Samsung Exynos 9810 First Results: Weird.

[DocSwag]

Sauce: https://www.anandtech.com/show/12478/exynos-9810-handson-awkward-first-results

 

Quote

As a refresher, early in the year Samsung LSI had dropped a bombshell in claiming an astounding 2x single-thread performance improvement with the new Exynos 9810. While this initially caused a lot of controversy and discussions on the validity of the claim, early this year we exclusively covered the high-level micro-architectural features of the new Exynos M3 core and by then it was clear that the performance claims were not just marketing claims. The new Samsung CPU core is the first “very wide” CPU microarchitecture to power Android SoCs and the first to finally follow Apple’s footsteps in the direction of maximising single-thread performance. As a result it stands to be a very interesting - and ideally very powerful - SoC for the Android market.

 

Quote

Firstly one of the biggest questions for me was confirming the final clock that Samsung would use on the Galaxy S9. We detected the clock as 2704 MHz, which is 200MHz less than the 2.9 GHz that Samsung's LSI division advertises for the chipset. What makes the story more compelling is that the 2.7 GHz clock is only achievable when one of the cores in the cluster is active - thus making Samsung employ scalable maximum frequencies depending on active core numbers in the big cluster. At two active cores the frequency drops down to 2314 MHz while three and four active cores the cores clock down to only 1794 MHz.

 

We can also confirm that the Mali G72MP18 GPU is running at a very conservative 572MHz. This is not what we had expected - the previous generation Exynos 8895 had a larger MP20 configuration, running at a similar 546MHz. The resulting performance gains for the GPU thus seem to be even lower than we had expected, as I was betting on a ~650-700 MHz clock for the graphics.

Quote

I was also able to confirm the cache configurations of the CPUs with help of our latency test. The L1D cache of the M3 cores is 64KB, up from the 32KB on the previous generation. The M3 cores also come with 512KB of private L2 caches, and a shared 4MB L3 cache.

 

The little A55 cores came at a surprise as they look to be in a separate cluster, rather than in a single DynamIQ cluster with the big cores. This creates something similar to a big.Little design, but each part of the 4+4 is its own DynamIQ cluster. So here it looks like Samsung has decided not to employ the optional L2 caches for the Cortex A55s, and instead the cluster solely relies on a shared 512KB L3 cache of the DSU. The latency scores to DRAM are outlandishly good and the best we’ve ever seen among current Android SoCs, so Samsung has definitely introduced a new generation of interconnect or memory controllers.

Geek bench

Spoiler

I think it's fair to say these numbers are really impressive. No other non-Apple Soc comes even close to touching the M3 cores, and the exynos has a good ~60% IPC lead compared to the SD845. The exynos has performance a whole generation ahead of Qualcomm and is pretty close to the performance of the A10.

 

However, what about some real world performance? And this is where it falls apart...

Spoiler

Wait what? Performance not much better than last gen's 8895, and that can't even match the 835?

Quote

What seems clear is that there is something is very very wrong with the Exynos 9810 S9+ that I tested. It was barely able to distinguish itself from last year’s Exynos 8895, let alone the Snapdragon 845 in the Qualcomm Reference Device which we previewed earlier this month. I looked through the system and monitored frequencies and indeed the big cores were reaching the maximum 2.7GHz core frequency. The only explanation I have right now is that it’s possible that the DVFS configuration, as well as the scheduler, are currently so conservatively tuned that there is barely any activity on the big cores.

 

I dug a bit more through the system and found out Samsung uses some new scheduler called “eHMP”. I’m not sure if this is something based on EAS but the system did use schedutil as a frequency governor.

 

One of the Samsung spokesmen confirmed to me that the demo unit were running special firmware for MWC and that they might not be optimized. I’m having a bit of a hard time believing they would so drastically limit the performance of the device for the show demo units and less so that they would mess around with the scheduler settings. I did get confirmation that Samsung is planning to “tune down” the Exynos variant to match the Snapdragon performance – however the current scores which I got on these devices make absolutely no sense so I do hope this is just a mistake that will be resolved in shipping firmwares and we see the full potential of the SoC.

