Smartphones
Different smartphone companies have different processors today. For example Apple makes it's own processors and Samsung also makes it's own processors called Exynos but it also uses processors made by snapdragon. Apple: Apple's current processor is called A13 Bioinic it's made using 7 nano meter architecture which by far the fastest chipset ever used on a smartphone. Apple crammed 8.5 billion transistors in the A13, an increase of 23 percent over the A12. The chip is estimated to be around 20 percent larger—about 98 mm², compared to 83 mm² for the A12. Thus, Apple achieved only a small increase in density, but a big increase in maximum performance and power efficiency. This is not Apple’s biggest chip ever, however. The A12X found in the new iPad Pro has more transistors (10 billion) and is estimated to be around 135 mm², and both the A5 and A10 Fusion were over 120 mm². If there’s one area where competing smartphone processors have been able to beat Apple, it’s in graphics performance. Apple’s GPUs deliver real-world performance that is among the best in the business, but some cross-platform benchmarks will favor the latest Android phones. The GPU in the A13 is still the a custom quad-core Apple design, which Apple claims is “metal optimized” and 20 percent faster. In our tests, it was much faster than that. The compute performance in Geekbench, which uses the Metal API on the GPU, ran about 40 percent faster in both Geekbench 4 and Geekbench 5. If you look at the combination of GPU compute performance and gaming-class 3D benchmarks, it’s fair to say that Apple’s go the fastest overall GPU in a smartphone right now. It is a closer call than CPU performance, though, and easier to see a competitor stealing the crown before the introduction of the A14 in the fall of 2020. It’s really interesting that we’re seeing real-world performance far in excess of Apple’s 20 percent improvement claim. It’s hard to know what’s responsible for this, but I theorize that it has to do more with memory bandwidth than peak GPU theoretical performance. Apple claimed that the A12 delivered up to 50 percent faster GPU performance than the A11, but we saw nothing of the sort—some tests didn’t show any real improvement at all. We guessed, at the time, that limitations of memory bandwidth were keeping the chip from achieving its top theoretical performance in some 3D graphics tests. Perhaps this year, though the chip isn’t 50 percent “faster,” it is better able to use the available memory bandwidth. Whatever the reason, Apple is realizing way more than the promised 20 percent improvement in GPU performance. LaptopsLaptops generally have intel and AMD processors we mainly talk about the latest processors from intel and AMD
Intel: Intel i5 Processor: The Intel Core i5 processors are together with the Core i7/i9 processors the most powerful processors for private use from Intel. All Core i5 desktop processors offer 4 cores and no hyperthreading while in the mobile sector all Core i5 processors have 2 cores and hyperthreading. Mobile versions of Core i5 CPUs can be recognized on the attached M, U or Y. Only Desktop versions with an attached K are overclockable, they have a unlocked multiplier. The cache is compared to the Core i7 1-2MB smaller but that has usually only a very little effect in performance. In addition, the clock frequency is slightly lower (usually in the range 100-200MHz in direct comparison). Since many PC games still have problems to use all CPU cores - this is especially true for hyperthreading - the Desktop Core i5 processors area quite enough for current games. In the mobile space, however, Core i7 processors make sense for PC gamers, because they have 2 times more physical cores, which can clearly make their mark in games. Intel i7: The Intel Core i7 series includes Intel's most powerful desktop and mobile processors for private use. In 2017 Intel released new Intel Core i9 processors for enthusiasts wich have more CPU cores than Intels Core i7 processors. All Intel Core i7 processors have hyperthreading technology that doubles the displayed CPU core number, while the performance increases by up to 40% compared to processors without hyperthreading. All desktop processors have at least 4 cores, mobile and ultra-mobile versions have 2 to 4 cores. The mobile versions with 2 cores can be recognized by the endings M and U, the variants with 4 cores are marked with MQ or HQ. Desktop versions that are marked with a K, have an unlocked multiplier and can be overclocked. The processors of the Core i7 series are very fast and suitable for very complex office tasks, video editing, image editing, or the latest video games. |
Samsung:
