redeem

If you have ever built a PC, it is very likely that you have purchased some component from the Crucial brand, owned by Micron. The RAM ‘pills’ or SSDs were of quality, but Crucial ceased to exist at the time when Micron decided that the segment of the artificial intelligence It was the priority. They focused on creating high-bandwidth memory for the platforms of the data centersbut now they have just announced their new generation of GDDR7 chips for gaming GPUs. And it is an example of how far behind they have fallen compared to the South Koreans. In short. In a post on his blogMicron has confirmed that it is starting mass production of 3 GB GDDR7 chips with a density of 24 Gb. They have done so with pride, as they complete an objective for which they have been fighting for months: to get on par with Samsung and SK Hynix, the leaders of the DRAM market. GDDR7 with asterisk. As detailed tomshardwarethese new Micron chips are 12.5% ​​faster than the first GDDR7 chips that hit the market. They have a bandwidth of 36 Gbps compared to the 32 Gbps from those original modules. However, although the density is the same as its competitors, the bandwidth is noticeably lower. Samsung chips have a bandwidth that can reach 42.5 Gbps and SK Hynix is ​​on par with its 40 Gbps modules… and is already working on 48 Gbps ones. To put it bluntly, the more memory a GPU has, the more textures it can hold, but bandwidth is the amount of simultaneous data it handles, which directly impacts performance in games. The third in contention. The more bandwidth the memory has, the better also for calculations in artificial intelligence applications, something that is becoming essential in video games with techniques such as Nvidia DLSS. And here we have to clarify something: although Micron’s is slower than its competitors, it doesn’t really matter that much in video games because even the most powerful cards from Nvidia and AMD move below 40 Gbps of bandwidth. However, and here comes another asterisk, the fact that Micron is announcing this now shows that it is months behind the two South Korean companies. This is something that is important because we are seeing that, especially in this AI race, whoever comes first is the one who takes the lead, a cat called Nvidia. It already happened a few weeks ago with Samsungbeing the first with the capacity to deliver Mass HBM4 memory to Nvidia for its new Vera Rubin platform and, precisely, being the one chosen by the AI ​​giant over SK and Micron. Nvidia always wins. But hey, here is a win-win. Micron is already in line with its competitors, at least as far as 3 GB GDDR7 memory production is concerned. And Nvidia manages to have a third manufacturer that can deliver those 3 GB chips for its GPUs. With a complicated market due to scarcityand with a gaming segment that remains important, having three memory manufacturers working on your platform can help unclog the GPU market. If Nvidia launches new products this year, what remains to be seen. That, speaking of Nvidia and the three big memory manufacturers, both the South Korean companies and Micron are already mass creating the aforementioned HBM4 for Vera Rubin. In Xataka | The US is investing a fortune in creating its own sovereign chips. Behind it is a South Korean company: Samsung