Comments Locked

130 Comments

Back to Article

  • lilmoe - Sunday, February 25, 2018 - link

    I wasn't really surprised. Samsung has always used a relatively conservative DVFS/scheduler compared to others for various reasons. I wouldn't bet on this being "fixed" in production units, unless Android gets a major overhaul in UI rendering efficiency and touch latency.

    What this has been confirmed to me, though, is that all the benchmarks where the M3s are showing "odd" results employ burst loads that are too short to make any tangle conclusions of platform performance.

    Did the devices feel "slow" to you though? Slower than last year's GS8 or the SD 845 reference?
  • Andrei Frumusanu - Sunday, February 25, 2018 - link

    I didn't spend enough time with it. As an anecdote, the iPhones web performance is very noticeable so I would expect at least similar improvement on the 9810 if the synthetic scores are to have any meaning at all. Currently none of our system tests show this, which is bad.
  • Anon123Anon - Monday, February 26, 2018 - link

    So..When are we the getting iPhone X review? The previous in-depth reviews were a very good read, why don’t you do the same with iPhone X? By the looks of test results for the Samsung, the iPhone X has had some testing, why not just post the review?
  • Samus - Monday, February 26, 2018 - link

    The iPhone X is literally an iPhone 8 with a different screen and the 8 Plus camera. Just assume it’s a decent OLED screen and accept the limitations of Face ID. That’s all there is to say.
  • Anon123Anon - Monday, February 26, 2018 - link

    Funny. OK, How about reviews of iPhone 8, 8plus and X highlighting the differences between them?
  • lucam - Tuesday, February 27, 2018 - link

    Unfortunately it seems that we won't get neither iPhone X (8 and 8plus) nor iPad Pro 2017 review. Anandtech for some unknown reason has decided like that.
  • Vanguy79 - Friday, March 2, 2018 - link

    We probably won’t ever see deep dives into chip architecture anymore cuz iFixit (the chip tear down company) probably charges a high syndication fee to publications that wants to use its findings but Anandtech can’t afford the higher fees. Samsung and Qualcomm can and probably do uses their marketing budgets to “subsidise” these syndication costs.
  • 192168ll - Tuesday, September 11, 2018 - link

    That is Really Sad. Hope AnandTech will publish a Review all these 3 models!!

    _______________________________________________
    https://192168ll.co/
  • melgross - Monday, February 26, 2018 - link

    That’s all for you to say, but the phone has extra co processors and other improvements, as well as better cameras. FaceTime works extremely well.

    Limitations of Samsung’s crappy ID system are major, and make it pretty much unusable. The OLED screen is the best one out there, according to testing.
  • lovex5050 - Tuesday, July 16, 2019 - link

    tanx

    http://www.baran-room.ir
  • lovex5050 - Tuesday, September 3, 2019 - link

    very good

    http://nazchat.info
  • lovex5050 - Thursday, September 5, 2019 - link

    http://baranchat.pw

    tanx
  • lovex5050 - Saturday, September 21, 2019 - link

    very good
    http://www.chatbaran.us
  • tipoo - Tuesday, February 27, 2018 - link

    There's no review of the 8 either so your comment doesn't help anything.

    Been itching for an A11 deep dive.
  • MrUNIMOG - Wednesday, February 28, 2018 - link

    We've yet to see even the long-promised A10 deep dive...
  • fred666 - Monday, February 26, 2018 - link

    iPhone reviews are less interesting because those in the market for an iPhone will usually buy it no matter what.
    On Android, a bad review can mean I'll go with another brand.
  • Anon123Anon - Monday, February 26, 2018 - link

    I don’t think everyone would agree there. A lot of tech savvy people will buy iPhones. I’m not asking for an Android v’s iPhone flame war, I could go elsewhere for that. I come here for good quality technical reviews that I usually find interesting. I’m asking if we are going to see a technical, in-depth review if the 2017 iPhone releases like we have for prior years releases. I don’t think my request is unreasonable.
  • MrUNIMOG - Wednesday, February 28, 2018 - link

    Same here. Already got the X so it's not about a buying decision, just want to know as much as possible about this impressive device, especially about its SoC.

