i7 4770k IMC is Aida64 enough????

[Johan45]

Ok I got the 2400 to run for an hour with no changed to the BIOS aside from the ram. I know it says 98 but someone forgot to crank the fans up again.



So Woomack I think you were correct in that it's not so much the IMC but just not enough V_Core to run prime95 blend.
Trying to up the cache frequency now.

 

[Ajay57]

IT has been noted that 1600mhz is the sweet spot for Ram on Haswell, depending on what Chip you have and how far you can tweak it Johan.

So try working from a base of 1600MHZ and just up it a notch at a time!!

AJ.

 

[Itsumo]

What I'm saying is the only way I could get P95 stable was by using the slower ram speed. We use P95 Blend test on the AMDs all the time since it will put a lot of stress on the IMC/memory. I find this 4770k very similar, by taking the load off of the IMC I was finally able to stablize the CPU with P95. I can run the high mem speeds using Aida64 but it won't stabilize with P95. So it's not the kits. I swapped the 2400 ram for 1866 just to make it easier on the CPU. Now that I have it stable I may try the 2400 ram again but I have a feeling that P95 won't run for long.

I see what you're saying, although I'm not sure what speed you were running your 2400 kit at. Have to take issue with your being disappointed by your chip. It's running real nice man. You're seeing some great speeds for that voltage, I'd need another .09 volts to sustain 4.6.

Okay.. I came to this thread as an actual noob in need of help on this topic.. Just stop me if I say something wrong. The IMC gets stressed with higher RAM frequencies, but not lower timings, right? Also a lot of potential with these DIMMs rated for higher rated frequencies is that they can be downclocked, 1600 or even 1300, right? And this leaves room for the timings to be tightened.. while at the same time giving more headroom for the CPU. I see potential here, really a great chip you got there.

 

[Johan45]

IT has been noted that 1600mhz is the sweet spot for Ram on Haswell, depending on what Chip you have and how far you can tweak it Johan.

So try working from a base of 1600MHZ and just up it a notch at a time!!

AJ.

I'm running at 2400 now AJay that last screen was for an hour, I'm running my Cache at 4300 now as well testing with P95. 4500 was a no go just lock up.

I see what you're saying, although I'm not sure what speed you were running your 2400 kit at. Have to take issue with your being disappointed by your chip. It's running real nice man. You're seeing some great speeds for that voltage, I'd need another .09 volts to sustain 4.6.

Okay.. I came to this thread as an actual noob in need of help on this topic.. Just stop me if I say something wrong. The IMC gets stressed with higher RAM frequencies, but not lower timings, right? Also a lot of potential with these DIMMs rated for higher rated frequencies is that they can be downclocked, 1600 or even 1300, right? And this leaves room for the timings to be tightened.. while at the same time giving more headroom for the CPU. I see potential here, really a great chip you got there.

I dropped the ram speed just to take it out of the equation. I thought I had a stable clock at 4.7 using Aida64 but for P95 I had to drop to 4.5 with roughly the same voltage but , I'm getting my ram speed and NB back up so it's working. The ram speed/timings is a 2 way street. low speed tight timings will net the same throuput as high speed with loose timings. I swapped kits since I was more familiar with the ripjaws and their speed/timings. I haven't run these Tridents below 2000 really and didn't want to guess at timings.

 

[Johan45]

OK I've come to the end aside from possibly tweaking the ram a bit.





One thing I did learn is I should've stuck with P95 in the first place and saved my self a day of backtracking, what a PITA that was. I should have just started over using P95, Aida is a good piece of software but I really feel it lacks something crucial for real world stability.

 

[Ajay57]

You can not beat a tried and tested method like P95 for showing up problems with your OC!! Great job my friend enjoy your new Intel set up and welcome to our world Johan!!

AJ.

 

[Johan45]

Thanks bud! I got off to a bit of a rocky start but I think I'm on top of it now.

 

[Ajay57]

NOW go get them Boints Brother 2D, 3D, and the rest you keep telling me, will not see you for the dust or Snow as you go bye LOL!!!

AJ.

 

[Woomack]

If you are using TridentX 2400 CL10 then at the beginning check it at 10-12-12-26 1N , tRFC 110-128, 1.65V. If you have TridentX CL9 then check 9-11-11-21 1N, tRFC 110-128 1.65V. If you make it stable then you can play some more with sub timings.
That 363 tRFC looks like on AMD

 

[Johan45]

Okay I'll have to take a look at the timings later Woomack. I was just happy it was running, I had read something about the new ram training at post so it seems that my Mobo may be missing something.

Edit: So far I've tried 128 and 174 , it's a no go. The ram resets. Also tried 2600 with the same CL10 timings and 1.8v my read drops by 10,000 everything else goes up but P95 won't run. Just an update.

