Hey guys I just currently pre-ordered the Pixel and am a little worried about the benchmarks that have been released. Do you guys think these are accurate? On some of the articles I have read the clock speeds they are claiming it is running are the speeds of the 820 not the 821. I mean the 6p scored higher on benchmarks than the pixel. How can these be right with the newest processor?
Look at the hands on videos. You won't be worried about performance after that. Looks like Google has done a lot of optimization. Benchmarks don't tell the whole story.
Well, seeing as the 821 is to an 820 the same as an 801 is to an 800... i.e., its the same damned chip, not really sure why you would expect there to be a dramatic performance change?
The 821 shows a peak cpu frequency spec a bit higher than 820, but this doesn't mean that everyone who uses it is obligated to use the highest frequency.
So here is a little bit of information about CPU manufacturing;
Every CPU core is a little bit different. Some of them are stable at lower voltages and higher frequencies than others. The CPU specification indicates a MINIMUM frequency that it MUST be stable at while operating within the designed power envelope. In other words, another CPU may be able to operate at the higher frequency, but it won't do so within the designed power envelope -- it will require OVER VOLTING.
The CPUs are separated according to their levels of stability. Call that "binning". One of these CPUs that bins poorly might be called a Snapdragon 820, and one that bins well will be called a Snapdragon 821. Within each model name, there are further levels of distinction that are used to set the baseline voltages being applied, in order to minimize the voltage that they are fed, such that you can reduce the power consumption as much as possible.
So you can think of an underclocked Snapdragon 821 as a SUPER DUPER AWESOME binned Snapdragon 820, operating at a lower voltage, and therefore consuming less power.
Don't worry about benchmarks! What it matters is the SoC you have, how well disipated is the SoC, and most important, how the software is done (kernel, drivers, android, binaries, etc).
There could be many devices with same SoC and better scores, but at the end, they lag more etc.
For instance, my previous Z5 Compact (with Sony Android, which is similar to AOSP) and a much better SoC than my current N5X, imo lags more than my current Nexus 5X with a worse SoC.
There's no way you can choose a device based on the benchmark, you must try both devices by yourself (ideally with your apps) and see the difference.
Giving another example...A Nexus 5 2013, is extremely fast in KK (with ART) and even in MM (but not in Lollipop).
However, it still throttles much more than a 5X because of the frequency, nm, and many other things.
doitright said:
Well, seeing as the 821 is to an 820 the same as an 801 is to an 800... i.e., its the same damned chip, not really sure why you would expect there to be a dramatic performance change?
The CPUs are separated according to their levels of stability. Call that "binning". One of these CPUs that bins poorly might be called a Snapdragon 820, and one that bins well will be called a Snapdragon 821. Within each model name, there are further levels of distinction that are used to set the baseline voltages being applied, in order to minimize the voltage that they are fed, such that you can reduce the power consumption as much as possible.
So you can think of an underclocked Snapdragon 821 as a SUPER DUPER AWESOME binned Snapdragon 820, operating at a lower voltage, and therefore consuming less power.
Click to expand...
Click to collapse
There actually are some differences in the 821 vs the 820. It's not the same chip exactly. A pretty great breakdown is here: https://www.gizmotimes.com/comparison/snapdragon-821-vs-snapdragon-820/16403
But essentially, slightly better power savings, improved camera performance, and a VR SDK.
Thanks for all the replies guys. I was just confused as to why a chip the snapdragon says should have a 10% increase in performance over the 820 is benchmarking lower than most 820's.
Good info, thanks guys!
We know nothing yet, time will tell obviously. The videos in the early previews look great, but we'll see under heavy load how these perform.
jbrooks58 said:
Hey guys I just currently pre-ordered the Pixel and am a little worried about the benchmarks that have been released. Do you guys think these are accurate? On some of the articles I have read the clock speeds they are claiming it is running are the speeds of the 820 not the 821. I mean the 6p scored higher on benchmarks than the pixel. How can these be right with the newest processor?
Click to expand...
Click to collapse
I'd like it if you could actually find something that claims that the 6p is anywhere near pixel in performance benchmarks. Reality is that it is more than 2x faster across the board.
As far as comparing it with 820, there are two things you can accomplish with the "1" -- more speed, or less power. They seem to be opting for the latter.
All the benchmarks I could find show it against either apple, or samsuck. Samsuck is well known for building TO the benchmarks (sometimes even *cheating*), which causes their scores to be unnaturally high, and comparing against apple is just stupid, since there is no baseline between them due to architectural differences and a complete lack of a common software stack. In other words, in a comparison between pixel and anything made by apple, you could have a smaller number, despite *actually* being considerably higher. The number doesn't equate across platforms.
---------- Post added at 08:18 PM ---------- Previous post was at 08:09 PM ----------
jbrooks58 said:
Thanks for all the replies guys. I was just confused as to why a chip the snapdragon says should have a 10% increase in performance over the 820 is benchmarking lower than most 820's.
Click to expand...
Click to collapse
That 10% is an interesting figure.
The SD820 has clock rates of 2.15 GHz on 2 cores, and 1.59 GHz on the other 2 cores.
Multiply by 1.1 (add 10%) and you get 2.365 and 1.749 GHz.
The SD821 has clock rates of 2.34 GHz on 2 cores and 2.19 GHz on the other 2 cores.
On those first two cores, that is marginally more the 10% higher clock rate. On the other 2 cores, it is considerably more than 10%. Note that a system's performance does NOT scale linearly with CPU frequency.
The other thing to note is that the pixel specs show it operating at 2x2.15+2x1.6 GHz, just like the SD820.
So what we can read from that, is that the pixel's CPUs are **underclocked**. That will allow it to use less battery power, and run cooler, while still running *really really fast*. If you want more, unlock and clock it up to 821 spec, I think you will find that this phone is an "overclocker's" dream, even if it isn't really overclocking.
That 10% figure comes directly from Qualcomm's publications on performance for the 821 vs 820.
craig0r said:
There actually are some differences in the 821 vs the 820. It's not the same chip exactly. A pretty great breakdown is here: https://www.gizmotimes.com/comparison/snapdragon-821-vs-snapdragon-820/16403
But essentially, slightly better power savings, improved camera performance, and a VR SDK.
Click to expand...
Click to collapse
Good read, thanks.
Related
It seems like every review I read tells me the Hummingbird processor is much better and a more capable processor then the snapdragon. If this is true why hasn't HTC produced a phone with a Hummingbird CPU? HTC is way ahead in the android game experience wise but they haven't seemed to make any major hardware changes other then upping the rom and ram and speed of the cpu. Would someone please tell me that samsungs new CPU isn't as good as all the reviews say it is
I can't comment on the CPUs but I would just like to mention that costs, failure rate etc. all play a role in the decision about what CPU to use. There may be reasons for HTC not using the Hummingbird yet other then performance.
