The follow-up to the Samsung Galaxy S6 is already being rumored, which people are obviously calling the Galaxy S7. The last two phones were unveiled at Mobile World Congress in Barcelona, Spain, which takes place in late February / early March. Information coming out in November is a bit… early. Some sites believe that Samsung will announce the phone in January, but who knows? Some of the rumors are interesting, though.
The one that catches my attention is the potential inclusion of a microSD card slot. External storage is rare these days, with Google removing it from their Nexus line and severely limiting what apps can do with the contents. That said, Android 6.0, recently released for a few devices, made further changes to increase its capabilities. You can now use SD cards as internal storage, but only if you agree to format and encrypt the storage to use only on that device. While the recent batch of Nexus phones don't include a microSD card slot, the changes might be enough to sway third-party manufacturers to include a slot.
As a developer, it would certainly be nice, especially if you intend to develop software that uses an SD card. Makes sense, right? Purchasing a developer phone that has all the features you might want to target?
Speaking of developer phones, the upcoming device should have a top-of-the-line processor in it. Reports are split between the Snapdragon 820 and the Exynos 8890. If it's the latter, availability is expected Q1 2016; the former started sampling a few months ago and was launched on November 11th. As such, SoC availability should be ready if Samsung intends to launch the phone early, regardless of the chosen chip, but that's probably not the limiting factor. It is also entirely possible that Samsung could include different processors for different markets. Qualcomm was absent from the Galaxy S6 line, but the S5 had some sub-models using Qualcomm processors and others Samsung's own implementation.
Either way, they are fast processors that support OpenGL ES 3.1 + AEP at the very least. The Adreno 530 is rated for about 550 GFLOPs, which is a tiny bit faster than a GeForce 9800 GT, although with Vulkan-level feature support (provided correct drivers). Thankfully Google has been more friendly to Khronos-based standards, and Samsung even more so.
When will we know for sure? Don't know. How much will it cost? Don't know. What will it be officially called? Don't know, but anything other than Galaxy S7 would be surprising. Would it make sense for Samsung to shake up the date and other long-running details? Well, the Galaxy S6 launch was lackluster, so this would be the most likely time for them to be squirrely. We'll see.
It would be nice to see if
It would be nice to see if these chip’s CPU cores were actually analyzed for the Snapdragon 820 and the Exynos 8890. How about each core’s reorder buffer, how many bytes?, or numbers of instruction decoders, or number of integer pipelines/units, or FP pipelines/units, or address generation units AGUs. Where is the CPU’s core hardware reports, they are custom cores with their micro-architectures engineered to run the ARMv8A ISA, so what about the instruction cache on these custom cores and their data cache, and any L3 cash sizes.
Here is the proper breakdown on the Apple A7 custom Cyclone core:
CPU Codename—————-Cyclone,
ARM ISA———————ARMv8-A(32/64),
Issue Width—————–6 micro-ops,
Reorder Buffer Size———192 micro-ops,
Branch Mispredict Penalty—16 cycles (14 – 19),
Integer ALUs—————-4,
Load/Store Units————2,
Load Latency—————-4 Cycles,
Branch Units—————-2,
Indirect Branch Units——-1,
FP/NEON ALUs—————-3,
L1 Cache-–——————64KB I$ + 64KB D$,
L2 Cache——————–1MB,
L3 Cache——————–4MB,
see refernce(1)
It would be nice if the online press would stop miss-informing their readers about the online press’s actual technical knowledge of any Custom, or non-custom core that are engineered to run the ARMv8a ISA.
There are some very nice custom core micro-architectures that are engineered to run the ARMv8A ISA, the Apple A7 as properly reported by (1)Anand Lal Shimpi, appears to be the last CPU core analysis done properly, and the so called “technology” press online sure made use of Anand’s work! So who is taking up Anand’s work now in the “technology” press, now that Anand is no longer working as a technology reporter. It’s intersting to note that the Apple A7’s instruction issue width is twice as wide as the ARM Holding’s refrence design cores(A53, A57, A72) that only issue 3 IPC.
So what about the Snapdragon 820’s and the Exynos 8890’s instruction issue width, how do they compare to the Apple A7/A8/A9 or the ARM holding’s refrence design(A53, A57, A72). All these chips cores, custom design and refrence design, can run the ARMv8a ISA, but some have much better execution resources compared to the others. So what about actually analysing the cores, instead of only the benchmarks!!!!
Time for Ryan to invest in an
Time for Ryan to invest in an electron microscope.
Anand Lal Shimpi, once of
Anand Lal Shimpi, once of AnandTech, did not have an electron microscope, he had the proper compiler optimization manuals for the CPUs/SOCs SKUs in question so that he could use software methods to tell some of the CPU’s hardware secrets. It’s easy to make an execution pipeline stall, or a Cache to thrash, etc. with the proper code and give up the CPU core’s inner hardware secrets. There is plenty of benchmarking software out there specifically engineered to tax the execution resources on microprocessors, with specific routines to suss out CPU core reorder buffer sizes and cache sizes, and cache efficiency, and other hardware features of the modified Harvard architecture microprocessors that make up the majority of CPU cores in modern microprocessors, both RISC ISA and CISC ISA based alike! Those ChipWorks images that they release online are utter garbage compared to what Chipworks provides to those with the money to get the better product.
I’m still wondering if the Apple A9X’s CPU cores are using SMT, because it’s not just the doubling of interface bus widths on the A9X that are the only thing responsible for the A9X’s extra performance relative to the A8X! And what of the PowerVR GPU cores that the A9X is using, do they have the built-in ray tracing circuity that Imagination Technologies offers as an option on their high end GPU SKUs. It looks like Apple is using the top end The PowerVR Series7XT family with configurations of 2 to 16 clusters offering scalable performance from 100 GFLOPS to 1.5 TFLOPS. The A9X and the A8X still have no information released relative to the information Anand Lal Shimpi was able to get on the A7 Cyclone SKU, and Anandtech does not have their namesake anymore do the write-ups on any of Apple’s CPU cores post the Apple A7’s seller writeup by Anand himself!
edit: seller writeup
To :
edit: seller writeup
To : stellar writeup
GOD-Fing Damn it OpenOffice with your agressive spell checking! Just highlight to miss-spelled words and let me choose the correct one!
I don’t care about any specs
I don’t care about any specs other than these two things.
1. Can I use a microSD card?
2. Is the battery removable?
I don’t care about any specs
I don’t care about any specs other than these two things.
1. Can I use a microSD card?
2. Is the battery removable?