Thanks for this extended test. The i7 performs as suspected and quite similar to the 2012 i5 vs. i7.
Have you had a chance testing idle performance of both?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
11" i7 vs i5 Test(s)
- Thread starter Jazwire
- Start date
- Sort by reaction score
Thanks for this extended test. The i7 performs as suspected and quite similar to the 2012 i5 vs. i7.
Have you had a chance testing idle performance of both?
>leslie11 ...
>Playing 3 movies, 1 on VLC, 2 on QT
>Handbrake encoding the same movie
Kudos for your careful test, but IMO your battery life and average temperature results are really distorted by including an unfinished task-oriented job like encoding a movie. A more realistic test would be to wait until the i5 finished encoding the movie (during the last period of which Handbrake would be inactive on the i7), then compare battery percentages. Also the temperatures should be averaged over the full time period of the test (including the final inactive Handbrake time on the i7). Otherwise you are comparing different amounts of total work on the two devices. Or simply omit Handbrake encoding from this particular battery test.
>Starting at the same time, the Handbrake encode initially
>showed a 10% difference in progress (i.e I5 at 5%, I7 at 15%).
>But when the I5 was at 50% done, the I7 showed 55%,
>ie the gap got smaller.
That surprises me, because the i7 should be about 30% faster than the i5 for CPU-bound tasks. My guess is that the 3 movies playing in the background drove up the temperature, so the i7 couldn't rev up to full Handbrake capacity without exceeding the 15W TDP cap. I'd be curious to see Handbrake performance compared between the i5 and i7 with the rest of the system mostly idle.
Thanks again for reporting your test... all these data points are interesting.
>Playing 3 movies, 1 on VLC, 2 on QT
>Handbrake encoding the same movie
Kudos for your careful test, but IMO your battery life and average temperature results are really distorted by including an unfinished task-oriented job like encoding a movie. A more realistic test would be to wait until the i5 finished encoding the movie (during the last period of which Handbrake would be inactive on the i7), then compare battery percentages. Also the temperatures should be averaged over the full time period of the test (including the final inactive Handbrake time on the i7). Otherwise you are comparing different amounts of total work on the two devices. Or simply omit Handbrake encoding from this particular battery test.
>Starting at the same time, the Handbrake encode initially
>showed a 10% difference in progress (i.e I5 at 5%, I7 at 15%).
>But when the I5 was at 50% done, the I7 showed 55%,
>ie the gap got smaller.
That surprises me, because the i7 should be about 30% faster than the i5 for CPU-bound tasks. My guess is that the 3 movies playing in the background drove up the temperature, so the i7 couldn't rev up to full Handbrake capacity without exceeding the 15W TDP cap. I'd be curious to see Handbrake performance compared between the i5 and i7 with the rest of the system mostly idle.
Thanks again for reporting your test... all these data points are interesting.
I'll be using the I5 for the next few days, then back to the I7. Currently your guess is as good as mine!
Glad to help, it might save some of you the time and effort. I'll test idle perf if I have the time, maybe someone else can chip in here?
Yep I agree, the batteries weren't identical to begin with so it's already a biased test. My main gripe when using my old 2010 MBA was the heat, battery life was seldom an issue for me so forgive the less than perfect test scenarios, I was trying to push the machines beyond the limit to see what the worst case scenario was for heat generated.
The Handbrake test on idle machines are easy enough, I just need to find the time to run it. Honestly, if it's only 10min between encodes, I'll save the $$$ and go I5. I hardly encode stuff so it's no biggie to me.
Posted "League of Legends" test. (On 1st page.)
I will do a general usage test tomorrow.
I plan on the following:
I am going to pick 10-15 websites to browse (all tabs open, same sites on each machine.)
Do some Spotify streaming, (same amount of time)
Watch 5-10 Youtube Videos (Same Videos)
While having Word, Excel,Mail, Evernote,Text Edit, Skype,Acrobat Reader all open in the background.
Using each program for 5 minutes each.
For primarily a battery/heat test.
And if i have time i'll do a handbrake or video encode test as well.
Good night!
