Question abut clock speed options on new 2013 Air?
So I know the battery life is amazing, but let's just say on a particular day you just wish to use the Air as a glorified typewriter; writing your future Oscar winning script on the train.
Let's say for this use case the 1.3Ghz model gets the same 15 and a half hours they got on the recent PCMag.com review: would the 1.7Ghz model get any less battery life at all in this scenario?
Source:
http://www.pcmag.com/article2/0,2817,2420468,00.asp
Let's say for this use case the 1.3Ghz model gets the same 15 and a half hours they got on the recent PCMag.com review: would the 1.7Ghz model get any less battery life at all in this scenario?
Source:
http://www.pcmag.com/article2/0,2817,2420468,00.asp
Comments
http://www.engadget.com/2013/06/13/macbook-air-review/
They assumed the i7 would use more power but both CPUs have the same 15W TDP so under load, they both stop at the same maximum. The i7 has better performance per watt than the i5. A few people seem to have mixed experiences:
https://discussions.apple.com/thread/5113765?start=0&tstart=0
https://discussions.apple.com/thread/5116094?start=0&tstart=0
https://discussions.apple.com/thread/5101953?start=0&tstart=0
The 13" has a 54-watt-hour battery so to get 15 hours of battery life, the whole machine has to draw an average of 3.5 Watts for that whole time. The best way to achieve that would be running the display at low brightness and of course running Mavericks when it's ready as it will throttle background web page processes. You'd turn off bus-powered drives etc too.
It will also depends on the browser you use as they have different rendering methods. One with hardware accelerated compositing will use the GPU instead of the CPU to render the page. The i7 has a 10% faster GPU.
This site does quite a good job comparing the specs:
http://cpuboss.com/cpus/Intel-Core-i7-4650U-vs-Intel-Core-i5-4250U
The i7 doesn't cost much more at $150 but you probably won't notice the 10% performance difference (both CPU and GPU). The i7 has some extensions for improved multi-threading that the i5 doesn't but would require programmers to use:
http://software.intel.com/en-us/blogs/2012/02/07/transactional-synchronization-in-haswell
There's a benchmark test here:
http://www.sisoftware.co.uk/?d=qa&f=ben_mem_hle
Real world performance difference would be impossible to say for sure but for basic laptop use negligible. I'd say go with 8GB of memory, the i5 and whatever SSD size is big enough.
Quote:
Originally Posted by Marvin
I'd say go with 8GB of memory, the i5 and whatever SSD size is big enough.
That's a great response btw, but why are you recommending 8GB of memory?
I would have thought 1.7 GHz over 1.3 GHz would have importance over adding memory, considering how well OS X handles memory?
And what sort of difference might I noticed for the i7 over the i5 on stuff like 1. converting (and batch converting) to MP4 using Handbrake, converting to MP4 using VisualHub 1.34, and exporting an iMovie project to MP4?
P.S. The 128 GB SSD is big enough for me. In fact, knowing I'll be stepping down in HD size has really encouraged me to clean up the files, photos and videos on my whole Mac. It's a cleanest, most organised my Mac has been in 8 years.
Your OS kernel is normally going to be using at least 1GB. Apple says the HD5000 GPU reserves 1GB now:
http://support.apple.com/kb/HT3246?viewlocale=en_US&locale=en_US#5000
Safari can easily use 1GB and a word processor the same + miscellaneous apps. I'd say it'll be pretty easy to start paging the drive. With an SSD, it's not such a big problem for performance as the PCIe is so fast but having the SSD active will use more power. Of course, having more RAM chips will use more power too so it's hard to say what effect it has on battery life.
It's only $100, will increase resale value, will lower wear on the SSD. I think it's worth it.
Ah, makes sense. I think I will get it.
Though I don't know if it's a good investment from a resale POV. It's €100 extra for the additional 4 GB of RAM, I doubt if I sold the machine in 2 years time (which is my intention with the Air likely in its second Retina year by then) that it would add €100 onto the resale value versus a 2 year old 1.3 GHz i5 Air with 4 GB RAM. Counterintuitively, I'd bet it may not change the resale at all: so you'd lose that €100. But you'd have benefited for 2 years from it; so not a complete loss by any means.
I would be interested in your assessment of the potential difference in battery life of a new Core i5 Air with 128 GB SSD, plus 8 GB RAM, versus the same machine with 4 GB RAM.
And would love if you could weigh in on this too:
http://thenextweb.com/apple/2013/06/30/we-asked-8-new-macbook-air-owners-how-the-extended-battery-has-impacted-their-life-heres-what-they-said/6/
Interesting!
If you were selling it next to the same model with 4GB, I'd say you'd be able to price it higher, though you might choose not to. On older-style laptops with replaceable RAM, someone could buy the 4GB and if they found they needed more RAM (even up to 16GB), they'd decide later but with soldered RAM, you don't get the option. For many upgrades, the extra cost would get absorbed somewhere and the only benefit might be that if your one was priced the same as a 4GB model, someone would take your one as it was better value but you lose the money on it.
