Hello. I just received the 2019 15" macbook pro with i7 2.6Ghz. The first run of cinebench is normal. The power is ~47 watts, the clock speed ~3.3Ghz and the temperature ~97C with score ~2500.
imgur.com/a/4laB5xM

After 2-3 cinebench r20 runs, it decreases the power to 30-35 watts, the clock speed to ~2.6-2.7Ghz and the temperature is 83-85 C with score ~2200. Is this normal?
imgur.com/a/4TEMw12

Quote from: Andreas Osthoff on June 15, 2019, 21:44:07
Prime95 is basically a worst-case scenario and even more stressful than Cinebench, so the clocks will be a bit lower. With this in mind and the mediocre results we got from our unit, it is not surprising that the i7 in the MBP 15 drops below its base frequency.

Over on reddit someone mentioned that its due to the AVX instructions in Prime95 and that intel introduced "dynamic frequency scaling" anytime AVX instructions are used to keep power in check.

I may give volta a try to see if it will work with my mac, but I doubt it.

Prime95 is basically a worst-case scenario and even more stressful than Cinebench, so the clocks will be a bit lower. With this in mind and the mediocre results we got from our unit, it is not surprising that the i7 in the MBP 15 drops below its base frequency.

I just received my brand new i7 2.6ghz 2019 MBP a few days ago. After running a mid-2012 15" MBP for 7 years, this new machine is a dream to use.

A friend of mine suggested I run some stress tests to ensure that everything works as it should. Rather than go with the typical Cinebench runs, I opted to run a torture test with Prime95 to really push the CPU. I ran the max cpu/max heat stress test.  (Found under "Options --> Torture Test")

Surprisingly, after a few seconds my CPU falls under the base clock and hovers around 2.4ghz @ ~98C and never reaches base clock during the test.

I was wondering if anyone with a similar machine would be willing to run the same test to see if this performance is specific to my machine  or is generally applicable? Furthermore, do you think that using Cinebench to stress-test the CPU underestimates potential throttling compared to Prime95?

Thanks!

(p.s i would have liked to post the screenshots, but my post count is too low to allow links)

Has anyone seen any other test of the 6 core?  Or any other tests of whether 8-core units bought in stores are actually undervolted?

If yes,please post links here. We need more info and more tests.

If the store bought 8 cores are not undervolted, then the positive reviews are inaccurate and are hiding a sizable issue.

What about the idle temperature? Is the same as the 2018 mpbs or lower?

Macbooks have two 8-core processors, and LTT who reported undervolting with great performance were using the most expensive option.

I guess the base 8-core would work just as bad as your 6-core.

Excellent job!! Thanks for the report. I miss the good old days when apple products were no-brainier.