Sign in with
Sign up | Sign in
Your question
Closed

Which combination of RAM would be a better choice? (2x4, 4x2, or 8x2)

Last response: in Memory
Share
May 14, 2012 10:12:18 PM

Looking to upgrade my PC memory, as I've just upgraded the majority of my PC and the last parts I need to upgrade are the GPU and RAM. I'm just wondering which might be the best choice for me. 2GB's x 4 Sticks, 4GB's x 2 Sticks, or 8GB's x 2 Sticks.

I'm currently looking at these selections:

G.Skill Ripjaws X Series 8GB (2 x 4GB) 1600MHz - $42.84 (Total)
G.Skill Ripjaws X Series 16GB (4 x 4GB) 1600MHz - $82.04 (Total)
G.Skill Ares Series 16GB (2 x 8GB) 1600MHz - $104.83 (Total)

I've never been really sure how timings change the performance, but the two G.Skill Ripjaws X Series are 9-9-9-24, and the G.Skill Ares Series is 10-10-10-?.

Or is there a better deal than any of those for the same performance? I want at least 8GB's @ 1600MHz. Please note I'm also using a "15% Off All Memory Coupon" for Newegg, which brings the prices down a bit.

And if it matters, the Mobo/CPU I currently have are a GIGABYTE GA-Z77X-UD5H, and an Intel i5-3570k.

Thanks for any help!
May 15, 2012 3:02:32 AM

What will you be using the system for? If it's for gaming I'd recommend just 8 GB (2x4GB). The G.Skill Ripjaws X are VERY good and recommendable.
Score
0

Best solution

a b } Memory
May 15, 2012 3:28:55 PM

Unless you are doing something such as Photoshop, 16GB is more than necessary. For gaming, going beyond 8GB is known to not help at all. Where possible, it is recommended that you only have one RAM module per channel and since the Ivy Bridge platform has two memory channels (like most do), two memory modules are preferable.

Having one module per channel is better than having two modules (four RAM modules on a dual channel system, such as Ivy Bridge, or eight modules on Sandy Bridge E, or six modules on the X58 platform) increases the strain on the memory controller and that decreases performance very slightly, usually less than a 1% drop (games at say 53.46FPS might drop to 53.39FPS. I'm making up the numbers there, but they are representative of the usual difference) in performance.

So, unless you need more than 8GB for a certain task (or just want it for some reason), the best configuration, like dedekind said, would be an 8GB kit consisting of two identical 4GB modules, IE a 2x4GB kit. I don't like Ripjaws myself, but that's only because I don't like large heat spreaders (low profile heat spreaders increase the chance of RAM fitting beneath a large CPU cooler). G.Skill, Corsair, and Crucial are generally the best RAM companies to buy and G.Skill's Ripjaws series is pretty much always great.

I doubt that this will be a problem, but make sure that the modules that you buy are 1.5v modules. Higher voltage requires you to change a few settings in the BIOS because if you don't, the memory controller can't handle the high voltage and your CPU will probably die sooner than it would with RAM that has a lower voltage (having RAM with too high of a voltage can cut the life time of a CPU from a long 5 to 20 years down to less than one years, but like I said, it varies between different CPUs, even of the same model. The variance is very large). You will probably not get a warning about this, just one day find your machine (probably inexplicably) not working anymore.

You said you want 1600MHz and I'll say that you chose a optimal frequency. 1600MHz is generally the greatest performance for the money and going beyond it tends to not net much gain at all, except in specific situations that are not representative of usual usage by most people (archiving, rendering, and some similar workloads such as some folding, are usually what benefit the most and most other programs take little to zero benefit. Gaming is known to almost always take no perceptible benefit at all). Since you have a coupon, you might be interested in getting an 1866MHz 2x4GB kit, but I recommend against going above that.

About the timings, they refer to latency, measured in real clock cycles of the memory. For all system DDR RAM memories, the real clock frequency is half of what it is said to be (IE 1600MHz memory is actually 800MHz memory, we call it 1600MHz even though it is not) because they are DDR, IE Double Data Rate (they transfer data twice per clock cycle, so when DDR first came out, instead of explaining this to everyone, they just say the effective frequency because for the most part, it doesn't make a difference what we call it unless you are talking about something such as the latency).

The timings (9-9-9-24 or 10-10-10-x, x probably being 30) do not make much difference, especially when their numbers are this close. They usually make less of a difference than the frequency does and the frequency already doesn't make much difference. Since the timings are measured in clock cycles instead of in time (IE they aren't measured in nanoseconds or anything like that), if you were comparing timings between different frequencies, you'd need to do a little math to compare them accurately. For example, 1600MHz 9-9-9-24 is higher latency than 1866MHz 9-9-9-24, but it is lower latency than 1600MHz 10-10-10-30 and 1866MHz 11-11-11-30. These are just four different timings, there are many more, but they are the four most referred too and chief among them is the CAS, the first number in that sequence. It is usually the most important timing and has the greatest impact on performance of all timings, although it is still usually less of an impact than the frequency which is already usually of little importance for actual system performance.

