2018 i5 quad core MacBook Pro 13” throttling?

[Mockletoy]

But I don’t get your point. It is reaching its full potential (arguably more). It’s listed at 2.3, yet it hovers around 2.5...

If it's thermally throttling it's not reaching its full potential. I am personally not comfortable running a device that hot. It just bothers me, running a CPU up against its thermal limits all the time. It annoys me as a matter of principle. That graph shows a CPU struggling to do its work without melting into a puddle of slag. I'm not okay with that. Others might be.
[doublepost=1532052945][/doublepost]

What does it matter if the line is flat or fluctuating as long as the average speed is above the base clock speed?

I have a gaming laptop with a 7700HQ in it. That CPU has a 2.8Ghz base clock with a 3.8Ghz maximum single-core turbo. With all 4 cores engaged, it has a maximum turbo frequency of 3.4Ghz. If I start a CPU load test and max out 4 cores / 8 threads, the CPU will boost to 3.4Ghz and hold that speed indefinitely. That is the maximum possible speed this CPU is capable of with all 4 cores under load. During gaming benchmarks that don't hit the CPU cores as hard, it'll boost up to 3.5Ghz or even higher, very briefly. But it can run at a turbo frequency of 3.4Ghz all day and night if I want it to. In any case, it runs those speeds with average temps in the mid 60's, spiking to 70 or so every now and then, with the fans being managed automatically.

My 2016 13" MacBook Pro has a 2.9Ghz i5. Under full CPU load (2 cores / 4 threads) it will boost to a constant and steady 3.1Ghz, which it can maintain indefinitely. That CPU is also working to its maximum potential. It's doing all it can do.

A CPU that is bouncing its clock speed all over the place because it's thermally throttling, even if it is managing to stay on average at or even slightly above its rated base clock speed is not working to its full potential. And it's also constantly running up against what Intel considers to be the CPU's maximum safe operating temperature, over and over, up and down, up and down, hitting that thermal safety cutoff point.

I don't like running things at the limits of their design capacities. I also don't like knowing my machine isn't running at its full potential. That's just me, though. Others might be fine with it. It would bug the crap out of me, though.

 

[CodeJoy]

But I don’t get your point. It is reaching its full potential (arguably more). It’s listed at 2.3, yet it hovers around 2.5...

It's not reaching it's full potential, though it could be close. I don't know exactly what the full potential is, but I've suggested a way for you to find out. The base clock isn't it though. That's not what base clock means, and arguably it doesn't mean a whole lot in terms of actual performance.

Modern processors are quite complicated in how they manage thermals and clock speed. For simplicity in marketing, they often list just a few numbers, like base clock, turbo boost clock, TDP, etc. But these numbers don't really reflect what you can actually get out of the chip. And frankly, a lot of the benchmarks that you see from youtubers etc are also flawed. I'll see if I can produce a picture of what a non-throttled coffee lake system looks like.

 

[CodeJoy]

Here is an example of a (let's say imaginary) 6-core Coffee Lake system at full load that is not throttling. The first shows a sustained multicore (non-AVX) workload, and the second is first a multicore run of CineBench followed by single core (it finished at 210, because I know you're curious ). This is an average desktop system (8700K), not a laptop, so the frequency, power draw and temperatures are different. However, the mobile chips are effectively variants of this same chip, so you should expect a similar profile both for the 6-core and the 4-core chips.

For some reason, the power gadget shows the frequency as 4.97GHz when it's actually overclocked to 5.1GHz. I think this chip actually overclocks to 5.3 (and surely more with LN) if you really torture it, but 5.1 actually ends up being at a relatively low and friendly voltage.

 

[RobbieTT]

It is reaching its full potential (arguably more). It’s listed at 2.3, yet it hovers around 2.5...

It is encouraging for those with or have ordered a 13" that multiple users have now observed it comfortably running above its specification. We can probably park the cpu concerns on the 2018 MBP 13" i5 but these machines will probably get trickle-down firmware tweaks aimed more at ameliorating the i7 & i9 issues. The i9 really does look awful.

 

[CodeJoy]

I'll be doing a toy project to see if I can beat a 13" i5 for a tenth of the cost. Just for fun, nothing serious. But if you like fun, come join in! Parts should be arriving in a few hours. Throttling may happen. Corners will be cut!

https://forums.macrumors.com/threads/toy-project-to-beat-2018-mbp.2128473/

 

[mikethebigo]

For anyone with the i5 and the Intel Power Gadget installed, what frequency is your CPU idling at? The 13" i7 is idling at 1.2 GHz.

 

[The Mercurian]

For anyone with the i5 and the Intel Power Gadget installed, what frequency is your CPU idling at? The 13" i7 is idling at 1.2 GHz.

Oh do you have an i7 13”. What are you seeing for the cinebench test? Are you seeing much throttling?

 

[mikethebigo]

Oh do you have an i7 13”. What are you seeing for the cinebench test? Are you seeing much throttling?

