2.2. Performance Guide¶

2.2.1.1.1. LMBench¶

LMBench is a collection of microbenchmarks of which the memory bandwidth and latency related ones are typically used to estimate processor memory system performance.

Latency: lat_mem_rd-stride128-szN, where N is equal to or smaller than the cache size at given level measures the cache miss penalty. N that is at least double the size of last level cache is the latency to external memory.

Bandwidth: bw_mem_bcopy-N, where N is is equal to or smaller than the cache size at a given level measures the achivable memory bandwidth from software doing a memcpy() type operation. Typical use is for external memory bandwidth calculation. The bandwidth is calculated as byte read and written counts as 1 which should be roughly half of STREAM copy result.

Table: LM Bench Metrics

2.2.1.1.2. Dhrystone¶

Dhrystone is a core only benchmark that runs from warm L1 caches in all modern processors. It scales linearly with clock speed. For standard ARM cores the DMIPS/MHz score will be identical with the same compiler and flags.

Table: Dhrystone Benchmark

2.2.1.1.3. Whetstone¶

Table: Whetstone Benchmark

2.2.1.1.4. Linpack¶

Linpack measures peak double precision (64 bit) floating point performance in sloving a dense linear system.

Table: Linpack Benchmark

2.2.1.1.5. NBench¶

Table: NBench Benchmarks

2.2.1.1.6. Stream¶

STREAM is a microbenchmarks for measuring data memory system performance without any data reuse. It is designed to miss on caches and exercise data prefetcher and apeculative accesseses. it uses double precision floating point (64bit) but in most modern processors the memory access will be the bottleck. The four individual scores are copy, scale as in multiply by constant, add two numbers, and triad for multiply accumulate. For bandwidth a byte read counts as one and a byte written counts as one resulting in a score that is double the bandwidth LMBench will show.

Table: Stream

2.2.1.1.7. CoreMarkPro¶

CoreMark®-Pro is a comprehensive, advanced processor benchmark that works with and enhances the market-proven industry-standard EEMBC CoreMark® benchmark. While CoreMark stresses the CPU pipeline, CoreMark-Pro tests the entire processor, adding comprehensive support for multicore technology, a combination of integer and floating-point workloads, and data sets for utilizing larger memory subsystems.

Table: CoreMarkPro

2.2.1.1.8. MultiBench¶

MultiBench™ is a suite of benchmarks that allows processor and system designers to analyze, test, and improve multicore processors. It uses three forms of concurrency: Data decomposition: multiple threads cooperating on achieving a unified goal and demonstrating a processor’s support for fine grain parallelism. Processing multiple data streams: uses common code running over multiple threads and demonstrating how well a processor scales over scalable data inputs. Multiple workload processing: shows the scalability of general-purpose processing, demonstrating concurrency over both code and data. MultiBench combines a wide variety of application-specific workloads with the EEMBC Multi-Instance-Test Harness (MITH), compatible and portable with most any multicore processors and operating systems. MITH uses a thread-based API (POSIX-compliant) to establish a common programming model that communicates with the benchmark through an abstraction layer and provides a flexible interface to allow a wide variety of thread-enabled workloads to be tested.

Table: Multibench

2.2.1.1.9. Spec2K6¶

CPU2006 is a set of benchmarks designed to test the CPU performance of a modern server computer system. It is split into two components, the first being CINT2006, the other being CFP2006 (SPECfp), for floating point testing.

SPEC defines a base runtime for each of the 12 benchmark programs. For SPECint2006, that number ranges from 1000 to 3000 seconds. The timed test is run on the system, and the time of the test system is compared to the reference time, and a ratio is computed. That ratio becomes the SPECint score for that test. (This differs from the rating in SPECINT2000, which multiplies the ratio by 100.)

As an example for SPECint2006, consider a processor which can run 400.perlbench in 2000 seconds. The time it takes the reference machine to run the benchmark is 9770 seconds. Thus the ratio is 4.885. Each ratio is computed, and then the geometric mean of those ratios is computed to produce an overall value.

Table: Spec2K6

2.2.1.2.1. Boot media: MMCSD¶

Table: Boot time MMC/SD

2.2.1.2.2. Boot media: NAND¶

Table: Boot time MMC/SD

2.2.1.6.1. GLBenchmark¶

Run GLBenchmark and capture performance reported Display rate (Fps), Fill rate, Vertex Throughput, etc. All display outputs (HDMI, Displayport and/or LCD) are connected when running these tests

2.2.1.6.2. GFXBench¶

Run GFXBench and capture performance reported (Score and Display rate in fps). All display outputs (HDMI, Displayport and/or LCD) are connected when running these tests

Table: GFXBench

2.2.1.6.3. Glmark2¶

Run Glmark2 and capture performance reported (Score). All display outputs (HDMI, Displayport and/or LCD) are connected when running these tests

