How good is the IMC on the 6700k

At command rate 2 4000mhz is easy, at command rate 1 only a few boards can surpass 3733mhz and be stable, the ones that can achieve this are the extreme, gene, impact, m-ocf

Edit, the above was tested at 12-12-12-28, on the Gene i managed 3950mhz cas12 CR1 and 4100+mhz cas12 CR2

Mems used/tested were g.skill ripjaws 3600mhz C16 (samsung b-die), max oc achived was 4362mhz http://hwbot.org/submission/3093445_nickolp1974_memory_clock_ddr4_sdram_2181.3_mhz
 
Last edited:
Very strong IMC but u also need good mem.

Best boards are as Nick says especially OCF M and OCF. OCF M 4100+ 12-12 1t 4150+ with good sticks.
 
Board should be 4G C12 capable, i believe the teams are samsung e-die, whats it for Kaap? Benching or 24/7??

With b-die so g.skill ripjaws or tridents you are looking for 8gb sticks with timings all the same such as 16-16-16-36 3600mhz or maybe 14-14-14-34 3200mhz.
Nothing beats b-die on air for efficiency but there is a problem with them, to run C12 3733mhz+(64bit OS) you have to limit maxmem to around 4096mb, which is fine for benching scenario's. There may of been a bios update to boards to combat this and maybe Ian can coment there??
XMP will be fine at there full rated capacity and maybe a small OC.
Also with the b-die and asus boards i believe Dancop released profiles for 4ghzC12 which iirc can be found at Hwbot.
All of which depends on your needs and uses.
 
It will mostly be for benching where only a 4/8 core is needed.

I would like to use 16gb of RAM as I may use it for a bit of gaming as well.
 
As i said then Kaap, you want g.skill ripjaws or tridents with b-die on them, xmp with full 16gb for gaming and limit to 4096mb and clock em high for benching, obviously i cant post a link to them here but just look for 3600 16-16-16-36 2x8gb kit and they will be b-die, there awesome and way above anything else once tuned correctly
 
As i said then Kaap, you want g.skill ripjaws or tridents with b-die on them, xmp with full 16gb for gaming and limit to 4096mb and clock em high for benching, obviously i cant post a link to them here but just look for 3600 16-16-16-36 2x8gb kit and they will be b-die, there awesome and way above anything else once tuned correctly

Thanks.:)
 
To clarify 2x b-die kits
Ripjaws 2x8gb 3600 16-16-16-36
Tridentz 2x8gb 3600 17-17-17-37 (proving to be the better kit oc wise but only just)

One thing to note if you get some. RTL's must be trained correctly to achieve the best stable results, for example should look something like 58-7-59-7 or 57-6-58-6 and not 58-7-64-7, the larger number must ascend by 1, if not then re-train.
 
I don't think Kaap will concern himself with 90% of what is being said here. For casual benching / 24/7 RTL training is absolutely fine as-long as running a recent build

Unconditionally past 3866 can be tricky 1T. I've managed 4133 CAS 16 through 3 hours of Stress app test on the Impact. This is with voltage that can be considered acceptable 24/7, and using the TG 3866 2x4GB. Guardband on these isn't great to begin with, high bin b-die will go further depending on IMC
 
Last edited:
I don't think Kaap will concern himself with 90% of what is being said here. For casual benching / 24/7 RTL training is absolutely fine as-long as running a recent build

Your quite possibly correct, but we all know he likes his high end gear and they are the best sticks available if the user is prepared to push then benchmark physics scores etc etc will be greatly improved. Of course a fair amount of voltage is required to achieve timings/speeds generally speaking 1.82-1.9v, some have gone upto 2v with no issues and SA/IO 1.3v + IMC dependant
 
I don't think Kaap will concern himself with 90% of what is being said here. For casual benching / 24/7 RTL training is absolutely fine as-long as running a recent build

Could a Don please close this thread.

Replies like yours and 8 Pack's in the Firestrike threads have killed my interest in the project.
 
Could a Don please close this thread.

Replies like yours and 8 Pack's in the Firestrike threads have killed my interest in the project.


lol, what? So you no longer want to try Skylake because RTL/DRAM IOL training on later UEFI is pretty good.

I think that says all that needs to be said on that matter. Most people will not need to move these from AUTO, which pretty much is reflected by your response here. You're taking offense, but the truth is the values mentioned, there is little room past the initial settings once you start reducing CAS, at least stability will become pretty limited. If you have a grasp of what these timings do then there'd be no need to throw your toys out of the pram, or else don't ask for advice.
 
Last edited:
lol, what? So you no longer want to try Skylake because RTL/DRAM IOL training on later UEFI is pretty good.

I think that says all that needs to be said on that matter. Most people will not need to move these from AUTO, which pretty much is reflected by your response here. You're taking offense, but the truth is the values mentioned, there is little room past the initial settings once you start reducing CAS, at least stability will become pretty limited. If you have a grasp of what these timings do then there'd be no need to throw your toys out of the pram, or else don't ask for advice.

Why don't you try and do something positive for once.
 
Back
Top Bottom