G.skill ECO 2x2GB 1333 7-7-7-21 Review/ Why the Asrock A85X extreme 6 is a TERRIBLE motherboard

 I finally figured out how to review these and got my review system fully setup so here goes.

Methodology of testing.
For the frequency part of the testing I test to see how high the kits can clock a certain set of timings for different voltages. This set of timings will be decided after I get a little time to play with the kit and so it will be different for every kit depending on how it does in pre-testing. Once those the timings to test are decided I will test the kit at the following voltages:
1.35V 1.5V 1.65V 1.7V 1.75V 1.8V 1.85V
The testing of the kit will be me raising the clock of the kit as high it will go while capable of running SuperPi 4M on my install of windows XP without crashing or erroring.

Note: I had to swap boards between the 7-9-8-24 and the 8-10-9-27 tests because my F2A88X-D3H died after did something stupid. So the rest of the review was tested using the Asrock A85X Extreme 6 and it has a bout 25mv of overshoot on the RAM VRM so when I list 1.65V for the higher timing test it actually means 1.675V  and 1.35V is actually 1.375V. The Gigabyte on the other hand under shot so if the results going from 7-9-8-24 to 8-10-9-27 look weird that's why. I will this stuff in later reviews. Also the A85X Extreme 6 is absolutely awful with RAM. I couldn't get the 21.33 multiplier to run and certain BLCKs wouldn't run any other multiplier than 13.33. It also automatically changes the Vcore and NB voltages when you change RAM frequencies which explains how it killed my 6800K it probably fed the NB 1.5V when I was trying to boot 2600mhz. Basically this board deserves to get punched. It's also why this review is unfinished. If anyone is wondering I was using the LN2 BIOS for the board. I wanted to test way more timings however there is no point in that when the board is incapable of producing certain RAM frequencies.

Results
Frequency


BTW RAM reviews take really really long to do. Especially when you don't have an SSD to boot from.

AMD RAM Timing Guide

Voltages
The usual DDR3 voltages for 24/7 are anything up to 1.725V actual(checked with DMM). For benching I wouldn't go above 1.85V on air cooling but if but if the sticks scale well with voltage 1.9V is still reasonable depending on what sticks you have.
With AMD chips the north bridge plays an important role in just how high clocks you can run so once going above DDR3 2400 you will need to raise the voltage. For 24/7 on 32nm and 28nm CPUs stay below 1.35V and for benching stay bellow 1.4V.

Primary Timings
The primaries behave like primaries on intel and depend entirely on your kit of RAM. You can run lower ones with more voltage.
The only rule for the primaries is that CL+tRCD+tRP=tRAS ±3

Secondary Timings
The Secondary timings have a massive impact on how high you can clock your RAM. Higher numbers will usually yield higher clocks but going to far will make your system unbootable so do not just max all of them out lower timings will get you more performance for a set frequency.
A free timing is a timing that you can adjust independently from all the other timings without losing stability for any other reason than being too low.

Gigabyte	ASUS	MSI	Asrock	Dependancy
tRC	WIP	WIP	WIP	free
tRRD	WIP	WIP	WIP	free
tWTR	WIP	WIP	WIP	free
tWR	WIP	WIP	WIP	previous*2±2
tWL	WIP	WIP	WIP	previous-3
tRFC0	WIP	WIP	WIP	free
tRFC1	WIP	WIP	WIP	free
tRTP	WIP	WIP	WIP	free
tFAW	WIP	WIP	WIP	previous*4
Command Rate	WIP	WIP	WIP	free

AMD only settings


Processor on die termination.
Settings: 240ohms 120ohms 80ohms 60ohms.
Lower on die termination allows running higher frequencies.

Drive Strengths:
Higher drive strengths allow higher clocks and lower timings. The DQS and DATA drive strengths top out at 1.25x for me regardless of what I do. I set all the other settings to the absolute maximum of either 2x or 1.5x.

I do not have any pictures because different motherboards have different timing table layouts so you'll just have to read them all anyway.

Friday bench day: 5.16Ghz FX 6350 Cinebench Rampage

This is the secondary computer I have at my grandma's house. Based around an FX 6350 8GB of Corsair Vengeance RAM and a ASUS Sabertooth 990FX motherboard. And since it finally started to snow a little I decided it was the ideal time to do some benching on low ambient temperatures.

So after a few hours of having both windows in my small tower(seriously it's a tower) room open the temperature reached around 8C° at the intake of my Phanteks heatsink.

Cinebench was the first benchmark I ever used and since it's the only benchmark I have installed on this computer that doesn't greatly benefit from a clean install of windows I did Cinebench 15 and Cinebench R11.5. In R15 I got 624points at  5.16Ghz. That's a nice 3rd place on the HWbot.org FX6350 leader board. 

