Exynos Galaxy S10 visits GeekBench, still behind the iPhone XS in single-core test
- Moj
- au0
- 19 Jan 2019
Vegetaholic, 19 Jan 2019Omg another one trying to embarrass itself. Do you know tha... moreplease stop embarrassing yourself!
If it was the cheap problem himself it would not performed well under linux(it is a well known fact and i am truly shocked how some one who claims to know about forums flooded with threadriper problem DOSE NOT know this simple fact)
The problem caused by windows kernel lack of knowledge about which core at the moment is talking with which data lane (the actual scheduling issue). The Windows issues appear to all be due to the implementation of the scheduler's best NUMA(non-uniform memory access) node assignments which is a method of memory access in a multiprocessing system that allows threads to play nicely in accessing the available memory pool.(also need to keep in mind that windows is able to handle 2 NUMA environments but in threadripper we have 4 NUMA environments and here problems happen) also it is known a simple tool like CorePrio can fix this windows problem(also i need to point out this only happens in customers grade windows aka windows 10, not on windows server).
the same happens in quad-die EPYC CPUs too!
"Thats how multiplatform benchmarks works, software designed in such away, which is not affected by different kernel architecture, hense the purpose of benchmarking."
Than how this happens if you run linux under windows (which obviously uses windows kernel to do all the necessarily stuff) and you still get the performance degradation? and if you run exactly the same setup in real linux environments there would be no performance degradation?
i hope you learn to see the real world not the apple love world you live in!
- V
- Vegetaholic
- dQc
- 19 Jan 2019
Moj, 19 Jan 2019not exactly! ever seen threadripper test on windows and li... moreOmg another one trying to embarrass itself. Do you know that threadriper has many multithreading and schedulling issues? All internet is flooded with forums about crippled threadriper performance due to task schedulling bugs. It is not problem with benchmarks, it is problem with a chipset itself, nothing to do with a kernels or libraries smart boy. Benchmarks only touch basic mathematical operations, for purely reason to test how well cpu copes with them. Thats how multiplatform benchmarks works, software designed in such away, which is not affected by different kernel architecture, hense the purpose of benchmarking. Hope you learn something.
- ?
- Anonymous
- vry
- 19 Jan 2019
Fighting over performance when all of you are just gonna use both of them like a midranger that costs 300 usd. What a joke.
- N
- Nick
- 00d
- 19 Jan 2019
Do you know guys how to buy snapdragon variant galaxys on europe???If i find it can i have official warranty from Samsung?
- K
- Kevin T
- YTE
- 19 Jan 2019
Even after being a tech enthusiast who follows latest tech &
Different chip architecture & all
It really doesn't matter which chip is faster at this moment
Cause Snapdragon 855 gives comparable performance to Apples A12 bionic
The more defining factor would be the feature set of these Chips
Which is more efficient
- Kiyasuriin
- nYT
- 19 Jan 2019
Anonymous, 19 Jan 2019Wow. They didnt catch up even when a12 barely had any impro... moreThey are stiĺl closer than ever
- Z
- ZolaIII
- 3iH
- 19 Jan 2019
Hopefully they improved on efficiency that they were lagging behind everyone else (but I doubt it). S855 is a bit more efficient than A12 (performance/w) & has a leverage in a better thinking regarding design (first big core being 15~17% faster allows better [maximum] utilisation of other three). Actually I whose catched up by the fact QC engenears would listen someone like me & implemented recommendations (including both about CPU typology and GPU emphasis on efficiency). On the other hand S855 still have same bad L4 system wide cache implementation as S845 that is both costly and performance degrading. It would be much more useful in the future to focus on both RAM true output and lateness. It wins A12 in both sustainable performance and multitasking. Still A12 is rather great design regarding its performance/W when we take in consideration how Apple big core's are still one third wider than A76 it also has a greater typology as we really don't need more than two big cores on mobile handy thing (considering that much of the real multitasking won't ever be a thing on something like that) but it simply misses the power drive limit for a heandy (on given lithography).
We really need Scalable vector SIMD NEON extension on ARM for further performance gains, the integer performance whose always good.
- T
- The last Oracle
- vIc
- 19 Jan 2019
There is something that seems like a work in progress with the 9820. It's like Samsung is holding back something until the device is released or after a firmware update.
- Moj
- 0p}
- 19 Jan 2019
Vegetaholic, 19 Jan 2019Again, incorrect comment as usual poping up. Geekbench 4 is... morenot exactly!
ever seen threadripper test on windows and linux?
there was about 1.5 times performance penalty in windows cause of kernel and libraries so the same can happen to iOS and Android.
it seems you don't understand how OS, KERNEL and LIBRARIES work and you are really hardly embarrassing yourself!
