One concern I have with the test is his configuration was equipped with the slowest SSD possible, the 512GB SSD. I got much better results running his test on my iMac which has a 2 TB ssd, 184 seconds. Of course an i7 with the 2TB SSD would also be faster than the i7 with 512GB but faster I/o should increase the margin between the two CPUs.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Decision time i7 vs i9 for new 2020 iMac?
- Thread starter BasilFawlty
- Start date
- Sort by reaction score
One concern I have with the test is his configuration was equipped with the slowest SSD possible, the 512GB SSD. I got much better results running his test on my iMac which has a 2 TB ssd, 184 seconds. Of course an i7 with the 2TB SSD would also be faster than the i7 with 512GB but faster I/o should increase the margin between the two CPUs.
And I love how you project your inferiority complex for not being able to afford an i9 that you feel compelled to try to put anybody down who decided that 10 cores is better than 8.
Enjoy your i7. Be happy with what you can afford and don’t be jealous of those who can afford more. It’s not healthy.
You are right, I can’t afford the i9... 😂😂😂 I literally got the computer at my 10% discounted rate and bought 2TB storage INSTEAD of wasting my money on the i9. But you know, cannot afford the i9.... LOL.
Last edited:
Your build is similar to mine except I added the i9 as well because why not.
The only items I still have to purchase will be the 64GB RAM via amazon for 250 dollars. But hey can’t afford that i9 right? 😂. All jokes aside though, if you got the cash and just want it, go for it. I’m not trying to stop anyone from buying whatever they want to spend their hard earned money on because at the end of the day, does it really matter to me? Obviously not. I’m only trying to throw my cents in backed with facts and data from many many sources. Enjoy your i9 and enjoy the awesome performance this computer is going to give us for years to come.
As soon as I get my i7/5700XT I'd be more than happy to spend some time doing some comparison testing. It would be awesome to get a clear picture of more than the typical benchmarks.
Nobody is harping on people who bought the i9 - As I've said multiple times, if you enjoy the i9, then all is good. The point is still helping the people that are still trying to decide and which one to get. And especially helping them to not waste money based on false assumptions or FOMO.
In the i9 camp there is also a lot of "I don't care what the benchmarks show I still believe things are different", instead of basing it on facts. So far we have seen no facts supporting the claim or belief that the i9 is noticeable faster than the i7. If we can find it by doing some more interesting tests on VMs, coding etc. I'd be happy to help. If there are certain things the i9 is better at then it will be in everybody's interest to know it Based on facts we can make informed decisions.
But I will also make it clear that it's not just the Max Tech videos that show no significant difference - It's multiple videos and reviews from multiple sources and not all of them just focus on video encoding.
So I suggest that everybody stops seeing this as a fight or a competition. It's just a matter of finding the facts so everybody can come to their own decision based on data. But the current situations right now is that the raw performance difference in Geekbench shows a potential of up to a 10% difference in performance in situations where the CPU is fully loaded, but all real world tests show more or less identical performance. But to those of you interested in this subject, let's help each other to investigate more performance scenarios.
Well terms like noticeable and "more or less" are somewhat subjective. BTW What videos are you aware of that provide real world performance for workloads other than video editing (or other Adobe based tasks)?
This guy's mad he wasted a couple of hundred dollars on an unnecessary upgrade. LOL. Hindsight is 2020 my man. Sometimes we just make mistakes. The first step is acknowledging it in order to move on.
He wasted $400 dollars and is still trying to justify it to everyone when we have consistently shown him that he’s just wrong. I think that’s why he’s been silent since I nuked him yesterday claiming I can’t afford an i9 lol. I linked my purchase Just to make my point.
Awesome! I’m still working out stuff with my own office setup too but I’ll start researching the best ways to do VM performance testing and would be open to do other tests alongside you once you get your machine.
You could try running the Xcode benchmark featured in the Max Tech video:https://github.com/devMEremenko/XcodeBenchmark
I ran the benchmark again on my iMac: i9, 64GB of RAM & 2TB SSD. Completed in 178 seconds (vs 217 for the i9 and 229 for the i7 in the Max Tech video).
So for a $600 SSD investment and $400 i9 you are looking at a 22.2% improvement in build times. Of course, I don't know how much of that was the SSD and how much was the i9.
But you don’t compile the entire Chrome base or Firefox browser multiple times a day.
In fact, I don’t compile anything, I work in Python 😝😝 But in my anterior dev life I did have to build Java web apps.
But some benchmarks at compiling « real » things like this would be interesting, and running VMs too but I don’t know how to measure performance with VMs regarding CPUs. Usually the bottleneck is RAM more than CPU. Maybe try to find something huge to deploy through Docker with a Compose script and measure the time required when images are local. But then again it might not be threaded in the most efficient way.
