Threads
- NVIDIA is introducing a new RTX 4070 with GDDR6 memory instead of GDDR6X memory with the same GPU name.
- The company has done similar things in the past with the GTX 660, RTX 3050, and other GPUs.
- This time, the move was made to prevent a GPU shortage due to quality concerns with the August GDDR6X memory batch.
NVIDIA has a notorious past when it comes to branding PC components. Whether you look at the GTX 660 or the RTX 4080, you’ll see similar branding issues. One would think that consumer outrage would help, but we just got reports of another case where NVIDIA is once again misleading customers with a new RTX 4070. What’s the issue? NVIDIA is launching a new RTX 4070 with GDDR6 memory instead of the original GDDR6X. This isn’t the first time NVIDIA has pulled a shady trick; let’s investigate further.
Mishaps With The GTX 660
In 2009, NVIDIA launched the GTX 660, which was a relatively decent GPU at the time. Later, NVIDIA decided to launch a new mystery GPU with a few performance hits here and there. Now, I wouldn’t mind if this new mystery GPU had another name, but the company decided to keep the same name, creating confusion within the masses.
LinusTechTips recently made a video highlighting similar issues. In it, he explains in detail how many GTX 660s he found on TechPowerup’s website. The real problem is that he couldn’t find the exact GTX 660 he was using, meaning NVIDIA must have launched even more variants that weren’t detected. I don’t know about anyone else, but I definitely wouldn’t want to buy a certain GPU and be sent something completely else.
There Are Two RTX 3050s
To avoid the media’s attention, NVIDIA started shaving down budget graphics cards, which are usually avoided. For example, the RTX 3050. We already know the original has 8GB VRAM, right? Well, there’s another variant with 6GB VRAM. There’s only a slight difference in specifications. Here’s the comparison between the specifications by NVIDIA itself:
I was intrigued by the spec difference, so I decided to take both GPUs on the test bench and check the performance difference myself. There’s not a big difference in specifications, but the performance hit makes me believe the new RTX 3050 should have been branded as the RTX 3040 or worse. Here are my results:
The Latest Controversy With The RTX 4070
Moving forward to the latest controversy, NVIDIA has created a new RTX 4070 variant that uses GDDR6 VRAM instead of the original GDDR6X. According to their blog post, no other specifications will be changed, and the performance hit will be minimal. The older VRAM means slightly slower memory modules, going from 21Gbps to 20Gbps. This further translates the memory bandwidth to go down to 480GB/s from 504GB/s.
So far, I don’t think a reduction in memory bandwidth will affect gaming performance too much, but we’ll only know the actual results once both SKUs are compared. Initially, I wouldn’t have a problem with the company’s move, but it’s suggested that the GPU won’t have a price cut. Also, it’ll be branded as the RTX 4070 and not a different variant.
So, if you’re out to purchase the RTX 4070 graphics card, there’s a good chance you might buy the GDDR6 version instead of the GDDR6X variant for the same price and get a slight performance cut. NVIDIA should have come clean and created a new variant, even if the performance difference is within a margin of error. If I’m spending around $500, I don’t want to purchase something that has been compromised thanks to cost-cutting.
Sometimes There’s A Good Reason
Although I’d like to speak ill of NVIDIA and its tactics, sometimes there’s a good reason to reduce the performance of already-existing GPUs. In the RTX 4070 case, according to reports, NVIDIA’s August GDDR6X batch quality was unsatisfactory, so they introduced a new GPU variant by reducing it to GDDR6.
https://x.com/harukaze5719/status/1817807114901073960
Understandably, NVIDIA pulled this move to avoid GPU shortage in the market, but the audience, at the very least, deserves a proper explanation and an announcement of the move. Instead, NVIDIA wrote a short paragraph and buried the announcement within the patch notes of the 560.94 WHQL driver update, misleading the customers with the new RTX 4070.
NVIDIA Needs To Stop
While the new changes aren’t drastic, hiding information from customers is where I draw the line. If it was once, we could have overlooked it, but it has happened so often that it’s not even remotely funny anymore. Sure, the company has avoided a GPU shortage this time, but new customers might not be aware of the changes in the RTX 4070.
Of course, we don’t know the performance hit so far, but your audience deserves to know about it, even if it’s a minuscule 1% drop. It’s high time NVIDIA stopped these shady practices. Instead of misleading customers with the new RTX 4070, they can own up to whatever issues they are facing and how they plan on countering them.
Thank you! Please share your positive feedback. 🔋
How could we improve this post? Please Help us. 😔
[Comparisons Expert]
Abdemanaf is a skilled creative writer who has been honing his craft since 2011. While initially working in different fields, he found a passion for technology and has been exploring the tech world since early 2015. Over the years, he has developed an in-depth knowledge of the latest tech trends and product offerings by various companies.
Abdemanaf’s writing reflects his analytical mindset and ability to think critically. He has a knack for breaking down complex technical information into easily digestible pieces, making his articles engaging and accessible to readers from all backgrounds. In February 2022, he joined Tech4Gamers as a blog and product comparison writer, where he has been able to hone his skills further.
As a writer, Abdemanaf is dedicated to staying up-to-date with the latest technological advancements and trends, enabling him to provide readers with the most relevant and accurate information. He is always eager to learn more and is constantly seeking new challenges to improve his skills.
Get In Touch: manaf@tech4gamers.com
Join Our Community