The Hummingbird is a little faster CPU-wise, which does surprise me as the Snapdragon has higher quoted instructions per clock. However, its GPU is really in a different league to that in the current Snapdragons and that's where the big difference lies in benchmarks (though that will, of course, only help GPU-accelerated things).
HTC don't necessarily use the chip that benchmarks the best on any given day - they have a relationship with Qualcomm and presumably get preferential rates from them, and they have an established platform. The good news for them is that Qualcomm are bringing out new Snapdragons with higher clockspeeds and better GPUs (and dual cores, though it remains to be seen if/when they arrive on phones) so they should equalise and quite probably turn the tables soon.
Competition is good - we were stuck with 5-600MHz XScales and ARM11s for ages, and now things are finally moving along.
So we all know the Nexus S has a 1Ghz Cortex A8 Hummingbird CPU, which sounds unimpressive considering the Nexus One has a 1Ghz Snapdragon QSD 8250, but it's a known fact that clock speed often has little to do with actual computational power. Qualitative previews have said that the Nexus S "flies," but I'd like to see something more in the numbers. If anyone has a demo device, could you run a few benchmarks? Or perhaps comment on performance after quick opening/closing several computationally intensive applications?
QuacoreZX said:
So we all know the Nexus S has a 1Ghz Cortex A8 Hummingbird CPU, which sounds unimpressive considering the Nexus One has a 1Ghz Snapdragon QSD 8250, but it's a known fact that clock speed often has little to do with actual computational power. Qualitative previews have said that the Nexus S "flies," but I'd like to see something more in the numbers. If anyone has a demo device, could you run a few benchmarks? Or perhaps comment on performance after quick opening/closing several computationally intensive applications?
Click to expand...
Click to collapse
1gb humingbird is fast as galaxy S and iphone 4. both which are like 30% or more faster then snapdragon
I think the key improvement is in graphics performance. Here is a comparison.
QuacoreZX said:
So we all know the Nexus S has a 1Ghz Cortex A8 Hummingbird CPU, which sounds unimpressive considering the Nexus One has a 1Ghz Snapdragon QSD 8250, but it's a known fact that clock speed often has little to do with actual computational power. Qualitative previews have said that the Nexus S "flies," but I'd like to see something more in the numbers. If anyone has a demo device, could you run a few benchmarks? Or perhaps comment on performance after quick opening/closing several computationally intensive applications?
Click to expand...
Click to collapse
The S5PC11x (Hummingbird) has 2x the memory bandwidth of the MSM8250.
The MSM8250 gets about 2x the floating point performance of the S5PC11x.
I believe the SGX540 GPU in S5PC11x is on the whole a bit faster than the GPU in the 8250, but I don't have hard numbers on that in front of me. They're architecturally different GPUs and will have different strengths and weaknesses.
It's really hard to do a good apples to apples comparison of different SoCs -- memory interconnect, cache sizes, ARM architecture version, GPU, etc, etc all play into overall system performance.
Gingerbread, overall, tends to be faster than Froyo on the same hardware.
Not really too familiar with this stuff, but will the JIT compiler being optimized for snapdragon instruction set make a huge difference still? My Vibrant plays games way better than the MT4G (imo) but scores terribly on Linpack and is terribly slow at opening applications and things vs. the MT4G.
Read the post above you. Linpack is mainly a benchmark for numerical performance(floating point etc), where the Snapdragon chips are MUCH better.
But the Hummingbird(PowerVR) GPU is better than the Adreno GPU found in the Snapdragon line. That's why the gaming performance of your Vibrant is better than the MT4G.
Ronaldo_9 said:
1gb humingbird is fast as galaxy S and iphone 4. both which are like 30% or more faster then snapdragon
Click to expand...
Click to collapse
PhoenixFx said:
I think the key improvement is in graphics performance. Here is a comparison.
Click to expand...
Click to collapse
Yup, just anecdotally, hummingbird is MUCH faster than snapdragon IMHO
galaxyS/NS SGX540= 90 million triangles/sec
HTC G2 Adreno 205 =44 million triangles/sec
Nexus one = Adreno 200 = 22 million triangles/sec
nexus S is running on the fastest GPU out now. And another good thing about running on power VR GPU is that iphone runs on one also so when lazy iphone porting happens you will have optimal performance running on that GPU than you would on Adreno
Ive noticed this especially on gameloft games
Trust me im on a vibrant and came from nexus one with out a doubt the nexus S GPU smokes nexus one GPU even out performance 2nd gen snapdragon
Hummingbird > all atm.
Orion will be the same.
Don't make pre-assumptions about the dual core chips.. Orion has good competition from the TI OMAPS line.. Qualcomm looks like they'll stay behind GPU wise though.
Plus the Sound Quality of the Hummingbird chip is awesome. MUCH better than the Snapdragon chips.
Also, you have to be cautious of manufacturer specs for GPU pixels/sec and triangles/sec -- the "box numbers" are always under optimal conditions and often not representative of real workloads.
For modern non-fixed-pipe GPUs (gl ES 2.x, etc) compute capabilities (how many shader ops / pixel/ etc you can get away with) factor in as well.
Depending on what your workload is like (geometry heavy? fill heavy? texture heavy? shader heavy?) you will see different strengths and weaknesses when comparing GPUs.
All that said, the SGX540 is indeed quite snappy.
chip
I agree the sound chip is good in the NS, as is the GPU
http://www.engadget.com/2013/05/06/samsung-galaxy-s4-octacore-review/
The A15 in the octa-core isn't any better or worse than the krait in the snapdragon, either in performance or benchmarks. The extra four helper cores you get doesn't improve battery life. In fact, with the four a15 cores, it is actually WORSE than the more efficient krait cores. All those people bragging about how much the A15 is better than the krait based on thin air speculation really need to shove it up their butt.
It was already expected that the four extra low power cores in an octa core would not make much difference in battery life and actually be worse off than a regular quad-core processor. History has already shown with the Tegra 3's helper core that utilizing low power helper cores is a tricky and inefficient affair. It's not easy to switch between them, to prioritize when to use what, and instead of making a more efficient A15 design, Samsung relied too much on the chip's switching capabilities instead of making an overall better processor.
So if you want LTE, BETTER BATTERY LIFE, rom compatibility and dev support with the most widespread SoC, actual availability in stores everywhere, then stop waiting or worrying about the Exynos octa-core and pick the widely available snapdragon version of the S4. Anyone still spouting how great the octa-core version will be and still lies about it being EIGHT WHOLE CORES! when there's really only four are the biggest trolls in the S4 forum.
this video show more real compasion
http://www.youtube.com/watch?v=Wt5im3WAZYc
http://forum.xda-developers.com/showthread.php?t=2258519
Heh, load of bullcrap.
