Are the Core i9 chips throttling due to heat ?

[TheRealAlex]

I won’t mince words the current 15” MBP with 16GB DDR4 RAM, Vega 20 and a Core i9 CPU, can’t be beat especially with a 20% off edu discount.

But what stops my purchase is the issues with thermal throttling that’s a lot of heat and 14nm CPU and 14nm GPU in a small space.

It’s unknown and untested since no reviewers or testers have pushed these systems to the max SKUs with Vega 20 have just shippied this week.

I guess I’ll wait and see. Any good reviews out yet ?

 

[leman]

Nobody has that configuration yet so there are no reviews. If we assume that the new GPU has the same heat output than the previous one, the only way how you get the CPU to throttle is by running a torture test on both the CPU and the GPU, where two if they are pushed simultaneously. Since real-world workflows involve synchronization between the CPU and the GPU, you are unlikely to experience any throttling. Of course, it depends on your exact use case.

With throttling I refer to operation below base frequency while under load.
[doublepost=1542626658][/doublepost]P.S. While it might sound like nit-picking, the name of the product Vega Pro 20. The reason why I mention this is because Vega 20 is a codename for a different chip. The Vega Pro 16 and pro 20 in the MBP are both configurations of a chip codenamed Vega 12. It’s confusing, I know

 

[TheRealAlex]

Nobody has that configuration yet so there are no reviews. If we assume that the new GPU has the same heat output than the previous one, the only way how you get the CPU to throttle is by running a torture test on both the CPU and the GPU, where two if they are pushed simultaneously. Since real-world workflows involve synchronization between the CPU and the GPU, you are unlikely to experience any throttling. Of course, it depends on your exact use case.

With throttling I refer to operation below base frequency while under load.
[doublepost=1542626658][/doublepost]P.S. While it might sound like nit-picking, the name of the product Vega Pro 20. The reason why I mention this is because Vega 20 is a codename for a different chip. The Vega Pro 16 and pro 20 in the MBP are both configurations of a chip codenamed Vega 12. It’s confusing, I know

Ok so no reviews. Man was everyone just as surprised as I was with Vega 20 ?

 

[maflynn]

thermal throttling that’s a lot of heat and 14nm CPU and 14nm GPU in a small space.

Yes, and apple hasn't done any updates the thermal management to include the Coffee Lake processors. There's plenty of threads and posts where a base 2.2 i7 MBP is beating out the i9 on cinebench because of throttling. I can only imagine the Vega equipped MBP will be worse.

Apple isn't the only ones, Dell did the same thing, slapped the Coffee lake processor in a enclosure that cannot management the heat generated.

 

[leman]

There's plenty of threads and posts where a base 2.2 i7 MBP is beating out the i9 on cinebench because of throttling.

This has nothing to do with throttling, rather with the opposite of it. The reason why i7 and i9 perform similarly in sustained multithreaded tasks is not because the i9 underperforms, but because the i7 over-performs. Both CPUs are running well above their nominal frequency, so they are perfectly within (and above) their spec. But since they are essentially the same chip with similar thermal properties (just configured differently), and Intel put the turbo limiter very high, they can reach comparable frequencies at the 45W operation. An average i9 will still outperform an average i7, simply because of binning, but that depends on your luck in silicone lottery.

The funny thing is that Intel could have "fixed" it by placing a more strict turbo limiter on the i7 (e.g. not allowing it to boost beyond 2.6 in multicore operation). I think it was a customer-friendly move from them that they didn't. Offers better bang for the buck to the customer. The i9 is a special beast, you pay a premium for a certain guaranteed performance + higher burst performance.

 

[TheRealAlex]

Yes, and apple hasn't done any updates the thermal management to include the Coffee Lake processors. There's plenty of threads and posts where a base 2.2 i7 MBP is beating out the i9 on cinebench because of throttling. I can only imagine the Vega equipped MBP will be worse.

Apple isn't the only ones, Dell did the same thing, slapped the Coffee lake processor in a enclosure that cannot management the heat generated.

That seals it. I canceled my order. Period.

 

[maflynn]

This has nothing to do with throttling, rather with the opposite of it.

I disagree, I've seen enough anecdotal evidence to suggest that the i9 runs hotter then the i7 and it runs slower.

Even if it isn't throttling, the fact remains the i9 is slower then the i7, why spend top dollar on the i9 when the base is faster ¯\_(ツ)_/¯

An average i9 will still outperform an average i7, simply because of binning, but that depends on your luck in silicone lottery.

I'm not seeing that, people are complaining at the performance of the top shelf MBP.

 

[leman]

Even if it isn't throttling, the fact remains the i9 is slower then the i7, why spend top dollar on the i9 when the base is faster ¯\_(ツ)_/¯

Because i9 is faster in burst workflows. And its very much far from being a "fact" that "i9 is slower than i7". Their performance is similar in sustained multi-core workouts, thats is. And frankly, I've seen only one test that would compare the i9 to an i7 — that was a guy on youtube who had two laptops run side by side, and he clearly showed that i9 completes a multi-core workload faster.

I'm not seeing that, people are complaining at the performance of the top shelf MBP.