ou’ll always find two variants of each high-end Samsung smartphone, one packing a Qualcomm Snapdragon processor for the U.S. market and another equipped with the company’s latest high-end Samsung Exynos chip for the global market. In a first for Samsung, there is dedicated AI silicon in the company’s brand-new Exynos 9820, meant to power its 2019 flagships. The Exynos 9820 offers a neural processing unit (NPU) for machine learning tasks, with Samsung claiming it’ll handle AI-related tasks seven times faster than the Exynos 9810. But there’s more to the 2019 flagship chipset than AI smarts… The Exynos 9820 offers a tri-cluster CPU arrangement for the first time, as opposed to the dual-cluster arrangement of previous Samsung flagship chips. That means two fourth-generation custom cores for demanding use-cases, two Arm Cortex-A75 cores for medium lifting, and four Cortex-A55 cores for lightweight tasks. We’ll need to wait and see if this provides a big boost to power and battery life, but the two big cores certainly look much more powerful than anything its competitors are using. The choice of two A75s rather than A76 for the middle cluster is interesting, as they aren’t as powerful but do have a smaller silicon footprint. This three-tier performance idea isn’t dissimilar to Qualcomm’s latest Snapdragon 855 or Huawei’s Kirin 980, although the CPU configurations are different will give different performance/energy trade-offs. The deciding factor is how well these cores balance power consumption and task scheduling. Otherwise, Samsung’s 2019 chip also offers the latest Arm Mali GPU (Mali G-76 MP12), support for up to five cameras, 8K/30fps or 4K/150fps video recording, and compatibility with the latest UFS 3.0 storage standard. But the chip is poised to deliver better battery life in theory too, as it’s built on an 8nm manufacturing process. However, this isn’t quite as advanced as TSMC’s 7nm processed used by Qualcomm and Huawei. Samsung is the only Android chip manufacturer still working on a fully custom in-house CPU design for its flagship products, known as Mongoose or M cores. The Exynos 9810 features the third generation of this design. Samsung’s Mongoose boasts a wider throughput design than Arm’s Cortex cores and comes paired with significantly more cache than other chips in the Android ecosystem. This is a similar approach to Apple’s custom CPU design. Jargon aside, this makes the Samsung M cores considerably bigger than their Arm Cortex counterparts and higher performing too, but there have been some issues with performance and efficiency as a result. AMD: AMD RYZEN 5: There are currently two third-generation Ryzen 5 chips based on AMD's latest Zen 2 microarchitecture, the Ryzen 5 3600X and Ryzen 5 3600. Both have six cores with multithreading support, which means each processing core can handle two instruction threads at a time for a total of 12 threads. The Ryzen 5 3600X tested here is a 95-watt chip with a 3.8GHz base and 4.4GHz boost clock. The $199 Ryzen 5 3600, meanwhile, is a 65-watt part that comes in at 3.6GHz base and 4.2GHz boost. (A third Ryzen 5 CPU, the Ryzen 5 3400G, is technically part of the third generation but costs less and uses the older Zen+ microarchitecture. The six-core AMD Ryzen 5 3600X is an excellent mainstream CPU, offering proficient performance, multithreading, and overclockability. AMD RYZEN 7: The AMD Ryzen 7 3700X is a brilliant piece of hardware. With a TDP of just 65W, this chip is capable of delivering raw performance that would take other processors much more power to equal. The reasonable price tag is just a bonus. This decision to 7nm has brought a beefy 15% boost to IPC (instructions per clock) performance. Effectively, compared to a Ryzen 2nd Generation processor at the same clock speed, you will get a straight 15% increase in performance. That’s not big enough to be evident in day-to-day workloads, but it does still mean something. AdvertisementThe improvements don't just end at IPC. With Ryzen 3rd Generation, as the CPU cores are on their own chiplets, AMD was able to pack way more L2 and L3 cache into the Ryzen 7 3700X – with 4MB and 32MB, respectively. Essentially, this processor has a grand total of 36MB of Cache, which AMD lumps together as 'GameCache'. This GameCache isn't anything entirely new, but it does show that this will help boost gaming performance in some cases – especially in older 1080p esports games. The major addition to the 3rd Generation of Ryzen, however, is PCIe 4.0. When paired with an AMD Navi graphics card like the Radeon RX 5700 XT or RX 5700, you'll experience much better performance, thanks to increased bandwidth. However, the way we look at it, SSDs are the real stars of the PCIe 4.0 show. Through this superior connection, NVMe SSDs are potentially up to 51% faster than their non-PCIe 4.0 peers. In our own testing, the Aorus PCIe 4.0 SSD that AMD provided was able to get up to 4,996 MB/s sequential read speeds. That’s remarkably fast for an SSD. |