new MP15 (2018) i7 (2.2/2.6Ghz) or i9 (2.9Ghz)

[omarojo]

So Im looking at 3 processor options

(2.9Ghz) i9 8950hk

(2.6Ghz) i7 8850h

(2.2Ghz) i7 8750h

Im planning to configure it for 32GB RAM with the 560x GPU
I just dont know which processor.. by looking at the benchmarks I see the numbers but honestly.. how am I suppose to know if 200pts or 500 pts or 1000 pts are considerable differences.

Im inclined for the 2.6ghz or the 2.9ghz but I read the differences are minimal..and i9 will just heat up more.

Benchmarks from Macrumors

https://forums.macrumors.com/thread...t-yearly-performance-gain-since-2011.2127645/

and this other benchmarks

I do Xcode for mobile apps and recently work with 3d Augmented Reality. As well as TouchDesigner and WebGL 3d visuals., Madmapper, Resolume.

Any suggestion ?

 

[stylinexpat]

I got the i7 8th gen with 16GB of RAM, that should be plenty powerful and fast

 

[mr.anthonyramos]

So Im looking at 3 processor options

(2.9Ghz) i9 8950hk

(2.6Ghz) i7 8850h

(2.2Ghz) i7 8750h

Im planning to configure it for 32GB RAM with the 560x GPU
I just dont know which processor.. by looking at the benchmarks I see the numbers but honestly.. how am I suppose to know if 200pts or 500 pts or 1000 pts are considerable differences.

Im inclined for the 2.6ghz or the 2.9ghz but I read the differences are minimal..and i9 will just heat up more.

Benchmarks from Macrumors

https://forums.macrumors.com/thread...t-yearly-performance-gain-since-2011.2127645/

and this other benchmarks

I do Xcode for mobile apps and recently work with 3d Augmented Reality. As well as TouchDesigner and WebGL 3d visuals., Madmapper, Resolume.

Any suggestion ?

I’d get the i9 for your line of work. This should be a no brainer.

 

[ckeck]

Due to the potential for thermal throttling on the i9 (in which case, the additional on paper performance is lost), I think the i7-8850H is the real sweet spot here.

 

[tarsins]

The i7 should be fine. As mentioned, the i9 is going to run HOT in that small case and will quickly lose its advantage.

 

[TheTuesdaydude]

here are some frequency numbers:

i7-8750H:

Frequency 2,200 MHz

Turbo Frequency

4,100 MHz (1 core),
4,100 MHz (2 cores),
4,000 MHz (4 cores),
3,900 MHz (6 cores)

i7-8850H:

Frequency 2,600 MHz

Turbo Frequency

4,300 MHz (1 core),
4,200 MHz (2 cores),
4,100 MHz (4 cores),
4,000 MHz (6 cores)

i9-8950hk:
​

Frequency 2,900 MHz
Turbo Frequency

4,600 MHz (1 core),
4,600 MHz (2 cores),
4,500 MHz (4 cores),
4,300 MHz (6 cores)

so this is what I found. but don't know how apple is handling the heat.​

 

[afir93]

The i7 should be fine. As mentioned, the i9 is going to run HOT in that small case and will quickly lose its advantage.

Do we actually know though that it will throttle so much more quickly, or is this just speculation at this point? I'm also interested to know this and I couldn't find any comprehensive thermal comparisons so far.

 

[maflynn]

Do we actually know though that it will throttle so much more quickly, or is this just speculation at this point? I'm also interested to know this and I couldn't find any comprehensive thermal comparisons so far.

We have scant information on the performance and throttling of the 2018 MBP, but we have other data, and what we do know is that coffee lake cpus run hot, and the i9 runs particularly hot given how powerful it is. If you're going to push the MBP for extended periods of time, I suspect you'll be crossing that threshold and the CPU will throttle. As of today, its my presumption because I don't have any hard numbers.

