16 comments

[ 2.6 ms ] story [ 40.4 ms ] thread
How is the "e-core" scheduling going these days? I am not an Intel guy, but I remember reading about it causing a lot of issues for multicore workloads and people having to explicitly pin processes away from those cores.
This <insert next gen> will be a beast!
If anyone was confused like I was, it's (Xeon 7) (E-Core). It'll have 288 cores. The Diamond Rapids Xeon 7 P-core chip has 192 cores.
After Intel released a microcode update that turned my processor into one that retails for $100 less, I don't really trust claims or even benchmarks until they're > 1 year old.

Intel also, as far as I'm concerned, owes me $100.

Intel made some really good low power gpus with good driver support. How are their gpus coming along for LLM support of local first?
So, its just a clone of current gen Zen5C products, and will fall behind Zen6-based products.
"how many N100 minipcs can you pack in 1U?"

turns out the answer is "lots"

The hyper-scalers are going to love this, since this has just brought down the cost per vCPU massively.
Not really, these E-Core are 1vCPU, while other P-Core or Zen 6C core are 2vCPU.
That first paragraph made my head hurt. A lot of contortion to avoid saying 'Intel has slightly more than half the (edit) revenue in the server market'.

I thought I'd find a chart, best I can do is here:

https://www.techpowerup.com/322317/amd-hits-highest-ever-x86...

Intel's revenue (edit) is falling year-on-year, and AMD gaining. Does anyone have a better chart?

Intel was asleep at the wheel for a decade at best and realistically was just milking profits while sitting on innovation. Disgraceful.

I'm team AMD all the way.

I have no faith or expectation anything Intel does will matter.

Maybe 10 years of them being irrelevant will convince them of their ways?

288 cores and 12 memory channels. If I did anything with this chip, I would probably just swamp its memory bandwidth and get no performance gains over a 48 core chip.
Somehow I doubt it.

The last time I was excited about anything Intel did was the Xeon Phi. And Intel failed spectacularly to follow up on a very good idea.

Intel simply can't innovate.

288 "efficiency" cores.

That each have 26 execution ports, can retire up to 16 ops per cycle.

(comment deleted)
Now if only Intel would start packing their iGPUs (maybe more than one, or many for a high core count CPU like this) on their Xeon CPUs, you could basically build insane video streaming infra with little hardware (given how good the iGPUs are at things like transcoding on a watt for watt basis). This seems like a huge slam dunk slapping at least one iGPU on their Xeon CPUs like you used to be able to get...I have no idea why they stopped.