If you do video battery tests you do them on Wi-Fi, congratulations on testing modem battery life. (Same carrier? Same tower? Same time of day? Bravo on apples vs oranges Engadget)
The benchmarks ARE faster on the 9500. Let's not mention that Engadget are incompetent fools who don't understand benchmarking. The Linpack scores are a joke as is CF-bench, one because the benchmark literally takes 0.3s and if you're idling before you press the start button that's not even though time for the CPU to ramp up to highest speed. CF-bench fails due to thermal throttling. At least that's a valid negative point, but not performance, scores are far beyond the 30k mark. I'm also getting funny more realistic results on the other benches: 661ms vs 732 SunSpider, 10% higher Vellamo score, 300 more 3D rating, and I'm sure there's others. Funny how they suddenly don't use GeekBench.
Matter of fact: the 9500 is undoubtedly faster and that's a technical reality. They even state so in their subjective comparison.
As for battery life: I've already mentioned how the early firmware is unfinished. I'm getting roughly 10% per hour usage; right now at 61% and 3h30 screen, and that's with doing benchmarks for the last half hour which ate 9%.
The only correct *video* battery life tests I've seen came from GSMArena (9505) and Russian mobile-review who got 12h.
You're going to have to wait for AnandTech to do a review in a few months to be able to use it as argumental material in such discussions.
Engadget is the pinnacle of ignorance and technical non-reporting, and as they've proved in their botched S3 review last year, the benchmarking seems to be done by the principal of their local baboon academy.
katamari201 said:
All those people bragging about how much the A15 is better than the krait based on thin air speculation really need to shove it up their butt.
Click to expand...
Click to collapse
I don't think people thought that devices using Octa vs. S-600 wouldn't perform comparably. It would make no sense for Samsung not to optimize both to the best of their ability and S-600 is a powerful and efficient chip. You can't help but get the feeling that the s/w isn't "done" on either version based on comments on all the SGS4 forums about lag and display driver issues. That's echoed by some of AndreiLux's comments from looking at the code. Octa is literally the first public implementation of big.LITTLE. All of ARM’s future designs will be based on it. That means Qualcomm’s next generation of chips after S-800 will be based on it also as they license ARM’s designs. I'd expect over time that updates will continue to improve Octa's performance (power and efficiency) whereas S-600 is simply a massaged version of S4 Pro and the OEMs have a lot more experience working with it so there's less upside potential. I'd still buy the i9500 over the i9505 if I were going to get a SGS4 (I'm waiting for the N3) as I think its long-term potential is greater than S-600 and, going forward, I'd expect it to be used in more Samsung devices once Qualcomm's RF360 universal LTE baseband becomes available. Once that happens, unless there's production capacity issues, there no reason Octa wouldn't be Samsung go-to high-end chip. Just my opinion of course.
P.S. - The i9500 has about 250MB more free RAM (13%) than the i9505 as Adreno reserves 500MB for itself while PowerVR reserves a little over 200MB.
matheus_sc said:
this video show more real compasion
http://www.youtube.com/watch?v=Wt5im3WAZYc
Click to expand...
Click to collapse
Funny thing is you do the same comparison on another I9505 or I9500 and it will most probably yield different results... they are too close to compare
---------- Post added at 11:42 AM ---------- Previous post was at 11:38 AM ----------
BarryH_GEG said:
P.S. - The i9500 has about 250MB more free RAM (13%) than the i9505 as Adreno reserves 500MB for itself while PowerVR reserves a little over 200MB.
Click to expand...
Click to collapse
Really? Where is the source for this? Anyone like to show free ram comparisons between both devices? I am sitting around 750mb free ram,I am Stock Rooted, I have 5 active applications open... And I still have most of the samsung bloat
BarryH_GEG said:
P.S. - The i9500 has about 250MB more free RAM (13%) than the i9505 as Adreno reserves 500MB for itself while PowerVR reserves a little over 200MB.
Click to expand...
Click to collapse
None of that memory on neither SoCs is allocated to the GPUs. Video memory is reserved on-the-fly from user-space. That unavailable memory is dedicated to camera controllers, image processors, video decoder, and a bunch of other smaller buffers.
katamari201 said:
http://www.engadget.com/2013/05/06/samsung-galaxy-s4-octacore-review/
The A15 in the octa-core isn't any better or worse than the krait in the snapdragon, either in performance or benchmarks. The extra four helper cores you get doesn't improve battery life. In fact, with the four a15 cores, it is actually WORSE than the more efficient krait cores. All those people bragging about how much the A15 is better than the krait based on thin air speculation really need to shove it up their butt.
It was already expected that the four extra low power cores in an octa core would not make much difference in battery life and actually be worse off than a regular quad-core processor. History has already shown with the Tegra 3's helper core that utilizing low power helper cores is a tricky and inefficient affair. It's not easy to switch between them, to prioritize when to use what, and instead of making a more efficient A15 design, Samsung relied too much on the chip's switching capabilities instead of making an overall better processor.
So if you want LTE, BETTER BATTERY LIFE, rom compatibility and dev support with the most widespread SoC, actual availability in stores everywhere, then stop waiting or worrying about the Exynos octa-core and pick the widely available snapdragon version of the S4. Anyone still spouting how great the octa-core version will be and still lies about it being EIGHT WHOLE CORES! when there's really only four are the biggest trolls in the S4 forum.
Click to expand...
Click to collapse
Get ornery myself some times. Trust me on this, go out, find some nice young lady, make her see God. Later when you read this and are wondering what the hell you were thinking you can apologize. Everybody wins. :good:
Well Said, AMEN
Well said Krabman.....
katamari201 said:
http://www.engadget.com/2013/05/06/samsung-galaxy-s4-octacore-review/
The A15 in the octa-core isn't any better or worse than the krait in the snapdragon, either in performance or benchmarks. The extra four helper cores you get doesn't improve battery life. In fact, with the four a15 cores, it is actually WORSE than the more efficient krait cores. All those people bragging about how much the A15 is better than the krait based on thin air speculation really need to shove it up their butt.
It was already expected that the four extra low power cores in an octa core would not make much difference in battery life and actually be worse off than a regular quad-core processor. History has already shown with the Tegra 3's helper core that utilizing low power helper cores is a tricky and inefficient affair. It's not easy to switch between them, to prioritize when to use what, and instead of making a more efficient A15 design, Samsung relied too much on the chip's switching capabilities instead of making an overall better processor.
So if you want LTE, BETTER BATTERY LIFE, rom compatibility and dev support with the most widespread SoC, actual availability in stores everywhere, then stop waiting or worrying about the Exynos octa-core and pick the widely available snapdragon version of the S4. Anyone still spouting how great the octa-core version will be and still lies about it being EIGHT WHOLE CORES! when there's really only four are the biggest trolls in the S4 forum.
