logo
What is Rowhammer bit-flip attack which forced Nvidia to issue security alert

What is Rowhammer bit-flip attack which forced Nvidia to issue security alert

Time of India16-07-2025
Researchers discovered that Nvidia's A6000 GPUs are prone to Rowhammer attacks. This allows hackers to tamper with user data on shared GPUs. Nvidia has issued an alert, advising users to enable Error Correction Code. Newer GPUs with GDDR7 or HBM3 memory have built-in protection. This vulnerability poses a risk in multi-tenant environments.
Tired of too many ads?
Remove Ads
What Is Rowhammer bit-flip attack
Tired of too many ads?
Remove Ads
Why is this a problem for Nvidia GPUs?
How did Nvidia respond?
A team of researchers has revealed that Nvidia's A6000 GPUs using GDDR6 memory are vulnerable to Rowhammer attacks, which can allow hackers to interfere with users' data on a shared GPU in the cloud, like AI models, even if they don't have direct access to it. In response to the research, Nvidia issued an alert for users asking them to ensure system-level ECC is enabled across the following NVIDIA products.'Specific generations of DRAM devices starting with DDR4, LPDDR5, HBM3, and GDDR7 implement On-Die ECC (OD-ECC) to help with DRAM scaling. OD-ECC indirectly protects Rowhammer bit flips. Note: OD-ECC is not adjustable by users. If OD-ECC is present, it is always enabled,' the company said in its alertRowhammer is a security issue where repeated access to memory can silently change data. Researchers just showed that this attack now works on Nvidia GPUs, so Nvidia issued an alert and recommends using ECC to stay safe.Rowhammer for GPUs, also called GPUHammer, is a new hardware attack that targets graphics cards, specifically NVIDIA GPUs with GDDR6 memory. It is a hardware vulnerability found in computer memory chips (DRAM).Normally, data is safely stored in separate rows of memory cells. But if you rapidly and repeatedly access (hammer) one row, it introduces bit flips in adjacent memory rows. This can 'flip' bits of data even if no one directly accessed that data.'Since 2014, this vulnerability has been widely studied in CPUs and CPU-based memories like DDR3, DDR4, and LPDDR4. However, with critical AI and ML workloads now running on discrete GPUs in the cloud, it is vital to assess the vulnerability of GPU memories to Rowhammer attacks,' the researchers said.Rowhammer is a circuit-level DRAM vulnerability that lets attackers flip bits in neighboring memory rows. It had only been shown on CPU DRAM before, but now, researchers have demonstrated the first Rowhammer bit-flip attack on GPU DRAM, specifically on GDDR6 memory used in NVIDIA GPUs like the A6000.Attackers can potentially use Rowhammer to mess with another user's data on a shared GPU in the cloud, like AI models, even if they don't have direct access to it. The research proved that even a single bit flip could destroy the accuracy of an AI model.The attack works in shared environments where multiple people or programs are using the same GPU at the same time (multi-tenant setups).Nvidia confirmed the Rowhammer risk on some GPUs and advised customers to turn on Error Correction Code (ECC), a memory feature that can catch and fix single-bit errors. This helps stop Rowhammer attacks, though it might slow down some AI workloads by up to 10%.Newer Nvidia GPUs (like those with GDDR7 or HBM3 memory) already have built-in protections (on-die ECC) against this type of attack.
Orange background

Try Our AI Features

Explore what Daily8 AI can do for you:

Comments

No comments yet...

Related Articles

Chinese AI firms form alliances to build domestic ecosystem amid US curbs
Chinese AI firms form alliances to build domestic ecosystem amid US curbs

Mint

time38 minutes ago

  • Mint

Chinese AI firms form alliances to build domestic ecosystem amid US curbs

SHANGHAI, - China's artificial intelligence companies have announced two new industry alliances, aiming to develop a domestic ecosystem to reduce dependence on foreign tech as they seek to cope with U.S. export restrictions on advanced Nvidia chipsets. The announcements were timed to coincide with the three-day World Artificial Intelligence Conference in Shanghai ending on Monday. The conference also showcased a slew of new products, such as an AI computing system from Huawei that experts believe rivals Nvidia's most advanced offering, as well as consumer-friendly products such as several kinds of digital AI glasses. The "Model-Chip Ecosystem Innovation Alliance" brings together Chinese developers of large language models and AI chip manufacturers. "This is an innovative ecosystem that connects the complete technology chain from chips to models to infrastructure," said Zhao Lidong, CEO of Enflame, one of the participating chipmakers. Other manufacturers of graphics processing units in the alliance include Huawei, Biren, and Moore Threads, which have been hit by U.S. sanctions that block them from purchasing advanced tech made with U.S. know-how. The alliance was announced by StepFun, an LLM developer. A second alliance, the Shanghai General Chamber of Commerce AI Committee, aims to "promote the deep integration of AI technology and industrial transformation." Participants include SenseTime, also sanctioned by the U.S. and which has pivoted from facial recognition technology to LLMs. Others are StepFun and another LLM developer, MiniMax, as well as chipmakers Metax and Iluvatar CoreX. One of the most talked about products at the conference was Huawei's CloudMatrix 384 which incorporates 384 of its latest 910C chips and outperforms Nvidia's GB200 NVL72 on some metrics, according to U.S. research firm SemiAnalysis. Huawei's system design capabilities have meant that it has been able to use more chips and system-level innovations to compensate for weaker individual chip performance, SemiAnalysis said. At least six other Chinese computing firms showcased similar "clustering" chip technology. Metax demonstrated an AI supernode featuring 128 C550 chips designed to support large-scale liquid-cooled data centre requirements. Other events included Tencent's unveiling of its open-source Hunyuan3D World Model 1.0, which the company said enables users to generate interactive 3D environments through text or image prompts. Baidu announced what it said was next-generation "digital human" technology that helps businesses to create virtual livestreamers. It features "cloning technology" that can replicate a human's voice, tone, and body language from just 10 minutes of sample footage. Alibaba was among those announcing AI glasses. Its Quark AI Glasses are powered by its Qwen AI model and are due to be released in China by the end of 2025. They will allow users to access the tech giant's map service for easy navigating and to use Alipay by scanning QR codes with voice commands. This article was generated from an automated news agency feed without modifications to text.

