Community hub
0 subscribers7 pages, 0 posts
Recent from talks
All channels
Be the first to start a discussion here.
Be the first to start a discussion here.
Be the first to start a discussion here.
Be the first to start a discussion here.
Contribute something
Welcome to the community hub built to collect knowledge and have discussions related to Krait (processor).
Nothing was collected or created yet.
Krait (processor)
View on Wikipediafrom Wikipedia
| General information | |
|---|---|
| Launched | 2012 |
| Designed by | Qualcomm |
| Common manufacturer |
|
| Performance | |
| Max. CPU clock rate | 1 GHz to 2.7 GHz |
| Physical specifications | |
| Cores |
|
| Cache | |
| L1 cache | 16 KiB and 16 KiB |
| L2 cache | 1 MiB or 2 MiB |
| Architecture and classification | |
| Instruction set | ARMv7-A, Thumb-2 |
| History | |
| Predecessor | Scorpion |
| Successor | Kryo |
Qualcomm Krait is an ARM-based central processing unit included in the Snapdragon S4 and earlier models of Snapdragon 400/600/800 series SoCs. It was introduced in 2012 as a successor to the Scorpion CPU and although it has architectural similarities, Krait is not a Cortex-A15 core, but it was designed in-house.[1] In 2015, Krait was superseded by the 64-bit Kryo architecture, first introduced in Snapdragon 820 SoC.
Overview
[edit]- 11-stage integer pipeline with 3-way decode and 4-way out-of-order speculative issue superscalar execution
- Pipelined VFPv4[2][under discussion] and 128-bit wide NEON (SIMD)
- 7 execution ports
- 4 KB + 4 KB direct mapped L0 cache
- 16 KB + 16 KB 4-way set associative L1 cache
- 1 MB (dual-core) or 2 MB (quad-core) 8-way set-associative L2 cache
- Dual- or quad-core configurations
- Performance (DMIPS/MHz):
- Krait 200: 3.3 (28 nm LP)
- Krait 300: 3.39[3] (28 nm LP)
- Krait 400: 3.39 (28 nm HPm)
- Krait 450: 3.51 (28 nm HPm)
See also
[edit]References
[edit]- ^ Brian Klug; Anand Lal Shimpi (October 7, 2011). "Qualcomm's New Snapdragon S4: MSM8960 & Krait Architecture Explored". Anandtech. Archived from the original on October 9, 2011. Retrieved 2013-07-28.
- ^ Brian Klug; Anand Lal Shimpi (February 21, 2012). "Qualcomm Snapdragon S4 (Krait) Performance Preview - 1.5 GHz MSM8960 MDP and Adreno 225 Benchmarks". Anandtech. Retrieved 2013-07-28.
{{cite web}}: CS1 maint: deprecated archival service (link) - ^ Linley Gwennap (December 18, 2012). "Krait 300 Bumps Up Performance". Linley on Mobile. The Linley Group. Retrieved 2013-07-28.
External links
[edit]Krait (processor)
View on Grokipediafrom Grokipedia
History and Development
Origins
Qualcomm initiated the development of the Krait microarchitecture as a successor to its Scorpion CPU core, addressing key limitations in power efficiency and performance scaling for mobile devices. The Scorpion core, an in-order design introduced in 2007 and utilized in the Snapdragon S1, S2, and S3 series, struggled to meet the escalating demands of smartphones and tablets, particularly in balancing computational throughput with battery life under increasing workloads. By crafting a custom ARM-based architecture, Qualcomm aimed to overcome these constraints while maintaining compatibility with the burgeoning mobile ecosystem. The primary motivations for Krait centered on achieving superior efficiency metrics, targeting a DMIPS/MHz rating of 3.3—more than double Scorpion's 1.25—while reducing power consumption by up to 65% compared to the Snapdragon S3 at equivalent performance levels. This focus was driven by the need for ARMv7-A instruction set compatibility in power-sensitive applications, enabling higher clock speeds without proportional increases in energy use. Qualcomm's engineering efforts emphasized optimizations for mobile workloads, such as improved instruction throughput and reduced leakage in advanced process nodes, to support the transition to multi-core configurations in devices like high-end smartphones and tablets.[2] Initial design choices for Krait included a fully custom implementation of the ARMv7-A instruction set, incorporating Thumb-2 for denser code execution, and shifting to an out-of-order execution model to extract greater parallelism from applications. This approach was intended to surpass the capabilities of standard ARM Cortex-A9 cores, which, while also out-of-order, offered lower efficiency in mobile contexts (around 2.5 DMIPS/MHz). By prioritizing these elements, Krait was positioned as a foundational leap for Qualcomm's Snapdragon platform, with early announcements in February 2011 highlighting its potential for up to 2.5 GHz per core in future SoCs.[2]Evolution and Releases