I will do a general usage test tomorrow.
I plan on the following:
I am going to pick 10-15 websites to browse (all tabs open, same sites on each machine.)
Do some Spotify streaming, (same amount of time)
Watch 5-10 Youtube Videos (Same Videos)
While having Word, Excel,Mail, Evernote,Text Edit, Skype,Acrobat Reader all open in the background.
Using each program for 5 minutes each.
For primarily a battery/heat test.
And if i have time i'll do a handbrake or video encode test as well.
Good night!
Last edited:
If both the i5 and i7 both ran at a base clock of 1.3ghz then yes you'd be right. But both CPUs idle at the same clock speed but under use, the i5 has a base of 1.3ghz and the i7 is a base of 1.7ghz... When they are running their base speeds, the i7 draws more power, period. You think the i7 pulls more at idle? I doubt it, they should both run the same idle MHZ and pull the same amount of power. Its when the cpus run their base clocks when the power changes since they are different.
It may be true that the 1.3ghz base close (not talking about the CPU's ability to downclock for power) is not high enough for Silverlight to run at 100% normalization and the i7 could very well be more efficient in certain tests when comparing power draw but I doubt Silverlight requires either of these CPUs to run into turboboost speed to keep up.
I'm not defending my i5 in anyway, i'd rather have the i7 because for CPU intensive games it seems to fair much better as pointed out in your LoL test..
Whew...all that heat!
Fortunately, I doubt if I will be running more than one or two "intensive" tasks at a time. It's been many years since I have been able to watch and follow three movies at a time while doing other stuff, too.
Interesting results, but I'd be more interested in real world testing and analysis, such as encoding a movie in the background while perhaps working on writing and editing a manuscript and occasionally checking email.
Fortunately, I doubt if I will be running more than one or two "intensive" tasks at a time. It's been many years since I have been able to watch and follow three movies at a time while doing other stuff, too.
Interesting results, but I'd be more interested in real world testing and analysis, such as encoding a movie in the background while perhaps working on writing and editing a manuscript and occasionally checking email.
Ya, I am not sure what is going on for sure, it was just a hypothesis.
I have now ran 2 tests that have required medium CPU power, and in those tests the i7 both times has ran cooler.
Now it is totally possible there is something wrong with my wife's i5, maybe it has 3 inches of thermal paste on the CPU. I don't know for sure.
But also for the people saying their i7's run hot, I would have to assume the same for them.
My i7 under low & medium CPU tasks does not get hot, nor have I had the fans kick in.
I don't think I have proven that the i5 is inferior (though my i5 may be inferior).
But that people claiming the i7 gets burning hot (under low/medium usage), is completely not always the case.
In a way I wish my i7 was this hot useable beast, I send it back get the i5 version and have an extra $150 in my pocket. My i7 stays cool to the touch when I am just doing normal usage. And even if i was to do some gaming LoL, it really would not be uncomfortable to have on my lap.
Honestly at this point I am keeping my i7, however , I do want to determine if I need to return my wife's i5 (for another i5).
To do this test a great service you should indeed return the wife's i5 and try another and re-test... If it acts the same way then we at least know there's a tipping point where the i7 models begin to be more efficient and that would make for great conversation indeed.
New in forum and waiting for my Air, hope having it tomorrow...
First of all, great tests, thank you!
As a suggestion, if I remember correctly, Coconut Battery tells the wattage used by the battery in real time, maybe could help to confirm the difference in consumption in idle, mid-CPU and full CPU usage.
The link:
http://www.coconut-flavour.com/coconutbattery/
First of all, great tests, thank you!
As a suggestion, if I remember correctly, Coconut Battery tells the wattage used by the battery in real time, maybe could help to confirm the difference in consumption in idle, mid-CPU and full CPU usage.
The link:
http://www.coconut-flavour.com/coconutbattery/
Just a minor update, I've just been doing light word processing for 2hrs, split equally between both machines.
There's really nothing in it, both performed the same and got slightly hot after a while. I was using my palms as a thermometer so don't put too much thought into this.
