PDA

View Full Version : The Config Dilemma




Terwal
Feb 11, 2008, 04:50 AM
Hi,

I have ordered the 1.8/80 custom config, but while waiting for shipment, I keep on going back and forth regarding the optimal config:

- 80 HDD vs 64 SSD: since I am not ready to pay the price difference at this stage (and since 64 GB is a bit tight for me), pretty comfortable with getting the 80 HDD now and then swapping for a larger SSD at some point in the future when it is available in larger size at a more reasonable price

- 1.6 vs 1.8 Ghz: the BIG outstanding question for me. On one hand, I like the “mental” comfort of having the 1.8 Ghz, the fastest MBA currently available: to be able to handle more/ better apps in the future. On the flip side, when I read the bench comparisons available, the difference in speed seems marginal (and mostly coming from the SSD as opposed to the processor). That said, aside from speed, would there be any other reasons to chose the 1.8 instead of the 1.6 Ghz processor ? :confused:

I have to say that I will use my MBA mostly for: surfing, itunes, watching movies, a bit of photo manipulation (pretty light) and music editing. As well, the MBA will be my main computer.

Thanks for your views, opinions, help, insight… for both the drive and processor parts… :)

w.



Ratty
Feb 11, 2008, 05:38 AM
Go for the 1.8 / 80, you can't change the processor but you should be able to change the drive to SSD in the future.

:)

saltyzoo
Feb 11, 2008, 05:45 AM
1.6 vs 1.8 Ghz: the BIG outstanding question for me. On one hand, I like the “mental” comfort of having the 1.8 Ghz, the fastest MBA currently available: to be able to handle more/ better apps in the future. On the flip side, when I read the bench comparisons available, the difference in speed seems marginal (and mostly coming from the SSD as opposed to the processor). That said, aside from speed, would there be any other reasons to chose the 1.8 instead of the 1.6 Ghz processor ? :confused:

Roughly speaking you'll be 12.5% faster running apps that require full cpu usage, no faster at apps that aren't using a lot of cpu, at the price of 12.5% less battery life and 12.5% more heat.

skyrider007
Feb 11, 2008, 05:47 AM
Go for the 1.8 / 80, you can't change the processor but you should be able to change the drive to SSD in the future.

:)

I agree, good thinking Ratty.

Terwal
Feb 11, 2008, 10:59 AM
Roughly speaking you'll be 12.5% faster running apps that require full cpu usage, no faster at apps that aren't using a lot of cpu, at the price of 12.5% less battery life and 12.5% more heat.

All that make sense... Would you say that 12.5% is "significant" enough to justify the extra $300 (which is more like 17% of the cost ;) ) ?

saltyzoo
Feb 11, 2008, 12:04 PM
All that make sense... Would you say that 12.5% is "significant" enough to justify the extra $300 (which is more like 17% of the cost ;) ) ?

Well, there are two ways to look at it. First, there's no way to increase cpu speed later, so hedge your bets and get it now so you won't be sorry later. But from a pragmatic point of view, I personally don't think it's worth it. Let me give more detail on my opinion, it's probably not the popular consensus.

Theoretically you're getting 12.5% "faster" cpu. However, you're only getting a true 12.5% increase if your cpu is running at 100% the entire time, which I hope is going to be rare. Let's throw some realistic numbers out.

Let's say you end up running at 5% cpu average. That means that the cpu is your bottleneck only 5% of the time, the rest of the time is spent waiting for RAM, HD, network, or user input. So really that 12.5% speed increase is only going to kick in 5% of the time.

Let's take an operation that takes 30 seconds and compare at different cpu usages.

5% average, 1.8Ghz : 30 sec
5% average, 1.6Ghz : 29.81 sec

If however, your typical tasks give you 10% average on the cpu. A whole 4/10 of a second faster:

10%, 1.8Ghz : 30 sec
10%, 1.6Ghz : 29.63 sec

If you're running at 50% it starts to get noticeable at 1.8 seconds faster:

1.8Ghz : 30 sec
1.6Ghz : 28.13 sec

And so on until finally at 100% you'd get the full benefit of the 12.5% increase in speed (3.75 secs faster on a 30 sec operation).

This is why I don't consider small cpu speed jumps important unless what you do normally causes more than 50% average load on your cpu(s). Accessing memory, network, and disk is much more important.

And, now that I've gone through my spiel, hopefully you'll understand why I make the next statement.

I would bet my paycheck that a 1.6Ghz model with SSD is considerably faster at most every real world usage than a 1.8Ghz model with 4200rpm HD. If you're doing graphics rendering or other high intensity cpu operations this statement may no be true, but I'm assuming few people will have that in mind if they are buying an Air.

tom peters
Feb 11, 2008, 12:22 PM
A couple of years ago I ordered a 2.5 GHz mac quad because the faster ( I believe 2.72Ghz or something) wasn't available for another month and I needed the computer for a job. I have long forgotton that I have the fractionally slower computer. I have done all the work I would have on the other computer just the same. Maybe the renders would have been a few minutes shorter here and there, but unless I read a post like yours I never even think about it.

