The cores are using 65w but it's the memory controller which bumps the total package power.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Anyone going to wait for Zen4 Vcache '7800X3D' / '7900X3D' CPUs?
- Thread starter g67575
- Start date
I don't understand the comparison here - isn't the 5800X (from @Joxeon ) at best a middle of the stack chip (and under £280) compared to the 12900K, which is top of the stack and almost twice the price by @Bencher .
So forgive me if I'm being dumb, but shouldn't this better performance be expected, considering the outlay?
Also, can we please not turn every thread that mentions AMD into a 12900k cinebench vs current zen 3, especially when the thread is about a different architecture that's got no definitive benchmarks.
@g67575 - If you can wait a couple more months for Zen 5's release date, reviews etc then I'd do that -if only to have the most choices when it comes to purchase. Plus if they (zen5) are competitive in price due to lack of absolute performance vs Intel, then Intel will probably drop the price a little, so win-win.
So forgive me if I'm being dumb, but shouldn't this better performance be expected, considering the outlay?
Also, can we please not turn every thread that mentions AMD into a 12900k cinebench vs current zen 3, especially when the thread is about a different architecture that's got no definitive benchmarks.
@g67575 - If you can wait a couple more months for Zen 5's release date, reviews etc then I'd do that -if only to have the most choices when it comes to purchase. Plus if they (zen5) are competitive in price due to lack of absolute performance vs Intel, then Intel will probably drop the price a little, so win-win.
I know, that's the problem with zen 3 and what makes the really inefficient compared to alderlake. On alderlake the rest of the package draws like 1 to 3 watts.
It's probably my fault, I think it set people off when I said "Well, per core Intel's P-cores use quite a bit more power. They couldn't increase the amount of Golden Cove cores over 8, because of the already very high power requirements".
This actually may not be strictly true, in the sense that if the clock speeds were limited to something like 4.4-4.6Ghz at a 'tighter' voltage setting, it might've been possible to design chips with more P-cores (maybe 12/16), which would be more similar to the server chips, which feature solely Golden Cove cores. The server chips can apparently boost upto 3.7ghz.
I would point out that Intel's 7nm EUV process in 2023, is likely to be somewhat better at handling higher voltages and clockrates, whist using less power, compared to the current 10nm process.
Regarding Zen 5, it's true that people could be waiting for what seems like forever... But the solution is simple, just buy a low or mid end CPU on AM5 when they release in Q4 2022 (assuming recent single core performance rumours are correct), then it should be easy to upgrade later.
Last edited:
I disabled the 8 ecores to compare 8gc vs 8 zen 3 cores. What difference would it make if he had a 5900x and i had a 12700f instead and still compared 8gc to 8zen 3?
I don't know and that's why I asked the question, I wasn't trying to trick anyone here. I saw a test of two CPUs where one was more expensive and newer and was more efficient, which just seemed like it should be the case and people are acting like it shouldn't.
In real word use, do Intel users disable the ecores if they are what cause any inefficiencies? Or is the score per watt still better will all of the cores on?
The reason i disabled the cores was to test 8gc cores with 8 zen 3 cores. Its a popular myth lately, especially in this forum, that the zen 3 cores are more efficient than the gc cores. So i wanted to figure out if thats true, and lo and behold, its not even close. The gc cores are way more efficient, which makes sense considering they are much bigger in die space.
The difference is big enough that most likely not even zen 4 can close that gap
There are too many variables unaccounted for to extrapolate what Zen 4 can or can't achieve efficiency wise from that particular 65w GC/zen3 8v8 cinebench comparison.
What you are testing in 5800X is Zen 3 desktop package configuration efficiency, and the point on the voltage frequency curve that can be achieved by the cores with whatever power budget is left of the 65w target. So much of that will change in Zen 4.
The closest apples to apples would be zen 3 Cezzane chips. You have 8 zen 3 cores in a monolithic die with integrated IMC - space and power efficiency, but then there are mobile arch specific compromises for space/power/cost efficiency that can adversely impact specific benchmarks such as reduced cache size and infinity fabric speeds.
It would be interesting if anyone has a 5700g they could run.