- Aug 14, 2017
- 1,153
- 982
- 146
What a failure of an SoC...
What a failure of an SoC...
Wow, the 9810 is a trainwreck.
Ugh people. Read the article. It's literally the paragraph above the picture.But what was the sock that overheated? Exynos or Snapdragon? Isn't clear.
I added a label underneath the picture to make it more obvious.I will also add that I encountered a very embarrassing issue on the Snapdragon Galaxy S9+.
It isn't any obvious.Ugh people. Read the article. It's literally the paragraph above the picture.
I added a label underneath the picture to make it more obvious.
It isn't any obvious.
I didn't see Andrei specifying in his article which is which. Which SOC are in his S9 and S9+.
Paying attention it's only vaguely, finally, suggested two paragraphs before that his S9+ is equipped with the Snapdragon. Coulnd't he made more clear which was the SOC that overheated?
It's on every single graph and table on that page. S9 (E9810) / S9+ (S845).It isn't any obvious.
I didn't see Andrei specifying in his article which is which. Which SOC are in his S9 and S9+.
Paying attention it's only vaguely, finally, suggested two paragraphs before that his S9+ is equipped with the Snapdragon. Coulnd't he made more clear which was the SOC that overheated?
Ugh people. Read the article. It's literally the paragraph above the picture.
I added a label underneath the picture to make it more obvious.
It's on every single graph and table on that page. S9 (E9810) / S9+ (S845).
I'm not sure what Andrei could have done more than mention it in the very paragraph that addresses the overheating.
When you say Andrei...you mean yourself correct?It's on every single graph and table on that page. S9 (E9810) / S9+ (S845).
I'm not sure what Andrei could have done more than mention it in the very paragraph that addresses the overheating.
Right on point, you know he is writing good articles when he starts referring to himself in the third person lol.You are speaking to Andrei himself! The one that wrote the article.
The power wall looks very Zen like doesn't it, these Samsung based processes really seem to have a problem delivering high clocks, it appears its a consistent trend across all types of devices, GPU's, large X86 CPU's and now large mobile CPU's. If these trend continues for them 7nm wont help for single thread perf if the competitors processes clocks better.Had the big cores eg started at 1450MHz more agressively and eg boosted to 2400-2500 at 150ms...?
I mean look at the specperf. Yes efficiency is bad but there is lots a performance potential left on the table.
This arch is build for 7nm. I have hopes. If not ...who cares a75 is plenty anyway.
I wouldn't be surprised if they boosted single core performance close to 2x in geekbench. I also wouldn't be surprised to find out that the various ARM SOC vendors are now optimising their designs around getting high geek bench scores at the expense of actual performance. They'd almost be insane not to when this single metric has somehow become the defacto benchmark of processor performance. Imagine what would inevitably happen if the GPU-buying market assessed the speed of GPUs almost entirely based on their 3DMark scores. GPUs would get very good at 3DMark in very short order.
This cpu is a mystery to me. Looking at Adreis analysis:The power wall looks very Zen like doesn't it, these Samsung based processes really seem to have a problem delivering high clocks, it appears its a consistent trend across all types of devices, GPU's, large X86 CPU's and now large mobile CPU's. If these trend continues for them 7nm wont help for single thread perf if the competitors processes clocks better.
That sounds like the phone's Power Saving mode in a nutshell.Another take on it. Had the cores run just 1450 performance would have equaled prior gen but having 30% better efficiency. Thats imo a solid result in itself. Even for the huge mm2 cost.
Interesting to see a high profile processor have such a mundane problem. Is it fixable in software? And why do android phone SoCs still rely on big.LITTLE instead of just conventional voltage and frequency scaling. Is there an actual measurable advantage?
That sounds like the phone's Power Saving mode in a nutshell.
After reading the review I pushed my own S9+ (S845) and I was able to get two messages from Samsung:
At least the GPU performs well on mine, even the sustained performance is somewhat better than AT's. On the other hand, the CPU performance is very underwhelming and inconsistent. Browser scrolling feels sometimes worse than on the S8, and camera shutter speed is slower. Though to be fair the capture speed was almost too fast on the S8, and sometimes I had to take multiple shots because I was not sure if shots were actually taken. (they were)
The upshot is that mine's battery life has been phenomenal. In real usage as well as in synthetics. (better than AT-reviewed unit by quite a margin in PCMark battery test - although screen brightness might play a role there)