Click to expand...
Click to collapse
AndreiLux said:
None of that memory on neither SoCs is allocated to the GPUs. Video memory is reserved on-the-fly from user-space. That unavailable memory is dedicated to camera controllers, image processors, video decoder, and a bunch of other smaller buffers.
Click to expand...
Click to collapse
People were saying the i9505 has 1.5GB of available RAM while the i9500 has 1.8GB. Is that true? If it is, what's contributing to the difference?
BarryH_GEG said:
People were saying the i9505 has 1.5GB of available RAM while the i9500 has 1.8GB. Is that true? If it is, what's contributing to the difference?
Click to expand...
Click to collapse
No, it's not true. My i9505 have 1.78gb of RAM available.
BarryH_GEG said:
People were saying the i9505 has 1.5GB of available RAM while the i9500 has 1.8GB. Is that true? If it is, what's contributing to the difference?
Click to expand...
Click to collapse
loollll no ... where are ppl coming up with this nonsense
Who cares really when are we gonna use the phone at max capacity? And 3 months later something better will be out so quit your *****ing and enjoy ya phone
Sent from my GT-I9505 using xda premium
By the time Verizon ships my S4, the next super phone will be out! Seriously, I've got to quit reading these forums.
Sent from my SCH-I510 using Tapatalk 2
1.78gb ram on I9505 here
In any device even low end device you will not find the full ram visible as some is reserved exclusively of the system.
Also the post was on difference between Octa and Quall why would you expect a significant different as if that happens samsung would be trouble because they are the same device right S4 so samsung would tuned both in a way that the performance battery life is almost the same that hows it should be right you cannot say my s4 is slower then yours bec I purchased it from US ????
BarryH_GEG said:
People were saying the i9505 has 1.5GB of available RAM while the i9500 has 1.8GB. Is that true? If it is, what's contributing to the difference?
Click to expand...
Click to collapse
It really doesn't made a huge difference on Android devices. The extra memory just allows you to have more applications paused in the background before the kernel kills them to free up space. With the S3 if you're playing a game and switch to a web browser it's very likely that the game will be closed as it only has 1GB. On the S4 it will stay open in the background.
katamari201 said:
http://www.engadget.com/2013/05/06/samsung-galaxy-s4-octacore-review/
The A15 in the octa-core isn't any better or worse than the krait in the snapdragon, either in performance or benchmarks. The extra four helper cores you get doesn't improve battery life. In fact, with the four a15 cores, it is actually WORSE than the more efficient krait cores. All those people bragging about how much the A15 is better than the krait based on thin air speculation really need to shove it up their butt.
It was already expected that the four extra low power cores in an octa core would not make much difference in battery life and actually be worse off than a regular quad-core processor. History has already shown with the Tegra 3's helper core that utilizing low power helper cores is a tricky and inefficient affair. It's not easy to switch between them, to prioritize when to use what, and instead of making a more efficient A15 design, Samsung relied too much on the chip's switching capabilities instead of making an overall better processor.
So if you want LTE, BETTER BATTERY LIFE, rom compatibility and dev support with the most widespread SoC, actual availability in stores everywhere, then stop waiting or worrying about the Exynos octa-core and pick the widely available snapdragon version of the S4. Anyone still spouting how great the octa-core version will be and still lies about it being EIGHT WHOLE CORES! when there's really only four are the biggest trolls in the S4 forum.
Click to expand...
Click to collapse
Amazing! Where were you all these days??
BTW, I don't have to wait, nor do I have to worry about getting a I9500. Stores near me don't have any Snapdragon variant. I think it's always a good practise not to take advise from a random person in the forums.
Nobody asked for your advise on what version to get. People are knowledgeable enough to make that decision.
Interesting results here. Everybody has been saying the G2 is quicker and better then Note 3 and I must say I am quite shocked with these findings so far
http://thedroidguy.com/2013/09/sams...-sony-xperia-z1-vs-lg-g2-benchmark-comparison
i dont care. n3 is the better phone.
oh i dont disagree i agree 100% that is why i have a note 3 coming and im not stopping at verizon today to see the overrated g2!
hah G2 is like a on screen buttoned Galaxy S4 LG is copying Samsung on many things these days -_-
Blackwolf10 said:
hah G2 is like a on screen buttoned Galaxy S4 LG is copying Samsung on many things these days -_-
Click to expand...
Click to collapse
I know right! everything almost looks the same. Its like there are a dev and just made a rooted s4 with some new ui looks!
Here's a potential difference. There are two versions of S-800; MSM8974 and MSM8974AB. Here's AnandTech's take...
Xiaomi makes the first (to my knowledge) public disclosure of MSM8974AB, which is analogous to the changes we saw between APQ8064 and APQ8064AB. From 8974 to 8974AB, Adreno 330 GPU clocks climb from 450 MHz to 550 MHz, LPDDR3 memory interface maximum data rates go from 800 MHz to 933 MHz, and the ISP clock domain (I think Xiaomi might mean the Hexagon DSP here) goes from 320 MHz to 465 MHz. 8974 comes in both a bin with the 4 Krait 400 CPUs clocked at 2.2 GHz (really 2.15 GHz) and 2.3 GHz (2.26 GHz) with slightly different pricing, while 8974AB comes with a Krait 400 clock available only at 2.3 GHz. Process is still TSMC 28nm HPM, but I suspect that the AB variant might have the high k dielectric and/or transistor mix tuned slightly differently based on a few rumblings I've heard recently.The S-600 in the SGS4 was "AB" so the the S-800 in the N3 might be also. We'll find out when more detailed reviews start to come out.
From AnandTech discussing the SGS4's S-600 chip...
That brings us to the Galaxy S 4. It's immediately apparent that something is different here because Samsung is shipping the Snapdragon 600 at a higher frequency than any other OEM. The Krait 300 cores in SGS4 can run at up to 1.9GHz vs. 1.7GHz for everyone else. Curiously enough, 1.9GHz is the max frequency that Qualcomm mentioned when it first announced Snapdragon 600.
Samsung is obviously a very large customer, so at first glance we assumed it could simply demand a better bin of Snapdragon 600 than its lower volume competitors. Looking a bit deeper however, we see that the Galaxy S 4 uses something different entirely.
Digging through the Galaxy S 4 kernel source we see references to an APQ8064AB part. As a recap, APQ8064 was the first quad-core Krait 200 SoC with no integrated modem, more commonly referred to as Snapdragon S4 Pro. APQ8064T was supposed to be its higher clocked/Krait 300 based successor that ended up with the marketing name Snapdragon 600. APQ8064AB however is, at this point, unique to the Galaxy S 4 but still carries the Snapdragon 600 marketing name.
