I know MSM8960 is made in 28nm LP from TSMC. But Im not sure whether APQ8064 is same or not
I heard the reason of that qualcomm removes modem from APQ8064, not like MSM8960, was 28LP cant handle that
Actually MSM8974(quadcore+lte modem in single chip) is set by 2.0~2.5GHz because of new 28nm HKMG from TSMC next year.
Its really hot and consuming so much battery compare to Exynos based on A9,
even if Krait has aSMP and many improvment which should provides more time and lower heat
Conclusion, I think APQ8064 used in N4, OG and others, was made in 28LP, and its not that good at battery life or temperature.
(I saw the CPU TEMP was even up to 65C in System Tuner)
Sent from my LG-F180S using xda app-developers app
The APQ8064 1.5G version, I believe, uses 28nm poly/sion LP process. What I want to point out is that under a heavy load which requires all 4 cores online, the APQ8064 does consumpt more power than the Exynos 4412, but when it comes to daily use and light load, the APQ8064 actually consumpts less power.
This year some companies will release phones powered by 1.7G version APQ8064. I am not sure what process the 1.7G version uses.
I only know that msm8974 and APQ8084 uses 28nm hkmg HPM process.
Related
Is it true that Qualcomm's dual-core CPU's will be based on the older ARM Cortex-A8 architecture set instead of the modern Cortex-A9 which is being used by Apple's A5 Chip and Nvidia'S Tegra 2 ?
Source:
http://smartphonebenchmarks.com/for...msm8660-12ghz-dual-core-snapdragon-processor/
The hardware benchmarks on the dual-core MSM8x60 1.2 Ghz chip used by HTC Pyramid (Sensation,Doubleshot) and the Evo-3D do not look pretty good.
Source:
http://smartphonebenchmarks.com/forum/index.php?showtopic=258
Need a bit of clarification on this issue why they didn't choose the Cortex-A9 path.
Ok so I just read this report from Qualcomm explaining this issue:
http://www.qualcomm.de/documents/files/linley-report-dual-core-snapdragon.pdf
Apparently their architecture set is compatible with ARM's instruction architecture set and they claim its better than the A9.
"The superscalar CPU uses a 13-stage pipeline to generate faster clock speeds than competing products can achieve using ARM’s Cortex-A8 or Cortex-A9"
Having said that still not sure why the hardware benchmarks are not near the Cortex-A9 dual-core processors.
Adreno-220 is pretty good though compared to other GPU's.
mjehan said:
Apparently their architecture set is compatibily with ARM's instruction architecture set and they claim its better than the A9.
Having said that still not sure why the hardware benchmarks are not near the Cortex-A9 dual-core processors.
Click to expand...
Click to collapse
Because bechmarks are meaningless and HTC have yet to put the work into fiddling them yet!
Quamcomm has been claiming that their design is better than ARM's Cortex A8 before but other than few special occasions, they are mostly equal at the same clock speed. Since MSM8x60 is also based on the identical cores, I don't see how it could be better than Cortex A9. In fact, Qualcomm is working on their own "equivalent to A9" version right now.
FYI, # of pipelines don't tell the whole story about the speed of CPUs. If not implemented well, it will simply cause longer stall delays. We have seen this in the old Pentium 4 architectures.
I think the 128bit fpu makes scorpion equivalent to a9 in floating points calculation
Sent via psychic transmittion.
Sets the bar at 2GHz
KOREAN HARDWARE GIANT Samsung has announced its latest processor, the Exynos 5250 that is based on the ARM Cortex A15 design.
The dual-core processor is designed for high-end tablets and will run at 2GHz. It is fabbed on a 32nm process and is set to appear in devices in the second half of 2012.
"The ARM Cortex-A15 brings unparalleled performance to our Exynos processor family and the exploding mobile marketplace," said Dojun Rhee, VP of system LSI marketing at Samsung. "The advanced low-power, high-performance processor technology of the new Exynos 5250 continues to deliver an unprecedented level of performance for users to enjoy a completely new mobile experience."