Table: Glmark2

2.2.1.7.1. H264¶

Table: Gstreamer H264 in AVI Container Decode Performance

2.2.1.7.2. MPEG4¶

Table: GStreamer MPEG4 in 3GP Container Decode Performance

2.2.1.7.3. MPEG2¶

Table: GStreamer MPEG2 in MP4 Container Decode Performance

2.2.1.8.1. TCP Throughput¶

Table: TCP Throughput

TCP Throughput Interrupt Pacing

Table: TCP Throughput Interrupt Pacing

2.2.1.8.2. UDP Throughput¶

UDP Throughput Egress

Table: UDP Throughput Egress

UDP Throughput Ingress

Table: UDP Throughput Ingress

• iperf version 2.0.5
• For receive performance, on DUT, invoke iperf in server mode.

iperf -s -u

• For transmit performance, on DUT, invoke iperf in client mode.

iperf -c <server ip> -b <bandwidth limit> -f M -t 60

2.2.1.9.1. PCIe-ETH¶

Table: PCI Ethernet

2.2.1.10.1. AM335X-EVM¶

2.2.1.10.2. OMAPL138-LCDK¶

2.2.1.11.1. AM654x-EVM¶

2.2.1.11.2. K2G-EVM¶

2.2.1.12.1. K2G-EVM¶

2.2.1.12.2. AM654X-EVM¶

2.2.1.13.1. AM57XX-EVM¶

2.2.1.13.2. AM654x-EVM¶

2.2.1.13.3. K2G-EVM¶

2.2.1.14.1. mSATA Driver¶

AM57XX-EVM

• Filesize used is : 1G
• MSATA Harddisk used is: SMS200S3/30G Kingston mSATA SSD drive

2.2.1.15.1. AM335x-EVM¶

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

2.2.1.15.2. AM57XX-EVM¶

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

The performance numbers were captured using the following:

• SanDisk 8GB SDHC UHS Memory Card
• Partition was mounted with async option

2.2.1.15.3. AM654x-EVM¶

The performance numbers were captured using the following:

• SanDisk 8GB SDHC UHS Memory Card
• Partition was mounted with async option

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

2.2.1.15.4. K2G-EVM¶

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

2.2.1.15.5. OMAPL138-LCDK¶

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

2.2.1.20.1. MUSB/XHCI Host controller¶

Setup : Inateck ASM1153E USB hard disk is connected to usb0 port. File read/write performance data on usb0 port is captured.

Table: USB Host VFAT

Table: USB Host EXT2

Table: USBDEVICE NCM IPERF TCP THROUGHPUT

2.2.1.21.1. OpenSSL Performance¶

time -v openssl speed -elapsed -evp aes-128-cbc

2.2.1.21.2. IPSec Performance¶

Note: queue_len is set to 300 and software fallback threshold set to 9 to enable software support for optimal performance

2.2.2.1.1. LMBench¶

LMBench is a collection of microbenchmarks of which the memory bandwidth and latency related ones are typically used to estimate processor memory system performance.

Latency: lat_mem_rd-stride128-szN, where N is equal to or smaller than the cache size at given level measures the cache miss penalty. N that is at least double the size of last level cache is the latency to external memory.

Bandwidth: bw_mem_bcopy-N, where N is is equal to or smaller than the cache size at a given level measures the achivable memory bandwidth from software doing a memcpy() type operation. Typical use is for external memory bandwidth calculation. The bandwidth is calculated as byte read and written counts as 1 which should be roughly half of STREAM copy result.

Table: LM Bench Metrics

2.2.2.1.2. Dhrystone¶

Dhrystone is a core only benchmark that runs from warm L1 caches in all modern processors. It scales linearly with clock speed. For standard ARM cores the DMIPS/MHz score will be identical with the same compiler and flags.

Table: Dhrystone Benchmark

2.2.2.1.3. Whetstone¶

Table: Whetstone Benchmark

2.2.2.1.4. Linpack¶

Linpack measures peak double precision (64 bit) floating point performance in sloving a dense linear system.

Table: Linpack Benchmark

2.2.2.1.5. NBench¶

Table: NBench Benchmarks

2.2.2.1.6. Stream¶

STREAM is a microbenchmarks for measuring data memory system performance without any data reuse. It is designed to miss on caches and exercise data prefetcher and apeculative accesseses. it uses double precision floating point (64bit) but in most modern processors the memory access will be the bottleck. The four individual scores are copy, scale as in multiply by constant, add two numbers, and triad for multiply accumulate. For bandwidth a byte read counts as one and a byte written counts as one resulting in a score that is double the bandwidth LMBench will show.

Table: Stream CoreMarkPro ^^^^^^^^^^^^^^^^^^^^^^^^^^^ CoreMark®-Pro is a comprehensive, advanced processor benchmark that works with and enhances the market-proven industry-standard EEMBC CoreMark® benchmark. While CoreMark stresses the CPU pipeline, CoreMark-Pro tests the entire processor, adding comprehensive support for multicore technology, a combination of integer and floating-point workloads, and data sets for utilizing larger memory subsystems.

