Brendan Gregg

SEPTEMBER 25, 2021

As a Xen guest, this profile was gathered using perf(1) and the kernel's software cpu-clock soft interrupts, not the hardware NMI. That chance encounter, coupled with the Netflix's fault-tolerant cloud, gave me enough confidence to suggest trying tsc in production as a workaround for the issue. But I'm not completely sure.

Speed 103

Speed Java AWS Virtualization 103

Brendan Gregg

MARCH 16, 2024

2015-2020: Overhead As part of production rollout I did many performance overhead tests, which I've described publicly before: The overhead of adding frame pointers to everything (libc and Java) was usually less than 1%, with one exception of 10%. I spent years as a cloud customer where I couldn't access many HW-level things.)

Java 137

Java Cache Google Hardware 137

Percona

JANUARY 17, 2019

On multi-core machines – which is the majority of the hardware nowadays – and in the cloud, we have multiple cores available for use. Multiple queries run at the same time will be using different threads and will utilize more than one CPU core. With faster disks (i.e. AWS Aurora (based on MySQL 5.6) select one row only, no index.

Cache 72

Cache C++ AWS Airlines 72

Brendan Gregg

JULY 4, 2021

It's an exciting time for developments in computer performance, not just for the BPF technology (which I often [write about]) but also for processors with 3D stacking and cloud vendor CPUs (e.g., This was a chance to talk about other things I've been working on, such as the present and future of hardware performance. Ford, et al., “TCP

Performance 52

Performance Latency Hardware Storage 52

Brendan Gregg

NOVEMBER 28, 2017

Hardware virtualization for cloud computing has come a long way, improving performance using technologies such as VT-x, SR-IOV, VT-d, NVMe, and APICv. At Netflix, we've been using these technologies as they've been made available for instance types in the AWS EC2 cloud. I'd expect between 0.1% small (thanks @cperciva ). ## 3.

Virtualization 40

Virtualization AWS Hardware Storage 40

Brendan Gregg

FEBRUARY 28, 2023

TCP Extensions for Multipath Operation with Multiple Addresses,” [link] , Mar 2020 [Gregg 20] Brendan Gregg, “Systems Performance: Enterprise and the Cloud, Second Edition,” Addison-Wesley, 2020 [Hruska 20] Joel Hruska, “Intel Demos PCIe 5.0 Ford, et al., “TCP

Performance 52

Performance Latency Cache Virtualization 52

Brendan Gregg

SEPTEMBER 25, 2021

As a Xen guest, this profile was gathered using perf(1) and the kernel's software cpu-clock soft interrupts, not the hardware NMI. That chance encounter, coupled with the Netflix's fault-tolerant cloud, gave me enough confidence to suggest trying tsc in production as a workaround for the issue. But I'm not completely sure.

Speed 52

Speed Java AWS Virtualization 52

Brendan Gregg

SEPTEMBER 25, 2021

As a Xen guest, this profile was gathered using perf(1) and the kernel's software cpu-clock soft interrupts, not the hardware NMI. That chance encounter, coupled with the Netflix's fault-tolerant cloud, gave me enough confidence to suggest trying tsc in production as a workaround for the issue. But I'm not completely sure.

Speed 40

Speed Java AWS Virtualization 40

ACM Sigarch

APRIL 26, 2018

HPU: Holographic Processing Unit (HPU) is the specific hardware of Microsoft’s Hololens. SPU: Stream Processing Unit (SPU) is related to the specialized hardware to process the data streams of video. TPU: Tensor Processing Unit (TPU) is Google’s specialized hardware for neural network.

Hardware 100

Hardware Artificial Intelligence Architecture Google 100

Brendan Gregg

AUGUST 8, 2017

Load averages are an industry-critical metric – my company spends millions auto-scaling cloud instances based on them and other metrics – but on Linux there's some mystery around them. They can be also useful when a single value of demand is desired, such as for a cloud auto scaling rule. I've never seen an explanation.

Latency 111

Latency Metrics C++ Systems 111

Alex Russell

MARCH 6, 2021

Devices and networks have evolved too: Alex Russell @slightlylate An update on mobile CPUs and the Performance Inequality Gap: Mid-tier Android devices (~$300) now get the single-core performance of a 2014 iPhone and the multi-core perf of a 2015 iPhone. Hardware Past As Performance Prologue. Mind The Gap.

Performance 127

Performance Network Mobile Metrics 127

Smashing Magazine

JANUARY 6, 2020

On the other hand, we have hardware constraints on memory and CPU due to JavaScript parsing times (we’ll talk about them in detail later). In real-life world, most products aren’t even close: an median bundle size today is around 417KB , which is up 42% compared to early 2015. Assets Optimizations.

Performance 112

Performance Cache Servers Network 112

Smashing Magazine

JANUARY 7, 2019

On the other hand, we have hardware constraints on memory and CPU due to JavaScript parsing times (we’ll talk about them in detail later). In real-life world, most products aren’t even close: an average bundle size today is around 400KB , which is up 35% compared to late 2015. We could also go beyond the bundle size budget though.

Performance 111

Performance Cache Network Metrics 111

Smashing Magazine

JANUARY 11, 2021

On the other hand, we have hardware constraints on memory and CPU due to JavaScript parsing and execution times (we’ll talk about them in detail later). compared to early 2015. Also check DNS Performance comparison and Cloud Peformance Comparison. From Fast By Default: Modern loading best practices by Addy Osmani (Slide 19).

Performance 145

Performance Cache Media Metrics 145