Update laysim-GR740 v0.10 for Evaluation

laysim-GR740 v0.10 Quad Core Processor Emulator

laysim-GR740 consists of two types of emulators: laysim-gr740 and laysim-gr740-dbt. 

laysim-gr740 is a graphic user interface (GUI)-based and cycle-accurate emulator for the GR740 quad core processor based on interpretation method. It includes an embedded source-level debugger and supports MMU for Linux-SMP, XtratuM-SMP and AIR/GMV. 

laysim-gr740-dbt is a GUI-based high-performance and cycle-approximate processor emulator with dynamic binary translation method. 

Both laysim-gr740 and laysim-gr740-dbt support Windows and Linux x64 platform including CentOS 7/8, Ubuntu 18.04 and Debian 10.3. Also two types of CLI (Command Line Interface) version, laysim-gr740-cli and laysim-gr740-dbt-cli, are provided especially for RTEMS5 testsuites. These versions have command line interfaces without a GUI, so the performance of CLI version shows higher than the GUI version of laysim.

laysim-GR740 v0.10 and its user's manual can be downloaded from Download article. (https://layright.blogspot.com/2020/02/download-laysim-leon3.html)

laysim-GR740 v0.10 Performance and Cycle Accuracy

The performance and cycle accuracy of laysim-GR740 were measured on Intel i7-6700K @4.00GHz CentOS 7 64bit (4 cores and 8GB RAM) under VMware Workstation 15 Pro.

In general, a processor emulator based on interpretation method cannot not satisfy the real-time performance of guest CPU when the guest CPU operates at 70MHz or higher. A processor emulator adopting binary translation method can achieve its real-time performance to 500 MIPS at maximum, so it can meet the real-time performance of single core of GR740 processor which runs at 250MHz. However, a processor emulator based on DBT can never satisfy real-time performance under multi-core environment.

When analyzing the performance of laysim-GR740 with various multi-core benchmarking programs such as NPB, SPEC 2012 OMP, RTEMS5-SMP, the real-time performance of laysim-gr740-dbt can achieve 45% performance at maximum and 23% performance at average. So KARI/FSW team are developing a new processor emulator which based on multi-thread framework with DBT to satisfy the real-time performance of GR740 multicores.

1. Dhrystone 

Dhrystone benchmark with 1000000 was compiled -O3 option using BCC-2.1.0. Only laysim-gr740-dbt (195.45%) outperforms the real-time performance of GR740 processor (250MHz). TSIM3-LEON4 Evaluation from Gaiser only operates as FAST UART. In terms of cycle accuracy, laysim-GR740 shows the cycle accuracy close to 100%.

2. RTEMS 5.1.0-RC1 SMP09

The SMP09 of RTEMS 5.1.0-RC1 is an example of RTEMS5-SMP. It uses approximately 77% load of each core. All emulators cannot meet the real-time performance of the GR740 processor. 

1) laysim-gr740 shows 10.60% real-time performance and 99.61% cycle accuracy, and it shows 28.98% real-time performance and 32.56% cycle accuracy on FAST UART option.

2) laysim-gr740-dbt shows 18.06% real-time performance and 96.90% cycle accuracy, and it shows 51.36% real-time performance and 30.58% cycle accuracy on FAST UART option.

3) TSIM3-LEON4 Evaluation only works as FAST UART. It shows 28.13% real-time performance and 31.79% cycle accuracy.

3. RTEMS 5.1.0-RC1 SMPATOMIC01

The SMPATOMIC01 of RTEMS 5.1.0-RC1 is an example of RTEMS5-SMP. It uses approximately 99% load of each core. All emulators cannot meet the real-time performance of the GR740 processor. 

1) laysim-gr740 shows 6.67% real-time performance and 100.74% cycle accuracy, and it shows 7.14% real-time performance and 94.92% cycle accuracy on FAST UART option.

2) laysim-gr740-dbt shows 44.92% real-time performance and 100.77% cycle accuracy, and it shows 49.72% real-time performance and 94.79% cycle accuracy on FAST UART option.

3) TSIM3-LEON4 Evaluation only works as FAST UART. It shows 16.81% real-time performance and 94.05% cycle accuracy.