2020 i7 MBPro 1068NG7 benchmarks

[Spectrum]

Now on Geekbench:

1068ng7 - Geekbench 5 CPU Search - Geekbench

browser.geekbench.com

Highest runs hitting over 1300 single core and almost 5000 multi core.
[automerge]1589119179[/automerge]
Pretty sure this is (one of) the fastest Macs at running single core Geekbench right now: https://browser.geekbench.com/v5/cpu/2102526

 

[Ocnetgeek]

Strange that there is such a variation in scores. That 1300 is not the norm so far

 

[KPOM]

Strange that there is such a variation in scores. That 1300 is not the norm so far

If my experience with an i5 is any indication, those low scores are people jumping the gun and running benchmarks while the Mac is still indexing in the background. The 1300/4900 seems reasonable to me, since I’m now getting about 1200/4400 on my i5.

 

[Spectrum]

If my experience with an i5 is any indication, those low scores are people jumping the gun and running benchmarks while the Mac is still indexing in the background. The 1300/4900 seems reasonable to me, since I’m now getting about 1200/4400 on my i5.

Yeah, I tend to find that the aggregate Geekbench scores reported in the main table are always ~5-10% lower than any matching machine I've owned due to the average being dragged down by poorly-controlled runs.

 

[magbarn]

Strange that there is such a variation in scores. That 1300 is not the norm so far

This is mine after 24 hours of running and no indexing

 

[KPOM]

This is mine after 24 hours of running and no indexing
View attachment 913912

Nice. That’s a decent increase from the base Ice Lake model.

 

[Spectrum]

What's amazing to me is that these are approaching (and exceeding in single core) the performance on my desktop 2018 6-core i7 Mac mini that has a SC turbo of 4.6 Ghz.

Below is a Geekbench run of the laptop I am hoping to replace...I just can't decide on MBAir or MBPro...and whether I'll need 2 TB and 32 GB within the ~6-year lifetime of the laptop...

MacBook Pro (13-inch Retina Mid 2014) - Geekbench

Benchmark results for a MacBook Pro (13-inch Retina Mid 2014) with an Intel Core i7-4578U processor.

browser.geekbench.com

 

[KPOM]

This is mine after 24 hours of running and no indexing
View attachment 913912

If you are planning to keep it six years, I’d go with the Ice Lake MBP. At 16GB it is $200 more than a base MBP. What is your budget? What types of applications do you use? 16GB would be sufficient for most users.

 

[w00tini]

suggest you get RAM. you can always store files in external drives and/or cloud services. we cannot add more RAM after the fact.
[automerge]1589152451[/automerge]
wow those i7 geek bench scores are awesome!

 

[magbarn]

What's amazing to me is that these are approaching (and exceeding in single core) the performance on my desktop 2018 6-core i7 Mac mini that has a SC turbo of 4.6 Ghz.

Below is a Geekbench run of the laptop I am hoping to replace...I just can't decide on MBAir or MBPro...and whether I'll need 2 TB and 32 GB within the ~6-year lifetime of the laptop...

MacBook Pro (13-inch Retina Mid 2014) - Geekbench

Benchmark results for a MacBook Pro (13-inch Retina Mid 2014) with an Intel Core i7-4578U processor.

browser.geekbench.com

View attachment 913935

Agree with KPOM, I had the 2020 MBA i7 for about 3 weeks before returning it. I liked the machine, and it was faster than the 2016 m5 MB 12 it was replacing, but the 2020 MBP 13 10th gen just blows it away in performance. I thought the performance gap would be smaller. I’ve made a benchmark for myself for what I use these machines for: editing/processing/converting DSLR RAW files to JPG for clients while traveling. My MBP 16 is faster still but just too bulky to take on a plane.

I posted this on another thread:

Time to process convert to jpg 50 45MP Z7 files

2019 MBP 16 i9/32/2TB 1:48
2020 MBP 13 i7/32/2TB 2:15
2017 MBP 15 i7/16/1TB 3:17
2020 MBA 13 i7/16/1TB 3:47
2016 m5 MB 12 7:25

What’s surprising is how close the 2020 MBP 13 is to the octacore 2019 i9 MBP 16 (I think Lightroom isn’t taking advantage of all the cores)
Sometimes a shoot will be close to 2000 images between my partner and I, so that processor difference becomes over 30-40 minutes in difference of processing between a loaded 2020 MBA and a MBP 13

 

[Spectrum]

If you are planning to keep it six years, I’d go with the Ice Lake MBP. At 16GB it is $200 more than a base MBP. What is your budget? What types of applications do you use? 16GB would be sufficient for most users.

