Discussion in 'Intel x86 CPUs and chipsets' started by vlinx, Dec 12, 2020.
Why be proactive when you can be reactive!
I'm confused, isn't the ring bus used in Intel currently and for the past while the same as Infinity fabric basically anyway?
You got the mem controller in the system agent which is like the cIOD in zen which communicates over the ring bus which is like Infinity fabric to the cores and cache?
I'm not even sure what I'm trying to say tbh.
Ring bus connection is like a big ole roundabout and Infinity fabric is like a point to point kinda thing right?
But in the end they both connect to the same things and Intel could still use ring bus but chuck a clock divider in there right?
Ice Lake is still using the ring interconnect (I know it's a completely different core)
All I could find was this which I think some guy just put labels on the diagram.
Yeah basically ring bus doesn’t scale. They never intended for it to go past 4 cores/8 threads, yet they squeezed 10 cores out of it. That’s how stretched it is.
The advantage they have is that since it’s so mature, they’re getting very good clock speeds out of it.
I don't think there's any indication that RkL has moved to a mesh. It's only 8 core after all.
It may have inherited Tigerlake's Dual ringbus, as part of porting the xe GPU over, and this might be a by-product of that, in order to scale to higher RAM frequencies . All speculation, until it's not
Nah we're not saying it has, it's just that they're moving towards the AMD "glue" design that they have hung so much shit on.
What do the xeons use?
So, fair to say the solution to ring bus not scaling already exists.
Newer ones use mesh.
First (legitimate) review of the 11700k.
Not looking good unfortunately. Power is nuts - 291w vs 204w for the 10700k and 141w for the 5800x.
Conclusion: The War of Attrition - Intel Core i7-11700K Review: Blasting Off with Rocket Lake (anandtech.com)
Oufff those gaming benchies look awful.
Hahahahahha welcome to Ryzen 1000 series Intel. Enjoy the glue.
That is insanely crap..wow
Don't know what's more oof , the results, or Ian losing the plot. Talk about destroying their Intel relationship in one attn grabbing review LOL . Not a good look honestly. How Anandtech have fallen since that numpty took over.
Before publishing this review, we gave Intel advance notice to respond to us having a full review ahead of the formal release. Our email seemingly generated some excitement inside (and to our surprise, outside) Intel, but we received a response from Intel stating that they had no comment to offer.
"No Comment" .. well, if there is a performance saving magic microcode update that everyone's waiting for, then they might want to comment soon? Anantech are still a big audience , despite killing themselves slowly . It'll be all over the place soon.
I’m sorry but No microcode update will make up for that much difference in performance.
Azzan Hope this isnt true???
I agree, this is pretty much it, but the results are so bad for gaming performance, I can't help but give some benefit of the doubt, at the least gaming perf might improve. People seem adamant the release date is what it is because of this uCode update , so we'll see.
Will get more noise about this pretty soon. They'll need to be seeding proper review samples with final BIOS/uCode very soon if NDA is the 30th
Anyway, overall nothing really unexpected. As I said before, Can't think of a single backporting job that wasn't suboptimal .
going to be hard for a lot of folks to cool that i would think?
yep, I alluded to this somewhere earlier in this thread, could see it coming a mile away, but it was brushed aside. Anyway, progress will come, Intel guys have to be patient, it is the Pentium 4 Netburstb to Dothan/Merom move all over again. It will come, but will take a couple of years.
‘In the meantime, there are faster and better alternatives in AMD at this point in time. That’s another topic altogether again, not without its issues, but at this point the fastest and most efficient setup you can buy.
Looking forward to a time when Intel will get on top of the new fab-process, it will come eventually, and competition is always good for us as consumers.
Maybe for those specific AVX workloads or a bad sample? Im not sure.
Compare this screenshot from hardwareluxx forum, R23 at stock. The 3rd image showing HWINFO temps below 64C
Edit: Using a custom w/c loop
Wow, looks like I'll be hanging onto my 10700k for a while then.