semiconductor engineer tom wasik come to know He considers the 3D V-Cache functionality on one of AMD’s best GPUs, the RX 7900 XT. Engineers took a peek inside the 7900 XT’s die with infrared imaging and found the same type of 3D V-cache connection points used on AMD’s Zen 3 and Zen 4 architectures. Wasik saw the connection dying at MCD.
Wasik couldn’t say whether these TSV connection points will be used exclusively for caching purposes, but AMD has no known plans at this point to expand its 3D packaging capabilities beyond vertically stacked caches. This makes it appear that these connection points will be used in conjunction with some sort of 3D cache to increase gaming performance and/or compute performance.
The discovery comes after several unconfirmed rumors that AMD will be adding 3D V-Cache technology to its GPUs.
So far 3D V-Cache has been used to great success on AMD’s Ryzen and EPYC CPUs. The technology relies on a hybrid bonding technique that fuses an additional 64MB slab of cache on top of the Ryzen or EPYC compute die to increase L3 cache capacity. Currently, this 3D stacking technology has allowed AMD to double the amount of L3 cache available in its desktop Ryzen 9 7900X3D and 7950X3D parts while allowing it to triple on its Ryzen 7 5800X3D, 7800X3D consumer chips, and EPYC Milan-X server processors. Have multiplied.
The performance gains from this technology have been impressive, with 3D-V-Cache chips providing an absolute generational increase in performance in applications that benefit heavily from large chunks of cache. A good example of this is with the Ryzen 7 5800X3D where we saw a 28% increase in gaming performance compared to the Ryzen 9 5900X, and 7% faster performance compared to the Core i9-12900KS.
AMD’s server counterparts are even more impressive, with AMD and Microsoft’s Milan-X benchmarks showing performance improvements of over 50% over standard Milan parts. However, this technique cannot magically increase performance as desired. Only cache-sensitive workloads will see this type of behavior.
What else can you see? A linear array of “spots” that look remarkably like the keep out zones on the X3D, and that are at the same 17-18 um pitch. Could they be considering stacked MCD functionality (or maybe they are something else)?January 27, 2023
See more
We have no idea how 3D V-cache will work in GPU applications. But in theory, the core principles of 3D V-Cache should still apply. Having more cache capacity enables faster processing of cache-sensitive workloads because the GPU has to make fewer trips to its slower GDDR6 memory.
We’ve already seen a good example of this with AMD’s Infinity Cache in the RX 6000 series, where AMD was able to use the slower GDDR6 memory and maintain performance similar to Nvidia’s RTX 30 series GPUs, which have power- There was hungry GDDR6X memory, thanks to its infinite cache to keep the GPU fed with data.
However, we don’t know if the same behavior will apply with 3D V-Cache. It will all depend on how sensitive AMD’s GPU architecture is to the extra cache capacity, and how many applications will benefit.
Another problem AMD has to deal with is thermal. We’ve seen this issue extensively on AMD’s Ryzen X3D processors, where the extra slab of cache hinders thermal dissipation, resulting in lower CPU frequencies and higher temperatures (compared to the non-X3D part) at the same time. Is. There’s a high chance AMD will deal with similar issues on the 3D V-Cache GPU, and will be forced to lower clock speeds to keep temperatures under control.
Still, it’s nice to see AMD considering the idea of adding 3D-Wache to its GPUs. We may be looking at AMD’s next silver bullet for “magically” boosting gaming performance.