    However, we might be out of luck since even the iPhone 7 technology deep dive hasn't shown up so far...
  • tipoo - Tuesday, February 27, 2018 - link

    Ironic given they've been shipping the most advanced ARM mobile core to date for months. A lot of technically inclined folks have been won over.
  • msabercr - Tuesday, February 27, 2018 - link

    That's typically because apple typically tries to occupy the premium space in every form factor due to them having three phones to compete with the entire Android universe. There is much more variability in android that doesn't always correspond to price point.
  • allfor.trash - Sunday, March 25, 2018 - link

    Please update the benchmarks now that the phones are out. Please.
  • raptormissle - Sunday, February 25, 2018 - link

    >unless Android gets a major overhaul in UI rendering efficiency and touch latency.

    Why would there need to be an overhaul in Android UI rendering efficiency and touch latency when the Pixel's EAS scheduler does just fine?
  • sonicmerlin - Sunday, February 25, 2018 - link

    It doesn’t,
  • raptormissle - Sunday, February 25, 2018 - link

    Do you have a Pixel 2?
  • tuxRoller - Sunday, February 25, 2018 - link

    It's very smooth but touch latency still drags behind iOS (about 110-120 vs 70-80ms). I don't believe the issue is a hardware one.

    https://danluu.com/input-lag/

    As far as "rendering efficiency" I don't know what the poster means. As long as both platforms are using gl you see similar "efficiency".
  • lilmoe - Monday, February 26, 2018 - link

    It's not nearly "similar".....
  • mmrezaie - Monday, February 26, 2018 - link

    Thanks for the link. It is a very nice read.
  • tipoo - Tuesday, February 27, 2018 - link

    Both platforms aren't using OpenGL, iOS now renders in Metal.
  • tuxRoller - Tuesday, February 27, 2018 - link

    Re-read what I said.
  • casperes1996 - Monday, February 26, 2018 - link

    iOS renders its UI with Metal
  • grahaman27 - Monday, February 26, 2018 - link

    and android with vulkan. Touch latency IS hardware related.
  • SydneyBlue120d - Monday, February 26, 2018 - link

    Are You sure?!? AFAIK no one is using Vulkan other than games...
  • lilmoe - Monday, February 26, 2018 - link

    Samsung uses vulcan for their launcher and some of their apps.

    Android doesn't use vulcan, and the problem isn't about which api is being "utilized". Ever used chrome on a desktop and compared it to naive browsers of the respective OSs? Yea. Google's hardware UI acceleration layer is crap. They've had industry leading software rendering, I'll give them that. But their software rendering layer is still present in the rendering pipeline and, coupled with hardware acceleration introduced with ICS (with sloppy improvements over time) , it uses a lot more system resources with more cpu time than competing mobile OSs.

    Samsung are doing the right thing with the Exynos gs9. 4 small cores should be plenty for rendering, workout even maxing them out. I bet Apple's monsoon cores barely fire at any UI related task.

    Modern Android phones should be able to last much, MUCH longer with all the sophistication in their SoCs, but Google ain't doing much to help at all, not even when nothing is being referred or displayed. Why don't we have internet permission native in Android yet? Shouldn't I be able to decide which app can use my network or even run in the background? Thanks for the small bone in Oreo Google?

    You'll hear lots of airheads saying that Android's hardware acceleration is just as efficient as iOS and Windows phone/mobile. Lol, funny.
  • tuxRoller - Monday, February 26, 2018 - link

    How do you think it differs between the two?
    I'm always interested in learning something new.
  • lilmoe - Monday, February 26, 2018 - link

    Overhead. Android's implementation relies more heavily on the CPU where it shouldn't. A pixel rendered on Android has a longer path to go before it reaches the screen compared to iOS or Windows Phone. As an example of Google's attempts at "fixing" that, they ramped up the CPU clocks to max as soon as the user touches the screen............ totally inelegant. That's not how you improve on a problem...

    In addition to that, there's also the fact that Android apps run in VM. They'll never be as responsive as native apps at the same power draw. Sure, they're "optimized" when installed on a device, but lets not mistake that with what Microsoft was doing with Windows Phones, this is where Google had a huge chance to improve, and failed.