Check out this thread https://forums.macrumors.com/thread...nce-throttling-testing.2128150/#post-26265188

Complete Cinebench benchmarks for my system

 

[The Mercurian]

Check out this thread https://forums.macrumors.com/thread...nce-throttling-testing.2128150/#post-26266184

Near the end I posted complete Cinebench benchmarks for my system

Right wow. Based on that I woulnd't pay for the upgrade. hopefully a few more i7 results roll in. Gotta say I'm loving the crowdsourcing effort here

 

[mikethebigo]

Right wow. Based on that I woulnd't pay for the upgrade. hopefully a few more i7 results roll in. Gotta say I'm loving the crowdsourcing effort here

Ya, it's a very minor difference. They end up maxing at the same frequency due to heat, but the i5 almost makes itself look more impressive because it turbos up more. The i7 for very brief bursts will of course outperform the i5, but it's kind of like, who cares? Haha.

The i7 has slight performance improvements, but they're so minimal I have no idea how anyone would notice them.

 

[CodeJoy]

Lisa had a pretty good and to the point review of the 13" i5. She didn't find it to throttle too badly, and to have very good performance relative to competition.

 

[BayouTiger]

I've been switching back and forth between the new 13" i7/16/1TB and my trusty Thinkpad X1C i7 (i7-7600U @2.8/16/512) for the past week while I've been getting everything moved over and after reading all the hand wringing over throttling and dongles, and many of the same topics that have been tossed around since 2016 on the design, I figured I would run some tests that matter to me rather than the usually meaningless benchmarks, or all the tests that indicate that every "pro" is doing video work.

I do a lot of lighting design, modeling and rendering. This is incredibly CPU bound, intensive work using a dedicated Windows application called AGI32. For grins I took a pretty simple model and ran the calc on my XiC and the time to run was 3.24 seconds. This is a flagship ultrabook that costs approximately $2100 new, so while it's not close to the new Mac's price tag, it's not a Best Buy bargain laptop.

I have a VM running in Fusion that pretty much duplicates the install I have on the Thinkpad in Fusion and with a VM set up for 4 cores and 8GB of ram - half the Mac's resources - the same model runs in 2:24. A full minute less (about 30% faster). This, while still leaving me plenty resources to continue working on the Mac side while the Windows app runs the calc.

This is why I am moving back to a Mac after trying very hard to move to the Windows side for the past year plus. I had previously tried a couple versions of Surface Pros as well as an XPS, after deciding that the 2016 15" MBP was far too big more the mobility I need. My life would be much simpler if I could just settle on a Windows laptop as my entire company is based on Windows Server and Office 365 infrastructure which has matured into an outstanding platform for my small business ( 20 employees).

I am just way more productive on MacOS that Win10. There are just way too many functions in MacOS that Windows just does not do as well with (QuickLook, Spotlight, Expose, etc) Yes Windows has things resembling these, they just don't work nearly as well when dealing with thousands of documents. It's also amazing to me what Windows users put up with as normal.
[doublepost=1532230188][/doublepost]

Oh do you have an i7 13”. What are you seeing for the cinebench test? Are you seeing much throttling?

I don't see any throttling on the Cinebench run. Runs about 2.9Ghz through the run. I actually saw more fluctuations earlier in the week running this, but it seems to have gotten better. What was curious before was that the clock was more erratic at the beginning of the run and would get more stable as it ran longer. I figured it was the fan coming on, but now I'm thinking that the OS housekeeping processes like mdworker were causing it and they are now all caught up (I was adding thousands of files to the HD for the first several days.

 

[RobbieTT]

I can now test my own machine with the new patch and the usual silicone lottery (2.3GHz / 16GB / 512GB SSD).

Turbo boost: 3.89 GHz peak, then 3.60 Ghz for ~30 sec.
Base / continuous: 3.00 Ghz (temp approaches 99 degC before dropping back and holding 94 degC)

Considering this is a base 2.3 GHz machine its ability to hold 3.00 GHz continuously at ~94 degC is pretty impressive. Ambient temperature in the UK is way above usual too.

Together with the amazing SSD performance, the T2 assistance and 3.00 GHz CPU I think I am a happy bunny.

 

[jiangning]

I can now test my own machine with the new patch and the usual silicone lottery (2.3GHz / 16GB / 512GB SSD).

Turbo boost: 3.89 GHz peak, then 3.60 Ghz for ~30 sec.
Base / continuous: 3.00 Ghz (temp approaches 99 degC before dropping back and holding 94 degC)

Considering this is a base 2.3 GHz machine its ability to hold 3.00 GHz continuously at ~94 degC is pretty impressive. Ambient temperature in the UK is way above usual too.

Together with the amazing SSD performance, the T2 assistance and 3.00 GHz CPU I think I am a happy bunny.