Qualcomm first publicly announced the Krait CPU architecture at Mobile World Congress (MWC) in February 2011 as the core of its next-generation Snapdragon S4 processors, fabricated on a 28 nm process node by TSMC.[2] The announcement highlighted Krait's design for enhanced performance and power efficiency in mobile devices, with initial variants like the MSM8960 entering sampling in the second quarter of 2011, followed by sampling in early 2012 for other models such as the MSM8930 and APQ8064.[2] Commercial integration began in 2012, marking Krait's debut in production Snapdragon S4 systems-on-chip (SoCs) for high-end smartphones and tablets. Key milestones followed with expansions across Qualcomm's product lines. In 2013, Krait powered the Snapdragon 600 series, featuring the refined Krait 300 cores and targeting mid-to-high-end devices with improved multimedia capabilities.[8] The architecture reached its peak in the Snapdragon 800 series, announced the same year but seeing widespread commercial adoption in 2014 devices, incorporating Krait 400 cores for higher clock speeds up to 2.3 GHz.[8] Development of Krait concluded around 2015, as Qualcomm transitioned to its successor Kryo architecture for the Snapdragon 820, emphasizing fully custom 64-bit designs to address evolving market demands.[7] Architectural refinements across releases focused on process node optimizations and efficiency gains. Early Krait implementations in the Snapdragon S4 utilized TSMC's 28 nm low-power (LP) process for balanced performance, while later variants in the Snapdragon 800 and beyond shifted to the 28 nm high-performance mobile (HPM) process, enabling higher clock frequencies and better power scaling without proportional efficiency losses.[9] Incremental instructions-per-clock (IPC) improvements were also pursued, with initial Krait cores achieving approximately 3.3 DMIPS/MHz, evolving to offer 5-10% performance uplifts in final iterations like Krait 450 for enhanced single-threaded execution.[10] These evolutions were shaped by intense competition from ARM's Cortex-A15 cores, which offered high performance but suffered from elevated power consumption, and Apple's custom A-series processors, which set benchmarks for efficiency and speed in iOS devices.[11] This landscape influenced Qualcomm's strategic pivot toward more bespoke architectures like Kryo, allowing greater optimization beyond licensed ARM designs.[12]Architecture
Core Design
The Krait processor implements full compatibility with the ARMv7-A instruction set architecture (ISA), enabling seamless support for the extensive ARM ecosystem and software base. This includes the Thumb-2 instruction set extension, which enhances code density by allowing a mix of 16-bit and 32-bit instructions to reduce memory footprint and improve instruction cache efficiency in resource-constrained mobile environments.[13] In addition to the base ISA, Krait incorporates the NEON advanced SIMD extension, providing 128-bit vector processing capabilities for parallel data operations essential in multimedia and graphics workloads. The core also integrates the VFPv4 floating-point unit, which supports double-precision operations and fused multiply-add instructions to accelerate scientific computing and signal processing tasks with higher precision and throughput.[13] At its structural foundation, Krait employs a superscalar design capable of dual or quad-core configurations, allowing for scalable parallelism in system-on-chip implementations while maintaining power efficiency through asynchronous multi-processing. The integer execution subsystem features a 4-way issue unit that handles load/store operations alongside arithmetic tasks, incorporating low-latency arithmetic logic units (ALUs) tailored for mobile-specific optimizations such as rapid integer computations in battery-constrained scenarios.[14] Krait's branch prediction mechanism utilizes an advanced dynamic predictor, including handling for indirect branches, to significantly reduce misprediction penalties and enhance control flow efficiency in branching-heavy code paths common to mobile applications.[15]Pipeline and Execution
The Krait processor employs an 11-stage integer pipeline designed for high-performance mobile computing, featuring a 3-way decoder that fetches and decodes up to three instructions per cycle, followed by a 4-way out-of-order speculative issue mechanism to maximize instruction-level parallelism by dispatching instructions dynamically based on resource availability and data dependencies.[14] This superscalar execution allows the core to issue up to four instructions per cycle, enabling efficient handling of complex workloads while maintaining compatibility with the ARMv7-A instruction set architecture. The pipeline's speculative nature supports branch prediction to minimize disruptions from control hazards, with recovery mechanisms to roll back incorrect speculations efficiently. Krait's execution units include multiple arithmetic logic units (ALUs) for integer operations, dedicated multiply-accumulate (MAC) units for fused multiply-add instructions, and load/store queues to manage memory access parallelism, collectively supporting the 4-way dispatch rate for sustained throughput in integer and floating-point computations.