Why isn't the M4 also another Rocket Lake?
I don't follow. Why would you expect it to be?
if they would have kept the memory controller on the compute tile. Would it have been perfect? No. But as a worst case it could have still been better than whatever benefits were achieved by using the underperforming tile layout they chose.
I'm honestly not sure how much the tile-based layout is to blame, here. Intel has prior experience with these technologies, even in the mainstream Sapphire Rapids, where the XCC configuration used like 12 tiles(!) and AMD is doing alright with chiplets using even older technology for inter-die communication.
I look forward to some memory latency benchmarks between Meteor Lake and Arrow Lake-H, using LPDDR5X memory with the same intrinsic latency. Then, we can actually see the cost of Arrow Lake's tile architecture vs. Lunar Lake's.
Why not keep latency low for the latency sensitive stuff and just tile off things like the iGPU and SOC stuff?
It's clear why Meteor Lake didn't do that. Intel said it was about power savings. Recall they put an e-core island and video decoder on the SoC tile, so they could power down the iGPU and even the CPU tile, when doing low-power stuff like video playback. I think they got really spooked by Windows-on-ARM and Qualcomm/Nuvia entering the laptop market, so Meteor Lake targeted power efficiency as its primary goal.
Why did they reuse the same tile architecture for desktop? Well, we don't yet know how much it's to blame, but we should also consider Intel might've been prioritizing development costs, time-to-market, and having a larger number of silicon configurations over delivering a solution that's most optimized for performance desktops.
The uplift in average thread IPC in ARL combined with CUDIMMs and the extra 50% L2 on the fast cores should have been enough to make RPL look like Skylake when compared to Arrow Lake but instead we got bungled node execution, bungled CPU component layout and bungled software and firmware deployment and wound up with a year or two of a CPU competing with the refresh of a previous generation. This seems like the sort of thing that the chief executive officer of a company might have to take responsibility for.
Yes, yes, and yes. However, I think you're probably overestimating how important the performance desktop market is to Intel's bottom line. It looks really bad and no doubt has an impact on their reputation in the industry, but Intel still seems to have had little trouble selling as many Arrow Lake CPUs as they can produce and the entire desktop market (not just performance desktops) is a small slice in their overall revenue pie.
Even if ARL on Intel 7 had to drop the clocks 10% from RPL
At this point, you're making unfounded speculation. If it's it's not tied to any hard data about how the two nodes actually compare, that makes it fantasy. I've said what I have to say about this and I'm not going to expound on that any further.
Facebook
Twitter
Instagram