It looks like common leaker Raichu (@OneRaichu on Twitter) has gotten his little Pokémon paws on a pair (a minimum of) of pre-release processors from each Intel and AMD. With the {hardware} in hand, he did what any correct fanatic would do, and benchmarked them towards one another.
Nonetheless, these benchmarks aren’t simply the same old Geekbench or CPU-Z, however as an alternative the server-oriented SPECCPU2017 suite, and he ran them utilizing fastened 3.6 GHz clock charges. This offers us an fascinating window into the true per-clock efficiency of Intel’s Raptor Lake processors in addition to AMD’s Zen 4 cores.
It is nonetheless fairly fascinating to see over the information, although. Raichu examined each chips with DDR5-4800 RAM, which is notably beneath the inventory pace for each chips. Then, he examined them utilizing DDR5 reminiscence operating at 6000 MT/s—which AMD beforehand confirmed to be the “candy spot” for Zen 4 processors.
Whatever the configuration, Intel’s Raptor Cove and AMD’s Zen 4 (represented right here by “RPH” for Raphael, the codename of the primary Zen 4 desktop CPUs) are neck-and-neck. Intel’s components usually have a small benefit apart from in SPECint with the slower 4800 MT/s pace, the place the distinction is so small it might as effectively be margin of error.
Opposite to what you may count on, Zen 4 really good points much less efficiency from the sooner reminiscence than Raptor Lake does. Raichu feedback that Zen 4 has a “nice” cache configuration, which makes it “not very depending on DRAM.”
Whereas testing on the larger reminiscence pace, Raichu additionally did the mathematics for the Intel chips’ Gracemont cores. Unsurprisingly, they arrive out far behind the large cores in floating-point efficiency, though they’re impressively shut in integer math. Raptor Lake presents a tiny IPC uplift for these cores too, doubtless as a result of doubled L2 cache obtainable to these cores.
Earlier leaks have promised that AMD and Intel’s next-generation chips might be neck-and-neck for efficiency, and this testing bears out these guarantees. If the IPC is that this shut, it is doubtless that the efficiency crown will come right down to whoever can clock the very best, and it seems to be like Intel’s gunning for that title by any means mandatory.
