Nvidia Turing with improved L1 and L2 cache decks

Until now, we did not know exactly how much cache the graphics cores based on the Nvidia Turing architecture would have. Well… now we know.

I realize that it doesn’t sound like something you need to know, but we can hook up the latest news from Videocardz to a curiosity. Hi! After all, Nvidia’s Turing architecture is “the biggest revolution since the creation of CUDA cores in 2006”. So far, we’ve learned not very detailed information about individual cores and units, which will suffice most knowledge-hungry consumers, but yesterday we got something special. I am talking about the exact specification of the TU102 core, which went to Quadro RTX 8000 and 6000 cards, and a bit more familiar to us – GeForce RTX 2080 Ti.

Also read: Why are GeForce RTX cards so expensive?

The news about the Nvidia Scanner feature, which is supposed to be built into Nvidia’s software, was also a big surprise. However, this is one big unknown because we are based here on leaks and rumors. But who wouldn’t want the most advanced OC feature ever? Let’s move on to the topic of this post, because it is at least one hundred percent sure.

It looks like Nvidia has doubled the bandwidth, lowered latency and increased the capacity of the L1 cache. This one is now 2.7 times larger than that of the Pascal architecture. The changes also affected the L2 deck, which can boast a whole 6 MB (3 MB more than the L2 capacity from Pascal). As Videocardz writes, “You shouldn’t see this slide until September 14th. So don’t tell anyone about it. “

Source: Videocardz
Photos: Videocardz