it is not a vague point when more than half of their Notes had a new SOC inside. And I gave you 3 more reasons. If they have 8nm, UFS 3.0 or LPDDR5 ready for 2H 2018 they will very likely introduce another SOC.
it is not a vague point when more than half of their Notes had a new SOC inside. And I gave you 3 more reasons. If they have 8nm, UFS 3.0 or LPDDR5 ready for 2H 2018 they will very likely introduce another SOC.That's a pretty vague way to make your point. Instead of looking 7 generations independently of each other, look at the past three years (which is clearly the most important cases, due to them being closer to the present). The past 3 years, the Note devices have used the same SoC as the S ones. Note 8 and S8, Note 7 and S7, Note 5 and S6...
The Adreno 540 makes the G71 and G72 look pretty bad.The Mali-G71 (E8895) is either as good or better than the Adreno 540 (SD835).
If you live in la-la-land.The A73 brought hardly any performance improvements to the table, over the A72.
This year.What is to say that their next architecture, which for all we know can come in 2020
See la-la-land above.won't do the same?
it is not a vague point when more than half of their Notes had a new SOC inside.
If you live in la-la-land.
This year.
.
Since you are so affirmative by claiming ARM won't "introduce the successor to A75 soon, which again will increase IPC 20-25% over A75", can you tell what your sources are?
Except I never claimed that ARM won't do it, I just contradicted his statement that they would. The actual truth is that we don't know.
Do not speak for others. The actual truth is YOU don't know. So you did not contradict my statement at all.
In addition you did two assumptions: 1) It could be 2020 2) there could be neglible performance increase.
Oh wow you're a lost cause. You didn't even read what I linked. You're exactly the kind of person who I describe as adding to the problem of GPU benchmarks in that article.That's a terrible comparison, as the G72 in the Kirin processor performs even wors than the G71, which it shouldn't. There should be a 20% gain. Or do you take ARM's supposed claims of performance improvement only when it fits you?
For a proper comparison, look at Galaxy Note 8 with E8995 vs Note 8 with SD835, like here:
https://www.gsmarena.com/samsung_galaxy_note8-review-1659p8.php#
G71 performs as good or better than Adreno 540. The G72 is supposed to perform 20% better than this, and 10% below Adreno 630.
WHY shouldn't it? Because G72 is 1 higher than G71 thus always better, RIGHT?That's a terrible comparison, as the G72 in the Kirin processor performs even wors than the G71, which it shouldn't.
Thank you for quoting me on the article that I wrote. Let's move on from marketing claims to tested performance.Cleary you're the one doing living in la-la-land: https://www.anandtech.com/show/10347/arm-cortex-a73-artemis-unveiled/3:
"the A73 is claimed to be up to 10% better performance than the A72 – on the same process and frequency."
That article was using SPEC2000 and GeekBench, both which are not very memory heavy. I come back at your hand-picked chart comment and point out that the K960 surpassed the K950 by 23-29% in some benchmarks: https://www.anandtech.com/show/10871/the-huawei-mate-9-review/7 PCMark in particular is very memory subsystem sensitive and this is where the A73 had huge upgrades.Now, let's look at an actual comparison between SoCs with A72 and A73, like the Kirin 950 and 960:
https://www.anandtech.com/show/11088/hisilicon-kirin-960-performance-and-power/2
"After accounting for the differences in CPU frequency, floating-point IPC for the Kirin 960’s A73 is 3% to 5% lower overall than the A72".
Let me repost that chart again, please put on your glasses before you look at it this time.I mean, even your own hand-picked chart proves that there's little to no performance improvement from A72 to A73.
If you actually follow what's going on then of course we know. The next ARM micro-architecture is the next big one from the Austin team. Expect a ~25% IPC increase over the A75 at similar high frequencies. It's coming out this year.The actual truth is that we don't know.
As for my assumptions, you say it yourself. I very clearly put the word "could" in them, and never hid the fact of my suppositions; I never stated anything as an absolute fact -- like you did.
