Testing 5G: Hype vs. Reality
- R
- Romeo
- MkX
- 05 Sep 2020
RickBro, 15 Jul 2020I've been subscribing to 5G from last year already. Av... moreWOW, you must be living in a good spot
- 6
- 6ix
- r3H
- 26 Aug 2020
RickBro, 15 Jul 2020I've been subscribing to 5G from last year already. Av... moreYou’re capping bro
- R
- RickBro
- NsB
- 15 Jul 2020
I've been subscribing to 5G from last year already. Average monthly usage is between 1.5 to 2 terrbytes per month only streaming between 4 user's. Got my S20 Ultra around 2 month's ago and got an add on 5G Sim in it. I've never got speeds of below 350Mbps. I'm an happy chappy 😀
- T
- The Albion
- Jix
- 26 Jun 2020
Sean, 15 Jun 2020Vodafone UK recently on an S20 Plus I got 475mbps downloadIn which postcode may I ask ?
- ?
- Anonymous
- 4Sd
- 23 Jun 2020
just thought, 22 Jun 2020gsmarena team. Thanks for the review. Raw speed figures ... moreGet Gb/s speeds, that's what
- D
- AnonD-754814
- p{8
- 23 Jun 2020
Mediatek sux, 22 Jun 2020Oh i havent seen many people here talking about kryo cores.... moreI think the Kryo core gave an advantages to the Qualcomm.
Like I said I have seen people saying it.
- M
- Mediatek sux
- ypY
- 22 Jun 2020
AnonD-754814, 22 Jun 2020It was my bad. I mistakenly wrote M5 to all. But then I cha... moreOh i havent seen many people here talking about kryo cores. The people i know just look at core count and ghz. Those that are a little more into it care about the benchmarks. Personally i dont care about the names and dont think it affects the popularity since its barely mentioned (if any) in phone advertisements. And the usual oems would go for qualcomm chips regardless since they are tried and tested.
The 4 x a55 are just there for marketing too i guess, since its a cheaper way to advertise more cores. 10$ extra really might get you an extra a78 core, but who knows how cheap the a55 are.
- j
- just thought
- n5B
- 22 Jun 2020
gsmarena team. Thanks for the review.
Raw speed figures are useful but not a true indication of real life speed improvement. I would have thoughts one should look at How much faster will 5G load a website or start streaming a vidéo.
What is the point of pure download speed to say get a film in a few seconds when you can stream it instead?
What can you an average user do with 5G that you cannot do with 4G?
- D
- AnonD-754814
- p{8
- 22 Jun 2020
Mediatek sux, 22 Jun 2020Yeah i started the m5 comparison and then you put the m5(ex... moreIt was my bad. I mistakenly wrote M5 to all. But then I changed back.
You should have realized that as all is M4 that one is M4 too.
Besides I wrote Exynos 9820 too. I guess Exynos 9820 doesn't use m5.
About using only 2 A55 core. I will give 10$ more any day any time to get that performance.
I know at very low frequency A55 does better, But do we really need 4 of A55 core in 2021 ?
Anyway, Thanks for your opinion. I was thinking SD might be in a disadvantage than Samsung just because of the cortex X1 announcement.
If there were no cortex X1 core Qualcomm would have a clear advantage over the Exynos.
About the Cryo naming scheme , I know and saw a lot of people(on this website) who have decent knowledge about phone but think it to be custom core.
I'm also afraid that we might not see a custom GPU until Q3 next year. That means no in house GPU for S21.
- M
- Mediatek sux
- ypY
- 22 Jun 2020
AnonD-754814, 22 Jun 2020Was I talking about the mongoose M5 ? I was saying mongoose... moreYeah i started the m5 comparison and then you put the m5(exynos 9820) and then the rest as m4. well i guess i just got confused by all the numbers.
Yeah the 990s gpu sucks. Maybe it really is as you say. Cant think of another reason as the benchmark for that has pretty low cpu utilisation and 100% gpu utilisation.
Oh yeah now you gotta use the arm cortex x1 brandname on everything. So no kryo i guess. But was the kryo really much of an advertisement though? Everyone who cared a little knew everything after the sd820 were just rebrands with little customisation. And those that didnt know dont care much other than the core count, frequency and benchmarks. I believe those are the main selling points and the x1 is going to deliver just that.
About the a55, its still useful for low power tasks since they are mostly just idle at 1ghz. At such low frequencies they use lower power than a77 atleast, prolly around 10% less. Its at 2ghz they use like 5% more power than a77 at lowest clock while also behind in perf.
So theoretically it would be be better to just use a78 for the little cores too. But that would make the soc expensive.
- D
- AnonD-754814
- p{8
- 22 Jun 2020
Mediatek sux, 22 Jun 2020Yeah its a lot better, like m5 uses 25% more power for 35% ... moreWas I talking about the mongoose M5 ? I was saying mongoose M4 being efficient enough.
Exynos 990 is a mess, no argument on that. But was it really the fault of M5 ?
It's on Samsung. They shouldn't have fired the mongoose custom core division that early. For lack of accurate knowledge or lack of people who had it Samsung couldn't optimize the SoC. They should have kept the team till February 2020.
