Testing ASRock Z77 OC Formula

[SF3D]

Hi all!

I was able to remember my password

It have been long time since I got any decent results in any benchmark. This post will not change that at all, but at least I have put up a lot of steam in my benchroom

So I decided to test how this new ASRock Z77 OC Formula is doing with memory overclocking and what I can get out of my best CPU.

Setup was this:
ASRock Z77 OC Formula
Intel Core i7 3770K
G.Skill PI series PSC memory
Kingston HyperX SSD + HDD
MSI GTX 670 PE VGA
Cooler master benchtable
Corsair 1200W AX PSU

SF3D OC Inflection point CPU pot
SF3D OC Triple point memory pot

Memory voltage up to 2V and temperature between -100C and -140C. More cold did not seem to do anything after -100C...

Here are some random results and pictures:

Bling bling...



LN2 setup



Some Ln2 scores (Yes, I suck on efficiency, but no idea to spend much time with this kind of CPU )

Superpi scores were run with CPU-Z windows open and program was not closed between runs. There was no tweaks applied in to these memory tests.

First DDR3 2700MHz CL5 8M run


1300MHz CL5 32M run



Then a bit more serious runs with better CPU speed.






That's all!


Article in SF3D.fi website (Finnish): ASRock Z77 Oc Formula
SF3D OC Facebook page, if you like to get some updates from our testing etc.

[EliteGeek91]

WOW nice!!! Very sweet... ONLY if they made for that board for AMD chips..

[Dumo]

Awesome run SF3D

1300 cas5 @5G seems kinda slow though. Its probly revert to cas7

[EliteGeek91]

Awesome run SF3D

1300 cas5 @5G seems kinda slow though. Its probly revert to cas7

I thought it's actually 2600MHz? Right?

[jimba86]

Nice run!

So what gskill Pi memory is that?

[flesheatinvirus]

Very nice indeed, awesome board.

[l0ud_sil3nc3]

Awesome run SF3D

1300 cas5 @5G seems kinda slow though. Its probly revert to cas7

If you do not set CAS 5 in the bios before memory training, and TWCL is not adjusted accordingly you are not really seeing CAS 5.

[l0ud_sil3nc3]

I thought it's actually 2600MHz? Right?

2600mhz effective, Dumo is correct it is 1300mhz

[fuzz3l]

Nice pics and results Petri!

Christian

[SF3D]

Yes CL5 results are very bad because those runs are just rerun after another rerun without any tweaking. Main issue is, that + TWCL timing must be set at same level as CL latency to have correct efficiency.
The memory bandwith and performance will be bugged if they are not the same value. Then results will be really bad like in these CL5 runs. The CL latency is changed for real, but it have minor effect on end result.
CL4 can be also used, but it is not real value and will only show up in CPU-Z but it is actually CL7 for real.

Here is basic run with some tweaks on air with loose mem timings:


7-11-7 and TWCL at 7 is giving best performance for me. I got 6-11-6 TWCL at 6 almost stable, but it would require too much IMC voltages for my "safe" benching.

I can not risk my chips nowadays

[ZoLKoRn]

Nice bench and nice to see you here again SF3D

well can you try compare CL5 vs CL6 with this RAM kit again ? i have problem it for my results CL6 better over CL5 every time
and Nick told me TWCL is key and i already tried it to but still same CL6 better

btw. what IMC voltage on your used for CL5 ?

Thanks if you have time

[Motley]

Did it snow inside your house? Good lord that is a lot of frost? That doesn't leak/melt onto any components?

Never seen anything like that before.

[nickshih]

Nice bench and nice to see you here again SF3D

well can you try compare CL5 vs CL6 with this RAM kit again ? i have problem it for my results CL6 better over CL5 every time
and Nick told me TWCL is key and i already tried it to but still same CL6 better

btw. what IMC voltage on your used for CL5 ?

Thanks if you have time

I will try all my PSC here next week. I think CL5 eff is related to the IC also .

[SF3D]

Nice bench and nice to see you here again SF3D

well can you try compare CL5 vs CL6 with this RAM kit again ? i have problem it for my results CL6 better over CL5 every time
and Nick told me TWCL is key and i already tried it to but still same CL6 better

btw. what IMC voltage on your used for CL5 ?

Thanks if you have time

To have good efficiency you would need to have TWCL set to 5 or 6 if you would like to use 5-11-6 or 6-11-6 settings. In my case TWCL 6 was almost stable, but 32M crashed every time in some point. Scores would have been very good with same clocks I used in my best run. TWCL 5 will be impossible with my setup. If someone can make their system run TWCL 5 and 5-11-5 settings with 1300MHz speed, that would be sick bandwidth for sure I hope someone will try it.

Did it snow inside your house? Good lord that is a lot of frost? That doesn't leak/melt onto any components?

Never seen anything like that before.

It was very humid in our house at that point, so in this long bench session there was a lot of frost forming over the parts. It is not dangerous if insulation is done properly.
Session was almost 4 hours long without any issues. Usually memory benching with Ln2 pot on them is causing troubles, but I guess insulation methods are better now.

With triple point pot memory chips and sticks are cooled down for real, so condensation will be issue around the memory socket area and pcb around them, if there is any parts without insulation.

[dernettemann]

Wow that board looks realy nice

[SoF]

very nice benching petri!

[Oliver]

ya know what SF3D means ..

but this is seriously SF2D .. Seriously Flying 2D

KEEP PUSHING IT