- Moj
- 0p}
- 19 Jan 2019
Shanti Dope, 19 Jan 2019Barely nowadays, but more poweful individual cores are.
Ev... morei like to have source for "open and run Crysis in maximum possible graphical settings"
cause a TITAN X would strangles and faces frame drop sometimes on playing Crysis at 4K at max setting!
something is telling me you fall for that nova BS being shown as Crysis!
Also show me a single phone which is capable of 4K@60 that has restriction!
Also a source for "To put into perspective on how ahead Apple chipsets are, a single Vortex core of the A12 Bionic is as powerful as dual Kryo 360 Gold cores of the Snapdragon 845. Let that sink in...." would be good!
- A
- Arif238
- XTt
- 19 Jan 2019
I saw video on youtube from @GaryExplains, in that video he showed that Exynos 9810 was siginficantly slower SD 845, I didn't knew that single core score was that high of Exynos 9810 than SD845 and I Exynos was much slower than Snapdragon counter part some reason. Don't know where Exynos Lack. In India we stuck with Exynos.
He used Note 9 BTW for test Exynos and Snapdragon variant.
- ?
- Anonymous
- H4X
- 19 Jan 2019
I expected so much for the 10th anniversary flagship i even saved money to upgrade from note 4 to S10 as they said it will have so many innovation but now slowly its turning out to be just another yearly update for series not a 10th year special
should have bought the note 9 but now this S10 is coming in Feb 20, lets just wait and see how it turn outs to be.
- D
- AnonD-824731
- 0mU
- 19 Jan 2019
Anonymous, 19 Jan 2019Lol fanbnoys. Back when androids were actually faster in... moreHow am I a fanboy? I don't care about Geekbench scores and I don't judge you for what you buy. Buy what you like and what best suits you. I don't care about which OS is better, both Android and iOS are great in their own ways. Android is customizable and it gives you freedom to do what you want with your device. iOS is clean, polished and integrates into Apple's ecosystem along with their other devices.
- ?
- Anonymous
- 4QP
- 19 Jan 2019
Jaeger, 19 Jan 2019Increasing the number of cores in the GPU won't make a big ... moreAlmost no difference?
And actually in multiple generations the exynos was superior in gpu whether in brute force or longevity. Let alone being superior in most generations in cpu, audio, and efficiency.
- ?
- Anonymous
- 4QP
- 19 Jan 2019
Anonymous, 19 Jan 2019Their multi-core would have been a lot better if they actua... moreI can't believe they skimped on the a76s for a75s.
No excuses.
This device will cost a lot and is supposed to be a special 10 year edition flagship release. Especially with how little they are doing already to make it anything but another regular minimal iteration release.
- ?
- Anonymous
- 4QP
- 19 Jan 2019
Abhi Dahiya, 19 Jan 2019Sadly the world doesn't run on your beliefs.
You just r... moreI almost for a second thought you knew what you were talking about.
9810s biggest problem was its scheduler. And of course being more power hungry. With the proper scheduler and clock speeds it's still a more powerful chipset.
Inefficient 8nm? What?
Multi core is just fine and already explained. A75 compared to a76.
And you have absolutely no idea how either chip will actually be. Or Ai.
Gpu will be fine and plenty close enough even if not quite as powerful.
And if 855 is more efficient and can sustain a little longer you know what? It's about time the tables turned that way for two times let alone at all because it's been that way for almost every generation of exynos socs.
So whoopdy doo Qualcomm finally made a couple strides there so now let's talk crap about it?
- b
- bobby
- tRH
- 19 Jan 2019
My recent S9 scores after oneUI update were Single Core 3742, Multi - 9072, these numbers seems to be close to S10. So what's the significant difference that we are looking at?
- J
- Jaeger
- 8xA
- 19 Jan 2019
Shanti Dope, 19 Jan 2019Yes sure, most of the tasks today are now optimized to work... moreIncreasing the number of cores in the GPU won't make a big difference. Have you ever compared a Mali-G72 MP12 (Kirin 970) and a G72 MP18 (Exynos 9810)? The difference is minuscule (roughly 5%) and with six extra cores. There is more to GPU power than number of cores, and Qualcomm knows that. That's why Adreno GPUs are always a step ahead, at least in the Android world (I won't make the mistake of comparing Android phones with iPhones. Is like comparing oranges with apples).