Performance overview - Qualcomm Dragonwing Documentation

Optimize the performance of Qualcomm^® Linux^® with the following key areas:

Performance-impacting features such as the CPU scheduler, CPU frequency governor, dynamic voltage and frequency scaling (DVFS) governors, Userspace Resource Manager (URM), and memory.
Tools for performance analysis to identify and diagnose issues.
Configuration and customization options for Linux settings to achieve enhanced performance.
Troubleshooting methods to resolve performance-related problems.
Measurement procedures to evaluate system performance.
Performance dashboards for QCS6490, QCS5430, and Qualcomm Dragonwing^™ IQ-615 for real-time insights.

Qualcomm Linux is a single Linux distribution that works with many system-on-chips (SoCs) for development and evaluation kits. See Hardware SoCs that are supported on Qualcomm Linux.

The following guides contain supplementary information, such as device specifications, supported resource opcodes, chipset-specific configuration and customization settings, performance dashboards, and measurement procedures:

Subsystem dependencies

The performance of the software depends on the CPU, GPU, and DDR subsystems. Qualcomm Linux uses the Qualcomm^® Kryo^™ CPU, which has the following clusters:

Prime cluster for high-performance CPU cores.
Gold cluster for balanced power and performance.
Silver cluster for low-power CPU cores, ideal for light-weight applications.

Cache memory is categorized into the following three levels:

L1 is the smallest and fastest cache level, storing both instructions (L1 I) and data (L1 D).
L2 and L3 are larger but slower cache levels, mainly for data storage.

The following tables list the specifications for the subsystems on QCS6490, QCS5430, and Dragonwing IQ-615:

QCS6490
QCS5430
Dragonwing IQ-615

Table : Specifications for QCS6490 CPU, GPU, and DDR subsystems

Specifications	QCS6490
Core type	Kryo Prime	Kryo Gold	Kryo Silver
Number of CPUs	1	3	4
CPU maximum frequency	2.7 GHz	2.4 GHz	1.9 GHz
L1 I cache	32 kB	32 kB/core	32 kB/core
L1 D cache	32 kB	32 kB/core	32 kB/core
L2 cache	256 kB	256 kB/core	128 kB/core
L3 cache	2 MB
GPU	Qualcomm^® Adreno^™ 643 GPU
GPU maximum frequency	812 MHz
DDRSS	- Supports dual-channel non-package-on-package high-speed LPDDR5/LPDDR4 SDRAM. - LPDDR5 SDRAM is designed for a 3200 MHz clock (2 × 16‑bit). - LPDDR4 SDRAM is designed for a 2133 MHz clock (2 × 16‑bit).

Table : Specifications for QCS5430 (FP 1 and FP 2) CPU, GPU, and DDR subsystems

Specifications	QCS5430 FP 1 (feature pack 1)	QCS5430 FP 2
Core type	Kryo Gold	Kryo Silver	Kryo Prime	Kryo Gold	Kryo Silver
Number of CPUs	2	4	1	3	4
CPU maximum frequency	2.1 GHz	1.8 GHz	2.2 GHz	2.1 GHz	1.8 GHz
L1 I cache	32 kB/core	32 kB/core	32 kB	32 kB/core	32 kB/core
L2 cache	256 kB/core	128 kB/core	256 kB	256 kB/core	128 kB/core
L3 cache	2 MB
GPU	Qualcomm^® Adreno^™ 642L GPU
GPU maximum frequency	315 MHz
DDRSS	- Supports dual-channel non-package-on-package high-speed LPDDR5/LPDDR4 SDRAM. - LPDDR5 SDRAM is designed for a 3200 MHz clock (2 × 16‑bit). - LPDDR4 SDRAM is designed for a 2133 MHz clock (2 × 16‑bit).

Table : Specifications for QCS5430 (FP 2.5 and FP 3) CPU, GPU, and DDR subsystems

Specifications	QCS5430 FP 2.5	QCS5430 FP 3
Core type	Kryo Prime	Kryo Gold	Kryo Silver	Kryo Prime	Kryo Gold	Kryo Silver
Number of CPUs	1	3	4	1	3	4
CPU maximum frequency	2.38 GHz	2.4 GHz	1.8 GHz	2.38 GHz	2.4 GHz	1.8 GHz
L1 I cache	32 kB	32 kB/core	32 kB/core	32 kB	32 kB/core	32 kB/core
L2 cache	256 kB	256 kB/core	128 kB/core	256 kB	256 kB/core	128 kB/core
L3 cache	2 MB
GPU	Qualcomm Adreno 642L GPU
GPU maximum frequency	550 MHz
DDRSS	- Supports dual-channel non-package-on-package high-speed LPDDR5/LPDDR4 SDRAM. - LPDDR5 SDRAM is designed for a 3200 MHz clock (2 × 16‑bit). - LPDDR4 SDRAM is designed for a 2133 MHz clock (2 × 16‑bit).

Table : Specifications for Dragonwing IQ-615 CPU, GPU, and DDR subsystems

Specifications	Dragonwing IQ-615
Core type	Kryo Gold	Kryo Silver
Number of CPUs	2	6
CPU maximum frequency	1.9 GHz	1.6 GHz
L1 I cache	64 kB/core	32 kB/core
L1 D cache	64 kB/core	32 kB/core
L2 cache	256 kB/core	64 kB/core
L3 cache	1 MB
GPU	Qualcomm^® Adreno^™ 612 GPU
GPU maximum frequency	845 MHz
DDRSS	Two-channel high-speed memory – LPDDR4X (2 × 16‑bit) 1555 MHz.

​Subsystem dependencies

​Next steps

Subsystem dependencies

Next steps