Quote from: Terror Byte on Today at 00:50:42

Quote from: ArsLoginName on Today at 00:29:39'...beats the Max+ 395..." Single core 3057/2929 =1.0437 or by 4.3% coming out 1 year later and manufactured on more advanced nodes. Wow. Shocking.... As for multi-core the 388H gets 17687 while Ryzen Max+ 395 has 19000.... So it loses by 7% but that is to be expected since it only has about 14 cores as the 4 LPE probably count as 2 cores or so....

Single core performance is mainly about clock speeds. This Panther Lake looks like it will perform very well at much lower total power draw than Max 390/395/285HX and should only really be compared to 285H or 370.

I'll bet it trashes both the current Intel and AMD on efficiency too.

It's not always about clock speed. Improving IPC will also improve single core performance at the same clocks. Apple's relatively wide front-end is pretty good at that, so it doesn't need high clocks. x86 designs are moving in that direction too, but only insomuch as the wider execution units can remain fed. If they're sitting halfway idle under load, that's a bad design that has a bottleneck elsewhere in the architecture.

Pushing higher clocks with narrower execution units is a tried and true approach, but even that requires some clever design tricks to scale performance up.

Quote from: ArsLoginName on Today at 00:29:39'...beats the Max+ 395..." Single core 3057/2929 =1.0437 or by 4.3% coming out 1 year later and manufactured on more advanced nodes. Wow. Shocking.... As for multi-core the 388H gets 17687 while Ryzen Max+ 395 has 19000.... So it loses by 7% but that is to be expected since it only has about 14 cores as the 4 LPE probably count as 2 cores or so....

Single core performance is mainly about clock speeds. This Panther Lake looks like it will perform very well at much lower total power draw than Max 390/395/285HX and should only really be compared to 285H or 370.

I'll bet it trashes both the current Intel and AMD on efficiency too.

'...beats the Max+ 395..." Single core 3057/2929 =1.0437 or by 4.3% coming out 1 year later and manufactured on more advanced nodes. Wow. Shocking.... As for multi-core the 388H gets 17687 while Ryzen Max+ 395 has 19000.... So it loses by 7% but that is to be expected since it only has about 14 cores as the 4 LPE probably count as 2 cores or so....

Quote from: davidm on Yesterday at 20:18:38We need to know the memory speed. Fast memory is a big part of why Macs have been ahead of mainstream PCs for so long, and four channel RAM is why Strix Halo is so fast for GPU and general operations, relatively speaking (it's actually midrange fast compared to the Macs).

Does it matter? With modern memory prices and still going up, this generation is a wash. Hopefully prices will go down by the time Medusa Halo is out.

We need to know the memory speed. Fast memory is a big part of why Macs have been ahead of mainstream PCs for so long, and four channel RAM is why Strix Halo is so fast for GPU and general operations, relatively speaking (it's actually midrange fast compared to the Macs).

I would prefer to see more laptops with AMD Halo, which is, in my opinion, the best processor on the market.
I don't understand why there aren't more laptops with AMD Halo when it's been on the market for a year now.
Let's hope that in 2026 there will be more laptops with AMD Halo.