    On a side note, there was a very interesting sub-test in the GFX benchmark called driver overhead. I'm not sure if its still there, but even during Lollipop, the difference between Android and iOS were massive, not sure where it is now, but I wouldn't vouch on the gap being massively reduced.
  • tuxRoller - Tuesday, February 27, 2018 - link

    Both iOS & Android use software and hardware rendering. Android has a nice list of supported hardware-accelerated drawing operations (https://developer.android.com/guide/topics/graphic... but I've been unable to find a similar list for iOS though I know it doesn't support a fully accelerated pipeline.
    The native vs jit (though art is more of a hybrid jit/aot runtime) difference is very small, and really only shows up in ram (gc had has become much better).
    The driver overhead is only a test of drivers (ie, pure graphics api test -- direct rendering in Linux speak).
    So, if you've any details you can offer to support your claim, again, I'd love to read them.
  • lilmoe - Tuesday, February 27, 2018 - link

    @tux
    Partial VS Full acceleration, and that's from the link you provided you provided. Flexibility in hardware acceleration is nice, but the fact that it's there means that there is an extra layer of overhead. Common sense? Not sure what exactly you don't understand here.

    JIT will always be that. What they've improved was load times and the latency in actually reaching the rendering pipeline, not the pipeline itself, at least not to a very significant degree. Why does Android need significantly more memory than iOS per app if there's no significant difference?

    Drivers make a huge difference in everything dependant on the GPU. Again, not sure what you don't understands here.

    Why do you need more details when Google is "fixing" a hardware acceleration issue with more aggressive cpu and memory clocks than is needed? Doesn't that raise any flags for you? I've provided a lot of details throughout my arguments here in the comment section on anandtech for several years. I don't have them saved, but if you still think Android's rendering pipeline is just as efficient, then I have nothing else to say. The difference in the visual quality of apps between Android and iOS speaks for itself.
  • ladyanita22 - Tuesday, February 27, 2018 - link

    Apple chipsets have already more CPU power than Android ones, so that's absolutely not true. Actually, iPhones have much more power left than Android phones for UI rendering. Have you tried iOS11 on an iPhone 5s? That's one hell of a powerful device and still, its UI performance is worse than Androids of similar power.
  • lilmoe - Wednesday, February 28, 2018 - link

    @ladyanita22
    I never believed Apple had "more CPU power" than chips powering Android. Powerful? Yes. More powerful than others? Heck NO, not at ~3 watts per core, on a DUAL core (multi-core power draw is guaranteed to be MORE than 3 watts). In comparison, at 3 watts, Exynos and Snapdragon deliver WAY more performance.

    Their javascript lead mostly comes from their javascript engine, and that's software, not hardware. When it comes to media heavy workloads and rendering tasks which constitutes the absolute majority of media-consumption workloads, multi-core performance is the deciding factor. This also includes web browsing, since javascript is only a small part of rendering a webpage.

    Yes, the iPhone 5S has a great CPU, and I have reasons to believe the problem isn't from the CPU, but from Apple. They do have a proven history of slowing down older phones, and while power delivery problems (evident from recent events) is one of the reasons of the slowdowns, I don't buy that that's the only instances where slowdowns occure one bit, especially for older devices.
  • tuxRoller - Tuesday, February 27, 2018 - link

    No, that's not what it means, AND iOS ISN'T fully accelerated, and no, I'm not going to look through your posting history. If you've actual evidence, let's see it. The rather broad way you answered my post suggests I might not find anything pertinent.
  • lilmoe - Wednesday, February 28, 2018 - link

    @tux

    Hah! That's rich coming from someone providing zero evidence to their claims. Others have also argued that screen resolution doesn't impact perceived performance and battery life.

    Google has a history of sloppy optimizations for Android, from touch and audio latency, to UI rendering, which Google actually discussed in multiple videos. They've long argued how "inefficient" it was to add hardware acceleration until they were forced by the community to support hardware acceleration in ICS. With all their "efforts", it wasn't until CPU power improved that it was actually possible to brute force a more "fluid" UI in Android, and that's with aggressive scheduling, and reduced useful features compared to other Android skins.

    You know, we're bit different. I don't give companies the benefit of the doubt so easily.
  • ladyanita22 - Wednesday, February 28, 2018 - link

    @lilmoe you certainly have no idea what you're talking about. Actually iPhone chips are really, really powerful, and they've been ahead in single thread for many years. That's what matters the most for UI performance and JS, and it IS the reason why iPhones perform so much better at it than Android phones. Nitro helps, that's true, but strong SC performance is key here. Drawing more power doesn't have any effect on these simple UI tasks.