Mostly just your standard office and online stuff. Very little Adobe (I avoid if I can). No video editing or music. But I do do genomics analysis in RStudio, and the datasets are getting more complex all the time.

1 TB would (just) fit all my stuff in now. I'm currently managing (OK) with a 512GB SSD on the 2014 laptop by smart syncing lots of stuff to Dropbox. It works, but isn't ideal.

My "issue" with 16 GB RAM, is that I've been using laptops (and desktops - since a 2011 Mac mini) with 16 GB RAM for the past 6+ years. My 2018 Mac mini has 32 GB (self-installed).

I'm wary of only getting 16 GB RAM again in a laptop even if most of the time it probably will never get used. It just seems out dated when looking 4-5 years ahead. It's an expensive upgrade though...

 

[KPOM]

Mostly just your standard office and online stuff. Very little Adobe (I avoid if I can). No video editing or music. But I do do genomics analysis in RStudio, and the datasets are getting more complex all the time.

1 TB would (just) fit all my stuff in now. I'm currently managing (OK) with a 512GB SSD on the 2014 laptop by smart syncing lots of stuff to Dropbox. It works, but isn't ideal.

My "issue" with 16 GB RAM, is that I've been using laptops (and desktops - since a 2011 Mac mini) with 16 GB RAM for the past 6+ years. My 2018 Mac mini has 32 GB (self-installed).

I'm wary of only getting 16 GB RAM again in a laptop even if most of the time it probably will never get used. It just seems out dated when looking 4-5 years ahead. It's an expensive upgrade though...

Perhaps consider the Ice Lake i5 with 32GB vs. the i7 with 16GB.

 

[sdedalus1]

What's amazing to me is that these are approaching (and exceeding in single core) the performance on my desktop 2018 6-core i7 Mac mini that has a SC turbo of 4.6 Ghz.

But that's because Geekbench measures short, bursty workloads, and doesn't run long enough to really stress TDP or thermals. Things will be very different with rendering, encoding, gaming, etc., that use all cores, and the hexacore will do much better then.

Can people post:
1) Cinebench R20 scores
2) The max turbo boost (measured by Intel Power Gadget) during the second half of the Cinebench run?

For the i5, Max Tech on YT saw scores of 1708 (perhaps while still indexing) and 1917 (later). Boost settled around 3.0–3.1 ghz under Cinebench load.

That will give a better sense of multicore performance for heavy workloads, and if the i7 is worth the 10% cost increase.

 

[Spectrum]

But that's because Geekbench measures short, bursty workloads, and doesn't run long enough to really stress TDP or thermals. Things will be very different with rendering, encoding, gaming, etc., that use all cores, and the hexacore will do much better then.

Can people post:
1) Cinebench R20 scores
2) The max turbo boost (measured by Intel Power Gadget) during the second half of the Cinebench run?

For the i5, Max Tech on YT saw scores of 1708 (perhaps while still indexing) and 1917 (later). Boost settled around 3.0–3.1 ghz under Cinebench load.

That will give a better sense of multicore performance for heavy workloads, and if the i7 is worth the 10% cost increase.

Perhaps, but I would have still expected an 8th gen 4.6Ghz turbo (65 Watt CPU) to outpace a 10th gen 4.1 Ghz (28W CPU) for bursty loads too...

Out of comparison, Cinebench R20 gets ~2700 on my i7 mac mini: Stabilises at 65 Watts and 3.5 Ghz turbo. I think ~1900 for a quad core CPU inside a small laptop is rather outstanding. The cooler inside the 2018 mac mini is certainly not the best in class IMO.

 

[gl3lan]

I just found this benchmark video comparing last year's i7 vs this year's :

Cinebench R20 scores around 2000 for i7 10th gen (see @10:07), vs. ~1917 for i5 10th gen.

 

[FastLaneJB]

Impressive scores on Geekbench, that's faster than the MacBook Pro 16 on Single Core for all processors (The i5 is also faster of course). Because of that jump in single core the Multicore isn't that far off the 6 core MacBook Pro 16. Very nice indeed.

