9800x3D 3D v-cache verified beneath the CCD. Intel will never recover from this...
9800x3D 3D v-cache verified beneath the CCD. Intel will never recover from this...
TLDR - 9000x3D offers superior performance compared to previous 3D v-cache setups, likely delivering the most powerful and reliable gaming experience. The design showcases the SRAM die beneath the CCD, confirming their understanding of the leaks. This update effectively eliminates 3D v-cache constraints and supports overclocking. It’s hard to imagine Intel matching this without adopting comparable tech. The 7800x3D was already strong; now you can match regular non-3D Ryzen speeds while achieving extreme frequencies and direct-die overclocking. I’m curious about revisiting a desktop with this chip for hands-on testing. How could AMD possibly compete?
This seems really intriguing. I can't wait for it to be released.
Considering this, I wonder if placing 64MB of L3 beneath the CCD would make sense, just as it did before. Adding another 64MB of L3 on top could potentially improve things even further!
Intel is simply copying it, and that's acceptable. It seems they'll be able to justify the thicker design in the future. AM6?
Incorporating a small adjustment like 1mm has minimal impact; they can just refine the IHS thickness or remove material from the top or bottom of the CCD, similar to previous work with 5000x3D and 7000x3D. It's much simpler to redesign the IHS than to add that extra step if it isn't needed. Clearly, what came to my mind is that there are no restrictions preventing them from implementing this SRAM 3D v-cache mod on their APU dies. One major constraint of those dies is the limited 16MB of L3 memory built into the die, so they could easily increase that to 64MB or more, or even replace part of the die with L3 cache and shift resources to the SRAM area. More room on the die for CPU or GPU functionality...
Just as they claimed AMD would never bounce back, it seems they’re holding up well in my view.
They haven’t paid much attention to Arrow Lake and Lunar Lake, have they? It seems they’re already lagging significantly. AMD is effectively eliminating the single 3D v-cache option, which previously imposed a major thermal constraint on the CCD. Lunar Lake was intriguing until you notice the CPU performance of those chips. It’s likely they allocated too many die to the NPU and iGPU, greatly limiting the multicore potential of the 258V and 288V models. Performance is now around 75% of what the 7840U/Z1 Extreme offers—something that truly matters. This level of impact means I won’t recommend these options except for pure iGPU setups.
The significant shift they implemented is understandable. It's possible they even mentioned it, though I shouldn't rely on that. It seems the upcoming version will likely be much better. While AMD certainly moved quickly, I'm not here to debate. What matters is that both sides can recover and grow again.