    You're the one providing ZERO sources, and looking at your history it's pretty clear you hate Google or something. Android is indeed fast and snappy if implemented properly, even on slower SoCs.
  • lilmoe - Wednesday, February 28, 2018 - link

    Sure, OK.
  • Netwern - Friday, June 29, 2018 - link

    Ignorant is a bliss.
  • Netwern - Friday, June 29, 2018 - link

    No, you're the one who is blabbering nonsense. I as a past loyal Android fan looking at what he explained and his past comment history, i can conclude that lilmoe knows well what's he's talking about. He even gets into all those technicalities maybe that's why you've missed the bus.
  • tuxRoller - Thursday, March 1, 2018 - link

    I've provided a link to the Android docs that explains how, and where, to use hardware accelerated rendering. That shows both the current capabilities of the canvas api. I've looked for, but been unable to find analogous iOS docs.
    You've not actually provided ANY EVIDENCE AT ALL.
  • ladyanita22 - Tuesday, February 27, 2018 - link

    By the way, it's been long since Android apps are turned into native code.
  • SydneyBlue120d - Tuesday, February 27, 2018 - link

    Do You have any official source proving Samsung Touchwiz is using Vulkan instead of OpenGL? I haven't heard anything after this 2016 news:
    https://www.androidcentral.com/samsung-testing-new...
    Thank You very much.
  • lilmoe - Tuesday, February 27, 2018 - link

    @sydney
    I really can't find it. I'm not sure to what degree it was utilized, but I'm sure I read last year that it's in actual use in the launcher and Samsung's browser after the nougat update.
  • ladyanita22 - Tuesday, February 27, 2018 - link

    Have you got any data to back up those claims?
  • SilthDraeth - Sunday, March 4, 2018 - link

    I can decide which apps use data in the background on my Samsung Galaxy Note 8...
  • tipoo - Tuesday, February 27, 2018 - link

    No...Android /has/ Vulkan, it doesn't render its own UI in Vulkan. Every launcher uses OpenGL.
  • Lolimaster - Tuesday, February 27, 2018 - link

    Right now, as in PC's for general use slow is no longer dictated by cpu (years from that) but by memory latency and type of storage used and gpu acceleration.

    eMMC on cheaper devices will always perform worse than UFS and this ones, worse than pci-e based like on the apples. There goes your perception of "slow".
  • lilmoe - Wednesday, February 28, 2018 - link

    Are we comparing smartphones to PCs now?
  • peevee - Monday, March 12, 2018 - link

    What is the point to compare to Qualcomm Reference Device? It is not a real phone, with real size, cost and power constraints. Just to make it look better?

    How hard is it to compare vs US S9?
  • jjj - Sunday, February 25, 2018 - link

    4 cores at 1.8 GHz remains pretty fast so that's ok if done at 3.5W and the turbos are upside.
  • jjj - Sunday, February 25, 2018 - link

    Any chance you got on screen GPU results, just in case the differences in CPU play a substantial role offscreen.
  • Andrei Frumusanu - Sunday, February 25, 2018 - link

    I didn't put any time in on-screen.
  • jjj - Sunday, February 25, 2018 - link

    Found offscreen and onscreen GFX 3.1 Car scene https://www.gsmarena.com/check_out_tfirst_samsung_...

    On screen we see around 90% gain over S8 while offscreen we see 12% so maybe at 1080p the big cores kick in hard while at native res the A55s do most of the job. Unless Samsung does something iffy and does to run it at native res when onscreen.
  • Andrei Frumusanu - Sunday, February 25, 2018 - link

    They probably forgot to change the default screen resolution of 1080 back to native 1440p.
  • jjj - Sunday, February 25, 2018 - link

    The 90% gain does seem ridiculous so maybe but it would be good to check if offscreen shifts load to the large cores.
  • Zanor - Sunday, February 25, 2018 - link

    I'm going to predict that these results will represent the final product. The S9 in general is a disappointment so this doesn't surprise me.
  • generalako - Sunday, February 25, 2018 - link

    2x single core performance uplift and 40% multi core performance uplift is your definition of "disappointment"? Lol.
  • lopri - Sunday, February 25, 2018 - link

    It is going to be worse than disappointing if those uplifts happen only in Geekbench, lol.
  • levizx - Monday, February 26, 2018 - link

    Don't forget Antutu.
  • grahaman27 - Monday, February 26, 2018 - link

    I generally try to.
  • lilmoe - Tuesday, February 27, 2018 - link

    Why? Wasn't geekbench the be all end all method of testing cpu performance?
  • North01 - Sunday, February 25, 2018 - link

