I predicted the future! - GRAID Graphics Accelerated Storage
0 up · 0 down · 0 ratings
Description
Get Canada's fastest internet tech with TELUS at telus.com Learn more about Kioxia and the BG5 Series at lmg.gg NVMe storage is blisteringly fast but has lacked a high-performing RAID solution for a long time. Today that may be changing with GRAID's new SupremeRAID GPU accelerated NVMe RAID solution that promises to deliver over 100GB/s of sequential read performance with next to no CPU usage. Let's see if it's up to snuff. Discuss on the forum: linustechtips.com
Check out the Gigabyte R282-Z9G GRAID Rack Server: lmg.gg
Check out the Micron DDR4 ECC Memory: lmg.gg
Promos
Check out the Liqid Element LQD3000 PCIe SSD: lmg.gg Purchases made through some store links may provide some compensation to Linus Media Group. ► GET MERCH: lttstore.com ► AFFILIATES, SPONSORS & REFERRALS: lmg.gg ► PODCAST GEAR: lmg.gg ► SUPPORT US ON FLOATPLANE: floatplane.com FOLLOW US ELSEWHERE --------------------------------------------------- Twitter: twitter.com Facebook: @LinusTech Instagram: @linustech TikTok: @linustech Twitch: twitch.tv MUSIC CREDIT --------------------------------------------------- Intro: Laszlo - Supernova Video Link: youtube.com iTunes Download Link: itunes.apple.com Artist Link: soundcloud.com Outro: Approaching Nirvana - Sugar High Video Link: youtube.com Listen on Spotify: spoti.fi Artist Link: youtube.com Intro animation by MBarek Abdelwassaa @mbarek_abdel Monitor And Keyboard by vadimmihalkevich / CC BY 4.0 geni.us Mechanical RGB Keyboard by BigBrotherECE / CC BY 4.0 geni.us Mouse Gamer free Model By Oscar Creativo / CC BY 4.0 geni.us CHAPTERS --------------------------------------------------- 0:00 Intro
I predicted the future! - GRAID Graphics Accelerated Storage explores a novel approach to NVMe RAID by offloading the RAID processing to a GPU-like accelerator card and bypassing the CPU for most data handling. The host hardware configuration centers on a Gigabyte server with high-end EPYC CPUs, abundant RAM, and a bank of Kioxia CD6-Rs NVMe drives. The video starts by introducing the SupremeRAID SR-1000 as a GPU-accelerated storage solution that promises over 100 GB/s of sequential throughput while keeping CPU usage low. The team explains the traditional RAID landscape, highlighting the bottlenecks of software-RAID and hardware RAID cards that fight for CPU cycles and limited PCIe bandwidth. They position GRAID as a middle ground where the data path stays directly between drives and system memory while the controller handles RAID logic, effectively removing the CPU as a bottleneck and enabling high throughput. The initial setup includes enabling Linux-based SupremeRAID software, installing the necessary NVIDIA drivers, and preparing a multi-drive array in a validated server environment. This first segment sets the stage for deep-dive benchmarking and practical exploration of configuration choices such as RAID0 versus RAID5 and the implications for performance and data protection.
Topics · technology · hardware · storage · servers · data-management
Questions answered
- What is SupremeRAID and how does it differ from traditional RAID cards?
- SupremeRAID is a GPU-accelerated NVMe RAID solution that offloads RAID calculations from the CPU and directs data flow directly between drives and system memory, reducing CPU overhead and enabling higher throughput than software-RAID while avoiding some bottlenecks of traditional hardware RAID cards.
- Can the SupremeRAID SR-1000 achieve 100 GB/s sequential throughput?
- According to the video, the SR-1000 is claimed to deliver over 100 GB/s of sequential throughput with minimal CPU usage, achieved by offloading RAID processing to the GPU accelerator rather than routing data through a traditional RAID controller.
- What were the initial test configurations used in the benchmark?
- The tests used a Gigabyte server with EPYC CPUs, a large RAM pool, and 12 Kioxia CD6-R4 NVMe drives configured into a drive group and virtual disks, with RAID0 and RAID5 tests performed to compare throughput and CPU impact.
- What is the practical takeaway about CPU usage for GPU-accelerated RAID versus software-RAID?
- The video suggests that GPU-accelerated RAID can achieve substantially higher read and write throughputs with very low CPU usage compared with software-RAID, though real-world results depend on the exact drive set, workload, and software maturity.