They should just make all the benchmark numbers 10 times bigger so everyone feels better.
Or something.
Or something.
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.
Geekbench 5 Released With Improved Benchmark Tests, Dark Mode Support, and More
- Thread starter MacRumors
- Start date
- Sort by reaction score
Easy, run Geekbench 5 on the old device...
Seriously, EVERY benchmark ever changed their scoring between major versions, be it CPU or GPU benchmarks, because they set a new score as a baseline.
So for all those people who think their scores got lower: that is totally normal and not a bug. Going forward all new devices will be benchmarked using the new scoring system.
I am sure that’s exactly how Apple was able to bring the first ever 64bit ARM chip to the consumer market before ARM even release their design.
Apple and a few other entities hold architectural license. They can use the ISA in whatever way they want. Saying Apple is just using ARM’s reference design is like saying Intel and AMD are just copy paste each other’s X86 designs. Do you really think chip design is like LEGO where you just glue cores together and get a bigger single core out of it? You either use an IP (reference design) or you aren’t. Any tiny change of design in chip making business require complete re-verification let alone the in house design Apple use.
Last edited:
Apple DOES NOT use ARM architecture. Samsung and Qualcomm do that (using ARM cores like the A76 or A53). Apple designs 100% custom cores (micro-architecture). They have a license from ARM to use the ARMv8 instruction set architecture (ISA).
Only BS is your post which shows a complete lack of understanding of "architecture".
And way faster then an amd threadripper with 32 cores.
This tells me enough about this stupid benchmark...
[doublepost=1567527340][/doublepost]
..... no words....
So did they intentionally change the scoring system, or is it a bug? It would make it difficult to compare to older machines if it was intentional.
In single core, yes, for obvious reasons if you’re into CPU architecture. In multi core, even an AMD 16 core threadripper scores 11000+ (https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=Threadripper), which is over 4x faster than the A11 in my iPhone X.
Keep in mind these scores are user-submitted, so you have to look over many of them to get a good idea. Some of these folks are obviously running Windows update, a virus scan, or who knows what other processes, leading to lower scores. Others are running overclocked systems, so you need to note the clock speed on the results.
[doublepost=1567527897][/doublepost]
This is intentional. They’re running different/updated tests that are allegedly more indicative of overall performance, so comparisons should only be made using the same versions across different devices/systems.
Last edited:
For me this synthetic benchmark is like it was sisoftsandra 2 decades ago. FAKE.
Show your data, I’m intrigued to see how you’re arriving at this incorrect conclusion.
Useless synthetic benchmarks. Much rather see real world performance from emulators, WinRAR file compression utility, etc.
Why? It can already handle a full day for basically all people. Make the phone thinner instead.
Don't get me wrong for my simple response with that fake word.
I'm sure a13 will be a nice processor, but, apple can't have something much better than arm designed. It can have something within 5-10%. But judging how greedy and lazy they are, I think it's with minus - 5-10% AIso Apple can't have something that arm has not designed yet. That will be like traveling into the future.
I'll suggest you look deeper into the arm thing. Yes, arm cores are nice processors, but nowhere near x86.
Only “useless” since Apple processors ran away with the performance lead.
Waaaaaaay back when Samsung and Qualcomm processors scored higher that’s all Android fans talked about. Now that Apple has left them behind suddenly benchmarks are “useless” and “don’t matter”.
How hypocritical.
I look forward to reviewing your data that you’ll surely provide to back up your claims. Thanks!
This is the dumbest thing I’ve seen posted today regarding Apple processors.
Anandtech had already show A12 run almost toe to toe in single core perf with 20 cores Xeon 8176 sever chip in “important industry standard benchmark”(as Anandtech put it), SPEC2006 benchmark test. Server chip usually trade single core performance for efficiency, and mobile chip has been laser focus on push single core perf for years. It isn’t really that surprising that it can run faster than a 32 core chip.
Sources:
https://www.anandtech.com/show/13392/the-iphone-xs-xs-max-review-unveiling-the-silicon-secrets/4
https://www.anandtech.com/show/12694/assessing-cavium-thunderx2-arm-server-reality/7
Last edited:
Wow that’s weird, it’s showing you quoted my post when you have someone else’s post in the quote...
You’re probably looking at Geekbench 4. Version 5 is free.
sorry. I quoted the wrong person. My comment was meant for whoever said it was showing up as costing $.99
If that's the case, why we don't see the a12 in servers, data centers, supercomputers, etc given that it consumes waaaay les power? Ask yourself!
Regarding the arm thing...
The numbers on GB5 will be a lot smaller than GB4 but it has to do with the changed baseline. The GB5 baseline of 1000 points is now an Intel Core i3-8100. The GB4 baseline was 4000 points on a Intel Core i7-6600U. Technology has improved and is getting better so the scores likely had to be shrunk. But I've run GB5 on all my devices that I ran GB4 on and they still compare in the same order tho the scores are very different. So GB5 is right on...