    I hope we get to see some sustained performance results for the Snapdragon 845, Exynos 9810, Kirin 970 and A11.
  • Wardrive86 - Sunday, February 25, 2018 - link

    Naturally it was assumed the M3 cores would be quiet power hungry at high frequency, still 1.8 GHz for 4 cores is impressive when they are that wide. I could be wrong but I'm still betting on the SD845 for all around performance and battery life. As multithreaded as Android is (OS and apps) I think SD will surpass the Exynos in many workloads and sustain that performance lead, not to mention the Adreno 630s' perceived advantage and traditionally better drivers.
  • tuxRoller - Sunday, February 25, 2018 - link

    That's been my thought as well.
    Even though I was expecting a much bigger pretty draw, that 3.1W number is still surprisingly high for, at best, twice the perf.
  • lilmoe - Monday, February 26, 2018 - link

    Lets not get ahead of ourselves. I wouldn't be so sure about the 845 being anywhere as efficient if I were you. 4xM3s at 1.8ghz should be pretty darn efficient, and when throttling does happen, I doubt the loss in performance would be as big as 4 A75 cores being throttled.

    While the massive jump in ST performance is impressive, I'm going to stick to my guns and say that it might not be too big of a "necessity".

    Anyway, the usual Youtube speedtests should be fun to watch when they're out.
  • Wardrive86 - Wednesday, February 28, 2018 - link

    Andrei posted power figures later on in the comments
  • lopri - Sunday, February 25, 2018 - link

    Very interesting. Thank you for such a quick and excellent write up!
  • Raqia - Sunday, February 25, 2018 - link

    The M3 looks impressive as a stand alone part in a lopsided SoC with bottlenecks the show up when running benchmarks using actual Android APIs. You have to wonder what the die size of the big CPU core is and if they would have been better served by going 2 big + 4 small overall to give the same single threaded performance and some thermal breathing room for better multi-core performance as well as die space breathing room for more GPU cores where parallelism more directly drives performance. It's nice to know what phone to get if you're only going to be running Geekbench though...
  • lilmoe - Monday, February 26, 2018 - link

    I was really hoping they would go with 2 M3's and 4 M2's. At 10nm LPP, wouldn't it be relatively efficient if last year's cores were heavily tuned down? I mean, the small cluster is usually always close to max clocks... I don't know, the chip would have ballooned causing it not to be economically feasible.
  • Javert89 - Monday, February 26, 2018 - link

    Reminds me of another SoC containing 810 in the name. Also in that SoC there was 'almost no activity on big cores'. Here at least throttling on big cores is going to be ~50% leading to maybe acceptable system performance. On SD810 throttling was up to 75%.
  • ladyanita22 - Monday, February 26, 2018 - link

    How can it be that both Samsung and Qualcomm had big fuckup moments whereas Apple always delivers? WTF? I'm starting to think the 845 is the better choice.
  • Spunjji - Monday, February 26, 2018 - link

    That's ignoring the transient load issues with Apple's A9 that regularly and repeatedly flatline the whole device after 6-9 months of normal use. A11 also struggles to maintain performance under a sustained load, although it seems churlish to complain when its overall performance still remains relatively high.
  • broxman - Monday, February 26, 2018 - link

    Please explain what you mean by "transient load issues with Apple's A9 that regularly and repeatedly flatline the whole device after 6-9 months of normal use"
  • id4andrei - Monday, February 26, 2018 - link

    The iphone 6, 6s and 7 have been found to permanently throttle the CPU within the reasonable 2 year operation of the device. Even a slight battery wear could cause the devices in question to shut down so Apple kneecapped them. The CPUs are good initially, but the numbers are not sustainable.

    Even the relatively new A10 is throttled already as the folks from GB have shown. According to a letter to a US committee of some sorts, the iphone 8 and the x have hardware in them that fixes this. Unfortunately, even after battery swaps the previous iphones will arrive to the same permanent throttle in unreasonably time length.
  • NetMage - Tuesday, February 27, 2018 - link

    Evidence that it is reasonable use over two years? I have plenty of anecdotal use over two years with no throttling, even going on 3.5 years. There is no fixed battery life, it all depends on use/abuse and charging patterns.
  • id4andrei - Tuesday, February 27, 2018 - link

    It was common enough for a given sample that Apple issued the kneecapping fix. Before the patch they issued a limited recall as shutdown issues started to multiply. They realized that the issue is wider than they thought and issued the patch instead of issuing a total recall.