Underwhelming... I want to see ARM design an ultra wide core like M3 and apple...keep A75 as intermediate perf core, even if you had say 1-2 A80?, 2- A75 and 4 low clocked A55, on 3 power planes that would be insanely efficient and fast all-round.Qualcomm 845
Gb st 2378
Unknown device
https://m.gsmarena.com/mysterious_snapdragon_845_device_shows_up_on_geekbench_4-news-29538.php
If you actually follow what's going on then of course we know. The next ARM micro-architecture is the next big one from the Austin team. Expect a ~25% IPC increase over the A75 at similar high frequencies. It's coming out this year.
Might depend on the particular use-case. However when compared to lets say a high performance desktop you will notice, that CPU performance is off by factor 3 while GPU performance is off factor 10 while at the same time screen resolution on high-end smartphones is only lower by factor around 2. This discrepancy is even more skewed if you take consoles as reference, which already have lower CPU performance. That should tell you, that GPU performance in current devices is the most limiting factor. In summary, higher single threaded CPU performance has diminishing returns while GPU performance has not. So when firing up a game CPU performance will gain you nothing while you see immediate benefits of higher GPU performance.
To be clear i am talking about user-experience and do not just go by numbers, which would indicate that 70%>10%.
Nothing wrong with throwing around assumptions before the fact was stated. However you started speculating after my statement. Do you see the problem?
I don't care what you think and I'm not going to tolerate such personal attacks.I'm sorry, but I have a hard time taking you seriously in anything, anymore. You clearly have a bias for Qualcomm and their devices, despite how much they lack in any kind of innovation for their products (I could say their GPU is impressive, but hell, even Apple are exceeding them here in overall performance, while maintaining an overall equally efficient SoC).
You recent article "The Snapdragon 845 Performance Preview" substantiates my opinions. 2 months ago, in your article about Snapdragon 845's launch, you made the ridiculous assertion that Snapdragon 845 would bring "overall expected performance improvement to 39-52%", based on your estimate of the A75 cores. You said this despite Qaulcomm very clearly stating in their launch "up to 25% performance improvement"; despite them repeating this several times even on their own website article about their own products. If the SD845 with the A75 cores produced ~25% performance improvement, according to the semiconductor company themselves, it was awfully strange of you to claim a ~45% performance improvement. But you still did.
I very clearly explain how I got to those estimates by factoring both the CPU frequency increases and the individual performance increases projected by ARM. Remember that Qualcomm published nothing besides that single 25-30% figure and they did not source on what benchmark or workloads those figures were based on. There's a wide swing between benchmarks and you can see it today's article, GB4 integer gained only 31% while FP came in at 45% - the browser benchmarks such as WebXPRT came as well on the higher end at 44%. It's a matter of fact that I remained sceptical about Qualcomm's claims as it would have meant that the A75 comes under the projected performance figures - and today we saw confirmation that indeed that's what happened and i very clearly express that in the article.The Kryo 385 gold/performance cluster runs at up to 2.8GHz, which is a 14% frequency increase over the 2.45GHz of the Snapdragon 835's CPU core. But we also have to remember that given that the new CPU cores are likely based on A75's we should be expecting IPC gains of up to 22-34% based on use-cases, bringing the overall expected performance improvement to 39-52%. Qualcomm promises a 25-30% increase which is at the low-end of ARM's projections.
For those of you who want the TL;DR on what the poster above is trying to accuse @Andrei. , here are the facts.You recent article "The Snapdragon 845 Performance Preview" substantiates my opinions. 2 months ago, in your article about Snapdragon 845's launch, you made the ridiculous assertion that Snapdragon 845 would bring "overall expected performance improvement to 39-52%", based on your estimate of the A75 cores. You said this despite Qaulcomm very clearly stating in their launch "up to 25% performance improvement"; despite them repeating this several times even on their own website article about their own products. If the SD845 with the A75 cores produced ~25% performance improvement, according to the semiconductor company themselves, it was awfully strange of you to claim a ~45% performance improvement. But you still did.
As you can see not only did he clearly point out Qualcomm's perf estimates, but he also explained why a more positive outcome was possible. The information was clear and to the point, with both conservative and optimistic numbers to pick from.The Kryo 385 gold/performance cluster runs at up to 2.8GHz, which is a 14% frequency increase over the 2.45GHz of the Snapdragon 835's CPU core. But we also have to remember that given that the new CPU cores are likely based on A75's we should be expecting IPC gains of up to 22-34% based on use-cases, bringing the overall expected performance improvement to 39-52%. Qualcomm promises a 25-30% increase which is at the low-end of ARM's projections.
