Nvidia Confirms Manufacturing Flaw in Early RTX 5080 Graphics Cards
Nvidia has officially confirmed that its RTX 5080 graphics card is also affected by the same rare manufacturing flaw that previously impacted the RTX 5090, RTX 5090D, and RTX 5070 Ti. The issue, which leads to one fewer Render Output Unit (ROP) than specified, results in a minor graphical performance drop of around 4% in affected GPUs. However, Nvidia reassures that AI and compute workloads remain unaffected.
Initially, Nvidia had only acknowledged the issue in three of its RTX 50-series cards, but following additional investigations, the company confirmed to The Verge that an early production batch of RTX 5080 GPUs also suffered from the same defect. Ben Berraondo, Nvidia's Global PR Director for GeForce, stated, "Upon further investigation, we’ve identified that an early production build of GeForce RTX 5080 GPUs were also affected by the same issue. Affected consumers can contact the board manufacturer for a replacement."
Nvidia also clarified that no other RTX GPUs, including the upcoming RTX 5070, have been affected by the problem. Additionally, the company reassured customers that the manufacturing anomaly has been corrected and that newer batches of these GPUs will not have this issue. Berraondo also emphasized that Nvidia was not aware of this flaw before launching the affected GPUs.
The full statement from Nvidia reads: "We have identified a rare issue affecting less than 0.5% (half a percent) of GeForce RTX 5090 / 5090D, RTX 5080, and 5070 Ti GPUs which have one fewer ROP than specified. The average graphical performance impact is 4%, with no impact on AI and Compute workloads. Affected consumers can contact the board manufacturer for a replacement. The production anomaly has been corrected."
The issue was first discovered by a Reddit user who reported performance discrepancies in their RTX 5080 unit. That user has since reached an agreement to provide the affected card to GamersNexus for further analysis. While Nvidia has reassured users that only a small number of GPUs were impacted, given the limited availability of RTX 50-series cards so far, this flaw has added to growing frustrations over the launch of Nvidia’s latest graphics lineup.
Despite the relatively minor performance impact, this incident raises concerns over Nvidia’s quality control in its latest GPU generation. With high expectations for the RTX 50-series, enthusiasts will be closely watching how Nvidia handles potential manufacturing challenges in the future.
Source: The Verge
Advertisement
That “4% average” raster loss is for the 5090 models. (176-168) / 176 ~ 4.76% loss in raster op throughput.
The loss is harder for the lesser models when that extra texture processing pipe is missing.
On a 5070 it amounts to a 10% loss.
https://en.wikipedia.org/wiki/GeForce_50_series
My PC is kind of old by the Moore’s law – i5 8400 and GTX 1650, bought it in 2019 and even by the standards of the time, it was something mid-range or below mid-range. But so far it runs fine on Linux Mint XFCE and I can play all the games that I want to so I don’t plan on ever upgrading it. Maybe when a time comes to get something new, whether it’s a laptop or another PC, I think I will still buy something that’s older and cheaper. Maybe i5 or i7 10th generation if I can find it brand new and maybe the GPU will be something like RTX 2080 or something like that, also if possible to find brand new.
I think by the time I have to get a PC with RTX 50×0, it will be several years old and maybe a 3rd party manufacturer, like in my case with the GTX 1650 by MSI, the flaws will have been ironed out and it will be a stable and reliable product, if not, there are always other options.
In my opinion, being on the cutting edge, comes with its risks and this is another example of them. I’ve seen articles about AMD’s recent CPUs and GPUs also having issues, Intel CPUs being mediocre and stale, nVidia focusing on AI crap more than anything else and Intel GPUs not making the impact they expected so staying behind the trends and using several years old hardware that is bought brand new is a more sensible approach than taking a leap of faith and buying the newest product ASAP. Same is with some countries where the iPhone craze has lead people to camp for days in front of stores or that Samsung phone whose battery would spontaneously combust.
Also, the times when I wanted a powerful PC so I can play all the newest games maxed out are long gone, I just want a PC with a fast CPU, some fast SSD and decent amount of RAM, GPU isn’t even a concern for me anymore. I put Linux on it and play some old school games on it like Age of Empires 2 the CD version, Heroes of Might and Magic 3, Vanilla World of Warcraft and other stuff from that era.
I think the newest game I tried to run with this PC was Cyberpunk 2077 and the game ran fine on low-medium, but beyond that and having my curiosity “will I run it?” be satisfied, I haven’t had a desire to try any new games on it since there is hardly anything new that interests me. Hence the lack of need to upgrade my PC or buy a new one, this will only happen if the PC breaks irreversibly or I need the mobility and get a laptop in that case.
It’s rare but happening across multiple models. The only rarity is that so few were available to buy on release day. Which only adds to the question why QC was so bad to miss these flaws. Nvidia wants more money from you but they can’t even make sure that you get what you pay for. Sure, it amounts to only a small percentage of performance loss. I wonder how fast the turnaround will be for these buyers with defective cards.