If we had to guess, we might be looking at an actual respin of the APQ8064 silicon in APQ8064AB. Assuming Qualcomm isn't playing any funny games here, APQ8064AB may simply be a respin capable of hitting higher frequencies. We'll have to keep a close eye on this going forward, but it's clear to me that the Galaxy S 4 is shipping with something different than everyone else who has a Snapdragon 600 at this point.
BarryH_GEG said:
Here's a potential difference. There are two versions of S-800; MSM8974 and MSM8974AB. Here's AnandTech's take...
Xiaomi makes the first (to my knowledge) public disclosure of MSM8974AB, which is analogous to the changes we saw between APQ8064 and APQ8064AB. From 8974 to 8974AB, Adreno 330 GPU clocks climb from 450 MHz to 550 MHz, LPDDR3 memory interface maximum data rates go from 800 MHz to 933 MHz, and the ISP clock domain (I think Xiaomi might mean the Hexagon DSP here) goes from 320 MHz to 465 MHz. 8974 comes in both a bin with the 4 Krait 400 CPUs clocked at 2.2 GHz (really 2.15 GHz) and 2.3 GHz (2.26 GHz) with slightly different pricing, while 8974AB comes with a Krait 400 clock available only at 2.3 GHz. Process is still TSMC 28nm HPM, but I suspect that the AB variant might have the high k dielectric and/or transistor mix tuned slightly differently based on a few rumblings I've heard recently.The S-600 in the SGS4 was "AB" so the the S-800 in the N3 might be also. We'll find out when more detailed reviews start to come out.
From AnandTech discussing the SGS4's S-600 chip...
That brings us to the Galaxy S 4. It's immediately apparent that something is different here because Samsung is shipping the Snapdragon 600 at a higher frequency than any other OEM. The Krait 300 cores in SGS4 can run at up to 1.9GHz vs. 1.7GHz for everyone else. Curiously enough, 1.9GHz is the max frequency that Qualcomm mentioned when it first announced Snapdragon 600.
Samsung is obviously a very large customer, so at first glance we assumed it could simply demand a better bin of Snapdragon 600 than its lower volume competitors. Looking a bit deeper however, we see that the Galaxy S 4 uses something different entirely.
Digging through the Galaxy S 4 kernel source we see references to an APQ8064AB part. As a recap, APQ8064 was the first quad-core Krait 200 SoC with no integrated modem, more commonly referred to as Snapdragon S4 Pro. APQ8064T was supposed to be its higher clocked/Krait 300 based successor that ended up with the marketing name Snapdragon 600. APQ8064AB however is, at this point, unique to the Galaxy S 4 but still carries the Snapdragon 600 marketing name.
If we had to guess, we might be looking at an actual respin of the APQ8064 silicon in APQ8064AB. Assuming Qualcomm isn't playing any funny games here, APQ8064AB may simply be a respin capable of hitting higher frequencies. We'll have to keep a close eye on this going forward, but it's clear to me that the Galaxy S 4 is shipping with something different than everyone else who has a Snapdragon 600 at this point.
Click to expand...
Click to collapse
so could be why we are seeing higher scores in the test note 3?
Why are people knocking the G2? It's the second fastest device on the market. It has an amazing screen area ratio and a very nice battery. It's camera is also one of the best. I would never consider it because I can never go back below 5.5 inches and I can't stand on screen buttons. But that phone should make a lot of people very happy.
Techweed said:
Why are people knocking the G2? It's the second fastest device on the market. It has an amazing screen area ratio and a very nice battery. It's camera is also one of the best. I would never consider it because I can never go back below 5.5 inches and I can't stand on screen buttons. But that phone should make a lot of people very happy.
Click to expand...
Click to collapse
im not saying its not a nice phone but nothing that "wows" me. It looks worse then Touch Wiz not a huge fan of but its ok (sense is my fav), the phone doesnt have sdcard and removable battery also a no no (why i didnt buy htc one), Note 3 has better specs with an spen and loads of new features. G2 looks like a rooted S4 running a launcher and i wasnt impressed by S4. So with that being said this is just a tad faster S4 with same look almost. Now Note 3 you may say is same look as S4 while it is, it at least carries an sdcard and removable battery and the dev support should be behind sammy. Also i do remember LG making an Intuition, revolution, lucid? whatever happened to those? oh thats right they fell through the cracks. LG just cant compete with samsung, htc, or even motorola right now
oneandroidnut said:
Interesting results here. Everybody has been saying the G2 is quicker and better then Note 3 and I must say I am quite shocked with these findings so far
http://thedroidguy.com/2013/09/sams...-sony-xperia-z1-vs-lg-g2-benchmark-comparison
Click to expand...
Click to collapse
Everybody? Who's saying that?
BTW, that article is useless. They are combining results from various places - PhoneArena/GSMArena etc.,
They took GN3 numbers from here: http://blog.gsmarena.com/the-first-benchmarks-scores-of-samsung-galaxy-note-3-are-in/
They also added some from PhoneArena: http://www.youtube.com/watch?v=NBwq0iAoVzQ
One major thing everyone forgets is that running benchmark from display models in launch events is plain wrong.
A] Most phones in such events (IFA, CES, MWC) are always charging. You should never benchmark when the phones is charging.
B] Have you ever seen any 'reviewer' in those shows to reboot the phone before running benchmarks? These display phones are abused by tech-journos. Tons of things would be running in the background. Yes, nobody bothers to clear the memory by rebooting it once. What's the point of such benchmark? Not to talk about thermal envelope after using these phones continuously.
C] G2 running release firmware, rest 2 phones running pre-release version.
(IMO) AnTuTu shouldn't be considered as a good benchmark. A benchmark tool must provide consistent repeatable result. If you run AnTuTu 5 times, I guarantee you that you will get variable result most times. No wonder AT doesn't like using AnTuTu.
Benchmarks never killed a phone :angel::angel:
CLARiiON said:
Everybody? Who's saying that?
BTW, that article is useless. They are combining results from various places - PhoneArena/GSMArena etc.,
They took GN3 numbers from here: http://blog.gsmarena.com/the-first-benchmarks-scores-of-samsung-galaxy-note-3-are-in/
They also added some from PhoneArena: http://www.youtube.com/watch?v=NBwq0iAoVzQ
One major thing everyone forgets is that running benchmark from display models in launch events is plain wrong.
A] Most phones in such events (IFA, CES, MWC) are always charging. You should never benchmark when the phones is charging.
B] Have you ever seen any 'reviewer' in those shows to reboot the phone before running benchmarks? These display phones are abused by tech-journos. Tons of things would be running in the background. Yes, nobody bothers to clear the memory by rebooting it once. What's the point of such benchmark? Not to talk about thermal envelope after using these phones continuously.