    In Geekbench HQ' example an iphone 7 was already throttled after one year. This constitutes in many parts of Europe a concealed defect. It is not reasonable to lose power after one year. It is reasonable to lose battery life. Furthermore no other devices does this.
  • Raqia - Tuesday, February 27, 2018 - link

    Given the perception of performance leadership from the Geekbench results of new iPhones, this rapid, inevitable but concealed performance degradation is especially misleading.
  • id4andrei - Tuesday, February 27, 2018 - link

    Think of the legal ramifications. Apple has gamed strict warranty and insurance terms. This permanent loss of performance instead of battery life is also a hidden flaw illegal in many parts of the world.
  • thunng8 - Thursday, March 1, 2018 - link

    None of the devices I have access to has throttled. This includes a 3+ year old iPhone 6 Plus and 2.5 year old 6s plus. I am not saying isn’t an issue for some people but I am just not seeing it.

    The problem they were trying to solve is shutdowns. If you do a quick search on htc or Samsung phones or just about any phone that has batteries there are large number or people with this issue and most people would assume it’s the battery and get the battery replaced or get a new phone. I had a HTC quite a while that randomly shuts down that was about 3 years old. I just chucked it away.

    What Apple did was prolonged the life of phone. What hey did wrong was not communicating it well and they didn’t give users a choice. Some might prefer shutdowns rather than throttling.
  • mobutu - Monday, February 26, 2018 - link

    "I did get confirmation that Samsung is planning to “tune down” the Exynos variant to match the Snapdragon performance"

    lol
  • BillBear - Monday, February 26, 2018 - link

    Are you running a benchmark while the device is sitting on a fan?
  • yhselp - Monday, February 26, 2018 - link

    Can you, please, add A9 to future comparisons along with A10 and A11?
  • krazyfrog - Monday, February 26, 2018 - link

    Maybe the Exynos version will finally be fast enough to brute force all those apps that developer don't bother to optimize for into scrolling smoothly.
  • grahaman27 - Monday, February 26, 2018 - link

    I don't think there will ever be enough to brute force some apps, its almost like lag is built in to some apps.
  • Holliday75 - Monday, February 26, 2018 - link

    You can't brute force timeouts.
  • tipoo - Monday, February 26, 2018 - link

    " I did get confirmation that Samsung is planning to “tune down” the Exynos variant to match the Snapdragon performance"

    Wat. That seems ludicrous to not make Qualcomm feel like the left behind they are.
  • lucam - Monday, February 26, 2018 - link

    When would be the IPhone X review?
  • warrenk81 - Monday, February 26, 2018 - link

    sooooo, are you guys just pretending nothing happened in 2017? Still missing any flagship device reviews from last year. iPad Pro, iPhone 8/X, Pixel 2...anything?
  • Lau_Tech - Tuesday, February 27, 2018 - link

    I'm not crazy about their lack of coverage of 2017, but at least it was uniform in that there was no coverage of anything at all (which is preferable, in my mind, to selective coverage of some products over others.)

    They've obviously decided to make a clean break and start over in 2018. With Andrei at the helm I think we have a lot to look forward to! I feel very encouraged by the even-handed tone and technical depth of these two S9 articles so far.
  • darkich - Monday, February 26, 2018 - link

    Andrei, you contradict yourself when talking about GPU. For years you've been repeating how it all should be about sustained performance and now you're making conclusions based on the very thing that according to you should be ignored - peak values.
  • Andrei Frumusanu - Monday, February 26, 2018 - link

    I can't test sustained performance on these devices so the option is to either post peak or to post nothing at all. It's a no-win situation.
  • jjj - Monday, February 26, 2018 - link

    Without testing in actual games, it is difficult to be certain how representative the load is and if the perf meets the users needs. Synthetic peak or sustained are not the metrics we need.
  • lilmoe - Monday, February 26, 2018 - link

    +1
    Been saying that all along. What matters is ACTUAL real world workloads.
  • GC2:CS - Monday, February 26, 2018 - link

    While not a definitive verdict, this seems suspicious.

