I can't believe the benchmark results!

T

Tummy

macrumors regular
I can't believe that the new Airs are getting results that good. My old 1.6ghz air got around 37, the new high end models are getting 137!

It's funny that I spent nearly $2k 8 months ago for my Air which gets the same score as a $350 MSI Wind with a Hackintosh OSX install.

I replaced my Air with a new MB 2.4ghz they day it was released. I upgraded the memory to 4gb and the HD to a 7200 RPM 320gb Seagate and it gets the same 137 score. That new SSD must make a huge difference.
 
Where are you pulling you benchmarks from? Xbench? I agree the new MBA scores seem high. I'm tempted to buy one.
 
I just got the 1.86/128SSD to replace my 2.4GHz Feb release MBP and the new Air is VERY impressive. I havent tried a ton of high end apps yet but this computer makes quick work of launching apps and multitasking. I don't notice the difference on the Air vs my MBP other than shorter battery life on the AIR.
 
The results have me thinking that I should return my MB and upgrade to the Air. The only thing holding me back is the price differential of around $800 for the 1.8ghz SSD version.
 
The scores are hugely different because of the performance of the SSD and in smaller part, due to the new integrated 9400m. Overall, this made the MBA a more well rounded system which is why the scores are higher.

The SSD is what is essentially throwing off the benchmark and giving it an absurdly high overall score. The overall computer is not 4.5 times faster as one might suggest by looking at the difference in scores.

The overall xbench score, IMO, is too heavily influenced by the SSD's performance and is not a practical comparison of performance between the new Air's and other laptops.
 
The biggest reason is because the cache is double (6mb vs. 3mb).

This is why the Air is owning the Macbook despite its lower clockspeed. The Air has a 6mb cache vs. a 3mb on the Macbook.

The 1.86 Air should leave 2 of the 3 Macbooks in the dust, and compete with the 2.4 Macbook and 2.4 Macbook Pro.
 
Tummy said:
The results have me thinking that I should return my MB and upgrade to the Air. The only thing holding me back is the price differential of around $800 for the 1.8ghz SSD version.
Click to expand...

That would just be silly.

While the read speed of the new SSD's in the Air surpass the HDD in MB and MBP's, it is not indicative of a more powerful overall system. Again, this is where I believe the Xbench overall score is flawed and cannot be used in a practical comparison between laptops.

Aside from the above, in a couple months when the prices fall on the SLC SSD's or Intel MLC SSD's, you'll be able to get purchase SSDs with greater performance than what the MBA 2.0 currently has at a much cheaper price for your MB or MBP.

Bottom line: Don't be fooled by the overly exaggerated Xbench scores.
 
Tummy,

Aside from screen, weight etc, you could put an SSD in your MB. The current HDD goes in an enclosure or is sold. Anyone got benchmarks for this? Waiting a bit for the SSD should save some $.

Cheers,
OW
 
nick9191 said:
The biggest reason is because the cache is double (6mb vs. 3mb).

This is why the Air is owning the Macbook despite its lower clockspeed. The Air has a 6mb cache vs. a 3mb on the Macbook.

The 1.86 Air should leave 2 of the 3 Macbooks in the dust, and compete with the 2.4 Macbook and 2.4 Macbook Pro.
Click to expand...

That is not true.

In tests performed by Anandtech, the difference between a 6mb cache (2.5ghz) vs a 3mb cache (2.4ghz) in real world performance was small (on average 5%) - and that's inclusive of the 100mhz advantage that the 6mb has over the 3mb cache. The difference is even smaller when comparing a Merom processor with the same clock w/ 4mb cache.

There is no way a 1.86 penryn, even with 6mb, can match the power of a 2.4ghz (3mb) penryn - the difference in clock speeds is just too great.
There is no "owning" of any sort.
 
Side by Side my new Macbook Air (1.86 ghz 128GB SSD) scores a little over 137. My 1st Generation Macbook Air (1.8 ghz 64GB SSD) scores a little over 54. What a difference a few month's make.
 
I had never run a benchmark until everyone started talking about the program in the past week. My 1.6/80 scored about a 40. Wish I could upgrade, but it's going to be a while.
 
Tummy said:
The results have me thinking that I should return my MB and upgrade to the Air.
Click to expand...

six.four said:
The scores are hugely different because of the performance of the SSD and in smaller part, due to the new integrated 9400m.
Click to expand...

I do not have the MBA, but do have one of the new MB alum models (2.4) and can confirm what appears to be a huge performance gain because of the ssd.

Using Xbench, my new MB gets a score of 108 with all boxes checked for the test, BUT, if I remove the MB's standard hard drive from the test,I get a whopping 142.

