Yeah I wouldn't think it would make a huge difference though. A 5 hour video export would take about 12 hours on my Mac Pro but 6 hours on my 2016 Macbook Pro.
Does that mean I am fine going with an i5, or is the i7 still recommended?
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.
The new iMac is a lot noisier!
- Thread starter Darajavahus
- Start date
- Sort by reaction score
Yeah I wouldn't think it would make a huge difference though. A 5 hour video export would take about 12 hours on my Mac Pro but 6 hours on my 2016 Macbook Pro.
Does that mean I am fine going with an i5, or is the i7 still recommended?
So you're saying that the perceived speed of your computer doesn't affect when someone feels it's time to upgrade? Interesting.
You know, it's usually me that makes snarky comments, although mine are typically better informed.
The difference with i7s is that hyperthreading is enabled. Hyperthreading generally won't make your machine feel snappier. It's a scheduling aid, and it's not a free speedup. It helps when your cpu has a high level of scheduling conflicts. If it's waiting on some other piece of hardware, then this won't do anything.
Depending on the hardware generation, it could be a $200-300 difference. It's typically 2-5 years before someone's machine "feels slow". Your results may vary, but it's often something other than the cpu. If you're buying one of these machines yet aren't doing anything terribly taxing, it won't make a difference. Apple might pull support. You might end up with hardware failure. You might decide you want a gaming machine. The cpu is among the least likely upgrade motivators here. That notion is held over from earlier decades. If everyone was cpu bottlenecked, laptops would be a lot less popular with the advent of the smartphone.
Update... I returned the i7 imac today and have an i5 3.4 with 1TB SSD coming in the next 2-7 days. As an audio guy I really wanted the i7 but in the end - I just did not like the temperature performance. Rarely below 70degC driving my whole system. The i7 also has a short term heat spiking "feature" that many have commented on. I sow this behavior everytime I would launch a new plugin or even simply opening a plugin GUI - the temperature spikes were 10 to 20degC. Just very odd for the EE in me. 100degC video renders were also just not what I want to see or hear.
The i5 will be a compromise for live plugins and VIs but I am going to spend a couple of weeks with the new machine to see about just accepting workflow modifications if necessary to make things go smooth. Been here before when my main DAW made a new plugin format that for years was badly supported.
I thought hard about just staying with the 2013 Mac Pro but I am addicted to the 5K screen. The screen is what you look at almost every second of using a computer! I can't find anything that compares. The $1299 LG in the apple store is hardly better than my $600 Dell at the resolutions I like to use (2nd lowest and middle one on the 5K retina choices)... $1500 with tax for a monitor just doesn't make sense to me at this time for a 2013 computer.
The i5 will be a compromise for live plugins and VIs but I am going to spend a couple of weeks with the new machine to see about just accepting workflow modifications if necessary to make things go smooth. Been here before when my main DAW made a new plugin format that for years was badly supported.
I thought hard about just staying with the 2013 Mac Pro but I am addicted to the 5K screen. The screen is what you look at almost every second of using a computer! I can't find anything that compares. The $1299 LG in the apple store is hardly better than my $600 Dell at the resolutions I like to use (2nd lowest and middle one on the 5K retina choices)... $1500 with tax for a monitor just doesn't make sense to me at this time for a 2013 computer.
I think the spiking is due to the thermal compound used between the core and the IHS (it's just really bad), when I delidded and replaced it the spiking stopped on my 7700K (and I even had the fans on a 2 second delay).
I built a PC a couple of years ago with an Intel 4790K, and still use it. At some point, I'm going to upgrade the graphics card. It's been a great computer. The stock 4790K just approaches 70 deg. C when running Prime95 with a Noctua air cooler on it. That CPU has finally been designated "end of life" by Intel after being around for a number of years. Too bad Intel seems to be cutting corners now. I'll see how the 7600K fares when I take delivery of my iMac.
Why would you be surprised a 2016 laptop can beat a 2010 Mac Pro?
Hyperthreading most definitely makes a computer feel faster. That’s why most of the dual-core machines have it now.
The 5K screen is definitely great. But why not order the 7600 non-K instead of the 7500?
Not true, you will have to look very hard and long to find something being able to utilize even two cores for UI-related tasks (which are becoming more and more GPU dependent), the only time you will really see the benefits is on work loads that can be parallelized.
