The Intel Execution Thread [2021]

Yes they are lol.

Yeah and then you start accounting for platform costs.
Lol.

No lol, stop defending the second coming of Prescott.
Just buy discounted CML (10400F's or 10850K's are good deals these days) if any AMD product is too sour grapes for you (oh, it is).

 

Platform costs are fine, there's more than Z590 for RKL.

I'm not defending anything, it's you who seem to be unable to grasp the concept of a good product - which isn't always the fastest one.

CML is slower than Zen3 while RKL isn't.
Both should be cheaper.
CML doesn't support PCIE4 for either GPU or storage.
I wouldn't buy CML now for precisely the same reasons why I wouldn't buy it a year ago. Nothing changed.

 

Lol.

Yes you are, it's cute.

Oh my-my how the tables have turned.

CML isn't that much slower than Z3 in client-relevant stuff while RKL is just too oven to justify touching it.

Irrelevant; more value-centric builds where you actually account for CPU pennies would never go pricey PCIe4 storage (aka the only bit of PCIe4 mattering in client).

Too bad, the market thingks discount CML is nice.

 

Go read the reviews please.

 

I already did; CML isn't that much slower over RKL while being cheaper and way way less of an oven.

 

Doesn't look like you did at all. Compare 10600K to 11600K here for example and tell me why I should buy the former against 5600X:



There are many such examples which make CML obsolete in this specific niche.
11700K also does fine most of the time.

Also "oven" is your own making. Nobody's forcing you to disable PLs to get +5% of performance with +100% of power draw.

 

Ah yes one specific benchmark.
Winning!

Lol.

Lol.

No, they're all nasty ovens from the get-go.
Just compare 11400 vs 10400.

 

As I've said, go read the reviews.
https://www.guru3d.com/articles_pages/intel_core_i5_11600k_processor_review,22.html
https://www.guru3d.com/articles_pages/intel_core_i5_11600k_processor_review,26.html

Nothing to add, sorry.

 

Yeah, time for you to look at them geomeans.
Fancy stuff I know.

Yeah.
Hopefully you have nothing to add in RDNA2 threads too so you stop shitting all over them.

 

Don't count on it Batman.

 

Here's my take on Rocket Lake:
Overall, this CPU family is disappointing for most power users. The architecture seems solid, but the actual products, the execution of said architecture is underwhelming, to say the least. Performance often regresses from the previous generation, and rarely approaches the "up to 19% faster" claim. It seems these chips are starved for power, and may even be hitting thermal limits in some tests. It would be interesting to see the maximum performance capabilities with these limits removed.

One interesting exception to the above is the performance in flight simulator, a notoriously CPU-bound, and even single-thread-bound application:
https://www.eurogamer.net/articles/digitalfoundry-2021-intel-core-i9-11900k-i5-11600k-review?page=2
24% increase in performance from 10900k to 11900k at 1080p, with diminishing returns as the resolution increases, of course. Flight sim is my jam, so to speak, so I am intrigued by this architecture and may in fact upgrade from my 9900k as a result of this standout performance lift over the *Lake CPUs.

 

My two cents...

I bet on those cheap B560 Boards RKL will throttle like crazy - you can count on that.

Also, for RKL you need a much beefier AIO liuquid cooler, otherwise .. more throtteling

I bet, if you pare a 11600k with a B560 for 100 USD and a cheap tower cooler for 30 USD, results will be MUCH different

You need an expensive Motherboard + AIO liquid cooler for results like in moste reviews.

 

One of the things pointed out by GamersNexus was several review motherboards which did not enforce the Intel recommendation of boost time limits. I'm wondering how many of these "great" benchmarks were done on firmware which permits basically unlimited boost clocks? While still technically within spec, it still isn't within Intel's guidelines as what should be used as a default configuration.

After having read and watched a number of reviews, and having no specific bone to pick either way, RKL seems disappointing at best. More than a few examples of the "equivalent" CML part actually being incrementally faster, and even more examples of it being approximately the same speed. In effect, RKL seems to only really "win" in a set of oddball and corner-case benches, a great example being MS:FS mentioned by Shaidar above.

All things considered, I'd probably do the cheaper CML chip and spend the bit of extra money on a better cooling setup to overclock the piss out of it.

 

I imagine the 10nm++ or whatever version will be much more worthwhile.

The single thread performance boosts are appealing to me. I still play a lot of Supreme Commander Forged Alliance online and my 4.9 GHz 8600K can drop to 50% game speed in extreme situations with thousands of units. Which is really every late game situation. The game has lots of threads, but eventually you are horribly bottlenecked by the game simulation thread.

 

There won't be anything like that.
Rocket Lake uses Cypress Cove, which is literally 14nm port of Sunny Cove, core used in 10nm Ice Lakes.
Their next architecture called Alder Lake is a step in another direction. It is 10nm++ (10nm Enhanced SuperFin), but it's in no way "version" of this. It uses Golden Cove cores (2 generations newer than Rocket Lakes Cypress/Sunny) paired with Gracemont "Atom"-cores. Topmodels will be "16C24T" by combining 8 of each and only Golden Coves supporting HT and they plan to launch it already this year (yeah, RKL is short lived)

 

Yeah I was sort of generally referring to Alder Lake. I'm excited to see how that turns out.