C] G2 running release firmware, rest 2 phones running pre-release version.
(IMO) AnTuTu shouldn't be considered as a good benchmark. A benchmark tool must provide consistent repeatable result. If you run AnTuTu 5 times, I guarantee you that you will get variable result most times. No wonder AT doesn't like using AnTuTu.
Benchmarks never killed a phone :angel::angel:
Click to expand...
Click to collapse
I hate benchmarks at events and real life situations is where it's at. We just need to wait till some more note 3 make it into the wild
Sent from my Nexus 7 using Tapatalk 2
oneandroidnut said:
Everybody has been saying the G2 is quicker and better then Note 3
Click to expand...
Click to collapse
Why would anyone say that? No one even has the Note 3, so we have to default to expectations. Why would anyone expect the the similar but faster clocked phone to be slower?
dscline said:
Why would anyone say that?
Click to expand...
Click to collapse
Show "anyone" this. All the tests were conducted by the same source; GSMArena.
Benchmark PI
AnTuTu
Linpack
Egypt (Offscreen)
T-Rex (Offscreen)
Sunspider
BarryH_GEG said:
Show "anyone" this. All the tests were conducted by the same source; GSMArena.
Benchmark PI
AnTuTu
Linpack
Egypt (Offscreen)
T-Rex (Offscreen)
Sunspider
Click to expand...
Click to collapse
no g2 on that list though
oneandroidnut said:
no g2 on that list though
Click to expand...
Click to collapse
Enjoy -- http://www.gsmarena.com/lg_g2-review-982p5.php
oneandroidnut said:
no g2 on that list though
Click to expand...
Click to collapse
Oops, I thought "anyone" was saying the N2 was faster than the N3. My bad.
Here's the G2 numbers, again all from a single source; GSMArena.
Benchmark PI
Linpack
AnTuTu
Egypt (Offscreen)
T-Rex (Offscreen)
Sunspider
In case anyone's bummed about the lower AnTuTu score here's a score taken from a production unit that was reviewed by a Russian site. GSMArena conducted their tests on demo units at the Berlin launch event. Based on these scores I'd bet anyone here the N3 is using a "AB" chip where the XZ Ultra and LG G2 aren't. So, at least for the time being, the N3's the fastest Android device on the planet.
But not to be a buzz kill, the SGS4 got fantastic benchmarks but had some lag in early s/w releases due to the ton-'O-crap Samsung had loaded on it. It improved over time and the N3 has more RAM so I'm hoping benchmarks translate in to "feel."
http://translate.googleusercontent....v.html&usg=ALkJrhha6VTm0y89eM70OxVC5rPRLSw6nw
BarryH_GEG said:
Oops, I thought "anyone" was saying the N2 was faster than the N3. My bad.
Here's the G2 numbers, again all from a single source; GSMArena.
Benchmark PI
Linpack
AnTuTu
Egypt (Offscreen)
T-Rex (Offscreen)
Sunspider
In case anyone's bummed about the lower AnTuTu score here's a score taken from a production unit that was reviewed by a Russian site. GSMArena conducted their tests on demo units at the Berlin launch event. Based on these scores I'd bet anyone here the N3 is using a "AB" chip where the XZ Ultra and LG G2 aren't. So, at least for the time being, the N3's the fastest Android device on the planet.
But not to be a buzz kill, the SGS4 got fantastic benchmarks but had some lag in early s/w releases due to the ton-'O-crap Samsung had loaded on it. It improved over time and the N3 has more RAM so I'm hoping benchmarks translate in to "feel."
http://translate.googleusercontent....v.html&usg=ALkJrhha6VTm0y89eM70OxVC5rPRLSw6nw
Click to expand...
Click to collapse
thanks man! and i cant wait to get my hands on one! and dont know who would keep a n2 over the n3 lol
All I know is that my S4 always benches higher than my HTC One. S4 using the "higher" binned S600.
In real world use, the HTC One felt twice as fast as the S4. Even rooted and running a custom debloated rom and kernel overclocked to 2.1GHz, the S4 still was laggy and much MUCH slower than a stock HTC One. The S4 would lag and stutter all over the place despite showing the superior numbers so I now take benchmarks with a grain of salt.
I'm really hoping Samsung gets it together and instead of just showing higher benchmark numbers, actually perform in real world use like the numbers indicate.
I'm using an LG G2 right now while waiting for my GNote3, so far I am IN LOVE with the G2. It's hands down the fastest device I've ever used, Nothing slows this thing down and I have yet to encounter a hint of lag or micro stuttering. Battery life matches or exceeds my Note 2 which I thought was incredible, I'm not too worried about the non-removable battery anymore. The screen is by far the best display I have seen, and the camera is amazingly good with OIS. In my opinion the S4 is not even in the same league as the G2, hardware or software wise. I really loved my Note 2 and have my fingers crossed the Note 3 doesn't have the incredibly frustrating laggy experience that plagued both my S4's. I would really love to keep the Note 3 as my main device because I actually use the S-pen a lot.
Dan37tz said:
I'm using an LG G2 right now while waiting for my GNote3, so far I am IN LOVE with the G2. It's hands down the fastest device I've ever used, Nothing slows this thing down and I have yet to encounter a hint of lag or micro stuttering. Battery life matches or exceeds my Note 2 which I thought was incredible, I'm not too worried about the non-removable battery anymore. The screen is by far the best display I have seen, and the camera is amazingly good with OIS. In my opinion the S4 is not even in the same league as the G2, hardware or software wise. I really loved my Note 2 and have my fingers crossed the Note 3 doesn't have the incredibly frustrating laggy experience that plagued both my S4's. I would really love to keep the Note 3 as my main device because I actually use the S-pen a lot.
Click to expand...
Click to collapse
The G2 could be considered a "next gen" phone because of S-800 and the additional features LG's provided. The One and SGS4 with S-600 are previous generation phones. Sadly for SGS_ owners, their device is released before the N_ is and Samsung learns from issues with the SGS_ what not to do in the N_. The SGS3 Exynos with 1GB of RAM vs 2GB in the N2 is a good example.
I share your fears though. The launch s/w on the SGS4 was pretty bad. But I'm hoping that 3GB of RAM, S-800 "AB," and "lessons learned" will make the N3 as big an improvement over the SGS4 as the N2 was over the SGS3. I had no issues with the stock unrooted performance of the N2.
As for "fastest" that's subjective. I don't personally get off on millisecond faster screen transitions as much as I do on 30% faster browser performance which Sunspider indicates the N3 achieves over the G2. Where Samsung phones are "fast" for me is in how, through their features, they allow me to get stuff done faster and in ways I can't with other manufacturer’s devices.