Adding it on dual core CPUs makes it easier to fully utilize the cores since threading is very hard (from a software perspective) and it's better to have the support of the hardware (and the OS scheduler) than to try to optimize it yourself.
Not read all 18 pages of this thread but wanted to add that without knowing the ambient temperature it's difficult to form a valid opinion.
My I7's fans kicked in much more two weeks ago when we were having uncharacteristically warm weather in the UK than they normally do.
Maybe the guy in the video could invest $100 on a portable air-con unit, I just bought one and it's awesome!
My I7's fans kicked in much more two weeks ago when we were having uncharacteristically warm weather in the UK than they normally do.
Maybe the guy in the video could invest $100 on a portable air-con unit, I just bought one and it's awesome!
It probably will cause a slight user-noticable speedup, independent of the app architecture. HT lets the OS assign two threads to the same core for any given timeslice so they pipeline in parallel. Those two threads can be two separate single-threaded app processes.
It might not make the UI snappier because UI can only do one thing at a time anyway, but any time spent on computation (including to inform the UI) would be accelerated. That said, IIRC the typical speedup is in the 10-15% range, nothing like doubling.
My ambient temps were never higher than 23C. However, even with short encodes my i7 fans would kick in. With my previous i7 870 (2010), the fans would kick in with encodes, but not as quickly and not as loudly initially.
The one thing that surprised me is that if I scrolled through my entire Photos library (several thousand photos) a couple of times, the i7 fan would come on moderate. Granted, I wouldn't do this often, but it surprised me nonetheless.
As propower stated, it was the lightning fast power up that was a bit disconcerting.
But again, I will reiterate for 98% of my usage, the i7 was at 1200 rpm, which for me is effectively silent. I just realized that the times I'm going to walk away from my desk to encode video, I don't mind if it takes 20 minutes or 30 minutes. But when I'm at my desk, even 2 minutes of fan noise is going to annoy me. I'd rather the encode take 3 minutes with minimal fan noise.
My 2012 i7 has gotten rather noticeably fan-happy in Sierra, especially if I touch photos. photoanalysisd loves to peg a core in the background, and the fan revs up to max pretty quickly. I'll be curious to see how the 2017 compares when it arrives.
Exactly. It makes multi-tasking better.
Well, it makes a more noticeable difference on lower clocked dual-core systems. On faster quad-core systems, the UI speedup is much less noticeable, since the UI is already fast enough.
That would only be relevant if the quad core was at 400% load, otherwise OS scheduling is far more advanced than to let cores idle if there's processing to be done, but then we're not talking about "feeling snappy" (which is a non-descriptive concept in the first place) but rather data processing, which I already mentioned. Or maybe you have some examples of UI rendering that consumes four threads in perfect parallelization?
I'm not arguing the benefits of HT on certain work loads, making your computer feel "snappy" is however not one of them.
It's not quite that simple, re: OS is better/400%. A given core can only do one thing at once. The OS schedules by time slicing between threads assigned to that core to rapidly context-switch what's running when--100% doesn't mean the core is "full" so much as all of its possible slices are allocated. That's the OS level.
But within any given timeslice you can get tiny periods of idle core due to the process stalling on resource access (memory bus, etc) which would normally be wasted time. HT allows the OS to say "do this or that" by scheduling two processes per physical core. That lets the core internally switch between the two choices based on which can proceed. The OS wouldn't otherwise be able to know that or do much about it within a single timeslice.
It's the recovery of those bits that result in the speedup and you'll potentially see it on any physical core with two or more threads or processes assigned as long as they're spread between the two logical cores.
You're right re: snappy, both in that it's a bs word and in that UI is usually bottlenecked on stuff other than cpu so won't generally see speedup itself. But if you hit a button, then stuff happens, then the UI changes, stuff happening faster means it'll potentially change quicker even if that delay is otherwise small. That can make a UI seem more responsive.
But within any given timeslice you can get tiny periods of idle core due to the process stalling on resource access (memory bus, etc) which would normally be wasted time. HT allows the OS to say "do this or that" by scheduling two processes per physical core. That lets the core internally switch between the two choices based on which can proceed. The OS wouldn't otherwise be able to know that or do much about it within a single timeslice.
