So, is there actually any "real-world" difference between 1067, 1333, 1600, 1800 RAM?

[mgartner0622]

I know there are threads saying that higher speed memory works in the newer 2011 MacBook Pros. However, as a power user, would I actually notice any difference between say 1333Mhz and 1800Mhz RAM? None of the other threads really address this.
I can buy 1333 for $29, while 1800 is nearly triple the price. Is it worth it? Does it affect the user experience at all besides increasing the geek bench scores?

 

[Hellhammer]

Faster memory rarely helps, mainly because the memory bandwidth and latency is not the bottleneck. Take a look at e.g. this test.

 

[bill-p]

Here are some graphs. For desktop RAM, but I think it applies to Macbooks and laptops somewhat.

http://www.tomshardware.com/reviews/ddr3-1333-speed-latency-shootout,1754-22.html
http://www.tomshardware.com/reviews/ddr3-1333-speed-latency-shootout,1754-23.html
http://www.tomshardware.com/reviews/ddr3-1333-speed-latency-shootout,1754-24.html

The short version of it is... 1333MHz gives a tiny gain to 1066MHz. Past 1333MHz, you're running into processor limitations, so your CPU is always slower than your RAM. I suspect that on laptops and Macbooks, the difference is even less pronounced as we have even slower processors than desktops.

So I think it is as you said, no benefit. Which was why I went for 1.35v instead of a higher frequency. I think lower memory voltage would benefit Macbook users far more than faster memory.

 

[mgartner0622]

In the Hellhammer's post, the 1333 memory was almost only 75% as fast as the 2100Mhz memory, while it seemed the difference between 1333 and 1600 really was not that great.

The reason why I was asking is I bought some 1333 2x4GB (8gb) from Newegg on sale for $29. I was wondering if I should have spent $60 to buy some 1600, or around 80 to buy some 1800.
After reading those graphs, I'm pretty confident in my purchase, plus I got a free 4GB microSD card w/ SD adapter, so all considering, I think I got a pretty decent deal.

Thank you to the both of you.

 

[HippieMagic]

My gaming PC has had 1333mhz and 1600mhz and there has been absolutely no real world difference in performance between them. It didn't even impact the actual benchmark scores enough to matter.

 

[drambuie]

Another factor affecting memory bandwidth, besides clock speed, is latency. The faster the RAM, the higher the latency, which somewhat offsets the gains from higher clock speeds. If you increase the clock frequency by 50%, but latency increases by 25%, you haven't gained a whole lot. You also have to consider the capabilities of the CPU memory controller, and bus bandwidth.

Latency is the delay, measured in clock cycles, for the RAM to perform addressing, and various internal functions.

In a desktop system with high quality overclockable RAM and a motherboard with an overclocking BIOS, it is possible to use higher clock speeds, while maintaining lower latencies. Combine that with an i7 Extreme CPU, and you have the potential for major memory bandwidth gains at higher memory clock speeds.

Unfortunately, laptop hardware operation is basically closed to user configuration. Also, SODIMMs have limited available frequency/latency configurations, compared to desktop DIMMs.

 

[mgartner0622]

Another factor affecting memory bandwidth, besides clock speed, is latency. The faster the RAM, the higher the latency, which somewhat offsets the gains from higher clock speeds. If you increase the clock frequency by 50%, but latency increases by 25%, you haven't gained a whole lot. You also have to consider the capabilities of the CPU memory controller, and bus bandwidth.

Latency is the delay, measured in clock cycles, for the RAM to perform addressing, and various internal functions.

In a desktop system with high quality overclockable RAM and a motherboard with an overclocking BIOS, it is possible to use higher clock speeds, while maintaining lower latencies. Combine that with an i7 Extreme CPU, and you have the potential for major memory bandwidth gains at higher memory clock speeds.

Unfortunately, laptop hardware operation is basically closed to user configuration. Also, SODIMMs have limited available frequency/latency configurations, compared to desktop DIMMs.

Interesting... thanks for explaining! I'm guessing latency is different for different brands of RAM?

 

[Hellhammer]

I'm guessing latency is different for different brands of RAM?

It depends on the specifications, not on the brand. You can calculate the latency on your own, as well.

1 / (f/2) * CL

Where f is frequency and CL is the CAS latency. The answer will be in nanoseconds (10^-9 seconds).

Example: The latency for 1333MHz DDR3 with CAS latency of 9 is.

1 / (1,333,000,000Hz/2) * 9 = 1.35*10^-8 = 13.5ns

So basically, the frequency and the CAS latency affect the final latency. Higher frequency and lower CL means lower final latency.

1333MHz DDR3 usually has CAS latency of 9 and hence latency of 13.5ns as calculated above.

1600MHz is usually CAS 11 unless you go for more expensive RAM where CL=9. 1600MHz DDR3 with CAS 11 has final latency of 13.8ns. With CAS 9, the latency drops to 11.3ns.

 

[theSeb]

Here are some real tests on different Mac configurations that I did. The faster memory does not make a difference in most things, but helps the HD 3000 IGP to be about a frame faster in OpenGL.

https://forums.macrumors.com/threads/1293809/

 

[PieterL]

I read that ram speed can help in some instances when making music (I use logic myself)

 

[theSeb]

I read that ram speed can help in some instances when making music (I use logic myself)

I would say you would probably see things 1 minute faster over a couple of weeks of use.

 

[hotgrease]

I got some 1600 ram from Newegg for $35 so I went with that. It seems good but I don't know if it is faster than 8GB of 1333 would have been.