I also don't consider the G2 in anyway a competitor to the N3. One's clearly a "phone" and the other's clearly a "phablet" with S Pen/S Note making the difference even greater. And the G2's lack of expandable storage is a step back not forward. That and the non-removable battery take it off my shopping list even if I were considering a "phone."
BarryH_GEG said:
I share your fears though. The launch s/w on the SGS4 was pretty bad. But I'm hoping that 3GB of RAM, S-800 "AB," and "lessons learned" will make the N3 as big an improvement over the SGS4 as the N2 was over the SGS3. I had no issues with the stock unrooted performance of the N2."
Click to expand...
Click to collapse
For the "AB" thing, I think, then, Note 3 is supposed to have Adreno 330 clocked at 550 MHz. Have you find any info regarding that?
BarryH_GEG said:
I also don't consider the G2 in anyway a competitor to the N3. One's clearly a "phone" and the other's clearly a "phablet" with S Pen/S Note making the difference even greater. And the G2's lack of expandable storage is a step back not forward. That and the non-removable battery take it off my shopping list even if I were considering a "phone."
Click to expand...
Click to collapse
Apart from your buying preference, if it were for the image stabilization how'd you see Note 3 over G2 in terms of "smart stabilization" vs OIS?
Why did they go with adreno gpu when nexus 6 that came out 1 year ago has adreno 420 already
Will you really be able to tell the diference? I doubt it. Its just a number game really
Sent from my GT-I9300 using Tapatalk
They had no choice. They could choose the 805, the 808, or the 810. If they chose the 805, everyone would complain that it's a processor from 2014. If they chose the 810, everyone would complain that it will overheat and get crappy battery life. The 808 is the best choice for the least number of complaints. Yeah, it has a slightly slower GPU than the 805, but the CPU is much faster than the 805, and even faster than the 810 in demanding situations because the 810 will completely turn off its BIG cores if it gets too warm, whereas the 808 doesn't get hot enough that it needs to turn off the BIG cores and switch to little.
Geordie Affy said:
Will you really be able to tell the diference? I doubt it. Its just a number game really
Sent from my GT-I9300 using Tapatalk
Click to expand...
Click to collapse
Cool story. If I use that logic my old lg g2 should be enough.
Sent from my LG-D800
gtg465x said:
They had no choice. They could choose the 805, the 808, or the 810. If they chose the 805, everyone would complain that it's a processor from 2014. If they chose the 810, everyone would complain that it will overheat and get crappy battery life. The 808 is the best choice for the least number of complaints. Yeah, it has a slightly slower GPU than the 805, but the CPU is much faster than the 805, and even faster than the 810 in demanding situations because the 810 will completely turn off its BIG cores if it gets too warm, whereas the 808 doesn't get hot enough that it needs to turn off the BIG cores and switch to little.
Click to expand...
Click to collapse
Yeah but it sucks that the whole android ecosystem has to depend on qualcomm. Imagine if next year they screw up again... It seems like samsung cpu rock this year and apple too..
Sent from my LG-D800
ambervals6 said:
Cool story. If I use that logic my old lg g2 should be enough.
Sent from my LG-D800
Click to expand...
Click to collapse
My point exactly lol. Whatever phone you buy it will be an upgrade in some way ... all this numbers game is becoming a tad OTT.
Sent from my GT-I9300 using Tapatalk
ambervals6 said:
Yeah but it sucks that the whole android ecosystem has to depend on qualcomm. Imagine if next year they screw up again... It seems like samsung cpu rock this year and apple too..
Sent from my LG-D800
Click to expand...
Click to collapse
Yep, it does suck. And it is a shame that Qualcomm could have made a great SOC instead of two meh ones. If they were smart, they would have put the Adreno 430 GPU in the 808 and marketed it as their flagship phone SOC, and marketed the 810 as a tablet only SOC, because tablets can better dissipate the heat. But none of that is Motorola's fault. I think Motorola chose wisely between the not so great choices they had.
Sent from my Nexus 6 using XDA Forums Pro.
gtg465x said:
Yep, it does suck. And it is a shame that Qualcomm could have made a great SOC instead of two meh ones. If they were smart, they would have put the Adreno 430 GPU in the 808 and marketed it as their flagship phone SOC, and marketed the 810 as a tablet only SOC, because tablets can better dissipate the heat. But none of that is Motorola's fault. I think Motorola chose wisely between the not so great choices they had.
Sent from my Nexus 6 using XDA Forums Pro.
Click to expand...
Click to collapse
Well as long as there is no serious competition out there, qualcomm will continue not to give a single **** and unfortunately upgrades will come in lame increments.
Sent from my LG-D800
I think it's funny how all the new 810 soc have the cores down clocked to 1.8ghz.
Sent from my HTC6525LVW using Tapatalk
ambervals6 said:
Well as long as there is no serious competition out there, qualcomm will continue not to give a single **** and unfortunately upgrades will come in lame increments.
Sent from my LG-D800
Click to expand...
Click to collapse
Competition is coming. Qualcomm should be worried. http://www.androidpolice.com/2015/0...qualcomm-begun-a-long-slow-fall-from-the-top/
gtg465x said:
They had no choice. They could choose the 805, the 808, or the 810. If they chose the 805, everyone would complain that it's a processor from 2014. If they chose the 810, everyone would complain that it will overheat and get crappy battery life. The 808 is the best choice for the least number of complaints. Yeah, it has a slightly slower GPU than the 805, but the CPU is much faster than the 805, and even faster than the 810 in demanding situations because the 810 will completely turn off its BIG cores if it gets too warm, whereas the 808 doesn't get hot enough that it needs to turn off the BIG cores and switch to little.
Click to expand...
Click to collapse
this isn't the full story + its a little misleading. here are the technical details:
the 418 is as good, if not better than the 420 for the following reasons:
1. The 418 has the same "system specs" as the 420, minus the down-throttling.
2. The 418 was fabbed on smaller architecture (20nm) vs. the 420 (28nm). This means greater power savings and less heat.
3. The 418/420 is to the 430 like the NVIDIA 960 is to the 980 GTX, but you wont get the 430 unless you get the 810.
Source: https://en.wikipedia.org/wiki/Adreno#Variants
640k said:
this isn't the full story + its a little misleading. here are the technical details:
the 418 is as good, if not better than the 420 for the following reasons:
1. The 418 has the same "system specs" as the 420, minus the down-throttling.
2. The 418 was fabbed on smaller architecture (20nm) vs. the 420 (28nm). This means greater power savings and less heat.
3. The 418/420 is to the 430 like the NVIDIA 960 is to the 980 GTX, but you wont get the 430 unless you get the 810.
Source: https://en.wikipedia.org/wiki/Adreno#Variants
Click to expand...