[14] These units are organized across seven execution ports, allowing concurrent execution of diverse instruction types, such as up to two loads and one store per cycle, which enhances overlap between computation and memory operations in out-of-order processing. The load/store queues buffer pending memory requests, preventing stalls from cache misses and enabling non-blocking execution until data dependencies resolve. To support speculation and ensure precise exception handling, Krait incorporates a reorder buffer with a capacity of 40 entries, which tracks in-flight instructions, retires them in program order, and facilitates recovery from branch mispredictions by squashing speculative results when necessary. This buffer size strikes a balance for mobile constraints, providing sufficient window for reordering without excessive power overhead, and integrates with the pipeline to commit results accurately even under high speculation rates typical of bursty application code. Power management is deeply integrated into the pipeline through dynamic clock and voltage scaling (DCVS), which adjusts frequency and voltage on-the-fly based on workload demands, allowing the core to throttle for low-activity periods and ramp up for intensive tasks in mobile environments. Complementing this, asynchronous symmetric multiprocessing (aSMP) enables independent per-core scaling, further optimizing energy use by isolating bursty workloads to active cores while idling others, resulting in up to 40% power savings compared to synchronous designs.Memory Hierarchy
The memory hierarchy in the Krait processor is engineered for efficient data access in power-constrained mobile environments, featuring small, fast per-core caches backed by a shared larger cache to support multi-core operations. This design prioritizes low latency for common access patterns while enabling coherence across cores through hardware mechanisms.[16] At the lowest level, the L0 cache consists of a 4 KB direct-mapped instruction cache and a 4 KB direct-mapped data cache, integrated directly into the execution pipeline for minimal access latency during instruction fetch and immediate data operations.[17] The L1 caches expand capacity with a 16 KB 4-way set-associative instruction cache and a 16 KB 4-way set-associative data cache per core, utilizing 64-byte cache lines and a write-back policy to reduce bus traffic by deferring writes to lower levels until eviction or explicit flushing.[17][18] The unified L2 cache is shared among cores, configured as 1 MB (8-way set-associative, 128-byte lines) per dual-core cluster in dual-core implementations or 2 MB total in quad-core setups, where each core has optimized access to half the capacity for balanced performance.[17] This level employs an inclusive design relative to L1 data in later variants like Krait 450, ensuring data consistency without excessive duplication.[19] Krait supports physically indexed and physically tagged caches at the L2 level, which aids in efficient address translation and hit detection in virtualized environments. For multi-core coherence, the architecture integrates the AXI Coherency Port (ACP), an AMBA AXI slave interface that allows external masters like DMA engines to snoop and maintain cache consistency with the L2 without software intervention.[16][20]Variants and Implementations
Krait 200
The Krait 200 served as the inaugural CPU core in Qualcomm's Krait family, debuting in the Snapdragon S4 series of system-on-chips (SoCs) launched in 2012.[21] This dual-core implementation marked Qualcomm's shift to a custom ARMv7-A architecture optimized for mobile devices, emphasizing a balance between computational performance and power efficiency to extend battery life in early smartphones and tablets. Built on a 28 nm low-power (LP) process node using high-k metal gate (HKMG) technology, the Krait 200 enabled significant reductions in power consumption compared to prior generations, achieving up to 40% lower energy use in typical workloads through techniques like asynchronous symmetric multiprocessing (aSMP).[21][22] In configurations such as the MSM8960 SoC, the Krait 200 operated at clock speeds ranging from 1.2 GHz to a maximum of 1.7 GHz for its dual cores, allowing dynamic scaling to match application demands while prioritizing thermal and power constraints.[3][22] It delivered a performance rating of 3.3 Dhrystone MIPS per MHz (DMIPS/MHz), representing approximately 30% improvement over ARM Cortex-A9 cores at equivalent frequencies and enabling efficient handling of multitasking and multimedia tasks in 2012-era devices.[23] This metric underscored the core's focus on delivering responsive user experiences without excessive power draw, with independent voltage and frequency control per core via aSMP further enhancing battery longevity during mixed workloads.