 

[supernaut42]

I averaged the single-core and multi-core score from the geekbench search result pages for the i5 (1038ng7) and i7 (1068ng7). These are from the first page of results, which has 25 results per page, sorted by Upload date, I believe.

	1038ng7 (i5)	1068ng7 (i7)
Single-Core	1155.32	1150.96
Multi-Core	4246.92	4272.92

I'm a bit dubious about some of the results on both pages, I'm guessing like KPOM said, people are running the benchmarks while other processes are eating cycles. If you want to try other pages, here's the code I used (open the Developers Tools Console in your browser and paste this in):

Single-core: Array.prototype.filter.call(document.getElementsByClassName("list-col-text-score"), (item, i) => { return i % 2 === 0 } ).map(item => Number(item.innerText)).reduce((acc, val) => acc + val ) / 25
Multi-core: Array.prototype.filter.call(document.getElementsByClassName("list-col-text-score"), (item, i) => { return i % 2 !== 0 } ).map(item => Number(item.innerText)).reduce((acc, val) => acc + val ) / 25

 

[sdedalus1]

Perhaps, but I would have still expected an 8th gen 4.6Ghz turbo (65 Watt CPU) to outpace a 10th gen 4.1 Ghz (28W CPU) for bursty loads too...

Ice Lake has 18% more IPC (instructions per cycle), so 4.1 * 1.18 = 4.8 ghz previous gen. About right, assuming that Geekbench isn't using those 2 extra cores.

Out of comparison, Cinebench R20 gets ~2700 on my i7 mac mini: Stabilises at 65 Watts and 3.5 Ghz turbo. I think ~1900 for a quad core CPU inside a small laptop is rather outstanding. The cooler inside the 2018 mac mini is certainly not the best in class IMO.

Thanks, so ~ 42% better on thermally limited hexacore with 50% more cores, where the IPC on sustained turbos roughly match (3.0 vs. 3.5 ghz). About right again.

And, yes, I agree the Ice Lake MBPs are a good machine and finally have all the power and features (no butterfly keyboard) I need to upgrade. I will be replacing both a quadcore 2018 mini and dualcore MBP with one. The question for me is just i5 vs i7.

I averaged the single-core and multi-core score from the geekbench search result pages for the i5 (1038ng7) and i7 (1068ng7). These are from the first page of results, which has 25 results per page, sorted by Upload date, I believe.

	1038ng7 (i5)	1068ng7 (i7)
Single-Core	1155.32	1150.96
Multi-Core	4246.92	4272.92

I'm a bit dubious about some of the results on both pages, I'm guessing like KPOM said, people are running the benchmarks while other processes are eating cycles.

Except that you'd expect both i5 and i7 scores to be lower. I have a hard time believing i7 buyers aren't as savvy about running benchmarks as i5 buyers; only thing different so far is we have more i5 data. I did a rough average in Excel and saw the same scores. I'm holding off until we get some reliable benchmarks to show whether the i7 can really do 10% better or not.

 

[FastLaneJB]

Except that you'd expect both i5 and i7 scores to be lower. I have a hard time believing i7 buyers aren't as savvy about running benchmarks as i5 buyers; only thing different so far is we have more i5 data. I did a rough average in Excel and saw the same scores. I'm holding off until we get some reliable benchmarks to show whether the i7 can really do 10% better or not.

Overall performance might not be the only item to consider. Normally are better binned chips so possible it can do it’s base clock at the same power draw as the i5 does its. Being that the base clock is a bit faster, that could result in quicker performance while keeping battery life the same.

I plan to use Endurance probably when I get mine and need maximum battery life.

 

[sdedalus1]

Overall performance might not be the only item to consider. Normally are better binned chips so possible it can do it’s base clock at the same power draw as the i5 does its. Being that the base clock is a bit faster, that could result in quicker performance while keeping battery life the same.

I plan to use Endurance probably when I get mine and need maximum battery life.

Interesting point. I wish we could undervolt these! I was able to do that with Windows on an older MBA, but it seems Apple is locking down all the new chips. I'll check out Endurance.

 

[KPOM]

Ice Lake has 18% more IPC (instructions per cycle), so 4.1 * 1.18 = 4.8 ghz previous gen. About right, assuming that Geekbench isn't using those 2 extra cores.



Thanks, so ~ 42% better on thermally limited hexacore with 50% more cores, where the IPC on sustained turbos roughly match (3.0 vs. 3.5 ghz). About right again.