    It feels like Samsung made the M3 core more for bragging in benchmarks than for real world use ?
    They pretty much promised to match Apple with those +100/40% numbers, but in order to pull out that great single core number, they went nuts on power ( Up to 3 W single core really ? And people calling intel core a hog ?), then they didn’t backed up from those core wars so they put 4 of them in there... to cut the clock speed by a third. And now those devices might be forced to use those little cores in most of the real life tasks ? Seems like a disaster to me.

    And oh they went lazy on GPU scaling too to work on M3 ? How much more mili meters went that way ?

    Nah can’t help but snapdragon feels like a more balanced platform to me.
    And people are angry there is no exynos in their country.
    Hope it gets better as more and more real world info shows up.

    Kudos to Anandtech for digging in so soon, while other sites might go on to report about unrivaled exynos “domination” for months.
  • Andrei Frumusanu - Monday, February 26, 2018 - link

    They didn't go nuts on power, Apple also has ~3-3.5W ST platform active power and more in FP workloads. The perf doesn't seem to be an issue of the core but of the software.
  • krumme - Monday, February 26, 2018 - link

    And by software we mean the wonders of the marketing department
  • thunng8 - Monday, February 26, 2018 - link

    On what cpu? I’d love a detailed review of the various Apple cpus.

    From my experience and from a few not so detailed reviews, the a10x uses much less power than the a9x. I can see the battery percentage move when doing a large batch export of RAW files on the a9x and the same operation on the a10x, the percentage barely moves.
  • Andrei Frumusanu - Monday, February 26, 2018 - link

    A10 and A11.
  • icalic - Wednesday, February 28, 2018 - link

    what about power consumption of high efficiency core a10 and a11 andrei, presentation apple said 1/5 power of high performance core, is it valid?
  • thunng8 - Thursday, March 1, 2018 - link

    I find it hard to believe that the A11 did not improve on the A10 since there was a die shrink and peak frequency was about the same. I would love to see a deep dive that shows your findings.
  • jvl - Tuesday, February 27, 2018 - link

    > And people are angry there is no exynos in their country.

    ... when it should be the other way around. I'd be pissed about Exynos-only in the EU, because Sammy has historically been very "hesitant" to open source drivers etc. - so for the EU galaxy models, few -if any- custom ROMs are available.

    I would only be pissed however, if I hadn't bought a Pixel 2 XL :D
  • Wardrive86 - Monday, February 26, 2018 - link

    The cache structure of the efficiency cores is certainly interesting, seems Qualcomm's implementation would have the upper hand here SD845: 4 x 128 KB L2 + 2 MB L3 + system cache vs 9810: no L2 ,512 KB L3, no system cache/L4? Though I see you said DRAM latency is exceptional so maybe that helps offset a bit. I wonder if this would push more workloads onto the big cores. Doesn't ARMs' in order cores increase in performance by 5%? with each doubling of cache?
  • Wardrive86 - Monday, February 26, 2018 - link

    I also wonder what potential performance outcomes will result from separate L3 caches/separate DynamIQ clusters (9810) vs Unified L3 cache/system cache and single DynamIQ cluster (845). Also what the quad core frequency is for the Kryo 385 Gold cores. I look forward to your detailed analysis in the future. Thanks again for this excellent article btw
  • Raqia - Tuesday, February 27, 2018 - link

    I would expect DRAM latency to be low given a simpler cache hierarchy between the DRAM and the smaller cores; there are common workloads with non-random access patterns where this less granular hierarchy could be a detriment. It sounds like they went scorched earth on the single threaded (Geekbench) performance to the detriment of many other subsystems and performance metrics. It could be software governor issues, but I suspect results are likely to be consistent with the SD vs Exynos variants of the S8 where end user performance of applications actually using real Android APIs on the Exynos tend to lag.
  • Wardrive86 - Monday, February 26, 2018 - link

    Phonearena got a Geekbench single core score of 3100 on the 845 varianthttps://www.phonearena.com/news/Samsung-Galaxy-S9-...
  • jjj - Monday, February 26, 2018 - link

    shows how competent that site is lol
    https://browser.geekbench.com/v4/cpu/search?utf8=%...
  • Tigran - Tuesday, February 27, 2018 - link

    >>>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<<<
    Andrei, do you have similar figures on Qualcomm chips (e.g. SD 845 or SD 835)?
  • Andrei Frumusanu - Tuesday, February 27, 2018 - link

    1.0-1.1W for both the 835 and 845. Of course the later needs more testing.
  • ladyanita22 - Tuesday, February 27, 2018 - link

    Those are some pretty nice numbers. I really don't understand the hatred against Qualcomm. It's true their chips don't have the greatest raw power, but they're extremely efficient and provide excellent performance at very low power requirement levels.