[21] A key unique aspect of the Krait 200 was its first implementation of an out-of-order execution backend within Qualcomm's custom ARM design, featuring in-order fetch and decode stages followed by dynamic instruction reordering for improved throughput—up to three instructions decoded and executed per cycle.[24] This architecture, which built on the general superscalar and VLIW elements of the Krait family (as detailed in the Core Design section), was paired with the Adreno 225 GPU in the Snapdragon S4 series to support integrated graphics acceleration for 1080p video playback and basic 3D rendering.[21][22] Overall, these features positioned the Krait 200 as a foundational step in Qualcomm's evolution toward high-efficiency mobile processing, setting performance baselines for subsequent variants.Krait 300
The Krait 300 is a quad-core capable CPU architecture developed by Qualcomm as an evolution of the Krait family, primarily integrated into mid-range Snapdragon 600 series system-on-chips (SoCs) such as the APQ8064T. Fabricated on a 28 nm low-power (LP) process node, it enables efficient operation in power-constrained mobile environments while supporting advanced multitasking.[4] Clock speeds reach up to 1.9 GHz per core in quad-core configurations, allowing for robust performance in multimedia and general computing tasks on devices like high-tier smartphones.[25] In terms of performance, the Krait 300 delivers approximately 3.39 DMIPS/MHz, reflecting minor hardware optimizations such as refinements to branch prediction accuracy.[26][27] A key unique aspect of the Krait 300 in quad-core implementations is its expanded shared L2 cache of 2 MB, which significantly boosts multi-threaded performance, particularly in multimedia processing and parallel tasks.[28] This cache scaling from the dual-core focused Krait 200 design supports better data locality and reduced latency in scenarios involving multiple cores, contributing to smoother operation in 2013-era devices powered by Snapdragon 600 SoCs.[29]Krait 400 and 450
The Krait 400 served as the high-performance CPU core in Qualcomm's Snapdragon 800 and 801 system-on-chips, marking a significant step forward in mobile processing capabilities during 2013 and 2014. Fabricated using a 28 nm HPm (high-performance minus) process node, it featured a quad-core ARMv7 architecture capable of reaching clock speeds up to 2.3 GHz in the Snapdragon 800 configuration. This design emphasized balanced power efficiency and performance for premium smartphones, enabling robust multitasking and graphics-intensive applications. In the subsequent Snapdragon 801 variant, the same Krait 400 core was refined to support higher clock speeds of up to 2.5 GHz, providing a modest boost in computational throughput without altering the underlying microarchitecture.[30][31] The Krait 450 emerged as the refined successor to the Krait 400, integrated into the Snapdragon 805 SoC to push the boundaries of 28 nm HPm-based mobile computing. Retaining the quad-core ARMv7 structure on the identical 28 nm HPm process, it achieved peak clock speeds of up to 2.7 GHz, allowing for greater overall processing capacity compared to its predecessor. The Krait 450 delivered a modest performance uplift over the Krait 400, attributed to enhancements in branch prediction accuracy and execution unit throughput, which improved instruction handling efficiency under load. These architectural tweaks enabled more consistent performance in sustained workloads, distinguishing the Krait 450 as the final evolution of the Krait lineage before Qualcomm's shift to 64-bit ARMv8 designs.[32][10] A key unique aspect of the Krait 450 was its optimized thermal throttling mechanisms, which mitigated heat buildup more effectively than prior iterations, supporting higher sustained clock speeds during prolonged high-demand scenarios. This made it particularly well-suited for resource-intensive tasks like 4K video encoding/decoding and advanced mobile gaming, where maintaining peak performance without excessive power draw or temperature spikes was critical. In contrast to the Krait 400's focus on general premium device use, the Krait 450's refinements prioritized endurance in ultra-high-resolution multimedia and graphics applications, influencing its adoption in larger-form-factor devices with better cooling potential.Performance and Features
Efficiency and Benchmarks
