With the release of Alder Lake less than a week away and the "Lakes" thread having turned into a nightmare to navigate I thought it might be a good time to start a discussion thread solely for Alder Lake.
Scroll a bit lower, the last table is a small table with efficiency in cinebench 23 that only has a few entries, the 12900k locked at 160w scores 25590 which is more than the 5950x although the 5950x does so at lower power, that's what he meant, the 12900k is perfectly fine at 125w and you only have to push it to 160w to beat the 5950x.
Less efficient still but not 241w to 105w as many are left to believe, 160w to 120w doesn't sound all that terrible anymore.
Good video here
But you can't, unless you void your warranty.and it wont happen if you push 5950x to 160W too.
But you can't, unless you void your warranty.
Yep, exactly the reason why I am passing on ADL for now. Even TVB being "done-ski" and not validated for 12th Gen CPUs points to the immaturity of the platform.
I am excited to see the face-off between Zen 4 and Raptor Lake as that will likely be my next upgrade point and hopefully by then DDR5 frequency and latency will be good enough to be clearly superior to DDR4. Because the premium for DDR5 is not worth it at the moment...
CB 15? What about the newer versions? I bet it isn't close. We've seen the numbers of the 5600x and 12600k (both run at 4.5GHz).This is it here.
View attachment 52835
Core i9 12900K processor review (Page 7)
Meet Intel's new flagship CPU, the Core i9 12900K. It is based on the Alder Lake architecture and is reviewed here. This time around, Intel was back at the drawing board, creating a completely new ar...www.guru3d.com
Both AMD and Intel chipsets are generally very stable, even at the beginning. Way back in the day when AMD left their chipsets to third parties like VIA were responsible for AMD chipsets, that was a whole different ballgame.Since you opened this door I will comment that we are talking about not only a new architecture, but a hybrid design, coupled with a new memory standard. Needless to say that's a lot of new variables. There are bound to be issues in the beginning. I don't think it's a big leap to write that generally Intel platforms are regarded as very stable once mature.
CB 15? What about the newer versions? I bet it isn't close. We've seen the numbers of the 5600x and 12600k (both run at 4.5GHz).
CB 15? What about the newer versions? I bet it isn't close. We've seen the numbers of the 5600x and 12600k (both run at 4.5GHz).
With the insane video card prices of 2020-2021 (and likely going forward from here on out), that wouldn't be too hard of thing to do.I prefer to set as much budget as I can to the GPU, with memory, motherboard, storage and CPU after that. In that order.
These threads are informative first, and Ford v. Chevy second. Nothing wrong with that, keeps the discussion lively.Since you opened this door I will comment that we are talking about not only a new architecture, but a hybrid design, coupled with a new memory standard. Needless to say that's a lot of new variables. There are bound to be issues in the beginning. I don't think it's a big leap to write that generally Intel platforms are regarded as very stable once mature.
Both AMD and Intel chipsets are generally very stable, even at the beginning. Way back in the day when AMD left their chipsets to third parties like VIA were responsible for AMD chipsets, that was a whole different ballgame.
However all new CPUs/chipsets/hardware in general has various growing pains to deal with when first launched. Throw Windows 11 just being released into the mix, and it's not unexpected for this new hardware to have some issues to work through.
That's why I personally wait to build/upgrade to any new hardware, and I generally wait 6 months or so before considering doing so. With operating systems, I generally like to wait about one year to let all the early adapters do all the beta testing stuff.
Good video here
How much silicon, area-wise are we looking at with a 5900X vs. 12900K? I'm thinking it's pretty close to the difference in efficiency.
what are you talking about man?
So it does get amusing to see people pouring over minutiae of the top end parts looking for gotchas.
VIA KT133. Horrible nonsense that didn't really get completely resolved until the awesome KT133A, with which I was able to run Kingmax BGA PC150 Sdram on finally. I seem to recall being so annoyed with KT133 that I rode a K7S5A Fry's special with a coked up modded bios for a while, and it was decidedly more stable for me until the KT133A fix. Went KT133A until ..
VIA KT266. Why not repeat the same mistake again? Mysteriously bad IDE and memory performance, stability issues, worse performance with DDR than my Sdram KT133A build so I abandoned it until I believe the KT333 and Nforce Era / thoroughbred stuff, even though the KT266A once again cleaned up the mess of the OG 266.