There's really nothing in it, both performed the same and got slightly hot after a while. I was using my palms as a thermometer so don't put too much thought into this.
That seems like a cool application. According to it, right now my computer draws around 3.5 - 4.5 watts from the battery. Thats truly impressive considering that I have my intelliJ, browser windows, spottily etc running.
No wonder this thing doesnt get hot. Machine is 13/i5/8/256
Wow, I have devices that consumes more on stand-by!
Is the screen on or do you have an external monitor?
iThink that the i7 has more transistors and can do therefore more in less time.
----------
Many.
Wow, I have devices that consumes more on stand-by!
Is the screen on or do you have an external monitor?
Screen was on (50% brightness), even the keyboard backlighting was on (lowest brightness setting)
@Mr. Retrofire, the i7 doesnt have more transistors.
http://ark.intel.com/compare/75114,75028
Its pretty much the same chip, operating at a higher frequency.
33% more L2 cache. And the i5 has some features disabled (not ones that we should care about anyways).
From the article:
Haswell ULT GT3 2C 22nm 2 GT3 1.3B 181mm2
Both the i5 and i7 found in the 2013 macbook airs are GT3 2C
Meaning 1.3B transistors in the i7 and i5
I just ran a simple Handbrake encode on both machines, side by side with nothing running in the background except Coconut battery.
Here are a few observations:
- a 10min encode on the I5 took 8min on the I7
- the I7 fan spun up to 6.5k rpm within 2min, the I5 kept at 1.2k rpm till the 5min mark then slowly increased to 1.6k rpm by the end of the test
- Coconut batt consistently showed the draw for the I7 at 22.5w and the I5 at 14.5w
Guess the I7 is going back to Apple then.
Here are a few observations:
- a 10min encode on the I5 took 8min on the I7
- the I7 fan spun up to 6.5k rpm within 2min, the I5 kept at 1.2k rpm till the 5min mark then slowly increased to 1.6k rpm by the end of the test
- Coconut batt consistently showed the draw for the I7 at 22.5w and the I5 at 14.5w
Guess the I7 is going back to Apple then.
My experience has been similar. Keeping the i5/8/512. Main reason - better battery life.
Still wish anandtech would do a formal i5 / i7 comparison, maybe it's coming?
Still wish anandtech would do a formal i5 / i7 comparison, maybe it's coming?
Sorry for not responding earlier, I haven't had full batt-depleting runs on both machines yet.
But I charged both MBAs to capacity and ran the latest Handbrake test, the higher draw on the I7 + fans at full blast resulted in the additional drain on a 8min encode.
It reasons that the I5 will last longer than the I7 at full speed, not taking into account the amount of work each computer produces when the batt is finally depleted.
----------
My main reason - less heat!
Yep I can only hope that Anandtech does one, those guys are certainly more accurate than most of us.
That seems fishy to me. Both cpu are designed to draw a max of 15 W (excluding the gpu).
11" i7 vs i5 Test(s)
Take into account memory, screen,etc. Coconut doesn't measure CPU power consumption, but instant battery usage.
For me, it seems that i7 uses the full 15Watt, but i5 don't, at least for the most part of the time.
Edit: thanks for the test leslie, I forgot to mention!
Take into account memory, screen,etc. Coconut doesn't measure CPU power consumption, but instant battery usage.
For me, it seems that i7 uses the full 15Watt, but i5 don't, at least for the most part of the time.
Edit: thanks for the test leslie, I forgot to mention!
Yup it's coming: from his twitter:
@anandshimpi: Finally back home, running some Haswell ULT experiments, working on MBA13 i5 vs i7 data...
When you said, "For me, it seems that i7 uses the full 15Watt, but i5 don't,at least for the most part of the time"
Or you saying your i7 uses the full 15w, almost all the time? like even web browsing / light work?
If so your i7 is messed up, right now i have about 7 browser tabs open, my email,Skype, which I would consider very light and Coconut reports I am using 3.3 watts.
Also opened up, Youtube, made it spike to 6w, and then settle in at 5.5 while playing.
If you are getting 15w on light stuff, there is something wrong with your i7.