True. I'd like to see some more real world tests and also looking forward to geekbench's statement that someone said they were going to make.
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.
My real world experience with TSMC vs Samsung
- Thread starter Markiie
- Start date
- Sort by reaction score
True. I'd like to see some more real world tests and also looking forward to geekbench's statement that someone said they were going to make.
the thing is that with every other benchmark the differences are within normal chip variations, the comment below was posted on toms hardware and does a good job of explaining it
Modern processors do not just run at a specific speed and voltage. They throttle their speed or scale their voltage based on how the CPU is performing. How well the CPU performs depends on random factors like contaminants in the silicon when it's being extruded, atomic-scale defects in the transistors, etc.
That's why so many other posters are calling this the silicon lottery. The differences you're measuring aren't differences between the Samsung vs TSMC processors. They're differences between the individual processors themselves. That is, if you were to test two Samsung or two TSMC iPhones, you'd still see this type of variance in the results.
the thing is that with every other benchmark the differences are within normal chip variations, the comment below was posted on toms hardware and does a good job of explaining it
Modern processors do not just run at a specific speed and voltage. They throttle their speed or scale their voltage based on how the CPU is performing. How well the CPU performs depends on random factors like contaminants in the silicon when it's being extruded, atomic-scale defects in the transistors, etc.
That's why so many other posters are calling this the silicon lottery. The differences you're measuring aren't differences between the Samsung vs TSMC processors. They're differences between the individual processors themselves. That is, if you were to test two Samsung or two TSMC iPhones, you'd still see this type of variance in the results.
That doesn't explain why tsmc consistently does better on geekbench
Yes, an article using their bench test....
The only one so far showing a real advantage in favor of TSMC...
Stop using YouTube as a reliable source.
Geekbench, geekbench and only geek bench so far...
Are you serious ? Do you understand something about testing and semiconductors ?
All the test besides Geekbench showed NO DIFFERENCES. When you have just a few minutes between results is called a normal test variation.
It is no difficult to understand. So far the chips are almost identical if not for Geekbench test
Lol op claims "real world usage", yet puts up benchmark scores. Hilarious
Dont be patronising to me. Yes geekbench was the only real difference, but my reply was to the person saying tom's hardware suggested samsung is better, and the toms hardware article also has only minor variations in battery life. So to summarise - no real evidence for samsung, (nearly) every geekbench result for tsmc.
I'd like to see more battery comparison videos across real world tasks to see whether there's an actual difference, as well as geekbench's statement. I'd rather this not be a problem than be one - I want the 6s!
http://tscholak.github.io/bayesianism/apple/chipgate/2015/10/14/chipgate.html
Interesting reading and results from Torsten Scholak
Postdoc at University of Toronto ⋅ Ph.D. in theoretical physics
Interesting reading and results from Torsten Scholak
Postdoc at University of Toronto ⋅ Ph.D. in theoretical physics
if you read that in context to when I said that I posted that in regards to the difference in benchmarks between the two chips in every benchmark except the geekbench battery test. someone mentioned the Samsung winning in the toms hardware benchmark and said it was weird when it was behind in most of the anandtech ones, I posted that quote and said the difference could be due to normal variance in every benchmark except obviously the geekbench battery test
the problem is in the end he bases his conclusion purely on the geekbench battery test still which sadly still doesn't provide any answers, he just states that based on the geekbench battery test the TSMC appears by far to have more battery life which is what we already knew but he uses big words
.in the end the conclusion he makes is as he states:
These results present strong evidence that the Geekbench battery performance of the TSMC chip is on average much better than that of the Samsung chip.
well we know that, the problem is we don't know why and if it affects real world usage or why this one benchmarks result to so different compared to every other benchmark performed on anandtech and tomshardware
Sigh.
That is an extremely in depth and interesting statistical analysis that states, once and for all, the thing that has been blindingly obvious to all of us: the TSMC chip appears to do better in the geekbench battery benchmark.
What it doesn't do is advance our understanding of *why* that is, or *why* that result is not reflected in any other test.
all credit to vietnamese in the other thread as he posted this originally
here is a link to his post, now this is interesting!
https://forums.macrumors.com/threads/which-chip-does-your-6s-6s-have.1922967/page-54#post-22084399
here is a link to his post, now this is interesting!
https://forums.macrumors.com/threads/which-chip-does-your-6s-6s-have.1922967/page-54#post-22084399
Last edited:
TL;DR summary... The Geekbench data shows that in the Geekbench battery test, the TSMC chip outperforms the Samsung chip. No conclusion or prediction can be applied to "real-world" usage based on the test.
So if your real-world usage is running battery benchmark tests, exchange away till you get the TSMC chip. If you want to use your iPhone as a phone (you know, to surf the web, send texts, send SnapChats of your junk), then you're on your own as far as determining if there's a difference between the chips.
C
lol, thanks..
but i might be falling in between your 2 options
(shoot a lot of 4K, etc.. which also been proven to be quite the same between them)..
This only compares the Geekbench scores. These are already known to "favor" the TSMC chip.
With Geekbench (and Geekbench only), the TSCM chip performs 20-30% better than the Samsung chip.
With Basemark (and Basemark only), the Samsung chip performs 10% better than the TSMC chip.
With ALL other tests, the Samsung chip and TSMC chip perform the same.
This is where things are at. No one is disputing that TSMC does better with Geekbench. I think the more interesting thing would have to be what exactly Geekbench is doing. And what Basemark is doing.
How do those benchmark programs differ compared to real-world work-loads?
As I understand it Basemark hammers the CPU, going as fast as possible (According to Anandtech's review of M8). Geekbench seems to put on 30% load on each core, thus 60% of maximum 200% load.
So the samsung chip is more efficient when maximum load is applied?
Yes it's bad.
Thermal throttling lowers CPU frequency and voltages, making it look like runs "cooler" and "more battery efficient"...
But yet the Samsung is consistently the faster of the two chips, admittedly by a small margin, even in Geekbench's test. So how can it be throttling?