Nvidia CEO Jensen Huang predicts: AI will create more millionaires in five years
Nvidia CEO Jensen Huang predicts: AI will create more millionaires in five years

Time of India

time2 hours ago

  • Time of India

Nvidia CEO Jensen Huang predicts: AI will create more millionaires in five years

Nvidia CEO Jensen Huang recently called AI the 'great technology equaliser of all time'. Speaking at an All-In podcast hosted by American venture capitalist Chamath Palihapitiya, Huang explained that in the era of AI, 'everybody is a programmer'. He further stated that the traditional gatekeeping of coding languages like C++, Python has faded, adding people now just need to talk to AI. 'Everybody is an artist now; everybody is an author now,' Jensen Huang said. Jensen Huang: 'AI will create more millionaires in 5 years' During the podcast, Jensen Huang also talked about the impact of AI on jobs. 'AI in my case is creating jobs; it causes people to create things that other people would like to buy. It drives more growth, more jobs, and all that goes together. AI is the greatest technology equaliser of all time,' the Nvidia CEO said. He forecasted that 'AI will create more millionaires in five years than the internet did in 20,' adding that he is worried that we are not moving fast enough to create jobs. Jensen Huang further warned that those who are not using AI will lose their job to someone with the knowledge of AI. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Indonesia: Unsold Sofas Prices May Surprise You (Prices May Surprise You) Sofas | Search Ads Search Now Undo Huang further predicted that in the future, every country will operate two factories – one physical and other digital. He gave an example of Elon Musk 's EV company Tesla, saying 'Tesla builds cars in one factory, and in another, it builds the AI that powers them'. In a related news, Huang discussed the scale of impact that small, focused teams of AI researchers can have and how Nvidia's success has translated into significant financial rewards for those in leadership. 'I've created more billionaires on my management team than any CEO in the world. They're doing just fine. Don't feel sad for anybody at my layer.' Jensen Huang said. He was responding to a question on the effectiveness and value of smaller AI teams. 'But the important big idea is that the impact of 150 or so AI researchers, with funding behind them, can probably create,' he noted. Jensen Huang gave examples of OpenAI which originally started with around 150 employees and China's DeepSeek which also has a similar workforce. Samsung Galaxy Watch8: AI on Your Wrist

Trump pauses export controls to bolster China trade deal: Report
Trump pauses export controls to bolster China trade deal: Report

Time of India

time3 hours ago

  • Time of India

Trump pauses export controls to bolster China trade deal: Report

The United States has temporarily eased restrictions on technology exports to China. This decision aims to facilitate ongoing trade negotiations with Beijing. It also supports President Trump's efforts to meet with President Xi Jinping. The Commerce Department has been instructed to avoid taking strict actions against China. Nvidia will resume sales of its H20 GPUs to China. Tired of too many ads? Remove Ads Tired of too many ads? Remove Ads The U.S. has paused curbs on tech exports to China to avoid disrupting trade talks with Beijing and support President Donald Trump 's efforts to secure a meeting with President Xi Jinping this year, the Financial Times said on industry and security bureau of the Commerce Department, which oversees export controls, has been told in recent months to avoid tough moves on China, the newspaper said, citing current and former could not immediately verify the report. The White House and the department did not respond to Reuters' requests for comment outside business U.S. and Chinese economic officials are set to resume talks in Stockholm on Monday to tackle longstanding economic disputes at the centre of a trade war between the world's top two giant Nvidia said this month it would resume sales of its H20 graphics processing units (GPU) to China, reversing an export curb the Trump administration imposed in April to keep advanced AI chips out of Chinese hands over national security planned resumption was part of U.S. negotiations on rare earths and magnets, Commerce Secretary Howard Lutnick has paper said 20 security experts and former officials, including former deputy US national security adviser Matt Pottinger, will write on Monday to Lutnick to voice concern, however."This move represents a strategic misstep that endangers the United States' economic and military edge in artificial intelligence," they write in the letter, it added.

DOWNLOAD THE APP

Get Started Now: Download the App

Ready to dive into a world of global content with local flavor? Download Daily8 app today from your preferred app store and start exploring.
app-storeplay-store