Graphics:

Spoiler

The gpu actually does pretty well. It's not quite as good as the SD845 but it's not too far off. However, keep in mind these are peak numbers, not sustained numbers. It's likely the gpu will throttle over the long term due to higher power consumption than the 845, so don't take these numbers as the be all and end all.

 

Power:

Quote

I wasn’t able to properly measure power on the event demo devices, as they had different interface settings than my tool had been programmed with, so I only was able to make some inaccurate estimates based on coarse current readout from the system.

 

For CPU workloads, our usual CPU power virus used up 3.1W at 1-core 2.7 GHz loads. 2-core 2.3 GHz seemed to have floated around 3.1-3.5W, and a 4-core load at 1.8 GHz maintained this power consumption.

 

Over the following days I will need more time, and hopefully get some SPEC figures to paint a more accurate picture. For now the results could swing either way and be either positive or negative for the M3 cores. It’s clear that the higher frequencies have a very large power penalty, and Samsung should want to operate more in the low-to-mid frequencies, hence the current frequency scheme.

 

On the GPU side for Manhattan fluctuated between 4.5 and 5.2W, which is an improvement over the Exynos 8895. But again, this is still at a disadvantage compared to the Snapdragon 845.

The exynos delivers impressive numbers in geek bench, but in actual real world scenarios it seems to fall short. While Samsung seems to be saying this is because of optimization stuff, the fact that we're seeing such terrible performance so close to launch doesn't give me much confidence. I really hope Samsung can iron these issues out but I'm not so sure about it...

 

On the gpu side the Mali gpu in the 9810 seems pretty good. It's not as good as those from the likes of apple and Qualcomm, but it's definitely getting there.

[IrshaadH]

I think we've reached a point where it would be more beneficial to consumers for manufacturers to prioritize efficiency rather than raw performance.  I don't care less if I can average 60fps in GFXbench when my phone can't last half a day without needing a recharge!

[Silentprototipe]

More of a reason not to upgrade 

[dizmo]

28 minutes ago, IrshaadH said:

I think we've reached a point where it would be more beneficial to consumers for manufacturers to prioritize efficiency rather than raw performance.  I don't care less if I can average 60fps in GFXbench when my phone can't last half a day without needing a recharge!

Efficiency and power reduction generally go hand in hand. You'll see a much greater improvement to battery life with an advancement in screen backlighting than with an SoC change. 

[LAwLz]

7 minutes ago, IrshaadH said:

I think we've reached a point where it would be more beneficial to consumers for manufacturers to prioritize efficiency rather than raw performance.  I don't care less if I can average 60fps in GFXbench when my phone can't last half a day without needing a recharge!

Raw performance does help with efficiency though.

In some scenarios it is worth having a higher peak power usage, if it means the task can be finished faster and the entire CPU cluster can go into deep sleep.

 

Also, higher peak performance and more efficient are not mutually exclusive.

Maybe these chips are faster and the average power consumption is lower, despite the peak being higher? We don't know yet.

 

 

Anyway, weird results. Geekbench looks nice and inline what we were expecting. The rest does not though...

Impressive architectural improvements between the G71 and G72 GPU but it's a shame seeing Samsung cut back on cores for it.

 

 

Side note, I think it's great to see that Andrei is back to writing for Anandtech.

I really like his articles.

[captain_to_fire]

Basically not on Apple's level just yet when it comes to graphics performance though in CPU performance it's close to the iPhone 7/7+'s A10 Fusion at least in Geekbench. Are there any tests on how fast it exports a 4K video? But the thing is, I don't think synthetic benchmarks are the deciding factor for most people.

[Eduard the weeb]

at this point I think phones speeds aren't super important anymore ( I don't games on my phone FYI ) but if it can stream video at the screens top res and feels pretty snappy if I have a few things open like spotify, news, and maybe downloading something then I'm happy.