Samsung claims the new chip can process nearly twice the number of instructions per second as the ARM 1.5GHz Cortex A9 chip. The Exynos 5250 will be able to support up to 2560x1600 resolution displays and will have four times the graphics capabilities of the Cortex A9.
To support this high resolution the chip has a doubled memory bandwidth of 12.8GB/s compared to existing chips, which also aids data processing speed. To help with power efficiency the Exynos 5250 also has an embedded Displayport (EDP) interface that is compatible with panel self-refresh technology (PSR). It allows a static image on the screen to be refreshed from a memory buffer instead of using processor cycles. µ
Source: The Inquirer
Click to expand...
Click to collapse
If that chip is in the Galaxy Note 2.....
I will never turn it off.
Sent from my GT-N7000 using xda premium
Exciting news, just as fun as way back when when following dual core PC chips move to quad core and the drool factor they had. I personally hope ARM keeps up this arms race so fast that x86 never has a chance...
Low cost, low power, high performance, small footprint.
This time next year is really going to be exciting...
Just when I had started enjoying my Galaxy Note :|
Its second half of 2012, long time till then.
Sent from my GT-N7000 using Tapatalk
Samsung just launched a new processor line up called Exynos5 Octa the story is still developing but the following are some details that are out
Samsung unveiled a brand new processor that will power its future mobile devices. Dubbed “Exynos5 Octa,” the new chip features two sets of four cores and a 28nm process that will eventually drop down to 10nm in the coming years. The chipset provides significant performance enhancements over Samsung’s previous-generation quad-core Exynos processor, which is now found in more than 53 million mobile devices. The company also said that the new Exynos5 Octa has a power-saving mode that helps improve battery life by reducing consumption when devices are idling and in other situations where full processing power is not required.
So between Exynos 5 Quad which is an A15 32nm chip that's supposed to be used in the SGS4, Krait v3 which is 15% faster version of S4 Pro, and now Octa, Nvidia may as well pack up their marbles and go home.
For geeks, here's some details on Exynos 5 which is Samsung's current "state of the art."
http://www.anandtech.com/show/6148/samsung-announces-a15malit604-based-exynos-5-dual
And S4 Pro
http://www.anandtech.com/show/6112/...agon-s4-apq8064adreno-320-performance-preview
Here's some disucssion of Teg4 from a chip forum. People who do chips for a living aren't impressed; especially with the GPU.
http://semiaccurate.com/forums/showthread.php?t=6224
Watch out. Demandarin might come around and tell us how great the Tegra 4 is Youg looks good, I wonder if it will be involved in devices before August 2013
Sent from my SPH-D710 using xda app-developers app
8 cores..wow...overkill.
It wasnt long ago (less than a year?) That reviewers were stating quad core is abit of a gimmich right now on phones and tabs etc.
Im all for it ofcourse..but wow 8 cores
The 8 cores will be device into 2 sub 4 cores - 1 sub 4 core will be an A7 1.2GHZ quadcore processor for lighter task and the second sub core will be 4 core A15 1.6GHZ of above quadcore processor.
The same will be built on 28 NM technology and later 10 NM technology over a period of time.
Again these are not confirmed specs
Simple question, I purchased a Moto G GPE and its on its way but i have the curiosity about how the SOC Snapdragon 400 Cuad Core compares to something like the S4 Pro that was in the Nexus 4 which ran great. I know its a newer chip but i havent found any real evidence whether the architecture is superior or inferior to the S4 Pro since it was a flagship chip but the 400 is budget. Anyone has any insight on this? Thanks.
best way to compare the two.
http://en.wikipedia.org/wiki/Snapdragon_(system_on_chip)
mike21pr said:
Simple question, I purchased a Moto G GPE and its on its way but i have the curiosity about how the SOC Snapdragon 400 Cuad Core compares to something like the S4 Pro that was in the Nexus 4 which ran great. I know its a newer chip but i havent found any real evidence whether the architecture is superior or inferior to the S4 Pro since it was a flagship chip but the 400 is budget. Anyone has any insight on this? Thanks.
