Geekbench as basis for making my purchase decision?

[SChimahusky]

My first post, so bear with me here...
I am in the midst of deciding which macbook I am going to buy. I am comparing MBA13", MBP13", MBPr13" and MBP15" in various configurations. Now to the question:

I have included a geekbench score in my computer analysis, and combining geekbench scores with computer price to create a metric. For example, MBA with 8GB RAM, 2.0 Dual Core i7 processor has an average geekbench score of 7157 and price tag of $1369, and so I have given it a "$ per geekbench" point value of .228. (i.e. It costs 22.8 cents per point on the geekbench scoring at this price point) In contrast, a MBP15" with 2.3 Quad Core i7 and 8GB RAM gets a "$ per geekbench point" value of .175 (i.e. it costs 17.5 cents per geekbench point.)

Seeing as how there are some significant similarities with the computers (specifically screen resolution, RAM), do people find geekbench scores reliable and valid enough to make this kind of comparison? Looking at these numbers, It is worth spending 2000 dollars on a MBP15" over the MBA because the power you get for the money is "cheaper" for the MBP than it is for the Air. Any comments and suggestions are appreciated.

Side note -- I relate this question to shopping at supermarkets, and looking at price per ounce for canned goods or cereal.... the question is, does the score mean enough to base a decision on it.

 

[duervo]

I wouldn't use GeekBench as the sole metric on which to base my purchasing decision, unless you know for a fact which scores were generated using the 32bit trial version, or the 64bit full version of GeekBench. The 64bit version usually achieves slightly higher scores than the 32bit one.

If you can verify that all your scores are either with the 32bit or 64bit version, then I think you can probably use it as a metric on which to base your decision. I wouldn't rely on it if there's a mix of both, though.

 

[SChimahusky]

I took the average of 32 and 64 bit scores for each configuration. I've had a lot of time on my hands today

 

[Suno]

If you're a casual user, anything above 9k is good and will last you for years, provided you take care of it. Beyond that, it's almost pointless crunching numbers and calculating per point cost because you won't see any notable difference in performance.

Just think about it -- Do you really think maximizing your "value" per point is going to net you better performance or more "worth" for the cost?

With that said, I got a 15" high-end cMBP because I valued performance and I got a better performing CPU then what I would have gotten for the same price on a Retina. In the grand scheme of things, however, I would have gotten the same performance on both.

 

[throAU]

Geekbench is pretty irrelevant for many people, as it is a CPU-only benchmark. I.e., it does not take into account SSD, RAM size, discrete video, etc.

Unless you know that your workload will be CPU-bound, geekbench as a metric for purchase is pretty irrelevant.

What most people do with their mac is not cpu bound for 99% of the time they are using it.


More important factors that will make a real world difference for most people is SSD vs non-SSD, amount of storage, amount of RAM and whether the machine has discrete video.

And of course, form factor.


Unless you know you need top level CPU power above all else (and most people do not - if you do, your only choice is 15" in the mobile platforms), you are over thinking this.

Pick the size machine you like (if you need discrete 3d, this is 15"), work out how much storage you need, buy as much ram as you can afford.


edit:
case in point - the mini running snow leopard in my sig feels mostly the same in web browsing use to the MBP in my sig, which geekbenches probably 8 times faster than it. Yes, the MBP is way faster at video transcoding, etc - but the point is, it all depends on what you're doing with the machine.

 

[duervo]

More important factors that will make a real world difference for most people is SSD vs non-SSD, amount of storage, amount of RAM and whether the machine has discrete video.

That's a good point that I forgot to mention.

To the OP, the weight of your scores could possibly be affected by things such as ability to upgrade the RAM and storage, which may be offset by the weight you give the retina display (for the MacBook Pro's.)

So, slightly changing my suggestion above, I would not use GeekBench scores as the sole metric. I would rather include them in with a few others. You will have to come up with an appropriate weight/worth for each option ... options such as ease up upgrading, retina or non-retina, smaller/lighter, larger/heavier, etc.

 

[BlaqkAudio]

Geekbench is pretty irrelevant for many people, as it is a CPU-only benchmark. I.e., it does not take into account SSD, RAM size, discrete video, etc.

Actually, Geekbench measures both CPU and RAM performance.

OP, if you're really set on using benchmarks, you should try Xbench. It test CPU, RAM, graphics, and disk performance.

 

[throAU]

Actually, Geekbench measures both CPU and RAM performance.

Well, sure. RAM/Processing performance are linked as well. But... RAM performance is even less relevant for most people

 

[SChimahusky]

thanks

Thank you all for your replies. As I see it, my metric is useful for a purpose, but maybe not the purpose I created it for. Rather than continuing here, I am going to start a new post with a more general question posed. I still cannot decide which model to go with.