I'll be honest I'm proud of my efficiency on this run because I was running on my Corsair Vengeance sh*t sticks.

I managed to get another 0.2 points in Cinebench R11.5 over my 5Ghz score. Unfortunately the cold got to me and I forgot to do a second run and so the score is ever so slightly lower than it could be.
I'll be doing another session soon on my 2333 9-11-10-32 G.skill Pis. Also it looks like this CPU could run 5.16Ghz all day everyday if I just had a good cooler because all these runs were done on just 1.6V with High LLC settings.
Here's a challenge to all my readers who have an FX 6350. The first to beat my score in either Cinebench R15 or R11.5 by the 27th of January 2015 and I'll give you 10,000 DOGE. !NO LN2 and leave a link to a pic of your system and your dogecoin address in the comments!

BTW I updated the support this blog page to include sources of free dogecoin. If every time someone goes on this blog they went to this site and set the dogecoin to this blog I could get or buy a fan to review every month and it just takes a few clicks.

The RAM and FAN test bench overview

 Built on an IKEA pine wood wine rack. For low weight.

 With WiFi for easy screenshot uploading and maximum portability.  (Mounted with zipties)

 Based on the Gigabyte F2A88X-D3H I reviewed ages ago.
(Also mounted with zipties)

 Insulated for use with LN2.

and with a crocodile clip for reading CPU core voltage directly

Running an Athlon II X4 750K cooled by a cooler master Seidon 120V closed loop cooler.
(The rad really is just sitting there)

With video provided by my modified R7 260X
(The cooler is mounted with zipties)
(If I get a bigger PSU and another CPU cooler I will replace it with the water cooled GTX 590)







 The RAM currently in this beast is the G.skill ECO 2x2GB 1333 7-7-7-21 1.35V kit I previewed some time ago.







 The HDD is a 350GB WD blue.
(Mounted with even more zipties)

The PSU is an EVGA 430W 80+ unit.
(Held in place by zipties)

This is built from scrap parts from other projects(GPU MB and CPU Cooler) and my friends old PC(wifi and HDD). The reason why I built this is that my 3960X is never ever going to boot with RAM above 2500mhz so I can't use it for RAM reviews. This 750K already managed to boot 2520mhz on these G.skill sticks so it can do a better job with RAM than my main system. Ideally this would be a Z97 system but good Z97 boards cost about 1.5-3X what this board cost and an i5 4690K is about 3X the price of the 750K. Yes I know there is the Pentium G3258 but IMO pairing a 3000czk MB with a 1700czk CPU is stupid. Plus AMD CPUs are funner to OC.
The idea for using the wine rack from IKEA is that. It can hold as much stuff as a full tower, cost only 199czk(10$) and weighs less than any other case you can buy.
I used the cooler master Seidon for cooling because it allows me to easily swap fans for fan reviews.
If I had to change any things I would get a better PSU(750W Seasonic/EVGA) and a better MB(Crossblade Ranger).

G.skill ECO 1333 7-7-7-21 2x2GB Early Expirience

Not long ago I bought a G.skill ECO 2x2GB 1333 7-7-7-21 1.35V ram kit. Now the stock settings on the kit may not sound like much but these things are great. Within 35 minutes of playing with them I had them booting 2400mhz. With better timings than many 2400mhz that you will find available in stores. Now they were not the miracles that I wanted because these sticks use PSC ICs and those are know to do as much as 2600 8-12-8-28 @ 1.85V on air cooling. Now this was early testing so I wasn't pushing very high voltages but it is very obvious that I have a pretty bad PSC kit. The settings I ended up with were 2400mhz 10-12-11-33-1T with sub timings of 5-160-12-6-24-6-7 and 1-1-1-1-3-3-2-1-3-1 using 1.75V. The primary timings are pretty bad as they are barely better than my 24/7 sticks which do 2400 10-12-12-34-2T. However the secondaries and tertiaries are pretty damn good.

I hope that I can get this kit to run CL 9/8/7 above 2400-2000mhz since that should put them way ahead of my 2000mhz 8-9-8-24 Ballistix Elite that got me my Super Pi 32M record.
I will have more updates after I'm done with the kit.

My plans for this kit are to see how high and how low it can go. Now frequencies above 2400mhz are really problematic for the IMC on Sandybridge-e CPUs so I doubt I will hit above 2500 even if the sticks are capable of that. However what I'm more intrested in is just how low I can take the CL because a low CL from my experience is key to getting a HWbot prime score so if these sticks could boot CL6/5 at 1333-1600mhz that be great.

UPDATE:
The kit is optimal at 2333 9-12-11-32-1T with secondaries of 5-156-12-6-24-6-6 using 1.8V. I still haven't done any benching with these because I need to setup my new benching HDD.
I still have to try top the frequency and try the low CL(5,6,7) max clock before doing a full write up this weekend.