And, yes, I agree the Ice Lake MBPs are a good machine and finally have all the power and features (no butterfly keyboard) I need to upgrade. I will be replacing both a quadcore 2018 mini and dualcore MBP with one. The question for me is just i5 vs i7.



Except that you'd expect both i5 and i7 scores to be lower. I have a hard time believing i7 buyers aren't as savvy about running benchmarks as i5 buyers; only thing different so far is we have more i5 data. I did a rough average in Excel and saw the same scores. I'm holding off until we get some reliable benchmarks to show whether the i7 can really do 10% better or not.

There are more i5s, and they were arriving a bit earlier, so my guess is that we'll start seeing more i7 data come in and the scores diverge. I've seen 1350/4900 for the i7.

 

[dhazeghi]

Would be great if somebody could post results for GeekBench 4 as well. EveryMac has included those in their tables for past models, so it makes it easy to compare.

 

[Spectrum]

Ice Lake has 18% more IPC (instructions per cycle), so 4.1 * 1.18 = 4.8 ghz previous gen. About right, assuming that Geekbench isn't using those 2 extra cores.



Thanks, so ~ 42% better on thermally limited hexacore with 50% more cores, where the IPC on sustained turbos roughly match (3.0 vs. 3.5 ghz). About right again.

And, yes, I agree the Ice Lake MBPs are a good machine and finally have all the power and features (no butterfly keyboard) I need to upgrade. I will be replacing both a quadcore 2018 mini and dualcore MBP with one. The question for me is just i5 vs i7.



Except that you'd expect both i5 and i7 scores to be lower. I have a hard time believing i7 buyers aren't as savvy about running benchmarks as i5 buyers; only thing different so far is we have more i5 data. I did a rough average in Excel and saw the same scores. I'm holding off until we get some reliable benchmarks to show whether the i7 can really do 10% better or not.

Good analysis - thanks!

Regarding the i5 vs i7 MBPro scores. I think there isn't much point looking at the overall averages. Too many of them are outliers due to indexing (or other background processes) that drag the average scores down. Importantly, these data are collectively *very* unlikely to be normally distributed around the mean because of this.

It is better (in my opinion) to look at the top 10-20% of reported results for both machines (you need more than 25 observations for this to be statistically reliable). Yet, even a quick glance reveals that there are no i5 results above 1300, with the highest around 1250/4500. By contrast, the i7 hits peaks of over 1350/4900. Thus, I'd give it about a 10% lead in both single core and multicore tasks (roughly proportional to the 10% clock-speed advantage and $200 price increase).

 

[sdedalus1]

Good analysis - thanks!

Regarding the i5 vs i7 MBPro scores. I think there isn't much point looking at the overall averages. Too many of them are outliers due to indexing (or other background processes) that drag the average scores down. Importantly, these data are collectively *very* unlikely to be normally distributed around the mean because of this.

It is better (in my opinion) to look at the top 10-20% of reported results for both machines (you need more than 25 observations for this to be statistically reliable). Yet, even a quick glance reveals that there are no i5 results above 1300, with the highest around 1250/4500. By contrast, the i7 hits peaks of over 1350/4900. Thus, I'd give it about a 10% lead in both single core and multicore tasks (roughly proportional to the 10% clock-speed advantage and $200 price increase).

I thought about trying to sort by the top 20% but that's way beyond my Excel skills (basically nil). And I know equally nil about statistics, so don't know if that's even a valid method. The problem with just looking at the peak numbers is the silicon lottery: although a few can hit that high, it doesn't tell us how many can, and how likely any individual buyer is to get that performance . . .

 

[KPOM]

I thought about trying to sort by the top 20% but that's way beyond my Excel skills (basically nil). And I know equally nil about statistics, so don't know if that's even a valid method. The problem with just looking at the peak numbers is the silicon lottery: although a few can hit that high, it doesn't tell us how many can, and how likely any individual buyer is to get that performance . . .

With CPUs there isn’t that much disparity in actual performance. Intel tests them so that similarly rated processors perform the same way. In other words, my 2.0 GHz Ice Lake i5 should perform the same as yours. It is true there is variability, but that’s taken into account when “binning” the chips. There is a chance that my i5 was capable of being finished as an i7 at 2.3 GHz, or could be overclocked so as to run as if it were an i7, but Intel certified it as an i5 at 2.0 GHz and the chip is programmed to run at those speeds.