Don't believe me ? Then tell me why is the Mali G77 GPU in so power Hungry than others where Samsung has never been less efficient while using the same GPU(MALI).
For example Exynos 9820's GPU were more powerful and more efficient than Kirin 980's GPU even though Exynos was manufactured at 8nm. Same history for the previous years. Samsung was able to make the GPU more powerful and efficient than other Mali GPU users.
But Somehow that has changed this year with Exynos 990. This puts a question mark on the topic that were actually M5 be that inefficient if optimized properly.
- M
- Mediatek sux
- ypY
- 22 Jun 2020
AnonD-754814, 22 Jun 2020Is it really that more efficient than mongoose core ?
Thi... moreYeah its a lot better, like m5 uses 25% more power for 35% less perf. Check it out https://www.anandtech.com/show/15813/arm-cortex-a78-cortex-x1-cpu-ip-diverging/4
About the little cores, its ok for them to be 2ghz for the slight boost in heavy multicore workloads, since the m5 cores already downclock to 1.37ghz during multicore tests due to huge inefficiency. It performs worse than sd855+ at multithreading.
- D
- AnonD-754814
- p{8
- 22 Jun 2020
Mediatek sux, 22 Jun 2020According to anandtech projections, x1 core is a lot more e... moreAnyway these weren't the topic I wanted to discuss.
My point was Qualcomm will lose advertisement advantage by using X1. So, will they use X1 or not ? Or will they pay royalties to ARM to put Kryo(may be something new) name on it ?
Also there was another topic I was thinking.
As A55 was announced back in 2017. So will it really be able to provide any better efficiency than new Cortex A78 ?
I think Samsung should go with 2 cortex A55 cores at 1.5GHz , 4-5 A78 and 1/2 cortex X1.
- D
- AnonD-754814
- p{8
- 22 Jun 2020
Mediatek sux, 22 Jun 2020According to anandtech projections, x1 core is a lot more e... moreIs it really that more efficient than mongoose core ?
Think about mongoose m5(exynos 9820) core.
M4 was more efficient than A75 core and A76 was also more efficient than A75 core.
That makes the efficiency difference between mongoose M4 and Cortex A76 less than 20 percent. The power is difference is more than 20 percent.
I thing Samsung was stupid(actually they might have cheaped out). Or why didn't they use 4 mongoose core in the Exynos 9820 ? 2 m4 at 2.7GHz and another 2 m4 at 2.3GHz.
4 power hungry core might sound a bit stupid at first. But I'm pretty sure it's certainly better than clocking cortex A55 core at 1.95(GHz) rather than clocking it under 1.8GHz. At this high frequency it consumes more power than performance cortex core. Also at high frequency the cortex A75 is less efficient than mongoose m4.
So, These weren't good decisions from Samsung unless they wanted to go cheap by using less performance core. So they had to clocked up the A55 core to up the multicore performance which was a hit to the battery.
I strongly believe clocking down the A55 from 2.GHz to 1.8 GHz can save up to 5% battery.
- M
- Mediatek sux
- ypY
- 22 Jun 2020
AnonD-754814, 21 Jun 2020I was wondering whether Qualcomm will use the cortex x1 cor... moreAccording to anandtech projections, x1 core is a lot more efficent than the current big cores like exynos m5 and apple a13. They say it will be 37% higher perf at 50% more power at 5nm 3ghz compared to sd865 prime core at 2.84 ghz 7nm. Thats less efficency but can be improved if they clock it a bit lower.
On 855 the prime core stays at 825mhz most of the time. I have only seen it go to max clock for a second on app launching and switching. On gaming it does go up when loading new areas and encountering other players etc. I assume its the same with 865. If its just this the efficency impact would be minimal plus the a77 cores and the gpu would be around 20% more efficient so it would all add up to a more efficent soc. Atleast thats my guess.
- A
- Aryan
- XS%
- 22 Jun 2020
I think we will only be able to decide about the 5G performance once we use it for ourselves and see what it entails.
- D
- AnonD-754814
- HxI
- 21 Jun 2020
Mediatek sux, 21 Jun 2020Oh i didnt see this until now, i guess theyll put atleast o... moreI was wondering whether Qualcomm will use the cortex x1 core or not.
I mean they have years of experience on regular cortex core and capable of overclocking it more than other manufacturers.
So, Isn't using a whole new power hungry X1 core be a little risky for them ?
- M
- Mediatek sux
- Dk@
- 21 Jun 2020
AnonD-754814, 16 Jun 2020I've gone a lot of searching to track you down. I gotta kn... moreOh i didnt see this until now, i guess theyll put atleast one x1 core and bump up the gpu quite a bit since amd samsung thing looks like a done deal for 2021. The leaks show adreno 660 or 665, i was hoping they would move onto 7 series since its been used 3 times already.
- P
- PeeWee
- swC
- 21 Jun 2020
Anonymous, 16 Jun 2020In China, you can watch full 1080P video without any interr... moreThat's a good idea, driving your car while watching a video.