[AnandTech] Memory Performance: 16GB DDR3-1333 to DDR3-2400 on Ivybridge IGP with G.Skill - Page 4

No my ram just went bad and downclocking is easier than rmaing lol.

Mine are set to 1333( default ) I'll gonna try to see if I can go to 1866 since my memory are rate 2000mhz

Here is something interesting found in d3. I still like to see more results of games that use physx (w/physx enabled) and games that maybe cpu bottlenecked.
http://www.rwlabs.com/article.php?cat=&id=691&pagenumber=5

AnandTech writes some great reviews and I count them as one of the best tech reviewers out there, but I am kind of surprised by this one. It has some interesting information but the actual test bed and some of the settings are not very relevant. Someone who cares or even knows about memory speed is not likely to be using the onboard HD4000 graphics. Even if they are they wouldn't be using unplayable settings as in the case of Metro 2033 (5.22-5.72 fps) and Civ V (5.98-6.38 fps), among others. He mentions not wanting to use synthetic benchmarks but it's just as bad to use settings and resulting fps that no sane person would tolerate.


Source

Then in the conclusion he mentions his first rig, an E6400 with 4GB of DDR2 (see full quote below). He says he got it sometime in 2005 but the problem is the Core 2 Duo E6400 2.13GHz was released on July 27, 2006 according to CPU World. He mentions his OCZ 4GB DDR2 kit being ~$240 in 2006 which definitely could be possible. He doesn't mention what exact speed kit he got, besides it being OCZ. Even if it is true it might exaggerate the expense of DDR2 RAM when possibility 1GB sticks were a better value. I don't have any proof of this and only my memory and as witnessed by this review one's memory isn't always reliable. So I won't push this point, it just seemed high to me. The tilde before the price indicates he wasn't positive about the price himself. Source

I know reviews are time consuming and take a lot of work, so I don't complain often about reviews unless they deserve it. This review is not up to AnandTech standards and I am honestly surprised they would publish it.
Edited by Ben the OCer - 10/19/12 at 6:18pm

I just oc my memory to 2202mhz and I get 1700 point more in 3Dmark 11
Default memory 1066 around 14600 now at 2202 I just get 16234 .

Many thanks to everyone who liked the review. I'm the writer, also a member of the OCN HWBot team.

Anand doesn't write everything on Anandtech That's why it says 'by ...'

Also the testing was done before the latest Piledriver was released. There are plans to do it again with Trinity.

It was mainly a review of the memory kits rather than the speed differences. Don't worry, you weren't the only one to think this

Common policy in benchmarking XYZ is to make it the limiting factor in testing. Hence the use of an OC'ed CPU in an ASUS P8Z77-Premium. Despite your comment regarding 'anyone who knows about memory speed is not likely to be using HD4000', you would be actually quite surprised - Intel have released a dual core 65W CPU with HD4000, the i3-3225, due to market demand.

With respect to the gaming settings, you cherry picked those two games. I tested a variety using none of the eye candy and all of the eye candy, to provide comparisons between CPU/iGPU intensive options such as lighting, high resolution textures and the like. The more important graph you should have been concerned about is the % rise against DDR3-1333:



To quote one of the comments on the article:

"The point of the test was simply: "Lets take a system that is generally fast and put it in a situation where the IGP is being stressed. This will be the best-case scenario for faster RAM helping it. Lets see if it does".

To me the idea was not showing everyone everyday situations where faster RAM will help them, instead it was to see where those situations might lay, by setting up a stressful situation and seeing the results. Most of the results were extremely small differences."

In the comments I mention it is not my first rig. Up until that point I had hand-me-downs or pre-builds. As a result, the first kit I purchased just as a kit rather than as a PC was that kit. I estimated around 2005 as it was around that time I built that machine (2nd year of university then, which I started in 2004. That would put my second year at 2005-2006).

What would you like to see in a memory review? Contrary to what you may think, reviewers are often open to suggestions. We are only individuals, not lords and masters of every different usage scenario. We attempt to cover the vast majority of use cases, and in this review, memory speed gets very important on IGP hence the focus. If you would like to see more of X, less of Y, or include Z, then please let us know. Clicking on the author's name at the top is a direct line to the email.
Edited by borandi - 11/12/12 at 9:00am

The overclocking results are embarassing for G.Skill.

Although a good company, I tend to discriminate against them for selling all their memory with 2T command rate. This is just a trick they use to barely pass binning and hence all 3 sets I've played with did not overclock whatsoever.

Moore's law is still quite accurate when it comes to memory. 7 years is ~ 2*2*2*1.4 = 11,4 according to moore's law. So 4 GB in 2005 is equivalent for 44 GB ram in 2012. 240$ for 44 GB sounds quite normal to me.

I can bump my 1600Mhz 9-9-9-24 1.35V hit to 1866, but I have to crank it up to 1.47V. I can't even post at 2000Mhz regardless of timings or voltage.