I don't care what you think and I'm not going to tolerate such personal attacks.
In the Snapdragon 845 announcement article, and let me quote the whole paragraph, state both my analysis as well as Qualcomm's targets:I very clearly explain how I got to those estimates by factoring both the CPU frequency increases and the individual performance increases projected by ARM. Remember that Qualcomm published nothing besides that single 25-30% figure and they did not source on what benchmark or workloads those figures were based on. There's a wide swing between benchmarks and you can see it today's article, GB4 integer gained only 31% while FP came in at 45% - the browser benchmarks such as WebXPRT came as well on the higher end at 44%. It's a matter of fact that I remained sceptical about Qualcomm's claims as it would have meant that the A75 comes under the projected performance figures - and today we saw confirmation that indeed that's what happened and i very clearly express that in the article.
If you simply to prefer to read marketing materials and take companies at their own word instead of having added independent analysis and trying to understand why performance and technology as it is I recommend simply reading press releases instead of my articles.
As you can see , the performance uplift range is quite variable, and while the average ends up bellow expectations, one can easily see why Andrei chose to be a bit more optimistic in the original article.
- some tests show a perf uplift over previous gen of 8-20%, with the lower 8-14% numbers being associated with memory heavy tasks, 17%-20% being PcMark Web and Performance scores
- other tests show a perf advantage of 37-44%, these are WebView and Photo Editing tests.
Who are you to decide what overall performance means and how that is calculated into a single figure? I specifically do emphasis on sub-scores and per-test analysis, and my estimates at the launch were simply based on what ARM said because we had no further information from Qualcomm at the time.Qualcomm said "up to 25% performance increase", not "up to 25% IPC increase". Claiming you didn't understand what Qualcomm meant is a bad argument, seeing as all you have to do is look at previous Qualcomm slides and launches from products like SD820, 810, 801, etc. By performance increase they mean the actual overall performance, not the IPC. On their own site, they very clearly rated the SD845 at "up to 2.8 GHz", and also very clearly said the performance increase would be ~25%.
Who are you to decide what overall performance means and how that is calculated into a single figure?
my estimates at the launch were simply based on what ARM said because we had no further information from Qualcomm at the time.
.
Also I'm sorry but you're just lunatic here - if we don't use marketing materials when we might as well pack up and simply ignore all announcements of unavailable products.
HAHAHAHAAHA ok buddy if that's your way of doing overall averages of CPU and devices then so be it! You're completely right.I am merely looking at the overall performance summary form your own test, Andrei. A limited test, for sure; but in any case of bias, it would be biased -- knowing you -- in favor for of the SD845:
PCMark Work. 2.0 overall Performance: 17% performance increase
WEBXRPT 2015 - OS WebView: 44% performance increase
Speedometer 2.0 - OS Webview: 37% performance increase
Overall: 33% performance improvement.
Here's a hint master genius: None of those are actually direct CPU benchmarks. You're taking an average of averages with no weighing into some stupid number. I know this is very hard to comprehend that the improvements are not linear across the board and there are many implications and factors at play, but maybe some day or somebody here will help you with it. I'm done here.Ok, let's do it the "Andrei way", then.
PCMark Work. 2.0 overall Performance: 17% performance increase
WEBXRPT 2015 - OS WebView: 44% performance increase
Speedometer 2.0 - OS Webview: 37% performance increase
Average is around 33%. But to hell with that. Instead we will look at the highest possible number and claim the overall CPU performance to be that. So Snapdragon performs 44% over Snapdragon 835.
Here's a hint master genius: None of those are actually direct CPU benchmarks. You're taking an average of averages with no weighing, I know this is very hard to comprehend that the improvements are not linear across the board and there are many implications and factors at play, but maybe some day or somebody here will help you with it. I'm done here.
It should be all cores, the power is more or less the same as the 835.@Andrei. Setting this nonsense apart... Did you confirm if the SD845 can achieve 2'8GHz in all cores or just one ?