Click to collapse
I'm not sure I trust that Wikipedia article. There are no references cited for the 418 information. Looking at Anandtech, the Adreno 418 is slower in EVERY graphics benchmark than the Adreno 420, even though it has the advantage of being paired with a faster CPU.
Here's a quote from Anandtech: "In GFXBench, we can see that the Adreno 418 GPU is a definite step up from the Adreno 330 in the Snapdragon 801, but not quite at the level of the Snapdragon 805's Adreno 420."
Look at the benchmarks for yourself here. The Nexus 6 and Note 4 (SD 805 / Adreno 420) both beat the LG G4 (SD 808 / Adreno 418) in every single graphics and gaming test performed. http://www.anandtech.com/show/9379/the-lg-g4-review/7
So I think it's safe to say the 420 is a little better than the 418. I don't think they would have named it the 418 if it was just a die shrunk 420. Usually a die shrink allows for faster clock speeds, and if a die shrink was the only difference, you would expect the 418 to match the performance of the 420, or even surpass it because the clock speed could go higher. That isn't the case, so I think there are some architectural differences as well that aren't shown in the Wiki article. I think Qualcomm naming it the 418 instead of the 422 even though it's newer is a pretty good indication that Qualcomm knows it isn't as good as the 420.
gtg465x said:
I'm not sure I trust that Wikipedia article. There are no references cited for the 418 information. Looking at Anandtech, the Adreno 418 is slower in EVERY graphics benchmark than the Adreno 420, even though it has the advantage of being paired with a faster CPU.
Here's a quote from Anandtech: "In GFXBench, we can see that the Adreno 418 GPU is a definite step up from the Adreno 330 in the Snapdragon 801, but not quite at the level of the Snapdragon 805's Adreno 420."
Look at the benchmarks for yourself here. The Nexus 6 and Note 4 (SD 805 / Adreno 420) both beat the LG G4 (SD 808 / Adreno 418) in every single graphics and gaming test performed. http://www.anandtech.com/show/9379/the-lg-g4-review/7
So I think it's safe to say the 420 is a little better than the 418. I don't think they would have named it the 418 if it was just a die shrunk 420. Usually a die shrink allows for faster clock speeds, and if a die shrink was the only difference, you would expect the 418 to match the performance of the 420, or even surpass it because the clock speed could go higher. That isn't the case, so I think there are some architectural differences as well that aren't shown in the Wiki article. I think Qualcomm naming it the 418 instead of the 422 even though it's newer is a pretty good indication that Qualcomm knows it isn't as good as the 420.
Click to expand...
Click to collapse
Did anyone else notice how high the 2014 moto x was in those benchmarks. Motorola must really optimize the kernel.
Sent from my HTC6525LVW using Tapatalk
Positive spin time!
The 808's gpu handles games fine and consumes less power than the 7420's gpu (S6 & Note 5). I would rather have a GPU that handles games as is, rather than drains more battery and prefer a more power economical GPU for a portable device. There is a reason you see a lot of complaints about the S6 battery life and others do not. Most correlates to those that use apps that are GPU heavy.
rushless said:
Positive spin time!
The 808's gpu handles games fine and consumes less power than the 7420's gpu (S6 & Note 5). I would rather have a GPU that handles games as is, rather than drains more battery and prefer a more power economical GPU for a portable device. There is a reason you see a lot of complaints about the S6 battery life and others do not. Most correlates to those that use apps that are GPU heavy.
Click to expand...
Click to collapse
Lmao this guy
Sent from my A0001
What was comedic besides my awareness it is spin? True that games perform fine on the 808 and the 7420 gpu consumes more power. As far as bigger fancier games that need even more power, not very practical on a portable device so kind of moot with a small battery.
Sent from my SM-N910V using Tapatalk
If it's a concern you should wait for the nexus to drop with its rumored snapdragon 820 and next gen adreno.
Also for the issue of this year's qcom products sucking, remember that market pressure forced them to release chips with generic ARM cores because their in-house 64 bit designs weren't ready. The 820 ditches the octocore big.LITTLE architecture for a quad core qcom design. Lots to look forward to.
And I think the 808 is probably the best chip they could have picked for the X this year.
ambervals6 said:
Cool story. If I use that logic my old lg g2 should be enough.
Sent from my LG-D800
Click to expand...
Click to collapse
It is, but you're all too spoiled to make it out
SchmidtA99 said:
I think it's funny how all the new 810 soc have the cores down clocked to 1.8ghz.
Sent from my HTC6525LVW using Tapatalk
Click to expand...
Click to collapse
I think 810 is way better than 808. Adreno 430 vs 418. The 430 is WAY BETTER. And if the 810 gets too hot, you can always turn off 2 high performance cores. But you can never have an adreno 430 in the 808
gtg465x said:
I'm not sure I trust that Wikipedia article. There are no references cited for the 418 information. Looking at Anandtech, the Adreno 418 is slower in EVERY graphics benchmark than the Adreno 420, even though it has the advantage of being paired with a faster CPU.
Here's a quote from Anandtech: "In GFXBench, we can see that the Adreno 418 GPU is a definite step up from the Adreno 330 in the Snapdragon 801, but not quite at the level of the Snapdragon 805's Adreno 420."
Look at the benchmarks for yourself here. The Nexus 6 and Note 4 (SD 805 / Adreno 420) both beat the LG G4 (SD 808 / Adreno 418) in every single graphics and gaming test performed. http://www.anandtech.com/show/9379/the-lg-g4-review/7
So I think it's safe to say the 420 is a little better than the 418. I don't think they would have named it the 418 if it was just a die shrunk 420. Usually a die shrink allows for faster clock speeds, and if a die shrink was the only difference, you would expect the 418 to match the performance of the 420, or even surpass it because the clock speed could go higher. That isn't the case, so I think there are some architectural differences as well that aren't shown in the Wiki article. I think Qualcomm naming it the 418 instead of the 422 even though it's newer is a pretty good indication that Qualcomm knows it isn't as good as the 420.
Click to expand...
Click to collapse
It's slower because Qualcomm halved the memory bus from 128-bit to 64-bit. The S810/A430 has the same bandwidth as the S805 because they doubled the speed of the RAM. So, 128-bit LPDDR3-800 (1600MHz effective) is equal to LPDDR4-1600 (3200MHz effective): 25.6 GB/s
Unfortunately, Qualcomm limited the S808 to LPDDR3-933 (1866MHz effective): 14.9 GB/s
The 418 and 420 are the same GPU, architecturally. The 418 could probably be slightly faster in non-bandwidth limited scenarios (low resolution 3D).
Memory bandwidth dropped from 25.6 GB/s to 14.9 GB/s. That's nearly a 25% loss and about equal to the real world performance losses. Hence, it's a 418.