Tuesday, May 13, 2025
Phonemantra
No Result
View All Result
  • Home
  • Mobiles
  • Tech News
  • Cars
  • Entertainment
  • USA News
  • Health
  • Cameras
  • Gaming
No Result
View All Result
  • Home
  • Mobiles
  • Tech News
  • Cars
  • Entertainment
  • USA News
  • Health
  • Cameras
  • Gaming
No Result
View All Result
Phonemantra
No Result
View All Result
Home AMD

Video cards against artificial intelligence. Testing shows how good modern adapters are at Stable Diffusion

Nvidia cards are the best

We know how good certain graphics cards are in games and even in various video processing applications. But our colleagues from the resource Tom’s Hardware decided to check how the adapters will show themselves in today’s popular tasks of accelerating AI.

Video cards
Video cards

For testing, the authors used the Stable Diffusion application, which can create images from a text description. This software was chosen because it can be installed and run directly on a PC.

 Stable Diffusion

 Stable Diffusion
Stable Diffusion

As you can see, the latest GeForce RTX 40 3D cards were the best at these tasks, which is not surprising given both the use of CUDA and the presence of dedicated tensor cores in Nvidia cards, although it is actually difficult to say whether they are used in this software.

The results of the new AMD Radeon RX 7900 are more interesting. Their results in comparison with Nvidia cards are worse than in games, but still very high overall. They are especially impressive against the background of the terrible results of the RX 6000 line.

However, here it is worth clarifying a number of points. The fact is that if you look at third-party news that was written on the basis of the source material immediately after the release, you can see a completely different diagram.

Stable Diffusion
Stable Diffusion

We can see significantly lower results for the RTX 40 and slightly lower results for the RX 7900. The bottom line is that initially they were like that, but then the authors got hold of various fixes with optimizations for these accelerators.

That is, we can conclude that the work of such software is very dependent on optimization. For example, it is not yet available for Intel cards, which is why they are much inferior to competitors.

  • 0Facebook
  • 0WhatsApp
  • 0Twitter
  • 0Pinterest
  • 0Reddit
  • 0Telegram
  • 0Skype
  • 0Facebook Messenger
  • Copy Link
  • 0Print
  •  shares

Related Posts

AMD
AMD

AMD will pack 32 processor cores into one chiplet.

December 4, 2023
Nvidia DLSS 3 and AMD FSR 3
Nvidia

Frame generation technologies in Nvidia DLSS 3 and AMD FSR 3 forced to work together

December 3, 2023
Nvidia
Nvidia

The head of Nvidia thinks so

December 2, 2023
Nvidia
Nvidia

Nvidia is turning into a monster.

November 30, 2023
Nvidia
Nvidia

Nvidia sells half a million H100 and A100 accelerators per quarter

November 30, 2023
Nvidia
Nvidia

Is Nvidia no longer a gaming graphics card company?

November 24, 2023

Recommended Stories

Asus ROG Ally

The Asus ROG Ally game console was priced at $600.

April 29, 2023
ventrogluteal injection

Mastering the Ventrogluteal Injection Technique: Precision and Safety

September 23, 2023

Intel admitted that the rejection of monolithic crystals will reduce the cost of production

December 6, 2020 - Updated on December 8, 2022

Ads

Popular Stories

  • Did You Gain the Quarantine 15

    Did You Gain the Quarantine 15?

    0 shares
    Share 0 Tweet 0
  • How to Protect Kids from Summer Insect Bites and Stings

    0 shares
    Share 0 Tweet 0
  • Six Savvy Tips for Snoozing More Soundly

    0 shares
    Share 0 Tweet 0
  • Reversing Prediabetes Naturally

    0 shares
    Share 0 Tweet 0
  • Six Simple Yet Powerful Steps

    0 shares
    Share 0 Tweet 0
Phonemantra

© 2025 Phonemantra

Navigate Site

  • Our Team
  • Sitemap
  • Legal Disclaimer
  • Privacy Policy
  • Contact Us

Follow Us

No Result
View All Result
  • Home
  • Mobiles
  • Tech News
  • Cars
  • Entertainment
  • USA News
  • Health
  • Cameras
  • Gaming

© 2025 Phonemantra