Windows 10 News and info | Forum
March 09, 2021, Loading... *
Welcome, Guest. Please login or register.

Login with username, password and session length
News: This is a clean Ad-free Forum and protected by StopForumSpam, Project Honeypot, Botscout and AbuseIPDB | This forum does not use audio ads, popups, or other annoyances. New member registration currently disabled.
 
  Website   Home   Windows 8 Website GDPR Help Login Register  
By continuing to use the site or forum, you agree to the use of cookies, find out more by reading our GDPR policy.
Pages: [1]
  Print  
Share this topic on Del.icio.usShare this topic on DiggShare this topic on FacebookShare this topic on GoogleShare this topic on MySpaceShare this topic on RedditShare this topic on StumbleUponShare this topic on TechnoratiShare this topic on TwitterShare this topic on YahooShare this topic on Google buzz
Author Topic: Nvidia's RTX A6000: 48GB of Memory Powers Twice The Performance  (Read 21 times)
javajolt
Administrator
Hero Member
*****
Offline Offline

Gender: Male
United States United States

Posts: 32288


I Do Windows


WWW Email
« on: February 13, 2021, 02:18:12 PM »
ReplyReply

When Nvidia launched its RTX A6000 48GB professional graphics card last October, the company said that it would offer at least twice the performance of the company's previous-gen Quadro cards. These types of claims are not unusual, but how fast is the $4,650 RTX A6000 really in real-world benchmarks? (Interestingly, that's only $650 more than Galax's flagship RTX 3090 GPU.)

Workstation maker Puget Systems decided to find out and ran multiple professional-grade benchmarks on the card. The results turned out to be impressive in fact, Puget says the Nvidia RTX A6000 48GB is the fastest professional graphics card they have ever tested.

An Ampere with 48GB of RAM

Nvidia's RTX A6000 48GB graphics card is powered by its GA102 GPU with 10,752 CUDA cores, 336 tensor cores, and 84 RT cores, and a 384-bit memory bus that pairs the chip with a beefy 48GB slab of GDDR6 memory. In contrast, Nvidia's top-of-the-range GeForce RTX 3090 consumer board based on the same graphics processor features a different GPU configuration containing 10,496 CUDA cores, 328 tensor cores, 82 RT cores, and a 384-bit memory interface for its 'mere' 24GB of GDDR6X memory.

While the Nvidia RTX A6000 has a slightly better GPU configuration than the GeForce RTX 3090, it uses slower memory and therefore features 768 GB/s of memory bandwidth, which is 18% lower than the consumer graphics card (936GB/s), so it will not beat the 3090 in gaming. Meanwhile, because the RTX A6000 has 48GB of DRAM on board, it will perform better in memory-hungry professional workloads. 

While all GeForce RTX graphics cards come with Nvidia Studio drivers that support acceleration in some professional applications, they are not designed to run all professional software suites. In contrast, professional ISV-certified drivers of the Quadro series and Nvidia RTX A6000 make them a better fit for workstations.

The combination of the Nvidia GeForce RTX A6000 drivers, 48GB of GDDR6, a slightly different GPU configuration, Quadro Sync support, enhanced reliability, a different display output configuration, and a blower-type cooler (which is preferable for multi-GPU configurations) create a solution that costs $4,650, which is considerably higher than a $1,500 MSRP of the standard GeForce RTX 3090 Founders Edition, though some custom RTX 3090's push the needle as high as $4,000.

Up to 92% Faster

As far as performance is concerned (read the full review at Puget's website), the GeForce RTX 3090 might get close to the Nvidia RTX A6000 48GB, but since the former is not a workstation-grade graphics card, Puget decided to compare the new professional board to Nvidia's Quadro RTX 6000 24GB (TU102 with 4,608 CUDA cores).

Technically, the Nvidia RTX A6000 48GB ($4,650) is the successor of the Quadro RTX 6000 24GB (~$4,000), even though the latter has only half the memory. Meanwhile, the previous generation flagship the Quadro RTX 8000 48GB (TU102 with 4,608 CUDA cores) is still priced at $5,000 ~ $5,500.

Not all professional workloads require enormous onboard memory capacity, but GPU-accelerated rendering applications benefit greatly, especially when it comes to large scenes. Since we are talking about graphics rendering, the same programs also benefit from GPU capabilities. That said, it is not surprising that the Nvidia RTX A6000 48GB outperformed its predecessor by 46.6% ~ 92.2% in all four rendering benchmarks ran by Puget.

Evidently, V-Ray 5 scales better with the increase of GPU horsepower and onboard memory capacity, whereas Redshift 3 is not that good. Still, the new RTX A6000 48GB is tangibly faster than any other professional graphics card in GPU-accelerated rendering workloads.

Modern video editing and color correction applications, such as DaVinci Resolve 16.2.8 and Adobe Premiere Pro 14.8, can also accelerate some of the tasks using GPUs. In both cases, the Nvidia RTX A6000 48GB offers tangible performance advantages compared to its predecessor, but its advantages look even more serious when the board is compared to graphics cards released several years ago.

Like other modern professional graphics applications, Adobe After Effects and Adobe Photoshop can take advantage of GPUs. Yet, both programs are CPU bottlenecked in many cases, which means that any decent graphics processor (and not necessarily a professional one) is usually enough for both suites. Nonetheless, the new Nvidia RTX A6000 64GB managed to show some speed gains compared to the predecessor in these two apps as well.

Fastest Professional Graphics Card

Unsurprisingly, Puget found the Nvidia RTX A6000 48GB is the fastest professional graphics card they have ever tested. Evidently, with the massive increase in GPU horsepower versus its predecessor, the card brings its biggest gains in GPU rendering benchmarks as well as in DaVinci Resolve. Other applications can also benefit from the new card, but actual advantages heavily depend on exact workloads.

source
Logged


Pages: [1]
  Print  
 
Jump to:  

Powered by SMF 1.1.21 | SMF © 2017, Simple Machines

Google visited last this page February 17, 2021, 12:35:28 PM