View attachment 772842

Nice. My i7 is rated 4.5 turbo boost, but for a week so far I’ve never seen it rise above 4.1 (before and after patch). But it can keep ~4.0 consistently so I’m not too dissatisfied.

 

[btt8]

So is the consensus here that the i7 is probably not worth it over the i5 on the 13 inch?

 

[Marketh]

So is the consensus here that the i7 is probably not worth it over the i5 on the 13 inch?

Sounds that way from what I've been reading - i7 not only gets hotter, but offers only a marginal boost in performance.

 

[joelhinch]

I have the i7 and i5 13" at the moment. Saw the massive CPU spiking pre-CPU throttle fix upgrade. MUCH better on both the i5 and i7 after update. But still the i5 and i7 are VERY close in performance. The i5 is an overachiever, constantly able to exceed base clock even when all cores/threads are in use, and the i7 underachieves.

Handbrake test:
i7 sits at 3.10 GHz
i5 also hitting 3.10 GHz (and a whole 1c cooler)

Converting a 1hr episode of OITNB in 1080p H264 - i5 beat the i7 by 30seconds

VMware test: 4 Windows 10 Pro VMs - 2CPU 2GB RAM each
i7 sits at 4.0GHZ
i5 sits at 3.6-3.8 GHZ

i7 wins this test. Bit more responsive and more room to breath. But not by much.

I'll be returning the i7 - no way is it worth the extra money.

 

[Marketh]

I have the i7 and i5 13" at the moment. Saw the massive CPU spiking pre-CPU throttle fix upgrade. MUCH better on both the i5 and i7 after update. But still the i5 and i7 are VERY close in performance. The i5 is an overachiever, constantly able to exceed base clock even when all cores/threads are in use, and the i7 underachieves.

Handbrake test:
i7 sits at 3.10 GHz
i5 also hitting 3.10 GHz (and a whole 1c cooler)

Converting a 1hr episode of OITNB in 1080p H264 - i5 beat the i7 by 30seconds

VMware test: 4 Windows 10 Pro VMs - 2CPU 2GB RAM each
i7 sits at 4.0GHZ
i5 sits at 3.6-3.8 GHZ

i7 wins this test. Bit more responsive and more room to breath. But not by much.

I'll be returning the i7 - no way is it worth the extra money.

Thanks for this - Makes me glad that I stuck with the i5.

 

[RobbieTT]

So is the consensus here that the i7 is probably not worth it over the i5 on the 13 inch?

 

[BayouTiger]

The i5 is almost always the better deal. My issue with the 13 is that the standard configurations with the i5 are all 8GB. If you want 16GB and 1TB, the only option at most stores seems to be the i7 if you want to buy from stock.

 

[justinf77]

The i5 is almost always the better deal. My issue with the 13 is that the standard configurations with the i5 are all 8GB. If you want 16GB and 1TB, the only option at most stores seems to be the i7 if you want to buy from stock.

I almost caved and bought the i7 / 1TB combo for this very reason. In the end I decided to be patient and just buy what I need (i5 / 16 / 512) and I'm glad I did. I think Apple should stock this model in stores as well, it seems to be a popular option.

 

[mikethebigo]

I almost caved and bought the i7 / 1TB combo for this very reason. In the end I decided to be patient and just buy what I need (i5 / 16 / 512) and I'm glad I did. I think Apple should stock this model in stores as well, it seems to be a popular option.

I'm kind of amazed they didn't. There isn't enough difference between i5/8/256 and i5/8/512. Make it worth it to pay more and people will. I would have picked up an i5/16/512 day one.

That being said, I'm now very used to my i7/16/1TB. I appreciate the better graphics as I'm using external monitors, and it's nice to have a little more single core burst performance. Not really worth the cash, but it's the Mac I got first and it's probably going to be the one I keep.

 

[BayouTiger]

I'm kind of amazed they didn't. There isn't enough difference between i5/8/256 and i5/8/512. Make it worth it to pay more and people will. I would have picked up an i5/16/512 day one.

That being said, I'm now very used to my i7/16/1TB. I appreciate the better graphics as I'm using external monitors, and it's nice to have a little more single core burst performance. Not really worth the cash, but it's the Mac I got first and it's probably going to be the one I keep.

I5/16/512 is a standard config for the 15”. I’ve never understood why they keep the 13 so “crippled” I figur3 they are just hitting a price target.

 

[JohnnyGo]

Nice. My i7 is rated 4.5 turbo boost, but for a week so far I’ve never seen it rise above 4.1 (before and after patch). But it can keep ~4.0 consistently so I’m not too dissatisfied.

My understanding is that “turbo boost” apply only for single threaded jobs, single core boost for short timespan. Turbo speed (lingo for all core top speed) is much lower on this new Intel chips

 

[joelhinch]

My understanding is that “turbo boost” apply only for single threaded jobs, single core boost for short timespan. Turbo speed (lingo for all core top speed) is much lower on this new Intel chips

In my testing, both i5 and i7 are turbo boosting even in multi core tasks.