz 24/7 its 100% stable, flying a lot the last two weeks no

problem at all.

 

By the way run the mems 2800mhz 10-12-12-28 1T

 

Almost forgot to say a have lapped the IHS , it worked great the temp on evap raised and temps on

cores get lower.

 

Cheers Hasse

By the way run the mems 2800mhz 10-12-12-28 1T

 

What volts do you think I'd need for that?

 

Hi Max

I done futher tuninig of the Cap tube , tune it for 210 watt have it for 300watt on begining.

it gain a 3-4 C , have -37c full load on the 3770k now.

raised the Vcore to 1,44v and run @5.4 ghz 24/7 its 100% stable, flying a lot the last two weeks no

problem at all.

 

By the way run the mems 2800mhz 10-12-12-28 1T

 

Almost forgot to say a have lapped the IHS , it worked great the temp on evap raised and temps on

cores get lower.

 

Cheers Hasse

 

Seriously awesome results! I sold my TridentX 2800 because my IMC wasn't strong enough, have some 2600 C10 on the way.

run @5.4 ghz 24/7 its 100% stable, flying a lot the last two weeks no

problem at all.

 

By the way run the mems 2800mhz 10-12-12-28 1T

I know who's having the fastest FSX rig around. Well done Hasse! I wonder/hope Haswell will bring this sort of performance for us "regular" overclockers next year...

I know who's having the fastest FSX rig around. Well done Hasse! I wonder/hope Haswell will bring this sort of performance for us "regular" overclockers next year...

 

Thanks,

I wait for the Haswell to think we ned it to run 3 monitor setup, it work ok now but think we ned 60-62 in FSMark11.

My systyem as it is now now killer settings high mem voltages, a plain every day setting produce 56-57fps in FSmark11 that

give me close 47 with 3monitors.

 

Get some hooks out to have a Haswell ES and a Giga or MSI board .

 

Hasse