As I've said, people are complaining not because the i9 is underperforming, but because the i7 is performing better than what one expects. Intel's 8-gen SKUs are very weird. I mean, just look at the specs. Kaby Lake mobile high-power lineup has the i7-7700HQ as its base — its a CPU with 2.8 GHz nominal frequency and maximal full-core boost of 3.4 Ghz. The high-end Kaby Lake is i7-7920HQ with 3.1Ghz nominal and max full-core boost of 3.7 Ghz. In each case, the maximal full-core boost is 0.6Ghz higher than the nominal clock. But with Coffee Lake, we see the 2.2 Ghz i7-8750H boost to 3.2-3.3Ghz — that is a whopping 1Ghz turbo it can sustain. The i9 reaches the same clocks, which are only 0.3-0.4 Ghz above its nominal spec — which is good (since its still above the spec) but not as amazing as the i7. If Intel were to use the same product differentiation techniques, the base Coffee Lake should have been limited to around 2.8 Ghz. I suppose they didn't do it since than the Coffee Lake wouldn't be only barely faster than the Kaby Lake quad cores... Again, its all about Intel's marketing.

So yes, I agree with you that the i9 is a somewhat dissapointig SKU (if you look at its performance relative to the other SKUs) and that an average user is better off skipping the CPU upgrade. There are valid reasons to get the i9 though (if you care about trust performance, like me) and it's certainly not the case that these CPUs are suffering from throttling.

 

[maflynn]

people are complaining not because the i9 is underperforming,

Actually they were complaining about the performance and throttling. At this point, we can agree to disagree.

 

[leman]

Actually they were complaining about the performance and throttling.

So you say, yet these complains are only based on the fact that the i7 performs very similarly. "My expensive CPU is not outperforming the cheaper one, so it must be throttling". A trivial misconception born from lack of understanding how the CPU works and the crappy Intel marketing.

At this point, we can agree to disagree.

Well, it's not my job to convince you. I have laid out all the basic facts about these CPU configurations and its up to the reader to look at all the available information and make their own conclusions. One can agree to disagree if the topic under consideration is a subject of faith or personal preference. We are discussing engineering and empirical facts from a known and well-understood scientific domain. There are no subjective opinions in this. There is either right or wrong.

 

[Thysanoptera]

I think the perception that i9 is slower than i7 comes from users who complain that their i9 is outright slow, unable to reach even base clocks and benches slower than what i7 can do. And I remember quite a lot of such posts, but their machines were clearly faulty. The disproportional amount of i9 complains is caused by the fact, that people who pay extra for i9 are more likely to verify if their machines perform as expected, while somebody with 2.2 which doesn't reach 2.6 GHz, who doesn't follow reviews/forums and uses it mainly to browse web at coffee shops may not even notice that something is wrong.

If the build quality is the same i9 will be faster than i7, it has to be, with diminishing difference going down close to zero the longer the load lasts and the more stable it is. But as evidenced by multiple posts the build quality is all over the place and ending up with i7 that's faster than i9 is not that uncommon, which is even more reinforced by the fact that the i9 is not that much faster than i7 to begin with, under ideal conditions and unlimited cooling. I mean 7% to 2.6, 12% to 2.2 (assuming usable max boost of 4.6 on i9).

 

[0-0]

This video shows the speed gains of the i9 vs the i7.

 

[Queen6]

Technically not throttling as base frequency is maintained, equally a poor show for such an expensive notebook. Apple could certainly do a lot more to cool the CPU & dGPU. Apple simply cares not too, likely as it's professional user base has dwindled to a point that such considerations are no longer relevant.

A base 8750H with adequate power supply & cooling will destroy an i9 MBP, yet Apple is all too happy to double up the upgrade price, sad but true...

Base 8750H with adequate power supply & cooling, want to pass this? you'll need an iMac Pro what a joke...

Q-6

 

[darksithpro]

I won’t mince words the current 15” MBP with 16GB DDR4 RAM, Vega 20 and a Core i9 CPU, can’t be beat especially with a 20% off edu discount.

But what stops my purchase is the issues with thermal throttling that’s a lot of heat and 14nm CPU and 14nm GPU in a small space.

It’s unknown and untested since no reviewers or testers have pushed these systems to the max SKUs with Vega 20 have just shippied this week.

I guess I’ll wait and see. Any good reviews out yet ?

I'm pretty sure apple will have it well tuned in macOS, where it probably won't throttle, or overheat, where base clocks and turbo boost can be tuned, but Windows would probably be a completely different animal. I'm sure someone will get one of these machines, boot up Windows and run something like Doom, or CS GO and the laptop overheats, or melts, lol...

 

[SDColorado]

Technically not throttling as base frequency is maintained, equally a poor show for such an expensive notebook. Apple could certainly do a lot more to cool the CPU & dGPU. Apple simply cares not too, likely as it's professional user base has dwindled to a point that such considerations are no longer relevant.

A base 8750H with adequate power supply & cooling will destroy an i9 MBP, yet Apple is all too happy to double up the upgrade price, sad but true...

Base 8750H with adequate power supply & cooling, want to pass this? you'll need an iMac Pro what a joke...
View attachment 805245 View attachment 805246

Q-6

Sure, the benchmarks look good from that big gaming laptop, but how does it look next to a Toasted White Chocolate Mocha ')

 

[Queen6]

Sure, the benchmarks look good from that big gaming laptop, but how does it look next to a Toasted White Chocolate Mocha ')

It's more of a meat eater

Q-6