When RAM or an SSD isn't active, it'll only use idle power. SSD ~300mW, RAM ~100mW. Assuming you don't use over 4GB of memory, 8GB would add 4GB of idle memory extra. The actual power consumption would depend on the voltages and how many chips were used.
There's a review here with the 8GB and they got over 13 hours playing video, 7-8 hours with average usage:
http://www.gizmodo.co.uk/2013/06/13-inch-macbook-air-2013-review-well-hello-battery-life/
The following review with the 4GB got around the same in daily usage:
http://mashable.com/2013/06/19/macbook-air-2013-review/
To hit something like a 15 hour battery life, the power usage has to be around 3.5W, which means doing very little with the laptop at all - that's probably iPad level of power consumption. Someone here tests the 11" and gets just under 7 hours:
http://www.tested.com/tech/mac-os/456567-testing-apple-macbook-air-11-2013/
Gaming in Starcraft 2 lasted just under 2 hours, which makes sense as the 11" has a 38Wh battery so would draw about 19W - most of which being the ~15W for the CPU/GPU. I just noticed Apple has different watt rating on the US site vs European sites:
http://www.apple.com/macbook-air/specs.html
http://www.apple.com/ie/macbook-air/specs.html
An imperial/metric conversion isn't required here Apple. Maybe they've been sued too much in Europe so they are taking it back in Watt-hours.
The i7 is 10% faster so 1.1x. Divide the time by the same ratio. So a 30 minute encode on the i5 would take 30/1.1 = 27 minutes on the i7.
Quote:
Originally Posted by Marvin
I just noticed Apple has different watt rating on the US site vs European sites:
http://www.apple.com/macbook-air/specs.html
http://www.apple.com/ie/macbook-air/specs.html
Pure guess, but I'd bet money this is purely a mistake on the Irish Apple Air web page, which has them using the size of the 2012 Air batteries by mistake. Considering they would have had to type the differences on the page by hand. Just human error.
Quote:
Originally Posted by Marvin
The i7 is 10% faster so 1.1x. Divide the time by the same ratio. So a 30 minute encode on the i5 would take 30/1.1 = 27 minutes on the i7.
I find it strange that it works out at a 10% difference. Wouldn't 1.7 be 30.77% more than 1.3 ?
Interestingly, someone on MR said the benchmarks seem to put the 1.7 at 16% faster.
I don't get how all this math doesn't seem to work out.
UPDATE:
From the MacWorld review tests:
Using our system performance testing suite, Speedmark 8, we found the CTO MacBook Air to be 23 percent faster overall than the $1099 stock 13-inch 128GB flash storage model. The CTO Air was faster in every test, with 24 percent faster times in our Handbrake, Aperture and file decompression tests. Even though both laptops used the same integrated Intel HD Graphics 5000, the CTO Air had 8 percent higher frame rates in Portal 2, and 14 percent faster performance in Cinebench’s OpenGL test.
Followed by:
"Check back soon for MacBook Air battery life results."
Grr.. because they are testing both models.
GREAT POST:
http://forums.macrumors.com/showpost.php?p=17510143&postcount=159
They have variable clock speeds, which complicates tests because after a while, the processor gets hot and so it clocks down. The 1.3 and 1.7 speeds are just the base clock rates. They dynamically overclock to 2.6 and 3.3. Each may only reach the maximum rating when 1 core is active but it's really whatever speed still allows it to stay within a certain temperature and power limit. It has to balance the graphics load too. Based on the maximum clocks, it would be closer to 27% difference.
Geekbench isn't an ideal benchmark to use for CPU performance because it measures memory performance too. Cinebench is mostly a CPU test. The MW Cinebench test shows around 15% faster:
http://www.macworld.com/article/2042347/lab-tested-ultimate-macbook-air-2013-holds-its-own-against-the-macbook-pro.html
Mathematica is 17%. Handbrake seems to do better for some reason at 30% but they use such short tests (~2 minute encode). The forum post at least was longer and shows 20% for Handbrake. It's a bit odd that both chips have the same power rating but the i7 was drawing more under load and running the fan faster. Perhaps Intel throttles the i5 more.
If you get 15-20% better performance for $150, that's not too bad value. I think that difference would be hard to notice in everyday usage. That 44 minute saving looks significant but that's on a 4.3 hour encoding. A 30 minute encoding would save 5 minutes. I'd still go with the i5 just to have the fan run quieter under load and save money.
Yeah I decided to go with the i5 last night because of the quieter fan and cooler setup (if the i7 fan is louder it means it needs to work harder to match the chip temperature of the i5: so it's technically hotter. And an additional 15 to 30+ minutes of extra battery life is a nice bonus. MacWorld found the machine to be 23% faster overall, but that 23% will only been seen in specific circumstances. I won't be going along on the train converting a video in the background. In reality I'd leave those tasks to when I have the machine is on charge and not in active use. So all round the savings of the €150 in my case makes more sense.
I've decided to get the i5 with 8 GB and 128 GB.
Phew!