 

[Johan45]

Ok did a bit more tweaking. I can not get the tRFC below 200 and remain stable at 180 the computers goes through a cycle at start up and sets the ram back to 1600. Also haven't found any subs that can be dropped at these settings and keep P95 running. I do need this profile to be stable. I also have the ram at 1.75v which helped
I'm hoping for a new bios at some point. I have this issue with the LLC, I enable it but it always shuts itself off when I re-enter the bios it says disabled, so there are dome things a bit squirrly at the moment. I've already updated to a beta but no difference.

 

[Woomack]

I think it's more board/bios fault than memory as all TridentX 2400 are based on samsungs and these chips are running at tight timings and low tRFC. XMP is probably ~190 tRFC. It's not any issue as performance is almost the same as on ~120.

Right now I'm running TridentX 2400 C9 @2600 10-12-12-26 ~1.65V, tRFC 110 on Z87 ( HYK0 ) and RipjawsZ 2133 CL9 @2520 10-12-12-26 ~1.635V tRFC 107 on X79 ( HCK0 ).

 

[Johan45]

I tend to agree with you on that one Woomack, I have now tried another bios version and still no change. LLC still will not stay active and the ram timings are the same as that last screen I put up. I'm going to contact Asrock about the LLC issue. Either I'm doing something wrong and need some other setting enabled or disabled. Or their BIOS is just a bit messed up or possibly even my board ??? Who knows but it is strange,

 

[Ajay57]

Well i hope not Johan, and i also hope they answer your questions quickly as you are aware they are not good at these things. And tend to take there own sweet time.

Happy New Year,

AJ.

 

[Johan45]

Happy New year to you to bud. I'm really starting to think I should have stuck with ASUS. If this board hadn't been on for $100, I never would have gotten it. Oh well live and learn. Sometime those little things make a big difference. Just take the manual for instance, what a joke.

 

[Ajay57]

You can see why Asus went on there own and left Asrock to they own devices-es, you live a bit and learn a lot lol!!!

YHM PM Soon!!

AJ.

 

[Woomack]

ASUS didn't leave them. Both companies were making motherboards in the same factories ... what more, ASRock had all boards from the same factory as ASUS ROG series and some other higher series while all other, cheaper ASUS boards were from worse factory and you could see much higher RMA rate for cheaper ASUS boards couple of years ago.
ASRock engineers had some really good ideas and their boards are pretty good but you have to also look at the costs. Cheaper boards won't have so high quality components and support won't have dedicated so much time to solve some less important issues.
Higher ASRock boards are not worse than ASUS ROG, they are just missing some additional features that aren't important for 99% users.
Also ASUS had really bad support for couple of years and many of their boards have never had fixed BIOS issues or other things. I won't even mention that some boards had 30+ BIOS versions and were still not working right.
There were boards with remote control which wasn't working and never had any patch. Boards were just discontinued. The same ASUS sound cards needed 3rd party soft to work as they supposed to. List is long and I wouldn't stick to any manufacturer but to single models.

If Shamino really quit ASUS team then I wonder how future ROG boards will look like. Most previous Intel ROG boards were improved till Shamino was testing them so usually ~3 months after premiere. After this time you could barely see new BIOS with anything more than " improves stability" or added new CPU microcode. Simply if board issues were not solved in that 3 months then were never solved.

 

[Johan45]

That's a good history lesson Woomack, and I was thinking along the lines of the Maximus VI Hero, I checked out the Formula but wasn't sure about all the plastic they've added to it plus it's $300. That Asrock board was almost half price at $100 so I wouldn't have classed it as cheap, and I always thought their Fatal1ty series were supposed to be a bit better.??

 

[Woomack]

I really like GB Z87X-OC as it's also one of the cheaper boards with many OC features and good support. At the beginning I wanted ROG board as MVG was great even though had couple of bugs ( but I got used to it after longer benching ) but GB Z87X-OC was pretty interesting, already available and cheaper than ASUS so I gave it a try.
I don't trust MSI hardware and I probably won't buy their gfx card or mobo soon ( let's say too many various issues ). From other brands there are good ASUS, ASRock and Gigabyte boards but more like single models and sometimes it's hard to to be sure what is good when you can't test it. To be honest , reviews are saying almost nothing about it.
Fatal1ty is a gaming series and was never best in anything no matter what brand was using it. It was made to improve sales but looking at popularity they simply failed. You can also notice that Wendel's face has been removed from ASRock product site and from UEFI start screen ... users were just asking to remove it.
ASRock OC Formula made some noise and some overclockers said it's the best series ... I had no chance to test it so I won't tell.

 

[Johan45]

I hear you on the Asrock, I'm certain that the CPU LLC issue is affecting the way this thing OCs since it's not compensating for the droop any longer. I'm thinking this one may just have to go back to the store. It's been 48 hrs and still nothing back from Asrock about this either.
I don't know if I mentioned this earlier but the board was on the backup bios when I got it didn't realize that for a while. When I swiched it back to main I had a heck of a time getting it started, but eventually it did. Three different Bios versions if you include the original and that setting still won't work. I'm really leaning towards a problem with the board itself.