Erm....do you think a modern 8c16t cpu is running less than 16 processes at any one time?
-
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.
Alder Lake-S leaks
- Thread starter Grim5
- Start date
Erm....do you think a modern 8c16t cpu is running less than 16 processes at any one time?
No but they certainly aren't all pegged at 100% while playing any games, there's more than enough spare cpu resources to slot in background processes.
Don
Context switches still incur a performance cost - and that occurs twice (once to switch away from your game thread, and again to switch back). Yes it's likely in the order of 0.1 microseconds, but these all add up when contributing to frame times etc.
(It's also part of the reason why people have said that Ryzen chips have appeared "smoother" at times - their context switches are noticably faster at ~0.04 microseconds)
https://techreport.com/forums/viewt...Hz processor (not,and 0.1 microseconds (Intel).
https://www.phoronix.com/scan.php?page=article&item=3900x-9900k-400&num=7
And actually that 0.1 microseconds is long time in scale of CPU.
For 5 GHz CPU clock cycle lasts 0.2 ns.
So that would be 1000 cycles + execution time of background thread down the drain and out from running game.
Having guaranteed exclusive use of CPU cores is the reason why console games can be optimized to squeeze out such high amount of performance in comparison to hardware.
Yep - it shall be a marketing meme for Intel to claim they have the first PCIE5 system, but it will still be 3 to 5 years from now before any graphics cards reaches the max bandwidth of PCIE4
The leakers are claiming the 12900k beats the 5950x in multithread benchmarks.
But it's difficult to ascertain if its accurate and also if its an apples to apples comparison. I suspect not because the 12900k system is using DDR5 memory which in many cases would give it a leg up against DDR4 system like the 5950x is limited to.
It is possible that's Windows 11 and Intel's new scheduler combined with this big.little architecture has in fact made a massive improvement to multithreaded system performance but it could just as easily just be the DDR5 that's doing it - only time will tell.
Either way it's still impressive. The 11900k has about half the multithread performance of the 5950x, so if by adding 8 little cores clocked at 3.7ghz along with a new scheduler, some IPC improvements and DDR5 has allowed it to more than double it's multithreaded performance then that's amazing.
But it's difficult to ascertain if its accurate and also if its an apples to apples comparison. I suspect not because the 12900k system is using DDR5 memory which in many cases would give it a leg up against DDR4 system like the 5950x is limited to.
It is possible that's Windows 11 and Intel's new scheduler combined with this big.little architecture has in fact made a massive improvement to multithreaded system performance but it could just as easily just be the DDR5 that's doing it - only time will tell.
Either way it's still impressive. The 11900k has about half the multithread performance of the 5950x, so if by adding 8 little cores clocked at 3.7ghz along with a new scheduler, some IPC improvements and DDR5 has allowed it to more than double it's multithreaded performance then that's amazing.
Don
So you want to see 1000w cooling solutions?
By allowing some parts of the package to use lower powered more efficient cores, it frees up some of the total package constraints to allow the higher performance cores to use extra power and that cooling budget.
Small cores might be useful if you could do all your non essential/ background tasks, like running browsers, playing music, running apps etc on those cores.
And save the large cores solely for playing games / demanding tasks. It does seem pretty niche though for desktop users, not sure why we would need more than 3-4 small cores for background stuff.
I hope the next CPU generation after Alder Lake just moves onto 12-16 larger cores instead.
Windows 10/11 update service will probably be setup to run constantly on the smaller cores
And save the large cores solely for playing games / demanding tasks. It does seem pretty niche though for desktop users, not sure why we would need more than 3-4 small cores for background stuff.
I hope the next CPU generation after Alder Lake just moves onto 12-16 larger cores instead.
Windows 10/11 update service will probably be setup to run constantly on the smaller cores
Last edited:
With Intel these days assuming their latest processor can beat their own (nevermind a competitors) last generation chip isn't a safe bet!
You'd hope with Alder Lake (and the move to DDR5 plus all the other platform/process changes) it'll make for the largest performance jump we've seen in several years as I'm sure it'll come with a noticeable hike in pricing.
It is just Intel not be able to make anything new these days, so this time instead of rehashing once again they try this approach, cores do not always run at 100%, they have steps even mine does and a core hitting 100% when opening browser or AV scan is not same as 100% running prime95.
Not sure where you got 1000w cooling from, nearly every Gen ran cooler than the last (even though smaller surface area) till recently and that is partly the crap TIM and now so called new solder (if it was proper solder like SB it would not delid via a tool and if you did try it would kill the CPU).
Last edited:
The 12900k is 25% faster than the 5950x in single core
https://wccftech.com/intel-core-i9-...nchmarks-faster-than-amd-ryzen-9-5950x-rumor/
https://wccftech.com/intel-core-i9-...nchmarks-faster-than-amd-ryzen-9-5950x-rumor/
The 12900k has a top boost frequency of 5.3ghz, so I think we can compare it to the 11900k at least as that chip also has a top boost of 5.3ghz.
So it's not like Intel has suddenly got a crap load of extra frequency out of their architecture that's allowing this large performance boost.
Associate
- Joined
- 29 Jun 2016
- Posts
- 541
That is quite the IPC uplift, if true. I'm currently putting it in the possible but unlikely pile of rumours.
I would love for it to be the case though, all this competition is forcing innovation on several fronts.
EDIT: Do we have any idea of the die size and transistor count? Maybe Intel have chucked a shed load of transistors into improving branch prediction and widened the pipeline?
Knowing Intel, improvement in Cinebench is not a good IPC measure.
There was a nice uplift in CB single thread between Intel 10 and 11 gen, but in games and real world the improvement was less. And when coupled with lower average clocks of 11 gen, almost no difference.
Whereas between Zen 2 and Zen 3, improved CB matched reported IPC across a range of apps (and of course game performance was even higher)
There was a nice uplift in CB single thread between Intel 10 and 11 gen, but in games and real world the improvement was less. And when coupled with lower average clocks of 11 gen, almost no difference.
Whereas between Zen 2 and Zen 3, improved CB matched reported IPC across a range of apps (and of course game performance was even higher)
Soldato
- Joined
- 31 May 2009
- Posts
- 21,456
benchmark makes no sense at all, even if the single is as powerful as it is, how on earth does it stack up that only 8 cores can pull that amount of multi
Because it has 16 cores, not 8. I dont' know why you thought it only has 8