The Krait processor family marked a significant advancement in power efficiency for mobile CPUs, primarily through adoption of the 28 nm process node and architectural optimizations including out-of-order execution. These changes enabled up to 60% higher performance while using up to 40% less power compared to the prior Scorpion core, allowing Krait to deliver higher computational throughput at comparable power levels.[22] Typical power consumption remained low, with cores operating under 2 W TDP during load in tested configurations.[33] Benchmark results highlighted Krait's strong integer performance, with DMIPS/MHz ratings of 3.3 to 3.51 in SPECint2000-like tests, reflecting efficient instruction throughput. In Geekbench evaluations, single-core scores typically ranged from 800 to 1200 points across variants and clock speeds up to 2.3 GHz, underscoring solid responsiveness for everyday tasks. Krait variants showed minor variations in these metrics, such as 3.3 DMIPS/MHz for the initial Krait 200. In workload-specific testing, Krait demonstrated superiority in JavaScript processing via the V8 engine and multimedia decoding tasks, leveraging its integrated NEON SIMD unit for accelerated vector operations. However, it trailed later architectures like the Cortex-A57 in sustained multi-core workloads, where thermal constraints limited prolonged high-performance operation. Comparative analyses revealed Krait was 20-30% faster clock-for-clock than the Cortex-A9, while offering burst performance on par with the Cortex-A15.Key Innovations
Krait processors featured a custom out-of-order execution engine that set them apart from contemporaneous ARM designs, which were predominantly in-order. Qualcomm engineered a 3-wide superscalar pipeline with dynamic scheduling, drawing inspiration from Intel's P6 microarchitecture while tailoring it for mobile constraints. The proprietary scheduler excelled at managing branch mispredictions and irregular control flows common in mobile applications, such as user interface rendering and multimedia processing, thereby minimizing latency stalls and boosting instruction throughput by up to 40% in mixed workloads compared to in-order predecessors. In terms of integrated heterogeneous computing, Krait enabled early CPU-GPU task handoff within Snapdragon SoCs, optimizing workloads like computer vision that foreshadowed AI acceleration. This capability leveraged shared memory coherence and low-overhead context switching to dynamically allocate compute-intensive tasks to the Adreno GPU, reducing CPU load and power draw in scenarios involving image processing or graphics overlays. By supporting unified programming models across CPU and GPU, Krait facilitated efficient heterogeneous execution, enhancing battery life without sacrificing responsiveness in premium mobile devices.[34] Security innovations in Krait included a robust TrustZone implementation augmented by Qualcomm's hardware root-of-trust for secure boot. The root-of-trust, embedded in the primary boot loader (PBL), verified firmware integrity using cryptographic hashes and keys stored in protected fuses, ensuring tamper-resistant initialization of the mobile OS. This hardware-enforced isolation partitioned sensitive operations—like payment processing or biometric authentication—into a secure world, preventing unauthorized access from the normal world and mitigating vulnerabilities in the boot chain.[35] Krait's scalability stemmed from its cluster-based design, which supported flexible dual- or quad-core configurations to balance performance and power in pre-Kryo Snapdragon platforms. Unlike rigid ARM reference designs, this allowed mixing core counts via coherent interconnects like the CCI, enabling big.LITTLE-like heterogeneity through clock and voltage scaling across clusters without distinct core types. Such adaptability optimized thermal envelopes in devices ranging from mid-range dual-core tablets to high-end quad-core smartphones, delivering sustained efficiency under varying loads.[36]Applications and Legacy
Integration in SoCs
The Krait CPU cores were primarily integrated into Qualcomm's Snapdragon S4, S4 Pro, 600, and 800 series system-on-chips (SoCs), forming the computational backbone for mobile devices during the early 2010s. These integrations leveraged Krait's ARMv7 architecture to deliver balanced performance and power efficiency within compact die sizes, often paired with complementary components for multimedia, graphics, and connectivity.[3][37] Specific SoC pairings included the Krait 200 in the Snapdragon S4 series, such as the MSM8960 (with integrated modem) featuring dual-core configurations clocked up to 1.7 GHz. The modemless APQ8064 in the S4 Pro series featured quad-core Krait 300 configurations clocked up to 1.7 GHz.[3] The Krait 300 appeared in the Snapdragon 600 series, notably the APQ8064T, which utilized quad-core setups at 1.7 GHz for mid-range applications. For higher-end tiers, the Krait 400 powered the Snapdragon 800 and 801 (MSM8974), with quad-core designs reaching up to 2.3 GHz, while the Krait 450 was incorporated in select variants like the Snapdragon 805 for enhanced performance.[37][22][38][39][40] At the system level, Krait cores were tightly coupled with Adreno GPUs for graphics acceleration—ranging from Adreno 225 in the S4 series to Adreno 320 in the S4 Pro and 600 series and Adreno 330 in the 800 series—enabling support for high-resolution displays and 3D rendering. The Hexagon DSP was integrated for offloading tasks like audio processing, image signal processing, and low-power always-on sensing, reducing CPU load and improving efficiency. Additionally, these SoCs featured built-in LTE modems in modem-equipped variants (e.g., MSM8960 and MSM8974), supporting multimode cellular connectivity including HSPA+ and 4G LTE for seamless data handling.