One thing I´d do is preprocessing a big amount of data, like a subset of ImageNet, in Numpy with, for example, a gaussian filter and image normalization. Numpy, when properly compiled, uses AVX instruction sets and could leverage CPU power. Multiprocess the list of files through 20 processes and the other with 16 processes. Measure time to process the whole thing. This would be a very good « real life » benchmark for a data scientist/ ML engineer. Maybe I could do a Docker image which do the whole thing automatically, assuming the Docker versions are identical and memory/CPU allocation is the same.
Last edited:
It is true, I don't compile the entire Chrome code base multiple times a day but I don't work in Python either and I don't like waiting for my code to build. Building anything substantial in Swift or C++ (or indexing a large project in Xcode or a JetBrains IDE) will have no problem maxing out the i9s 10 cores.
Last edited:
Good, go enjoy that i9 and stop wasting your time in this thread being salty about wasting your money and then attempting to justify your nonsense to others 😂😂😂.
This thread is getting sh*tty and is going freaking nowhere. Stop comparing your dick guys. Are we on a pre-school forum? Seriously.
Come on BuCkDoG, let's keep it clean. There is still the chance that further benchmarking might reveal certain types of workload where the i9 can make a difference. Let's focus on that discussion and let people enjoy whatever they bought.
I wasn't trying to make it toxic in here at all, just scroll up to that ZBoater guys posts towards me so I had to defend myself. In all honestly, I hope there is a workload that can actually fully benefit from it from the i9. I feel bad that people are spending that extra money to get such minimal gains when that should not be the case. Trust me, if that i9 provided sweet gains in a lot of areas, I would have bought the i9 and 1TB Storage instead of the i7 with 2TB since its the same price difference and just used an external SSD for my additional storage. I plan on keeping this iMac for quite a long time so that's why I waited for so long before purchasing for benchmarks, extensive testing, thermals, etc. Once I get my i7 5700XT in, I can do further testing if anyone wants to see certain workflows tested.Come on BuCkDoG, let's keep it clean. There is still the chance that further benchmarking might reveal certain types of workload where the i9 can make a difference. Let's focus on that discussion and let people enjoy whatever they bought.
Come on BuCkDoG, let's keep it clean. There is still the chance that further benchmarking might reveal certain types of workload where the i9 can make a difference. Let's focus on that discussion and let people enjoy whatever they bought.
Even that Max Tech video everyone keeps citing showed a 5% difference with the Xcode benchmark and that was with the 512GB SSD (I doubt anyone but Max Tech would spec the I9 & 5700XT and leave the SSD at 512GB).
I have the i9 with a 2TB SSD and saw a 22% improvement over the base i7 iMac with the 512GB when I ran the same benchmark. Of course the i7 with 2TB would also be faster, how much I don't know.
This is the problem with using one YouTuber's video as gospel. The machines he used to compare the i7 to i9 should have both had a 2TB SSD and 5700XT if he was trying to accurately compare the i7 to the i9.
We would need 6 machines.
i7 512 GB
i7 1 TB
i7 2 TB
i9 512 GB
i9 1 TB
i9 2 TB
Install boot camp on every of these machines.
Run some real SSD test benchmarks just like those of Anandtech or any real storage benchmarks, not only those stupid BlackMagic « benchmarks » that tell absolutely nothing.
Then come back on macOS.
Compile something big, like Tensorflow or Chrome base or Firefox. Measure time.
I would skip the 512GB configurations. I don't think anyone considering an i9 should be also considering sticking with 512GB. I agree with everything else though.
I now have my 10-core, 5700 XT, 2 TB iMac! I also added 64 GB of Timetec RAM.
Something I learned from 9to5 Mac's video about RAM upgrades is that the slot-choices matters. If you simply insert two new sticks in the open slots (I think 2 and 4) of the iMac, your RAM performance will actually decrease because the clock speed drops from 2,666 to 2,133. Instead, pair the RAM in proximity.
So instead of:
I know this counter's someone else's suggestion here to match 1 and 3, and 2 and 4, but it does make a difference, at least in the Apple System Profile information.
So I wonder if this makes a difference in the other performance tests. If you simply inserted the additional RAM into the open slots, all your RAM would perform slower.
Something I learned from 9to5 Mac's video about RAM upgrades is that the slot-choices matters. If you simply insert two new sticks in the open slots (I think 2 and 4) of the iMac, your RAM performance will actually decrease because the clock speed drops from 2,666 to 2,133. Instead, pair the RAM in proximity.
So instead of:
- 4 GB
- 32 GB
- 4 GB
- 32 GB
- 4 GB
- 4 GB
- 32 GB
- 32 GB
I know this counter's someone else's suggestion here to match 1 and 3, and 2 and 4, but it does make a difference, at least in the Apple System Profile information.
So I wonder if this makes a difference in the other performance tests. If you simply inserted the additional RAM into the open slots, all your RAM would perform slower.