Any idea why MacBook Pro 15 2016 has lower scores than previous?

[zhaoxin]

In Geekbench 4, https://browser.primatelabs.com/mac-benchmarks

MacBook Pro 15 2016, i7 2.7Ghz, is slower than previous models, like 2.5Ghz in 2015/2014, 2.6Ghz in 2013.

Is this a bug in Geekbench?

 

[littlepud]

Maybe because the 2.5 GHz 2015 (i7 4870HQ) turbos up to 3.7 GHz, whereas the 2.6 GHz 2016 (i7 6700HQ) only turbos to 3.5 GHz? Skylake is supposed to be faster than Haswell clock-for-clock, but that might not completely nullify the older model's 200 MHz advantage. The other thing to remember is that the Haswell chip is a Crystalwell with Iris Pro, which effectively has an additional 128 MB L4 cache.

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/23
http://www.anandtech.com/show/6993/intel-iris-pro-5200-graphics-review-core-i74950hq-tested/3
http://ark.intel.com/m/products/835...-Cache-up-to-3_70-GHz#@product/specifications
http://ark.intel.com/m/products/889...-Cache-up-to-3_50-GHz#@product/specifications

 

[leman]

Also, the seems to be some weird problem with RAM speed. Unfirtunately, we don't have any additional ram benchmarks so far except geekbench, so we don't know whether it's a problem with geekbench or the ram in the new models is indeed slower

 

[zhaoxin]

Maybe because the 2.5 GHz 2015 (i7 4870HQ) turbos up to 3.7 GHz, whereas the 2.6 GHz 2016 (i7 6700HQ) only turbos to 3.5 GHz? Skylake is supposed to be faster than Haswell clock-for-clock, but that might not completely nullify the older model's 200 MHz advantage. The other thing to remember is that the Haswell chip is a Crystalwell with Iris Pro, which effectively has an additional 128 MB L4 cache.

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/23
http://www.anandtech.com/show/6993/intel-iris-pro-5200-graphics-review-core-i74950hq-tested/3
http://ark.intel.com/m/products/835...-Cache-up-to-3_70-GHz#@product/specifications
http://ark.intel.com/m/products/889...-Cache-up-to-3_50-GHz#@product/specifications

As I remembered, turbo should not work when all cores were in use, right? So the multicore testing should not be affected.

 

[x-evil-x]

I saw some testers using geekbench 3 also. Doesn't seem like you can use this for free?

 

[zhaoxin]

I saw some testers using geekbench 3 also. Doesn't seem like you can use this for free?

It seems you just missed the point. The online score browser is the scores that people who used Geekbench to measure their Macs and uploaded the scores back to Geekbench server. You don't need to have a Geekbench license to view the scores. It is opened.

 

[NAYo2002]

Check out this video

 

[zhaoxin]

Check out this video

The video tested the tasks related to the video card. The Geekbench is purely basing on the CPU. So this video doesn't help.

Besides, the score in the video base on the CPU of 2.9 GHz i7, not the base model 2.7 GHz i7. And even with 2.9 GHz. It is still behind the scores of 2.6 GHz and 2.5 GHz in previous models.

 

[x-evil-x]

The video tested the tasks related to the video card. The Geekbench is purely basing on the CPU. So this video doesn't help.

Besides, the score in the video base on the CPU of 2.9 GHz i7, not the base model 2.7 GHz i7. And even with 2.9 GHz. It is still behind the scores of 2.6 GHz and 2.5 GHz in previous models.

so you're saying you want a higher score but slower computer when it comes to real world use? The scores are basically the same or very close to. most of the ones I've seen are with the 455

 

[MattSeven]

Check out this video

Thanks, that's a great video. There are a lot of comp sci wannabe's around here telling people "what's important" but the reality is more mundane: if the apps you use run significantly faster on the new machine, who the **** cares what the geek bench says? If your apps are all about the CPU, and you're running terminal or other text-dominant apps in clamshell mode all the time, obviously this is not the upgrade for you.