[3][37][38][39][41] Configuration options emphasized flexibility for power and performance trade-offs, with dual-core Krait setups in the S4 Plus (MSM8960) prioritizing efficiency for entry- to mid-range devices, while quad-core arrangements in the S4 Pro, 600, and 800 series targeted demanding workloads like multitasking and gaming, all constrained within 28 nm process nodes to manage thermal and battery constraints.[3][37][42][38] All Krait-based Snapdragon SoCs were fabricated by TSMC on 28 nm processes, including high-performance mobile (HPM) variants for premium tiers like the 800 series, which optimized transistor density for faster clocks and lower leakage compared to mid-range 600 series implementations. This manufacturing choice enabled cost-effective scaling across device tiers, with premium configurations like the 800 series tuned for flagship performance and mid-range ones like the 600 series focused on balanced efficiency.[43][9][40][38]| SoC Series | Krait Variant | Core Configuration | Key Components | Process Node |
|---|---|---|---|---|
| Snapdragon S4 (MSM8960) | 200 | Dual-core up to 1.7 GHz | Adreno 225, Hexagon QDSP6, LTE modem | TSMC 28 nm |
| Snapdragon S4 Pro (APQ8064) | 300 | Quad-core up to 1.7 GHz | Adreno 320, Hexagon QDSP6 | TSMC 28 nm |
| Snapdragon 600 (APQ8064T) | 300 | Quad-core at 1.7 GHz | Adreno 320, Hexagon QDSP6 | TSMC 28 nm |
| Snapdragon 800/801 (MSM8974) | 400 | Quad-core up to 2.3 GHz | Adreno 330, Hexagon DSP, LTE modem | TSMC 28 nm HPM |
Notable Devices and Impact
The Krait architecture powered several flagship Android smartphones during its peak adoption period, enabling advanced multimedia and connectivity features in premium devices. The HTC One, released in 2013, utilized the Snapdragon 600 SoC with quad-core Krait 300 processors clocked at up to 1.7 GHz, marking one of the first widespread implementations of Krait in a high-end handset with a full HD display. Similarly, the Samsung Galaxy S4, launched the same year, incorporated the Snapdragon 600 variant featuring quad-core Krait 300 cores at 1.9 GHz, which supported its 1080p Super AMOLED screen and became a bestseller in the Android ecosystem. The LG Nexus 5, introduced in late 2013, employed the Snapdragon 800 SoC with quad-core Krait 400 processors at 2.3 GHz, serving as Google's reference device for stock Android and highlighting Krait's role in delivering smooth performance for emerging 1080p experiences. Krait's integration significantly influenced the mobile market between 2012 and 2014 by facilitating the transition to high-resolution displays and integrated LTE connectivity in Android flagships, which drove consumer demand for premium multimedia capabilities. Devices like the HTC One and Galaxy S4 were among the first to pair Krait-powered SoCs with 1080p panels and LTE modems, enabling seamless 4G data speeds and video playback that set new benchmarks for mobile entertainment. This contributed to Qualcomm's growing dominance in the premium mobile SoC segment, as Snapdragon platforms with Krait outperformed competitors from Nvidia and Texas Instruments, capturing over 50% market share in high-end Android devices by 2013 and solidifying Qualcomm's leadership in integrated CPU-GPU-modem solutions. The legacy of Krait lies in its demonstration of Qualcomm's expertise in custom ARM-based designs, which laid the groundwork for more advanced heterogeneous architectures in mobile computing. As an in-house evolution of ARMv7, Krait's efficient per-core performance inspired the shift toward tailored CPU implementations, directly preceding Qualcomm's Kryo cores introduced in 2015 with the Snapdragon 820 SoC to support 64-bit computing and big.LITTLE configurations. This transition aligned with the broader industry adoption of ARM's big.LITTLE standard, where high-performance and efficiency cores coexist, becoming a staple in subsequent mobile processors from multiple vendors. Despite its successes, early implementations of the Snapdragon 800 series with Krait 400 encountered thermal challenges, particularly overheating under sustained loads that necessitated aggressive throttling to maintain stability. Reports from 2013-2014 highlighted issues in devices like the Nexus 5 and HTC One M8, where high clock speeds led to elevated temperatures, prompting Qualcomm to refine thermal mitigation strategies in later designs. These experiences influenced the evolution of mobile SoC engineering, emphasizing better power envelope management and cooling solutions in the post-Krait era.References
- https://en.wikichip.org/wiki/qualcomm/snapdragon_600