    Of course, I'd love Android to have powerful chipsets, but it doesn't make much sense if it can't really sustain its performance, and the Pixel proves you don't really need an extremely powerful CPU to have the smoothest experience.
  • tombombadil - Tuesday, February 27, 2018 - link

    Could it be that the cores were throttled? That would explain the first few tests were amazing, but subsequent ones were sub par due to the throttling.
  • Andrei Frumusanu - Tuesday, February 27, 2018 - link

    They weren't.
  • Javert89 - Tuesday, February 27, 2018 - link

    Did some rough calculations (based on past cores power curves, and core scores observed 'til now) and M3 (10nmLPE) seems to particularly suffer in comparison to A75 (10nmLPE) at low-mid frequencies. At SD801 peak performance level A75 seemingly is going to consume 14% less power than M3; at SD810 peak performance level A75 seemingly is going to consume 7% less power than M3; at SD835 peak performance level A75 remains more efficient by 6% in comparison to M3; for heavy workloads, M3 becomes more efficient. M3 at SD845 peak performance level could consume 12% less power than A75. The difference seems to not be so big, in my opinion, to justify a 3-cluster a.k.a. Mediatek configuration (e.g. 3 A55 + 3 A75 + 2 M3).
  • Lolimaster - Tuesday, February 27, 2018 - link

    845 is consistent with what it's supposed to be, no lies.

    For now, Samsung created a Geekbench bot.
  • Lolimaster - Tuesday, February 27, 2018 - link

    When will get an 845 Galaxy Tab S4 3000x2000 3:2
  • watzupken - Friday, March 2, 2018 - link

    Could it be due to poor optimization? I feel this is the first time someone decides to go down the path and make a big change to the ARM SOC. Usually every SOC makers, Qualcomm, Samsung, Huawei, Nvidia, etc, makes small tweaks to the reference ARM SOC and brand it Kyro, Exynos for example. With Samsung going wide with their SOC, it is quite a significant departure from the norm. Unlike Apple that have the advantage of integration of both hard and software which allow them to optimize very well, Samsung is at the mercy of Google since they are just using the OS that they have provided. I could be wrong, and this is just my guess.
  • Wardrive86 - Friday, March 2, 2018 - link

    I wouldn't think so. Most likely a scheduler issue. Going wide just means that they can extract more instruction level parallelism, and since core count hasn't went down it would seem that the ability of the Exynos to extract thread level parallelism hasn't been hurt in any way. Android and it's programs are generally well threaded. We may see TLP suffer if these big cores can't stay performant for long. Best to wait and see in shipping devices, Samsung may blow our socks off, or they may let us down.
  • NICOXIS - Monday, March 5, 2018 - link

    When can we get a commenting system that isn't from the 90s?
  • babadivad - Wednesday, March 7, 2018 - link

    Whoa, easy buddy. That's crazy talk.
  • anaidioschrono - Friday, March 16, 2018 - link

    I'd love an update to this. I'm definitely holding off on my purchase to see if it's worth the extra hassle of getting the Exynos version in the States.
  • Geir - Friday, March 16, 2018 - link

    Now galaxy s9 and s9+ ara available so please benchmark again please :)
  • bjelsmore - Tuesday, March 27, 2018 - link

    More than a month later and the S9 & S9+ has launched and you haven't re-benchmarked them??
  • khakstry - Saturday, July 20, 2019 - link

    It was very awesome thanks
    <a href="http://http://www.baranchatt.ir/">چت روم</a>
  • khakstry - Saturday, July 20, 2019 - link

    Very good
    Http://www.baranchatt.ir
  • khakstry - Saturday, October 5, 2019 - link

    Thank you.
    Please visit my site:
    <a href="http://www.ghoghnooschatt.com" rel="nofollow">چت روم</a>
    <a href="http://www.ghoghnooschatt.com" rel="nofollow">ققنوس چت</a>
    <a href="http://www.ghoghnooschatt.ir" rel="nofollow">چت روم</a>
    <a href="http://www.baranchatt.ir" rel="nofollow">باران چت</a>
    <a href="http://www.ghoghnooschatt.ir" rel="nofollow">ققنوس چت</a>

Log in

Don't have an account? Sign up now