It might be best to save your money.

Tom Peters

Halon X
Feb 11, 2008, 01:18 PM
I was in the same boat and went for the 1.6

It's been plenty fast enough for everything that I've thrown at it and I'm pretty sure I really wouldn't notice too much real world difference between the two processors. I'm coming from a Macbook Pro.

I felt the $300 difference was better spent on Applecare. In the end I have the exact same computer, a teeny bit slower with full protection for 3 years for $50 less than I would have paid for the 1.8.

mrJnC
Feb 11, 2008, 02:01 PM
A point of merit for the standard 1.6/80 configuration vs. custom... if you buy at an Apple store and ever have a warranty problem, they should be able to swap it out for another MBA in stock. Whereas if it's a custom build to order, you'll have to send it out for repairs.

bjdraw
Feb 11, 2008, 02:12 PM
I was in the same boat. I ordered a custom config 1.8/80, but when the 1.6/80 was in stock at the store I bought it and returned my custom config when it arrived.

In the end I just didn't think it was worth the $300 and based on my experiences during my first week with my MBA, I'm confident I made the right decision. I've yet to see the processor max out even when playing 1080p h.264 encoded HD videos. The only time I think I would've appreciated the difference is when I transcode video for my iPhone, but considering I usually do it before I go to bed, I doubt I'd appreciate it either.

netdog
Feb 11, 2008, 02:23 PM
You were right all the way up to here...


I would bet my paycheck that a 1.6Ghz model with SSD is considerably faster at most every real world usage than a 1.8Ghz model with 4200rpm HD. If you're doing graphics rendering or other high intensity cpu operations this statement may no be true, but I'm assuming few people will have that in mind if they are buying an Air.

I played with both stock models for at least an hour at Regent Street, loading apps, loading data, shutting down, restarting, cold starting.

The fact is that the 1.8 SSD wasn't considerably faster, and believe me, I was looking for the smallest excuse to buy the SSD model. Given what I found, if the 1.8 SSD wasn't considerably faster than the 1.8 80, then the 1.6 SSD won't be either. This experiment also supports your contention that there isn't much to be gained between the 1.6 and 1.8 processors.

As much as I was dying to get the SSD, in real world situations, other than with startup, the SSD has little to show for the price as Ars Technica and Walt Mossberg said.

While I would have been the first to defend the SSD model even if the gains still weren't substantial, the fact is that in real world computing, the 1.6 80 is the same as the rest. Heck, given its ability to awaken from sleep almost instantly, the startup times are a non-issue as well given that the SSD's startup times are nothing to get excited about. In fact, shutting down, the SSD took forever.
The smart money will buy a standard 1.6.

If you PM me, I'll give you the address to send that paycheck.

PS: FWIW, I use a MBP 2.33 and a Mac Pro 2.8 Octo with 8GB of RAM at home, so it isn't as if I don't know what a fast computer is like.

Terwal
Feb 11, 2008, 02:23 PM
A point of merit for the standard 1.6/80 configuration vs. custom... if you buy at an Apple store and ever have a warranty problem, they should be able to swap it out for another MBA in stock. Whereas if it's a custom build to order, you'll have to send it out for repairs.

Very relevant indeed... I had not thought about that. Do you know if, buying in one given store, you can return in a different store ? Say abroad, for example ?

Thanks,

w.

Terwal
Feb 11, 2008, 02:26 PM
Given what I found, if the 1.8 SSD wasn't considerably faster than the 1.8 80, then the 1.6 SSD won't be either.

Different question but: did you find the 1.8 Ghz + 80G HDD in store ?

saltyzoo
Feb 11, 2008, 02:32 PM
You were right all the way up to here...



I played with both stock models for at least an hour at Regent Street, loading apps, loading data, shutting down, restarting, cold starting.

The fact is that the 1.8 SSD wasn't considerably faster, and believe me, I was looking for the smallest excuse to buy the SSD model. Given what I found, if the 1.8 SSD wasn't considerably faster than the 1.8 80, then the 1.6 SSD won't be either. This experiment also supports your contention that there isn't much to be gained between the 1.6 and 1.8 processors.

As much as I was dying to get the SSD, in real world situations, other than with startup, the SSD has little to show for the price as Ars Technica and Walt Mossberg said.

While I would have been the first to defend the SSD model even if the gains still weren't substantial, the fact is that in real world computing, the 1.6 80 is the same as the rest. Heck, given its ability to awaken from sleep almost instantly, the startup times are a non-issue as well given that the SSD's startup times are nothing to get excited about. In fact, shutting down, the SSD took forever.
The smart money will buy a standard 1.6.