[Zando_]

1 minute ago, Eduard the weeb said:

at this point I think phones speeds aren't super important anymore ( I don't games on my phone FYI ) but if it can stream video at the screens top res and feels pretty snappy if I have a few things open like spotify, news, and maybe downloading something then I'm happy.

But.... you need speed because, because... um.... reasons I guess... 

 

My iPhone 6 Plus is getting a bit slow in it's old age, I really only get cranky when it doesn't load apps snappily, other than that I don't really care about the processor speed of the phone. My iPhone 4S can still run most mobile games, and I have an iPad Pro so if I wanna play some iOS games I have an epic screen with which to do so. And my Galaxy S6 is still snappy, so I don't see the fascination with faster and faster phones, though I suppose you'll notice if you game heavily or are editing and exporting video on your phone. But the iPhone 6S and Galaxy S7 are still blazing fast phones, and many people seem to gloss over them to rave over the newest phones....

[D13H4RD]

You know what really made me worry?

 

How they said they're probably going to pare back its performance to meet the 845

 

[michaelocarroll007]

50 minutes ago, IrshaadH said:

I think we've reached a point where it would be more beneficial to consumers for manufacturers to prioritize efficiency rather than raw performance.  I don't care less if I can average 60fps in GFXbench when my phone can't last half a day without needing a recharge!

also the difference in loading my facebook or webpage from 2 seconds to 1.5 seconds doesnt matter but extra 30 mins of battery does. phones now days are so fast making them faster doesnt benefit 99% of the users anymore.  Being a tech webisite most on here are that 1% that would or like the improvement in speed though. 

[Taja]

1 hour ago, IrshaadH said:

I think we've reached a point where it would be more beneficial to consumers for manufacturers to prioritize efficiency rather than raw performance.  I don't care less if I can average 60fps in GFXbench when my phone can't last half a day without needing a recharge!

So true. Most people dont need the performance (I have a moto g5s plus and it has plenty performance for me - and I use my phone A LOT. Never game on it tought.)

 

What I find almost offensive is the Focus on irrelevant stuff (like tons of RAM - which Will NEVER be used) instead of features that would actually help the user, and not just help the marketing.

[D13H4RD]

28 minutes ago, Taja said:

What I find almost offensive is the Focus on irrelevant stuff (like tons of RAM - which Will NEVER be used) instead of features that would actually help the user, and not just help the marketing.

Kinda funny, because this just happened

 

https://www.sammobile.com/news/after-galaxy-s9-samsung-no-longer-obsessed-with-being-the-worlds-first/

[Commodus]

1 hour ago, D13H4RD2L1V3 said:

Kinda funny, because this just happened

 

https://www.sammobile.com/news/after-galaxy-s9-samsung-no-longer-obsessed-with-being-the-worlds-first/

There's a certain irony to that statement, because one of the easiest ways to predict Samsung phone features is to see what Apple introduced months earlier.  Not that Apple is always a font of originality, but when AR Emoji conveniently behave a whole lot line Animoji, when Samsung Pay conveniently expands on the basic formula of Apple Pay, when the S5 conveniently has a fingerprint reader shoehorned in following the iPhone 5s launch... there's a pattern.

[LAwLz]

5 minutes ago, Commodus said:

There's a certain irony to that statement, because one of the easiest ways to predict Samsung phone features is to see what Apple introduced months earlier.  Not that Apple is always a font of originality, but when AR Emoji conveniently behave a whole lot line Animoji, when Samsung Pay conveniently expands on the basic formula of Apple Pay, when the S5 conveniently has a fingerprint reader shoehorned in following the iPhone 5s launch... there's a pattern.

Well the same can be said about Apple. Want to know what the next iPhone will have? Look at what Android phones introduced earlier.
 

I think it's a bit funny that Samsung have now been accused of:

1) Copying Apple

2) Just throwing things at the wall and seeing what sticks.

3) Just being iterative and not doing anything new.

 

When you pick and choose which things to highlight you can arrive at any predefined conclusion you might have.

