-snip-
The successor to the i5-9300H is the i5-10300H. The i5-10600T is intended for desktops
according to Intel themselves, and the final letter "T" (low-power desktop) clearly marks it as different from the ones that end in "H" (high-power laptop). They (i5-9300H and i5-10600T) occupy a different market segment and should not be compared.
I don't think it matters that the i5 has 4 cores and the Ryzen, 6. Let me put it this way: they're not even meant to compete with each other (one is for standard laptops, the other for high-power laptops), market-wise and the lower-power one
won. They occupy the same performance segment (i
5 and Ryzen
5), so it's not a terrible comparison to make; if anything, the Ryzen's at a disadvantage because it's a lower-power part. Maybe AMD's part isn't inherently better, but that process node advantage makes it more power-efficient. It's an impressive showing from the Ryzen, nonetheless.
That shows it's faster and more efficient than the i5, considering that my i5 is basically allowed to draw as much power as it wants in my testing, yet the Ryzen ends up with higher or at least equivalent scores relative to the i5 despite a smaller power budget. The i5 would lose even harder if the i5 was running stock; something like 1600 points in Cinebench R20 instead of 2050 points. This i5 is literally the best-case scenario, and it still lost against an apparently "smaller" target.
For an apples-to-apples comparison, let's compare the Ryzen 5
4600H against the i5-10300H. They're in the same power category, the same performance segment, and were released about the same time. They cost roughly the same. The number of cores shouldn't matter, given that these two are meant to compete with each other. You can buy gaming laptops with either an i5-10300H or a Ryzen 5 4600H.
i5-10300H:
8593 points (2663 single-thread) in Passmark's CPU benchmark.Ryzen 5 4600H:
14949 points (2497 single-thread) in Passmark's CPU benchmark.Doing 74% better than your competitor in multi-thread with like-for-like parts isn't a fluke, that's a kill shot. Maybe the Zen 2 arch of the Ryzen isn't inherently better than the Skylake-derived arch of the Intel, but TSMC's smaller process (vs Intel's 14nm) means that in power-constrained situations (such as in laptops), the Intel gets completely destroyed by the Ryzen because it's able to derive more performance from the same amount of power. The single-thread's slightly worse (the Ryzen gets 93% of the single-thread score of the Intel), but at that point, it's nitpicking.
I think it's worth mentioning that the Ryzen 5 4600H (and all 4000-series APUs) has only 8 PCIe 3.0 lanes instead of the full 16 PCIe 3.0 lanes of the Intel, so it's not like it's perfect. With higher-end GPUs, that might present a bandwidth bottleneck,
though from this Techspot article about PCIe bandwidths and how it impacts RTX 3080 performance, it shouldn't be much of a bottleneck for today's games even at x8 3.0, and that's for a
desktop 3080. That's for gaming performance, though. Anything bandwidth-sensitive might see a substantial decrease in performance, though I can't think off the top of my head anything that would rely so much on PCIe bandwidth. I know some that rely on VRAM bandwidth, but that's a whole different can of worms.