Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
In the meantime, have you learned how to read a scientific paper and how to use MLPerf (which is open source)?
I'm a bit surprised someone is asking for open source, but is then searching for open source software in an app store, because he can't figure out how to run the open source software... 😂😂😂
In the meantime any software that's not open source cannot be trusted. How do we know the displayed results are not altered compared with other machines? How do we know which instructions sets are used in cpu A and if they are used also in CPU B or C. How can we trust a software that's not open source? 😂 😂 😂
 
In the meantime any software that's not open source cannot be trusted. How do we know the displayed results are not altered compared with other machines? How do we know which instructions sets are used in cpu A and if they are used also in CPU B or C. How can we trust a software that's not open source? 😂 😂 😂
1.) MLPerf is open source. Geekbench describes what they're using. What they're using is open source, just download the code and run it manually.
2.) You still don't understand how this works. Different CPU architectures have different instruction sets, so you can't use the same instruction, it's literally impossible. Also, you're still confusing CPU and SoC. Again, what do you want to benchmark? The whole SoC, the CPU cores or the Neural Engine?
 
1.) MLPerf is open source. Geekbench describes what they're using. What they're using is open source, just download the code and run it manually.
2.) You still don't understand how this works. Different CPU architectures have different instruction sets, so you can't use the same instruction, it's literally impossible. Also, you're still confusing CPU and SoC. Again, what do you want to benchmark? The whole SoC, the CPU cores or the Neural Engine?
I get it with the core app of geekbench being the mlperf. But we can't be sure how geekbench publishes those final scores until we have access to their's source code. Regarding the what parts of the soc tests, that's for another discussion. Let's keep it to the source code ftm.
 
I get it with the core app of geekbench being the mlperf.
The core of Geekbench is not MLPerf. Two different things. If you don't trust Geekbench, use MLPerf which is entirely open source. If you don't want to run either, you can run each benchmark Geekbench is using manually and compare the results to what Geekbench is reporting. That way, you can be sure what Geekbench is reporting is true (or not).
 
There are people who euphemize everything Apple does and there is you who tries to disqualify Apple's chip power.
If Apple's declaration of their chips didn't match reality, could you imagine how many people would sue them?

Personally, I think it's too risky to state these information when they're actually false.
 
There are people who euphemize everything Apple does and there is you who tries to disqualify Apple's chip power.
If Apple's declaration of their chips didn't match reality, could you imagine how many people would sue them?

Personally, I think it's too risky to state these information when they're actually false.
You can call those people non believers. Yes I don't believe this fastest ever and on any phone. Because of what proof? Only Geekbench? Which I repeat, it's not an open source software, therefore cannot be trusted.
 
Wow you all have to look at total performance, not just AI performance. The A12 vastly outperforms the 865 on single core, and is ahead on multicore, and GPU. For the total package Apple is far ahead. Qualcomm did a good job on AI though.
 
And BTW Huawei just released their's mate 40 pro and a new soc saying also Fastest ever on any smartphone. Yeah right... I'm not saying snapdragon is the fastest chip, but don't come to me with console like quality when in fact any smartphone chip cannot compete with a pc graphics card from 2010,not 2020. Also those geekbench scores don'ttell me an a14 is better than a ryzen or an I7. They are not even at 30% of those cpus... I repeat geekbench it's not an open source software. And also if it was, we need to know how each cpu is implemented in the benchmark regarding the cross platform and also how the scores are measured and also published. It is using the AI in a14, and not in snapdragon? It is using full set of instructions 256 bit or 512 in Intel and AMD? Then what score it publishes? 128 bit for Apple and 512 for Intel floating point? We don't know this until "primate labs" (for God's sake it's in their's name) makes it open-source.
 
Wow you all have to look at total performance, not just AI performance. The A12 vastly outperforms the 865 on single core, and is ahead on multicore, and GPU. For the total package Apple is far ahead. Qualcomm did a good job on AI though.
Yes. Maybe even A10 Bionic or what was the name is better than even snapdragon 875 which is not released.... 😂
 
Wow you all have to look at total performance, not just AI performance. The A12 vastly outperforms the 865 on single core, and is ahead on multicore, and GPU. For the total package Apple is far ahead. Qualcomm did a good job on AI though.

Who cares about synthetic benchmarks. Let them do real world test to see which is best.

I am not saying Apple does it, but there have been manufacturers who have been caught boosting geekbench numbers artificially.
 
  • Like
Reactions: Coffee_Time
Who cares about synthetic benchmarks. Let them do real world test to see which is best.

I am not saying Apple does it, but there have been manufacturers who have been caught boosting geekbench numbers artificially.
Antutu also is full of garbage...
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.