[suicidalfranco]

3 hours ago, Zando Bob said:

My iPhone 6 Plus is getting a bit slow in it's old age

try replacing the battery 

[Zando_]

Just now, suicidalfranco said:

try replacing the battery 

Did that. I need to replace the lighting port as well, it's very hard to get it to charge, the port is really picky. In it's defense, I have had it since about mid-2014 or so, have dropped it multiple times, gotten it wet, and even accidentally drop-kicked it while walking and had to bend the corner back into place with pliers so the screen would sit flush. The screen has never cracked and everything works, except the aforementioned lighting port and sometimes the headphone jack. 

[suicidalfranco]

6 minutes ago, Zando Bob said:

Did that. I need to replace the lighting port as well, it's very hard to get it to charge, the port is really picky. In it's defense, I have had it since about mid-2014 or so, have dropped it multiple times, gotten it wet, and even accidentally drop-kicked it while walking and had to bend the corner back into place with pliers so the screen would sit flush. The screen has never cracked and everything works, except the aforementioned lighting port and sometimes the headphone jack. 

Now that's a legit beating 

[DocSwag]

3 hours ago, D13H4RD2L1V3 said:

You know what really made me worry?

 

How they said they're probably going to pare back its performance to meet the 845

 

And TBH, I don't even understand why. What advantages are there for Samsung? All I see are disadvantages, unless they're looking at hitting really good efficiency.

[Zando_]

37 minutes ago, suicidalfranco said:

Now that's a legit beating 

Yeet. Still alive, but I am planning on eventually getting an iPhone SE or possibly an Essential Phone.

[ScratchCat]

3 hours ago, D13H4RD2L1V3 said:

You know what really made me worry?

 

How they said they're probably going to pare back its performance to meet the 845

 

Note that this will mean it can run at insanely low clocks compared to the SD845 resulting in an amazing efficiency on the CPU. However I expect them to add a boost mode on the E9810 version, otherwise the investment in the architecture would be ill spent.

Using Geekbench SD845/E9870 2000/3500 ~ 0.54, 0.54*2.7GHz =1.54GHz. Given 1 core at 2.7GHz uses 3.1W and 1.8GHz on 4 cores uses ~3.3W we can expect a 4 core load power consumption of around 2.5W (sketch of graph estimation).

 

4 hours ago, IrshaadH said:

I think we've reached a point where it would be more beneficial to consumers for manufacturers to prioritize efficiency rather than raw performance.  I don't care less if I can average 60fps in GFXbench when my phone can't last half a day without needing a recharge!

Have a look at the Xiaomi Redmi lineup, 4000mAh with an SD650 on the Redmi Note 3 can get me through two or more days easily.

[BuckGup]

5 hours ago, IrshaadH said:

I think we've reached a point where it would be more beneficial to consumers for manufacturers to prioritize efficiency rather than raw performance.  I don't care less if I can average 60fps in GFXbench when my phone can't last half a day without needing a recharge!

Yeah it's called Apple. Literally the highest on all the benchmarks and I bet it runs the longest too

[Trixanity]

41 minutes ago, DocSwag said:

And TBH, I don't even understand why. What advantages are there for Samsung? All I see are disadvantages, unless they're looking at hitting really good efficiency.

Well, efficiency is probably one reason. Huge disparity between otherwise identical products isn't good either.

 

I also suspect even if the SoC is dialed back that the R&D invested in their custom architecture will pay off in potential new markets beyond phones. Just like Apple is doing.

[DocSwag]

11 minutes ago, Trixanity said:

Well, efficiency is probably one reason. Huge disparity between otherwise identical products isn't good either.

Higher performing Exynos variants should create a bigger demand for those products, which I would assume Samsung would make a higher profit off of (since they don't have to pay extra money to Qualcomm for the SoC).

[Dabombinable]

So this is another Tegra 3 type pig of a mobile CPU? Because it has to be drawing more power and putting out more heat.

[DocSwag]

18 minutes ago, Dabombinable said:

So this is another Tegra 3 type pig of a mobile CPU? Because it has to be drawing more power and putting out more heat.

It draws more power for sure, mostly when looking at single core. But it's still not anymore than what Apple tends to pull when it comes to power.