 

[Queen6]

One of our members has a personal comparison

https://forums.macrumors.com/thread...l-throttling-issues-lets-make-tables.2127132/

Q-6

 

[omarojo]

Due to the potential for thermal throttling on the i9 (in which case, the additional on paper performance is lost), I think the i7-8850H is the real sweet spot here.

The more I read the more I ditch the i9... seems like the money is not worth the i9, plus the heating problem.
Im definitely thinking getting the i7 8850H, but then again.. If you look at the charts... it is pretty much the same as the base model. Im perfectly fine paying for the 8850 but in real world... will I feel the difference with the base model ? You know.. compiling apps, opening programs, having multiple apps running at the same time, rendering in real time.

I just dont wanna be thinking later: "man.. I could have taken those extra bucks for more SSD or peripherals"

 

[igitz]

@omarojo hi, any news on this topic? did you get the i7 2.6 or the i9? i have a very similar line of work and i'm not sure what to do.

 

[stylinexpat]

The more I read the more I ditch the i9... seems like the money is not worth the i9, plus the heating problem.
Im definitely thinking getting the i7 8850H, but then again.. If you look at the charts... it is pretty much the same as the base model. Im perfectly fine paying for the 8850 but in real world... will I feel the difference with the base model ? You know.. compiling apps, opening programs, having multiple apps running at the same time, rendering in real time.

I just dont wanna be thinking later: "man.. I could have taken those extra bucks for more SSD or peripherals"

I don’t have the i9 and it gets hot as is. I could imagine how hot the i9 would get. I did notice though that if it is plugged in to a power outlet and also plugged in to an external monitor the temperature difference of how hot it gets is huge. If you are not plugged into anything any stock index browsers open then you will find that the temperature is normal and acceptable but otherwise not so. Before this I had a X1 Carbon and come to think of it now, compared to how hot this Macbook Pro 15” can get with 8th generation processor the other one was actuallly about the same.

I am looking for a nice little desk fan now. Little fan that is small and strong to place next to my laptop on my desk. Perhaps that will help keep things a bit cooler.

 

[Qaulity]

I have the i9, and I added thermal pads to the heat sink and redid the TIM on the CPU and GPU. I found there is no thermal throttling, but there is power throttling. My MBP won't go above 87W of power draw when running prime95 and temps are usually around 90C. This means when I run 12 workers and have nothing else running I can hit 2.9GHz consistently and stick at base clock. If I am using the dGPU and pulling power through that I will drop down to 2.4GHz or 2.2GHz to keep the total system wattage below 87W and my CPU temps will be around 70C. I think to turbo to 4.3GHz on the i9 you would need 120W, which is why the XPS15 comes with a 130W charger I assume.

I really think this laptop is going to be marketed for bursty loads and nothing continuous. If you had to run a simulation over night on all cores you'd be stuck at 2.9GHz giving up about 30% of the maximum performance of this chip. It's just not a workstation laptop.

 

[Thysanoptera]

This means when I run 12 workers and have nothing else running I can hit 2.9GHz consistently and stick at base clock.

That’s it?! 2.9 only? What was it before the repaste and pads? I can keep my 2.2 at 3.2 GHz all day long, and I was thinking about opening it up and doing the same. It should be higher than that, I think it is because the bootcamp throttling is still not optimized, please test it in MacOS. I’m really interested in your results, but doing it in bootcamp doesn’t really show the correct picture.

 

[Qaulity]

That’s it?! 2.9 only? What was it before the repaste and pads? I can keep my 2.2 at 3.2 GHz all day long, and I was thinking about opening it up and doing the same. It should be higher than that, I think it is because the bootcamp throttling is still not optimized, please test it in MacOS. I’m really interested in your results, but doing it in bootcamp doesn’t really show the correct picture.

My results were under macos.

 

[Thysanoptera]

My results were under macos.

I forgot there is a version for MacOS. It looks like the blend test magically limits the CPU to 2.9GHZ, while the temps are below 80C. I'm really curious how they did it. If I run the large FFT it behaves like other benchmarks, goes to 4.1, temps hit 100C, throttle down, and stays at 2.9 with temps at 90C. Apple changed the throttling setpoint, right now it is 90C, before it was bouncing form 100C.

