Clik here to view.
Intel Details Skymont
Previously Intel’s Skymont slides were published in low resolution, and I wrote a short article on them. Now, the presentation is public with higher resolution slides and presenter audio. Because...
View ArticleClik here to view.
Testing AMD’s Bergamo: Zen 4c Spam
Server CPUs have pushed high core counts for a long time, though they way they got high core counts has varied. Bergamo is AMD’s move to increase core counts beyond what scaling up an interconnect can...
View ArticleClik here to view.
Testing AMD’s Giant MI300X
Editor Note (6/26/2024): We have rephrased the acknowledgment section to make more clear that we got no direct support from AMD on this article. Our testing is fully independent, and AMD did not have...
View ArticleClik here to view.
Examining the Nintendo Switch (Tegra X1) Video Engine
The Tegra X1 SoC featured in Nintendo’s Switch was meant to fill a variety of market segments including Android set top boxes and automotive applications. Hardware video encode and decode are vital to...
View ArticleClik here to view.
The Snapdragon X Elite’s Adreno iGPU
Qualcomm is no stranger to integrated graphics. Their Adreno GPU line has served through many generations of Snapdragon cell phone SoCs. But Qualcomm was never content to stay within the cell phone...
View ArticleClik here to view.
China’s New(ish) SW26010-Pro Supercomputer at SC23
Computing power has emerged as a crucial national resource. Ever since the first general purpose computer, ENIAC, was used to calculate artillery and bomb ballistics, compute applications have...
View ArticleClik here to view.
Inside Kepler, Nvidia’s Strong Start on 28 nm
Nvidia’s Fermi architecture was ambitious and innovative, offering advances in GPU compute along with features like high tessellation performance. However Terascale 2’s more traditional approach...
View ArticleClik here to view.
GCN, AMD’s GPU Architecture Modernization
AMD’s Terascale architecture became very competitive as it matured with the HD 5000 and 6000 series. But using GPUs for general purpose compute was trending around the 2010s and AMD didn’t want to...
View ArticleClik here to view.
Cortex A57, Nintendo Switch’s CPU
In the early 2010s, Arm’s 32-bit cores had established themselves in cell phones and tablets. But rising memory capacities in those devices meant Arm would have to go 64-bit sooner or later. On top of...
View ArticleClik here to view.
AMD’s CDNA 3 Compute Architecture
AMD has a long history of vying for GPU compute market share. Ever since Nvidia got first dibs with their Tesla architecture, AMD has been playing catch up. Terascale 3 moved from VLIW5 to VLIW4 to...
View ArticleClik here to view.
Nintendo Switch’s iGPU: Maxwell Nerfed Edition
Graphics performance is vital for any console chip. Nintendo selected Nvidia’s Tegra X1 for their Switch handheld console. Tegra X1 is designed to maximize graphics performance in a limited power...
View ArticleClik here to view.
A New Year and New Tests: GPU L1 Cache Bandwidth
In my past articles on GPUs, I didn’t have good measurements for L1 cache bandwidth. Microbenchmarking cache bandwidth is harder on GPUs than CPUs. That’s because programming GPUs in assembly code is...
View ArticleClik here to view.
Maxwell: Nvidia’s Silver 28nm Hammer
Nvidia’s Kepler architecture gave the company a strong start in the 28nm era. Consumer Kepler parts provided highly competitive gaming performance and power efficiency. In the compute market, Kepler...
View ArticleClik here to view.
Previewing Meteor Lake at CES
Intel has been using a hybrid core strategy for years in a bid to leverage their bigger engineering budget to corner AMD. Specifically, P-Cores focus on maximizing per-thread performance. E-Cores...
View ArticleClik here to view.
Inside Qualcomm’s Adreno 530, a Small Mobile iGPU
GPU architectures vary drastically depending on their primary use cases. Mobile designs like Qualcomm’s Adreno face a daunting set of challenges, with smaller power and area budgets than even Intel’s...
View ArticleClik here to view.
Examining AMD’s RDNA 4 Changes in LLVM
As 2024 continues on, because time never stops, AMD has been working on their upcoming RDNA 4 architecture. Part of this involves supporting open source projects like LLVM. If done right, merging...
View ArticleClik here to view.
AMD RDNA 3.5’s LLVM Changes
Integrated graphics have been a key part of AMD’s strategy ever since they bought ATI. Bringing CPU and GPU blocks together in the same chip has given AMD substantial wins, including in Microsoft’s...
View ArticleClik here to view.
AMD’s Mild Hybrid Strategy: Ryzen Z1 in ASUS’s ROG Ally
Editor’s Note: ASUS sent us the ROG Ally sample – our first review sample from a company – in order to test the Ryzen Z1 SOC inside the device. So a massive thank you to them! CPUs with hybrid core...
View ArticleClik here to view.
LLVM’s Ampere1B Commit
Ampere Computing found a niche in creating ARM-based server CPUs. Ampere Altra saw service in Oracle Cloud, Google Cloud, and Microsoft Azure. For Altra, Ampere used Arm Ltd’s Neoverse N1 core design....
View ArticleClik here to view.
Ryzen Z1’s Tiny iGPU
Editor’s Note: Just like our prior Ryzen Z1 article, the ROG Ally was kindly provided by Asus to let us test the Ryzen Z1. ASUS ROG Ally comes in two configurations: AMD’s Ryzen Z1 Extreme and the...
View Article