Mihail Stoynov's blog

Most currently sold laptops are with DDR2 memory. Even the high-end ones, like Sony Vaio. It would seem that only Apple sells its current line of laptops (the aluminum body ones) with DDR3.

Why is that? Why isn’t there other high-end laptop with DDR3? It would only make sense to make it faster with DDR3. Or maybe it wouldn’t make it faster? I checked it out.

This is an excerpt from a post in the forum of macrumors.com

The polymac has 2gb of DDR2 @667mhz much like it’s predecessors. The aluminum mac has 2gb of DDR3 @1067mhz. At first glance you might say WOW! 1067 is a lot more megahertz than 667! I gotta have that.

This doesn’t figure in 3 things.

1. DDR3 increases speed at the penalty of higher latency. In real life terms, lets simplify it a lot and say that DDR2 and DDR3 are cars, and that the top speed of DDR2 in this case is 66 mph and the top speed of DDR3 is 106mph. But DDR2 has lower latency than DDR3, so lets say DDR2 only has to drive 5 miles and DDR3 has to drive 8 miles. So if the speed were the same it would take DDR3 longer cause it has a longer way to drive. DDR3 makes up for it by driving faster, and as you can see the faster you can get DDR3 the less difference the latency makes (i.e. that 3 miles makes less over all difference when you drive 160mph i.e. DDR3 1600mhz (no not available on macbook)) What this does mean though is that at lower speeds of DDR3, there isn’t much benefit as the extra latency makes the performance very similar.

2. According to others you can upgrade your ram to DDR2 800mhz
http://forums.macrumors.com/showthread.php?t=651330&highlight=ddr2+800
http://discussions.apple.com/thread.jspa?threadID=1877953&tstart=0

Now let’s go back to point 1. Now if DDR2 can go 80mph over 5 miles vs. DDR3 going 106mph over 8 miles, there’s probably even less of a difference.

3. The last and most important thing to remember is that computer performance is tied to all the hardware, and is bottlenecked by some things far more than others. In some benchmarking between DDR2 800mhz and DDR3 1067mhz on a PC with virtually identical hardware otherwise, the bandwidth increased from 6146 to 6613… or about 10%. Which sounds alot faster. But then they performed other benchmarks, such as super pi, which showed that
the DDR2 800mhz finished in 46.08 seconds and the DDR3 1067mhz finished in 45.11 seconds. The difference is 0.97 seconds or a whopping 2.1% performance boost.

The guy even supplied a lot of benchmarks:

To further illustrate this here are some benchmarks with the old polymacbook vs the new aluminum macbook

http://eshop.macsales.com/shop/Memory_Benchmark/Apple_MacBook/

To better demostrate apples to apples, (no pun intended we’ll look at the 2.4 ghz macbook aluminum vs the 2.4ghz polycarbonate) in photoshop CS3

Lower time is better
Macbook Aluminum 2GB ram 99.34
Polucarbonate macbook 2GB ram 107.63

which shows that the macbook aluminum is faster 1067 vs 667 here.

But what happens when you upgrade to a 320gb 7200 rpm drive?

Macbook aluminum 2gb ram stock 99.34
Macbook aluminum 2gb ram 320gb 92.28

how about when we upgrade the ram?

Macbook aluminum 4gb ram stock 79.06
Macbook aluminum 4gb ram 320gb 77.19
Polycarbonate macbook 4gb 78.41

Then the guy continues with comparing prices of the two macbooks.

Source: http://forums.macrumors.com/archive/index.php/t-654277.html

Since this is a forum post, one has to check the validity of the statements. So I went to wikipedia. This is what I found out:

CAS (column address strobe) latency (CL) is the delay time which elapses between the moment a memory controller tells the memory module to access a particular column of memory on a DRAM memory module, and the moment the data from given array location is available on the module’s output pins.

In asynchronous DRAM, the interval is specified in nanoseconds. In synchronous DRAM, the interval is specified in clock cycles, and must be multiplied by the cycle time (i.e. divided by the clock frequency in GHz) to convert to nanoseconds.

[…]

Because they have multiple internal banks, and data can be output from one during access latency for another, the output pins can be kept 100% busy regardless of the CAS latency; the maximum attainable bandwidth is determined solely by the clock speed. However, that only applies if the data to be read is known long enough in advance; if the data being accessed is not predictable, the latency matters. For a completely unknown memory access, the relevant latency is the time to close any open row, plus the time to open the desired row, followed by the CAS latency to read data from it. Due to spatial locality, however, it is common to access several words in the same row. In this case, the CAS latency alone determines the elapsed time.

[…] Note that memory rated for a certain maximum speed can always be operated at a lower clock rate. The CAS latency can be set lower in such a case. […]

[Then there’s a table showing how DDR3@1066Mhz has a CAS latency of 7 cycles and DD2@666Mhz – of 5 cycles]

Source: http://en.wikipedia.org/wiki/Memory_latency

So the guy in the forum seems to be correct.

One thing remains unfolded though – why is the latency higher with higher speed? This is not explained in either the wiki, or the forum post. And unfortunately I don’t have the time to investigate.

If someone knows, please share.

This brings me to the reason I investigated all that – I have a spare 2x 1GB DDR@1066 Mhz:

(I guess no one will want them because only Macbook guys have machines with DDR3 and the minimum for them is 2GB)

I have high hopes for the 160GB-5400rpm-Hitachi harddrive though.

If someone is interested, please let me know.