• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Samsung outs Exynos 9 Series 9810

Page 11 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
https://www.anandtech.com/show/12478/exynos-9810-handson-awkward-first-results

So the 9810 rocks at Geekbench and sucks at a bunch of other benchmarks.

Has it been engineered first and foremost to attain a high GB rating exploiting the fact that tech reviewers have put all of their eggs into that one proverbial basket? I have no doubt that the the big players in cell phone SOCs are doing this to a degree already since by now they would be dumb not to, but the sheer magnitude of the difference (combined with the fact that GB is fairly diverse in terms of workloads) makes me skeptical that this is the primary driver.

Maybe it's a power thing, where the default state of the 9810 is to significantly throttle itself (assuming clock frequencies aren't reported correctly), or run most workloads on the power efficiency cores, but opens the floodgates when GB is detected to get its high score.

Maybe it's the same behaviour as above, but instead of throttling because it needs to there's a bug in the scheduler.

Or maybe there's some combination of the above factors.
 
HurleyBird said:
https://www.anandtech.com/show/12478/exynos-9810-handson-awkward-first-results

So the 9810 rocks at Geekbench and sucks at a bunch of other benchmarks.

Has it been engineered first and foremost to attain a high GB rating exploiting the fact that tech reviewers have put all of their eggs into that one proverbial basket? I have no doubt that the the big players in cell phone SOCs are doing this to a degree already since by now they would be dumb not to, but the sheer magnitude of the difference (combined with the fact that GB is fairly diverse in terms of workloads) makes me skeptical that this is the primary driver.

Maybe it's a power thing, where the default state of the 9810 is to significantly throttle itself (assuming clock frequencies aren't reported correctly), or run most workloads on the power efficiency cores, but opens the floodgates when GB is detected to get its high score.

Maybe it's the same behaviour as above, but instead of throttling because it needs to there's a bug in the scheduler.

Or maybe there's some combination of the above factors.
Click to expand...

I recon its just stuff they have to work through.

But i think one of the interesting things is the memory latency. Has anyone done testing for GB performance keeping everything the same just adjusting memory latency? LPDDR4 is really fast, thats what crystal well level of latency.
 
I'd love to see better benchmarks than Geekbench, but right now the 9810 seems far faster than the competition.

But this SoC isn't available for Verizon and Sprint, which is quite a bummer. I am jealous!
 
itsmydamnation said:
LPDDR4 is really fast, thats what crystal well level of latency.
Click to expand...

Memory latencies of PC chips: https://www.pcper.com/reviews/Processors/Ryzen-Memory-Latencys-Impact-Weak-1080p-Gaming

It's hard to tell what Random Access within Window is comparable to. In-Page latency?

The Full Random Access of 7700K is a lot lower than the same for 9810.

Crystalwell gets 50 cycles, not 50ns according to the Sandra bench Anandtech ran. That equals to 14ns assuming 3.5GHz frequency for the CPU.
 
Marketing textbook chapter 3:

You cant have a product with similar name exhibit different quality.
Reason: Consumers judge variation very hard. Better to keep an on average lower objective quality than have variation.

Its been examined and its how consumers want it. From beer to now of all things; cpu. Better some consistent crap they can predict and know. For production: Better well controlled lager than haywire ale even if the result is boring.

Now if Samsung marketing department did fiddle with the cpu performance i look forward to read Andrei take on it ! I can see he is starting to take off.
 
Oh dear, personally I don't understand why they put such small batteries in, surely bigger batteries can be fitted, at least 4000mah.
 
When Samsung dual sources their soc for the s9 they get in a bad dilamma because its the lowest denominator that decides how the soc can perform due to marketing reasons.

Not only is the exynos perhaps limited for cpu perf worse is the limiting of the dsp for eg video camera. No digital stabilization on 4k 60Hz and no hdr 4k. I am pretty sure the 845 does it but its lacking.

All in all.
The phone seems to lose excactly the nessesary compettitive edge that is vital for the fat margins vs eg apple. I am not so sure the beancounters made the right decision here.
 
Does anyone expect ARM to release its own 'big core', a stock ARM core to expand up to custom big core category i.e. Mongoose, Monsoon, and K12?

ARM currently has no updated roadmap beyond A75. With ARM unveiling a new flagship core every year it makes one curious what it has in store for the future.

aside:

HP Envy x2 sold-out in significant quantities--https://www.fudzilla.com/news/notebooks/45787-hp-snapdragon-powered-envy-x2-sold-out-in-a-matter-of-hours

Supporting the market expansion potential for ARM SoC's.
 
Andrei. said:
https://www.phonearena.com/news/Sam...omparison-vs-s8-iphone-lg-pixel-note_id103071

Looks like a complete failure of a SoC. As fast and less efficient than the 8895 at first glance. Maybe they should have just continued with their old uarch.
Click to expand...
I got my Exynos S9 Plus alredy.

3ed80cdd2cbe7cd5fc71eaf1428255e6.jpg

Ok the quality is redacted. I got CPUCL0 bin 11, CL1 bin 9 and G30 7.

Profanity is not allowed
in tech areas.

AT Mod Usandthem
 
Last edited by a moderator:
Andrei. said:
And why is that funny? I'm getting the same results. The iPhone 8 lasts more than the S8's. I didn't finish battery life on the S9 yet but PA's tests seem accurate.
Click to expand...
I have a hard time believing that... maybe in a pure web browsing battery life it wins over the small S8 because of high APL and better performance/engine. But not better than the S8 in a mixed usage, to not mention the S8+. Every other more reputable sources claims this, Arstechnica reports a downgrade from iPhone 7.
 
Lodix said:
I have a hard time believing that... maybe in a pure web browsing battery life it wins over the small S8 because of high APL and better performance/engine. But not better than the S8 in a mixed usage, to not mention the S8+. Every other more reputable sources claims this, Arstechnica reports a downgrade from iPhone 7.
Click to expand...
Ars' test is very compute light and runs far longer into unrealistic hours. The web test I designed is fair in APL and has some darker websites, the iPhone 8 wins.
 
Exynos 9810 has a similar GB single-thread score (approx. 3700) as Ryzen mobile despite a 2.8Ghz vs 3.6Ghz max clock difference?

What is the most accurate way to interpret this? Can M3 claim a ST victory over Ryzen mobile with 800mhz lower clock and 3 watts lower tdp, or are there caveats?
 
The caveat is it's a single benchmark and so far the 9810 has been totally unable to post the kind of performance you might think it should be capable of based off of that one benchmark in any other benchmark.

But we'll know more as more people get their hands on the S9.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top