Click to expand...
Click to collapse
Snapdragon 400 is positioned more like S4 Plus, not Pro. As in marketing positioning.
Speaking in technical terms however, those two SoCs have a completely differnet architecture. S4 Plus always has Krait cores, while our S400 has Cortex-A7 cores, which is generally somewhat slower than Krait, partially because Krait is an Out-of-Order architecture, while Cortex-A7 is In-Order.
In terms of cores S4 Plus always had two cores (up to 1,5 GHz), while our S400 has four cores @ 1,2 GHz.
Now go figure what´s faster. I´d personally prefer two Kraits
Note: Qualcom had really messed up the naming in S400 series as the chips can contain either two Krait cores or four Cortex-A7 cores. Don´t get me wrong, Cortex-A7 is a great architecture , mostly because its performance/power consumption ratio, but the SoCs containg those cores should have been labeled as Snapdragon 350 or something like that.
So the Qualcomm Snapdragon 820 is quad core. Why did Qualcomm decide to go with a quad core one over a octacore or hexacore? How would that affect the GS7/GS7 Edge if it were octacore or hexacore? How much of a difference is there between the Exynos and 820?
Indeed the snapdragon 820 is a quad core SOC unlike most recent socs which have featured 8 cores (2 clusters of 4 cores). However most big.little socs like the exynos 8890 and 7420 use 4 low power slow processors and 4 high power, but power hungry processors. They are completely different architectures. This means that while the exynos 8890 for instance has 8 cores. Only 4 of them are really designed for performance. The other 4 are designed to save power. The 820 is different. It's also some sort of big.little setup with 2 clusters of 2 cores. However both clusters are identical architectures. The difference is one cluster is clocked lower and has a different l2 cache configuration in order to use less power. On top of that the custom cores in the 820 are faster per core than the exynos 8890. So clock for clock the 820 would win against the high power cluster of the exynos. In heavily multitheaded situations though. The exynos still can tap into all 8 cores at the same time which should give it an advantage in that scenario. For the rest of the time I would imagine the 4 faster cores of the snapdragon would be better suited to everyday stuff. As for why they only went with 4. My guess is cost, and power efficiency. Kyro is a brand new architecture. Krait went through many iterations. Kyro will probably see a noticeable reduction in its power envelope in the next iteration which would make shoving more of them onto an SOC a more viable option. As for gpu, all signs are pointing to the snapdragons adreno GPU beating the Mali in the exynos atm. Development will also be improved on the snapdragon device as Qualcomm releases the proprietary vendor binaries and Samsung does not. This means the likelihood of seeing cm or aosp on an exynos variant is slim. Hope this helps!
Actually, the Kryo cores are (slightly) better at running single threaded tasks while the Exynos cores are (slightly) better at running multi-threaded tasks. I doubt the everyday users will notice.
The Adreno is also more powerful than the MALI GPU, though everyday users will mostly notice a performance improvement on applications using the Vulkan API vs regular applications, than anything between both these GPUs.
Finally the memory management seems much better on the Exynos 8890 for some reason (about twice as fast), since the same chips are used I wonder whether it's a software or a hardware implementation difference, both units are plenty fast though.
The real difference between both these SOCs will seen in the power management efficiency, in fact both variants are overpowered in every aspects as far as regular usage goes, so there is little point in comparing which one's the fastest. Instead, you need to wonder which one is the most conservative with power consumption while achieving equivalent performances.
Both the GPUs on these SOCs support the Vulcan API. And, whilst the Adreno is faster in terms of pure benchmark numbers, I very much doubt there will be a noticeable difference on any game or application, Vulcan or otherwise, that will be released during the lifetime of these phones.
Yeah, that does help explain it. Thanks. I just hoped there wouldn't be a TSMC vs Samsung difference in the iPhone 6S/6S Plus SoC.
i was thinking 8 core snapdargon 810 was over heating and thermal throttling so they went for 4 cores instead on snapdragon 820.
just my thoughts.