This huge difference is all due to the read and write speeds of the hard drive.
 
six.four said:
Aside from the above, in a couple months when the prices fall on the SLC SSD's or Intel MLC SSD's, you'll be able to get purchase SSDs with greater performance than what the MBA 2.0 currently has at a much cheaper price for your MB or MBP.

Bottom line: Don't be fooled by the overly exaggerated Xbench scores.
Click to expand...

Well, I can never add the weight and size savings of the Air back into my MB. That is what I loved most and gave up to go to a MB 2.4.

I'm a consultant, fly twice a week, and stay at hotels 3-4 nights per week. So I really enjoyed the weight savings going to the Air, especially since my previous computer was the 15" MBP.

I think I am just amazed that they can get similar performance to the MB out of the new Air, which was so underpowered before. And they are able to sell them at the same price for the 1.6ghz and even cheaper for the top end model.

I still have my Air, but it's so slow and wimpy now, I can never go back to using it daily.
 
six.four said:
That is not true.

In tests performed by Anandtech, the difference between a 6mb cache (2.5ghz) vs a 3mb cache (2.4ghz) in real world performance was small (on average 5%) - and that's inclusive of the 100mhz advantage that the 6mb has over the 3mb cache. The difference is even smaller when comparing a Merom processor with the same clock w/ 4mb cache.

There is no way a 1.86 penryn, even with 6mb, can match the power of a 2.4ghz (3mb) penryn - the difference in clock speeds is just too great.
There is no "owning" of any sort.
Click to expand...
The thing is that its an accumulation of MANY advancements. I think the Cache increase of 50% actually helped a lot!!! Of course you might get low results in a TEST, but think about the real world "feel" of the computer. The cache increase is a significant part of the overall speed of the AIR. Actually, the whole point of the cache is its ability to work with the main memory!
The fact that there was a huge performance boost with the SSD, its vital that the cache can keep up. Its about the reads, not the Ghz. People put WAY to much emphasis on the Ghz when thats not the ONLY thing that makes a computer tick and you DON'T need 6 Ghz to run a basic app or get that "instant" result (like booting up an app). The big numbers are just for the people that don't understand computers (which is pretty a pretty large %)
Of course when testing speed, the Ghz does matter. But not necessarily the underlining factor. I bet you would see a major change with a 3 mb cache and the 6 mb cache becuase of the effect it will have on the amazingly fast reads in the SSD

Usually you have good critiques, but don't be so harsh and shoot someone down without the proper facts.
http://en.wikipedia.org/wiki/CPU_cache That will explain a little better.


TL;DR - Fetching the reads fast enough from the BLAZING fast SSD pretty much requires a "stronger" or larger cashe, AKA, 6 MB makes people happy.
 
six.four said:
In tests performed by Anandtech, the difference between a 6mb cache (2.5ghz) vs a 3mb cache (2.4ghz) in real world performance was small (on average 5%) - and that's inclusive of the 100mhz advantage that the 6mb has over the 3mb cache. The difference is even smaller when comparing a Merom processor with the same clock w/ 4mb cache.
Click to expand...

This is very true, cache size isn't that important for most applications.

On another note, people seem to be giving a lot of credibility to synthetic benchmarks (xbench, geekbench etc.). Synthetic benchmarks are limited by nature. For accurate comparions between Air v1 and v2 you need to do real life benchmarks, not synthetic ones.
 
Philflow said:
This is very true, cache size isn't that important for most applications.

On another note, people seem to be giving a lot of credibility to synthetic benchmarks (xbench, geekbench etc.). Synthetic benchmarks are limited by nature. For accurate comparions between Air v1 and v2 you need to do real life benchmarks, not synthetic ones.
Click to expand...

Read my post. Because of the SSD, larger cache happens to matter (for once).
 
twist2b said:
Read my post. Because of the SSD, larger cache happens to matter (for once).
Click to expand...

I just read your post. I respect fully disagree.

Unless you point me to some objective real world testing (not synthetic benchmarks or subjective impressions) that proves your point I will not be convinced.

six.four has expressed it pretty well in several post in this thread.
 
L2 cache accesses are around 7 nanoseconds, main memory accesses more like 70ns, and even the fastest SSDs are around 70 *microseconds* (70,000ns). If you're hitting disk, the CPU is going to be waiting. Applications should be prefetching relevant data from the disk so it's (at worst) a main memory load not a disk load.
 
OK I ran Geekbench for kicks. Final score was 2474. I am not familiar with it so I have no reference point.

I also ran that OpenGL thing but it didn't seem to give a composite score. Looked like a lot of different things you can play with.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Back
Top