Pretty much a similar path I took back then with the various hardware.What? You didn't enjoy being a beta tester and paying for the privilege? 🥲😂
Some notable experiences from my end :
AMD Irongate. Not going to lie, I had fun with goldfingered Athlon Slot A stuff, but the motherboards I used (an Asus and an Abit I think) both had nightmare level issues with regards to AGP speed, BSODs, memory timings, IDE performance, even hating my brand new PCI nic at the time, a 3Com card with the extra cache. Socket A couldn't come fast enough.
VIA KT133. Horrible nonsense that didn't really get completely resolved until the awesome KT133A, with which I was able to run Kingmax BGA PC150 Sdram on finally. I seem to recall being so annoyed with KT133 that I rode a K7S5A Fry's special with a coked up modded bios for a while, and it was decidedly more stable for me until the KT133A fix. Went KT133A until ..
VIA KT266. Why not repeat the same mistake again? Mysteriously bad IDE and memory performance, stability issues, worse performance with DDR than my Sdram KT133A build so I abandoned it until I believe the KT333 and Nforce Era / thoroughbred stuff, even though the KT266A once again cleaned up the mess of the OG 266.
Intel 845. Abominable performance, but stable enough I suppose. Work Dell Dimension something or other pairing a Willy with Sdram, a true low point. Northwood with proper DDR was so so much better.
Intel P68. Birth of UEFI era, and my Asus P8P67 Pro was just not a fun time at all, I seem to recall it being fixed to some extent with later revisions, but too many problems to list.
Ryzen 300 series mobos for me combined with Zen1 was mostly bleh, for the most part because getting Ram, even from the so called QVL lists to run at rated spec was often an exercise in futility. I even ran into cases where a 3000 kit would not run at 3000 in any way, but DID work at 3200 after loosening the timings slightly. Zen2 and 400 series was decidedly better with much fewer issues. Zen3 and 500 series has been painless for me in all areas other than I hate having a Chipset fan with all my heart.
I also had some older and more obscure stuff, but those were mostly trade show oddball nonsense I got for cheap to play with that turned out varying degrees of great, indifferent, or useless lol.
I had a 939 VIA chipset (A8V Deluxe) at one point and it was actually pretty solid. I recall it had much fewer issues than several nforce 2/4 boards I had around that time.
I am excited to see the face-off between Zen 4 and Raptor Lake as that will likely be my next upgrade point and hopefully by then DDR5 frequency and latency will be good enough to be clearly superior to DDR4. Because the premium for DDR5 is not worth it at the moment...
Same. While I am impressed by Alder Lake and acknowledge it a strong first strike to get Intel back in the game, the real fireworks are going to be when Raptor Lake takes on Zen 4. All Intel needs is an additional 10% IPC gain in microarchitectural improvements with Raptor Lake, and the enhancements to the cache, memory controller and power efficiency should make it very competitive with Zen 4. I actually wouldn't be surprised if Raptor Lake manages to beat Zen 4 in single thread performance.
Supposedly, Raptor Lake will have a Digital Linear Voltage regulator that might lower power consumption as much as 25%. The way the patent is described, makes it sound as though it will have an effect similar to an experienced overclocker undervolting a CPU without the laborious tuning, and we all know that Alder Lake is horribly overvolted because motherboard voltage references are always sky high.
Intel Raptor Lake’s Special VReg Tech Could Dramatically Reduce Power Consumption
So yeah, Raptor Lake will likely be much better and more polished as a product compared to Alder Lake. Amazing to finally see Intel making a comeback!
It's true, Alder Lake-S is weird. I'm beginning to see that people who claim Alder Lake is more of a mobile part in a desktop part's clothing are correct (which makes Intel's decision to release it first on desktop particularly strange), and that the 12900k in particular seems to be operating well outside of its efficiency range when pushing all-core clocks to such extremes.
Intel would have been better off with more P cores and fewer E cores on desktop, with lower power limits and lower clocks in MT workloads. Golden Cove isn't perfect, but trying to cover up for its failings by running up its clockspeeds and then adding some Gracemont (that isn't on a separate power plane!) leads to some bizarre behavior.
Or more to the point, if Intel were to release a 16c Golden Cove power limited to ~150W, I'm pretty sure it would beat the 5950X in nearly everything.
Well this is the proper way that it should work, heavy benchmarks that occupy all available cores have to run at low priorities to not lock up the system during the whole run and the hardware scheduler throws low priority threads to the ecores, linux 5.16 fixed the problem the previous version had, nothing to see here.Linux is just not happy with Alder Lake.. Which is strange since they have been doing great with big.Little
Perhaps a Smart thread Scheduler with AI?If you want something to run on all available cores you have to manually set this up, otherwise what's the point in having a thread scheduler.