Table: CoreMarkPro

2.2.2.1.7. MultiBench¶

MultiBench™ is a suite of benchmarks that allows processor and system designers to analyze, test, and improve multicore processors. It uses three forms of concurrency: Data decomposition: multiple threads cooperating on achieving a unified goal and demonstrating a processor’s support for fine grain parallelism. Processing multiple data streams: uses common code running over multiple threads and demonstrating how well a processor scales over scalable data inputs. Multiple workload processing: shows the scalability of general-purpose processing, demonstrating concurrency over both code and data. MultiBench combines a wide variety of application-specific workloads with the EEMBC Multi-Instance-Test Harness (MITH), compatible and portable with most any multicore processors and operating systems. MITH uses a thread-based API (POSIX-compliant) to establish a common programming model that communicates with the benchmark through an abstraction layer and provides a flexible interface to allow a wide variety of thread-enabled workloads to be tested.

Table: Multibench

2.2.2.1.8. Spec2K6¶

CPU2006 is a set of benchmarks designed to test the CPU performance of a modern server computer system. It is split into two components, the first being CINT2006, the other being CFP2006 (SPECfp), for floating point testing.

SPEC defines a base runtime for each of the 12 benchmark programs. For SPECint2006, that number ranges from 1000 to 3000 seconds. The timed test is run on the system, and the time of the test system is compared to the reference time, and a ratio is computed. That ratio becomes the SPECint score for that test. (This differs from the rating in SPECINT2000, which multiplies the ratio by 100.)

As an example for SPECint2006, consider a processor which can run 400.perlbench in 2000 seconds. The time it takes the reference machine to run the benchmark is 9770 seconds. Thus the ratio is 4.885. Each ratio is computed, and then the geometric mean of those ratios is computed to produce an overall value.

Table: Spec2K6

2.2.2.2.1. Shield (dedicated core) Case¶

shield_shell()
{
create_cgroup nonrt 0
create_cgroup rt 1
for pid in $(cat /sys/fs/cgroup/tasks); do /bin/echo $pid > /sys/fs/cgroup/nonrt/tasks; done
/bin/echo $$ > /sys/fs/cgroup/rt/tasks
}

2.2.2.3.1. Boot media: MMCSD¶

Table: Boot time MMC/SD

2.2.2.3.2. Boot media: NAND¶

Table: Boot time NAND

2.2.2.7.1. Performance (Fps)¶

Table: GLBenchmark 2.5 Performance

2.2.2.7.2. Vertex Throughput¶

Table: GLBenchmark 2.5 Vertex Throughput

2.2.2.7.3. Pixel Throughput¶

Table: GLBenchmark 2.5 Pixel Throughput

2.2.2.8.1. H264¶

Table: Gstreamer H264 in AVI Container Decode Performance

2.2.2.8.2. MPEG4¶

Table: GStreamer MPEG4 in 3GP Container Decode Performance

2.2.2.8.3. MPEG2¶

Table: GStreamer MPEG2 in MP4 Container Decode Performance

2.2.2.9.1. TCP Throughput¶

Table: TCP Throughput No Interrupt Pacing

Table: TCP Throughput Interrupt Pacing

2.2.2.9.2. UDP Throughput¶

Table: UDP Throughput Egress

Table: UDP Throughput Ingress

• iperf version 2.0.5
• For receive performance, on DUT, invoke iperf in server mode.

iperf -s -u

• For transmit performance, on DUT, invoke iperf in client mode.

iperf -c <server ip> -b <bandwidth limit> -f M -t 60

2.2.2.10.1. PCIe-ETH¶

Table: PCI Ethernet

2.2.2.12.1. AM574x-IDK¶

2.2.2.12.2. AM654X-IDK¶

2.2.2.14.1. AM574x-IDK¶

2.2.2.14.2. AM654x-IDK¶

2.2.2.15.1. mSATA Driver¶

• Filesize used is : 1G
• MSATA Harddisk used is: SMS200S3/30G Kingston mSATA SSD drive

2.2.2.16.1. AM574x-IDK¶

2.2.2.16.2. AM654x-IDK¶

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

2.2.2.21.1. MUSB/XHCI Host controller¶

Setup : Inateck ASM1153E USB hard disk is connected to usb0 port. File read/write performance data on usb0 port is captured.

Table: USB Host VFAT

Table: USB Host EXT2

2.2.2.22.1. OpenSSL Performance¶

time -v openssl speed -elapsed -evp aes-128-cbc

2.2.2.22.2. IPSec Performance¶

Note: queue_len is set to 300 and software fallback threshold set to 9 to enable software support for optimal performance

2.2.2.22.3. PRU Ethernet¶

Table: ** PRU UDP Throughput Egress**

Table: PRU UDP Throughput Ingress