9-9-9-24 are the most common values for the four main timings for 1333MHz, 1600MHz, and 1866MHz memory. Only care about the timings if they are higher than this (unless the frequency is greater than 1866MHz, then looser, IE higher, timings are not as bad), or if you are choosing between two memory kits that otherwise have the same voltage, capacity, amount of modules, and frequency. They just aren't even close to being the most important factor in choosing what kit that you want for most workloads.

Sorry about the length of the post and if it seems kinda like a lecture, it's a pretty big topic.
Share
Related resources
Can't find your answer ? Ask !
May 15, 2012 3:51:32 PM

blazorthon said:
Unless you are doing something such as Photoshop, 16GB is more than necessary. For gaming, going beyond 8GB is known to not help at all. Where possible, it is recommended that you only have one RAM module per channel and since the Ivy Bridge platform has two memory channels (like most do), two memory modules are preferable.

Having one module per channel is better than having two modules (four RAM modules on a dual channel system, such as Ivy Bridge, or eight modules on Sandy Bridge E, or six modules on the X58 platform) increases the strain on the memory controller and that decreases performance very slightly, usually less than a 1% drop (games at say 53.46FPS might drop to 53.39FPS. I'm making up the numbers there, but they are representative of the usual difference) in performance.

So, unless you need more than 8GB for a certain task (or just want it for some reason), the best configuration, like dedekind said, would be an 8GB kit consisting of two identical 4GB modules, IE a 2x4GB kit. I don't like Ripjaws myself, but that's only because I don't like large heat spreaders (low profile heat spreaders increase the chance of RAM fitting beneath a large CPU cooler). G.Skill, Corsair, and Crucial are generally the best RAM companies to buy and G.Skill's Ripjaws series is pretty much always great.

I doubt that this will be a problem, but make sure that the modules that you buy are 1.5v modules. Higher voltage requires you to change a few settings in the BIOS because if you don't, the memory controller can't handle the high voltage and your CPU will probably die sooner than it would with RAM that has a lower voltage (having RAM with too high of a voltage can cut the life time of a CPU from a long 5 to 20 years down to less than one years, but like I said, it varies between different CPUs, even of the same model. The variance is very large). You will probably not get a warning about this, just one day find your machine (probably inexplicably) not working anymore.

You said you want 1600MHz and I'll say that you chose a optimal frequency. 1600MHz is generally the greatest performance for the money and going beyond it tends to not net much gain at all, except in specific situations that are not representative of usual usage by most people (archiving, rendering, and some similar workloads such as some folding, are usually what benefit the most and most other programs take little to zero benefit. Gaming is known to almost always take no perceptible benefit at all). Since you have a coupon, you might be interested in getting an 1866MHz 2x4GB kit, but I recommend against going above that.

About the timings, they refer to latency, measured in real clock cycles of the memory. For all system DDR RAM memories, the real clock frequency is half of what it is said to be (IE 1600MHz memory is actually 800MHz memory, we call it 1600MHz even though it is not) because they are DDR, IE Double Data Rate (they transfer data twice per clock cycle, so when DDR first came out, instead of explaining this to everyone, they just say the effective frequency because for the most part, it doesn't make a difference what we call it unless you are talking about something such as the latency).

The timings (9-9-9-24 or 10-10-10-x, x probably being 30) do not make much difference, especially when their numbers are this close. They usually make less of a difference than the frequency does and the frequency already doesn't make much difference. Since the timings are measured in clock cycles instead of in time (IE they aren't measured in nanoseconds or anything like that), if you were comparing timings between different frequencies, you'd need to do a little math to compare them accurately. For example, 1600MHz 9-9-9-24 is higher latency than 1866MHz 9-9-9-24, but it is lower latency than 1600MHz 10-10-10-30 and 1866MHz 11-11-11-30. These are just four different timings, there are many more, but they are the four most referred too and chief among them is the CAS, the first number in that sequence. It is usually the most important timing and has the greatest impact on performance of all timings, although it is still usually less of an impact than the frequency which is already usually of little importance for actual system performance.

9-9-9-24 are the most common values for the four main timings for 1333MHz, 1600MHz, and 1866MHz memory. Only care about the timings if they are higher than this (unless the frequency is greater than 1866MHz, then looser, IE higher, timings are not as bad), or if you are choosing between two memory kits that otherwise have the same voltage, capacity, amount of modules, and frequency. They just aren't even close to being the most important factor in choosing what kit that you want for most workloads.

Sorry about the length of the post and if it seems kinda like a lecture, it's a pretty big topic.




Don't worry about the length, I actually enjoyed reading that, it was very informative. Based on your information I've decided to go with "CORSAIR Vengeance 8GB (2 x 4GB) 240-Pin DDR3 SDRAM DDR3 1600". It has a 1.5v as you suggested getting, and was also only $38.25 after my coupon, which didn't seem like a bad price at all. I do not use Photoshop, or anything like that, so for gaming it seems like based on your information 8GB's should be more than enough for me. Also since I have an Ivy Bridge and you suggest one memory module per each memory channel, it seems to work out best with two sticks as well.

Thanks for your help! Thanks to you as well, dedekind!
Score
0
May 15, 2012 3:52:08 PM

Best answer selected by ferrari91169.
Score
0
a c 146 } Memory
May 15, 2012 5:31:06 PM

This topic has been closed by Nikorr
Score
0
!