It's the recovery of those bits that result in the speedup and you'll potentially see it on any physical core with two or more threads or processes assigned as long as they're spread between the two logical cores.
You're right re: snappy, both in that it's a bs word and in that UI is usually bottlenecked on stuff other than cpu so won't generally see speedup itself. But if you hit a button, then stuff happens, then the UI changes, stuff happening faster means it'll potentially change quicker even if that delay is otherwise small. That can make a UI seem more responsive.
Even with HT a core can just do one thing at once, it's just a much more efficient way of context switching. Unless you have a bug in the VM/Runtime, a core at 100% is doing all the work it can do (as long as you're looking at the right values and not things like waiting for I/O, like memory, disk, network).
And now we're talking about things that are measured in ns or faster, the human perception of instant is around 10ms, tens of thousands of times slower...
When you press that button, if you're crunching data HT can help (if it's a parallelized work load that actually sees the benefit of HT), if you're moving some text around in the window, scrolling a webpage etc, it won't.
I just don't think confusing people with their computer feeling snappier with a i7 is correct, nobody will ever be able to tell the difference browsing the web on an i5 vs an i7.
Well I just got my iMac last week and I can't contribute to the very technical details of fan speed and noise and all that because this is my first Mac computer ever, but I can say that I'm quite impressed. I have the top tier 27" with i7 and ssd. I bootcamped windows last week and just installed battlefield 1 and titanfall 2 trials to test the graphics power and fan. Needless to say it's incredible compared to my 2012 windows laptop. I ran the games off an HDD and there was no lag in sight. The fan DID kick on, not sure of the rpm but it was noticeable. Nowhere near as loud as a white nose machine, and I could easily hear games and music over it with 3 clicks of volume. If the machine gets to hot I assume it will shut itself down, but with my AC on in the house the fan barely breaks a sweat on max settings. Not sure if these games are a good stress test but it's a real world example, and it seems to me that this fan noise issue is blown out of proportion
I really just looked at this option. I can see it might bring something between 5 and 10% speed increase (certainly helping in the work I do). I was originally thinking that a less than 3% base speed was just not significant for me. The 575 vs 570 same story. The 7600 does Turbo 300MHz higher but in my case something less since I always have processes going on all 4 four cores. The non HT will be the squeeze for me - not a sheer CPU speed thing. I also have never tested this "faster" chip to see what the thermal impact will be. I would think this will be minimal but I remain concerned about the iMac thermal design so chose the part I have been testing versus a new character.
PS - I agree that the thermal compound ay be the issue with the 7700K which is certainly a major mistake if true. Regardless - in the iMac 2017 world - no way I attempt to fix that!
Re HT: Multitrack Audio recording and mixing is a real time application. Resources (cores) rule what the most that can happen in real time is. For us Audio folk HT allows for way more plugins and Virtual Instruments to be used in a session. Not about perceived speed or snappiness - its a go - no go situation.
Agreed. I have the same computer but with 2TB FD. If I run four instances of "yes > /dev/null" the four cores go to 100% pretty quickly and the fan goes to full revs (2700 RPM) shortly after. Temp maxes out at 53C (room temp is around 20C - winter here). Shut down the 'yes' instances and things quickly go back to 1200RPM and around 37C. Not a big deal given what it is doing. I am very happy with the noise profile.
I chose the i7 because I had a 2012 i5 at home and a 2011 i7 at work. The i7 was always much more 'snappy'.
Isn't the 2012 i5 a dual-core and the 2011 i7 a quad? If so, no wonder the i7 is more snappy.
Has anyone measured the dB of the i7 fan at max load from a couple of feet in front of the screen? I'm from the bad old days of 50-60dB at nominal CPU and GPU fans and am curious what we're really talking about here.
I have a sound meter so I'll measure it myself when mine comes in and compare to the 2012 i7, but I'm awfully curious.
I have a sound meter so I'll measure it myself when mine comes in and compare to the 2012 i7, but I'm awfully curious.
OK, I think you drop 6dB per radius doubling so that's ~39dBA at "ergonomic" distance. Most scales compare that to library ambient, or about half as loud (perceived) as average house ambient. How annoying it is will depend on pitch, but assuming it's similar pitch to the 2012 I can live with that as a worst case.