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.
According to a PostDoc's paper, TSMC is much better than SAMSUNG 6s' battery efficiency
- Thread starter Prince134
- Start date
- Sort by reaction score
A real thing. Heh. A whole new thread to bring attention to the stunningly new news that TSMC chips do better than Samsung's on the Geekbench battery test. Because we needed Prince134 and a theoretical physicist to point this out.
You sir are a troll.
Was there a reason to create a duplicate thread https://forums.macrumors.com/thread...msung-is-slightly-better-overall-lol.1928358/
So you guys mad?
Because this is from a difference source, and it's a detailed paper. And also, it's based on Geekbench result that Tom omitted.
I trusted Toms hardware before, but not this time.
According to the thermo picture, there already were some one pointed out something wrong with the picture. It took me a while to notice the problem of the picture. Can you see where the problem is?
It's interesting that despite all the proclamations that say the two chips are approximately equal, everyone is saying that the TSMC chip is still uniformly slightly better. The argument is how much better and how much of a magnitude is worth caring about. But no one is claiming that the Samsung is actually a better chip. To me that's saying that the TSMC lovers are maybe OCD, but they are not flat out wrong that the Samsung and TSMC chips perform identically, b/c the TSMC are at least 2-3% better. There is something to this.
Actually the Tom's Hardware test did say the Samsung was slightly better. If you take Geekbench out of the equation, the results are very mixed across the board.
Could be due to slight differences in internal layout, thanks to the smaller chip - the Apple is plastic so not a good heat conductor. The OPs notion that they stuck a fan on it, and then took thermal pictures, is ludicrous (if they were going to cheat like that, why not just take the pictures when the phone was cooler?)
Oh my mistake. But I saw the PHD dude's paper and he used stats to conclude that all of these isolated reports of the TSMC having a higher geekbench battery score than the Samsung are NOT due to random chip variation, but it all actually means what we think it means: TSMCs on average will score better on this particular test than the Samsung. That's not really in debate anymore. Prior, people were saying that oh you got a really great TSMC chip vs a really bad Samsung chip, that's why there's a big variation. Luck of the draw. But this paper is saying that in aggregate, the TSMC performance is clearly separable, and by a significant margin, on the geekbench test. (1 hour and 50 minutes of extra gaming or video editing battery life to me sounds large)
The only remaining area for debate is what the geekbench battery test itself is measuring. Is it completely out of touch with reality? If so, forget chipgate. If, however, the battery test does measure use cases for which some of us use the phone frequently, then I think that it would matter. Even if I did not game for 4 hours straight, many may gave for an hour. Now, if a TSMC chip left me with 20% power vs 0% power left, that can make a difference between having enough juice to last the rest of the night vs having to go find an outlet. Others, they will say just relax because that scenario will happen like once a year and you can just plan ahead to avoid it. I say let's leave it like this: there is a difference and let the customer have 14 days to go to town on this issue and decide whether to return.
Yeah, that statistical thesis is a bit annoying since it spends a lot of time explaining in very authoritative and scientific terms something we all already know - TSMC chips perform better in the Geekbench battery test. That hasn't been in question for a while now; it's a simple fact and can be observed by anyone on the Geekbench site. The margin is by no means small or subject to any great degree of variation.
What that doesn't seem to translate into is any real world experience, or even any other benchmark result. At the moment it's an outlier, with no other real evidence to back it up.
No, Samsung just thermal throttles. Or then that test is fake and there's a fan blowing.
SO tipical of Samsung ....
BTW all I can read here is just another test demonstrating that IN GEEKBENCH the TSMC chip is better.
Are you going to run GB all the day or to actually use the phone ?
Tom's HW explained very well why they didn't use GeekBench Test, but you didn't read it....So you guys mad?
Because this is from a difference source, and it's a detailed paper. And also, it's based on Geekbench result that Tom omitted.
We told you several time that is just BS.
But you seem to be very like Prince134.
The Samsung's A9 DOESN'T THROTTLE, otherwise you would have had difference PERFORMANCES in the tests. Performances were roughly the same between the two chips, so the two chips didn't throttle (or both throttle, but Anandtech already demonstrated that TSMC A9 doesn't throttle).
Period.
And the test's author explained in the comments below the article the reason of the 2° F of difference in the background. It is totally irrelevant to the test....
But keep ignoring it.
I don't know which Anandtech throttling article you are talking about, because I can't find one...
But here's Ars Technica showing throttling...
Now please tell what that is if not throttling?
I can admit they both could throttle when we get more data, but saying A9 doesn't throttle at all... is just plain wrong.
Anyway, since you stopped by here could you Max(IT) explain us, now when n=1 is no longer, these Geekbench results, because at least for me they are still a mystery?