https://liliputing.com/2021/06/amds-3d-chiplet-tech-to-bring-192mb-of-l3-cache-to-next-gen-ryzen-processors.html
AMD’s 3D chiplet tech to bring 192MB of L3 cache to next-gen Ryzen processors AMD is showing off technology that could dramatically increase the amount and speed of cache memory available for next-gen chips, with up to 192MB of L3 cache on upcoming AMD Ryzen desktop chips and support for bandwidth of over 2 terabytes per second.
In terms of real-world performance, AMD says that will bring the same kind of performance boost we’d normally expect from a move from one chip architecture to the next (from Zen 3 to Zen 4, for example).
In a demo during the company’s Computex 2021 keynote event, AMD showed two computers running the same video game. Both systems featured Ryzen 9 5900X 12-core processors running at 4 GHz and both were using the same graphics card.
But one card was a standard Ryzen 9 5900X with 64MB of L3 cache, while the other was a modified version with 192 MB of L3 cache. That model achieved frame rates that were 12-percent higher. And that’s on the low end of what AMD is promising – the company says across a range of games, it’s seeing an average 15-percent boost in frames per second.
AMD says the increase in memory capacity and speed comes from its 3D chiplet technology, developed with TSMC. It allows the memory to be stacked in a way that allows big increases in density and bandwidth. At 2TB/s, the L3 cache is actually faster than the L1 cache, although the latency is higher.
In a chip like the Ryzen 9 5900X with three CPU core clusters, the new 3D chiplet technology allows AMD to place a stack of 64MB of L3 cache on top of each core cluster for a total of 192MB.
AMD says it will begin production of its “highest end products with 3D chiplets” by the end of this year, suggesting we could see next-gen processors featuring the new technology in late 2021 or early 2022.We saw a false hyper-clocking based announcement two days ago from Intel, one that depended on very extreme hyper-clocking stuff that runs so VERY hot that the special cooler designs it requires don't even exist yet from Intel.
Is this the definition of BOGUS Intel stuff or what ........Then 2 days later AMD routinely drops their next generation of TSMC lithography right on time along with new copper bonded chiplet stacking techniques that come direct from TSMC. AMD's new products will fully support higher (chilled water) cooling levels, but will still outperform Intel using simpler fan and fin cooling systems.
Rumor has it that AMD will get still more TSMC wafer allocation along with this 4-5nm move as AMD is keeping all their existing generations of wafer allocations rolling right along since Intel hasn't yet outperformed any of the existing AMD products based on the oldest 7nm TSMC technology. Intel is still tossing up 14nm units against the oldest AMD 7nm processors. Intel is throwing up hyper-clocked 10nm Intel chipsets against the very latest 4-5nm AMD chipsets. Nether of these match ups go to Intel, but mostly go to AMD on almost all benchmarks, including the Intel "special tests" that can only be run by Intel's special little test cheat houses.
Intel loudly claims these "partial victories" which are pretty dubious due to Intel lying and cheating on their little specialty houses benchmarking, using specialty cooling methods that are not available in the marketplace yet to set their "record performances".
AMD is expected to continue to take more and more market share from Intel since they can make up more finished chipsets now using the latest TSMC shipping generations with the 6mn lithography (and the upcoming next generations at 5nm and at 4nm will also clearly outperform Intel's 10nm products and will come with new wafer allocation increases as well).
ARM and Mediatek are expected to take more and more market share from Intel using the new, better grades of Chromebook chipsets that can be built with these new smaller lithography modes.
Apple is taking Intel market share with the M-1 chipset (and the M-2 when it gets here will take more again).
Intel has their product lines all announced now, but none have shipped yet for the independent testing that will make lies out of Intel's dubious claims. AMD is telling all the test houses to simply go buy their vendor's assembled products and test them against Intel's assembled products using the cooling systems that the vendors are providing in those production units (by doing so AMD is injecting some reality right past Intel's latest BS specialty house testing program that uses cooling systems that cannot possibly ship inside finished laptops).
AMD is confident that if cooling systems are both REAL and equivalent, their 15 watt draw units will still win out over Intel's 15-28 watt units quite handily in most finished laptop designs where Intel is sucking up 28 watts (or even more when Intel is hyper-clocking).
To combat Intel's program of seriously bogus hyper-clocking AMD now allows their AMD chipsets to be run on up to 28 watts now if the laptop's cooling system is up to it.
It is rumored that late this year AMD will now allow hyper-clocking spikes to 48 watts on their most modern processors
if and only if the laptop builder thinks his cooling system can cool that sort of action down quickly enough for the hyper-clocking to be practical.
Yes, AMD thermal loads like this may be supported now by laptop builders in some cases, but SOME REALLY STRONG cooling systems have to be really available (and fully functional) at that point in time as
AMD encourages all such products to be tested with the cooling systems that are actually used in the finished units.
AMD plays it straight, if you do hyperclocking your cooler system has to be up to the job as the complete laptop will be tested and sold (and reviewed by Tom's, Anandtech and the other real test houses as a packaged set).
AMD has much better thermal characteristics, more cores/threads, more on chip L2 and L3 cache memory and much better set of on board graphics than Intel does.