 

[arbitrage]

I think anyone who wants to upgrade but isn't sure should take advantage of Apple's generous return policy and get the computer and then test it against there old one in the apps they use. That is my plan and why even though I wasn't sure I wanted this new laptop after the keynote, I ordered anyways and now with Holiday Return Window have almost 2 months to evaluate it against my 2012 rMBP.

Benchmarks give data but they don't tell the whole story.

 

[zhaoxin]

so you're saying you want a higher score but slower computer when it comes to real world use? The scores are basically the same or very close to. most of the ones I've seen are with the 455

I am saying different people has his own everyday use. For me, I am a programmer. I just use CPU. I could get alone with a Mac just with an Intel graphic card, which is provided in previous MacBook Pro generations.

Also, I want to check out on CPU solely, if it is the CPU's problem or the benchmark's problem.

 

[maflynn]

The video tested the tasks related to the video card. The Geekbench is purely basing on the CPU. So this video doesn't help.

Besides, the score in the video base on the CPU of 2.9 GHz i7, not the base model 2.7 GHz i7. And even with 2.9 GHz. It is still behind the scores of 2.6 GHz and 2.5 GHz in previous models.

I believe the computer as a whole is a lot faster, thanks the to GPU, and SSD, and while the CPU may (or may not) be faster, the computer itself is faster and at the end of the day, isn't that what counts?

The issue with synthetic benchmarks is they don't really test holistically how well a computer performs, and the video linked in this thread, shows that running some fairly intensive tasks, the 2016 MBP is in fact faster.

 

[CasablancaDR]

The video tested the tasks related to the video card. The Geekbench is purely basing on the CPU. So this video doesn't help.

Excuse me ? The Handbreak test showed in the video is 100% CPU related. The transcode / import test in FCP X showed in the video is 100% CPU related. And the Premiere test is a mix between CPU and GPU if there is graphic effect to render which may not be the case for this test.

Geekbench is a test disconnected of the reality and is always to be taken lightly.

 

[zhaoxin]

Excuse me ? The Handbreak test showed in the video is 100% CPU related. The transcode / import test in FCP X showed in the video is 100% CPU related. And the Premiere test is a mix between CPU and GPU if there is graphic effect to render which may not be the case for this test.

Geekbench is a test disconnected of the reality and is always to be taken lightly.

No, you are wrong. Converting video can only test the CPU partially. FCP X use GPU to accelerate. All of the tests are video converting related.
[doublepost=1479317994][/doublepost]

I believe the computer as a whole is a lot faster, thanks the to GPU, and SSD, and while the CPU may (or may not) be faster, the computer itself is faster and at the end of the day, isn't that what counts?

The issue with synthetic benchmarks is they don't really test holistically how well a computer performs, and the video linked in this thread, shows that running some fairly intensive tasks, the 2016 MBP is in fact faster.

Please see the thread title, I open this thread to talk about multicore cpu scores, which the link shows.

If you insist the mbp as a whole is faster, sure, maybe. But you are off the topic. I just want to know, if the CPU is lack of ability or not.

 

[maflynn]

If you insist the mbp as a whole is faster, sure, maybe. But you are off the topic. I just want to know, if the CPU is lack of ability or not

What does it matter if you're getting your work done faster. Looking at the big picture you're getting a faster computer, why worry about the CPU

 

[zhaoxin]

What does it matter if you're getting your work done faster. Looking at the big picture you're getting a faster computer, why worry about the CPU

I had answered this question before. My work is solely on CPU.

 

[terraphantm]

As I remembered, turbo should not work when all cores were in use, right? So the multicore testing should not be affected.

On my 2012 rMBP (Ivy Bridge), despite the official documentation saying the max speed is for one core, all 4 of my cores always boosted to 3.6GHz (from the 2.6GHz base). I don't have one to test, but I wouldn't be surprised if the Haswell/Broadwell ones were the same. I wonder if they changed that behavior with Skylake.