If you PM me, I'll give you the address to send that paycheck.

PS: FWIW, I use a MBP 2.33 and a Mac Pro 2.8 Octo with 8GB of RAM at home, so it isn't as if I don't know what a fast computer is like.

That's disappointing and indicative of a problem actually. Perhaps a bad driver, or a bottleneck in the interface. But alas, I'm sorry, one person playing with it in the store isn't quite enough evidence for me to sign over the check. ;)

abijnk
Feb 11, 2008, 02:40 PM
Well, there are two ways to look at it. First, there's no way to increase cpu speed later, so hedge your bets and get it now so you won't be sorry later. But from a pragmatic point of view, I personally don't think it's worth it. Let me give more detail on my opinion, it's probably not the popular consensus.

Theoretically you're getting 12.5% "faster" cpu. However, you're only getting a true 12.5% increase if your cpu is running at 100% the entire time, which I hope is going to be rare. Let's throw some realistic numbers out.

Let's say you end up running at 5% cpu average. That means that the cpu is your bottleneck only 5% of the time, the rest of the time is spent waiting for RAM, HD, network, or user input. So really that 12.5% speed increase is only going to kick in 5% of the time.

Let's take an operation that takes 30 seconds and compare at different cpu usages.

5% average, 1.8Ghz : 30 sec
5% average, 1.6Ghz : 29.81 sec

If however, your typical tasks give you 10% average on the cpu. A whole 4/10 of a second faster:

10%, 1.8Ghz : 30 sec
10%, 1.6Ghz : 29.63 sec

If you're running at 50% it starts to get noticeable at 1.8 seconds faster:

1.8Ghz : 30 sec
1.6Ghz : 28.13 sec

And so on until finally at 100% you'd get the full benefit of the 12.5% increase in speed (3.75 secs faster on a 30 sec operation).

This is why I don't consider small cpu speed jumps important unless what you do normally causes more than 50% average load on your cpu(s). Accessing memory, network, and disk is much more important.

And, now that I've gone through my spiel, hopefully you'll understand why I make the next statement.

I would bet my paycheck that a 1.6Ghz model with SSD is considerably faster at most every real world usage than a 1.8Ghz model with 4200rpm HD. If you're doing graphics rendering or other high intensity cpu operations this statement may no be true, but I'm assuming few people will have that in mind if they are buying an Air.

This is a very good analysis of the two processors, and exactly what I considered when picking one out. I personally went for the base model and have found it to be more than adequate for what I do, which is office documents, web, email, and some light to moderate programming for classes.

dahcheet
Feb 11, 2008, 02:45 PM
I decided to upgrade the processor. My rational thought process was that I'm already spending $1800, I might as well spend $300 more for a faster processor. Of course, in my irrational frenzy to pre-order it may of clouded my rational thought. It's 17% more for 15% potential computer power. Personal preference really. If you decide not to upgrade you could always spend the money on the Super-Drive and accessories.

netdog
Feb 11, 2008, 02:46 PM
That's disappointing and indicative of a problem actually. Perhaps a bad driver, or a bottleneck in the interface. But alas, I'm sorry, one person playing with it in the store isn't quite enough evidence for me to sign over the check. ;)

I think that SSDs that don't cost an arm and a leg just aren't there yet. I expect that they will be soon. I doubt that is was the driver, and PATA shouldn't be the issue. PATA can do more than this drive did to increase the subjective experience while doing real world computing.

The write times on that shutdown were just atrocious.

netdog
Feb 11, 2008, 02:48 PM
I decided to upgrade the processor. My rational thought process was that I'm already spending $1800, I might as well spend $300 more for a faster processor. Of course, in my irrational frenzy to pre-order it may of clouded my rational thought. It's 17% more for 15% potential computer power. Personal preference really. If you decide not to upgrade you could always spend the money on the Super-Drive and accessories.

It's 12.5% faster .001% of the time. Not a criticism. I wanted the SSD after all. No doubt you will love using your MBA, so what else matters?

Terwal
Feb 11, 2008, 03:56 PM
Thanks to all for your input. Very useful...

And actually, while going through those posts, just checked the status order of my MBA ! Just went on "prepared for shipment" (1 week ahead of schedule) ! :D

So I guess I will it will be a 1.8/80 for me (with future SSD upgrade) :)

saltyzoo
Feb 12, 2008, 09:08 AM
I think that SSDs that don't cost an arm and a leg just aren't there yet. I expect that they will be soon. I doubt that is was the driver, and PATA shouldn't be the issue. PATA can do more than this drive did to increase the subjective experience while doing real world computing.

The write times on that shutdown were just atrocious.

I've been thinking about this and I'm wondering if maybe Apple didn't put a huge cache on the disk controller to try to improve the performance of a 4200rpm drive. If the cache were big enough it might hide the difference between a SSD and HD except on large reads and writes (boot up and shut down for example)