What were your results before repaste? Was it worth it?

 

[Qaulity]

I forgot there is a version for MacOS. It looks like the blend test magically limits the CPU to 2.9GHZ, while the temps are below 80C. I'm really curious how they did it. If I run the large FFT it behaves like other benchmarks, goes to 4.1, temps hit 100C, throttle down, and stays at 2.9 with temps at 90C. Apple changed the throttling setpoint, right now it is 90C, before it was bouncing form 100C.

What were your results before repaste? Was it worth it?

I would normally idle around 55C now I idle in low 40s/high 30s and the cool down time is significantly lower, like <1sec from stopping a benchmark. Worth it? I didn't mind, it won't void the AppleCare, but if I didn't take electronics apart for fun I don't think I would have done it. It's not enough difference for the average person to put that much effort into repasting.

 

[Thysanoptera]

I would normally idle around 55C now I idle in low 40s/high 30s and the cool down time is significantly lower, like <1sec from stopping a benchmark. Worth it? I didn't mind, it won't void the AppleCare, but if I didn't take electronics apart for fun I don't think I would have done it. It's not enough difference for the average person to put that much effort into repasting.

Thanks, I think I skip it then, mine doesnt go above 40C at idle with 21C ambient. Looks like I got a decent machine, even looking at benchmarks when fully loaded. I don't think it matters that you have i9, they should idle at the same frequency, mine fluctuates between 1.3 and 1.5GHz.

When I got the first redesigned XPS it took me a lot of effort to make it usable, repasting was probably the easiest task .

 

[Qaulity]

Thanks, I think I skip it then, mine doesnt go above 40C at idle with 21C ambient. Looks like I got a decent machine, even looking at benchmarks when fully loaded. I don't think it matters that you have i9, they should idle at the same frequency, mine fluctuates between 1.3 and 1.5GHz.

When I got the first redesigned XPS it took me a lot of effort to make it usable, repasting was probably the easiest task .

Ambient here is 25C so seems reasonable. I tried the full torture test on prime95. It’s worse for me. Spikes to 3.7 ghz then drops to 2.7ghz or less within 10s. Pretty sure it’s a power throttle first then a heat throttle.

 

[Naimfan]

I went with the 2.2 after buying a 2.9 and the 2.2 at the same time. After running cinebench, and letting them chew on some of the data analysis I do, I sent the 2.9 back as the 2.2 was outpointing it on benchmarks and completing the same amount of work as the 2.9 (more, in cases where the GPU was also used - I theorize that it's because the GPU dumps a significant amount of heat into the case, requiring the 2.9 to throttle back more than the 2.2).

 

[Thysanoptera]

Ambient here is 25C so seems reasonable. I tried the full torture test on prime95. It’s worse for me. Spikes to 3.7 ghz then drops to 2.7ghz or less within 10s. Pretty sure it’s a power throttle first then a heat throttle.

I wonder if the larger i9 cache size has anything to do with it. I guess looking for primes results in a lot of missed cache calls, and frequent replenishing of 30% larger cache will lead to increased heat production and more throttling. Before you open it up again try a different benchmark and see if you get results similar to reported by others. Your ambient is also quite a bit higher than mine.

I went with the 2.2 after buying a 2.9 and the 2.2 at the same time. After running cinebench, and letting them chew on some of the data analysis I do, I sent the 2.9 back as the 2.2 was outpointing it on benchmarks and completing the same amount of work as the 2.9 (more, in cases where the GPU was also used - I theorize that it's because the GPU dumps a significant amount of heat into the case, requiring the 2.9 to throttle back more than the 2.2).

We've run some tests comparing 555x and 560x with the same CPU. If the CPU was loaded at the same time as GPU the machine with 555x was faster or at least almost equal in benchmarks, although when CPU was idle the 560 was almost 30% faster. Yeah, there is not too much thermal room in this box.