xSOSxHawkens Jul 16, 2020 @ 7:21pm
And so it begins... Death Stranding Core Scaling...
And so starts the onlsaught that will be high thread count games... Last gen pushed devs to widely support 8 threads (due to console limitations). Current gen is pushing the standard to 16 threads... And now we are starting to see modern games hitting the market that make *MASSIVE* use of highly threaded CPU's...

https://twitter.com/CapFrameX/status/1283460107809030149

Just look at the frame time comparison between the 4t vs 24t bench....

Anything 4c/4t is frankly dead at this point. 4c/8t is holding on by a string, and I would highly suspect to see games in the next 12-24m pushing 6c/12t minimums, either in listed min requirements, or in real world needed power for playable purposes regardless of listed minimum.
< >
Showing 16-30 of 36 comments
Tiberius Jul 17, 2020 @ 10:32am 
Ehh.. i was expecting a real comparison between 9700k and 3700x benchmark. This is pointless
r.linder Jul 17, 2020 @ 1:04pm 
Once more titles start using more cores, it's going to become an all-core frequency battle, and single core boosting will mean nothing. Hopefully, boosting all cores will become a standard so stock isn't useful to just single core loads.
Tiberius Jul 17, 2020 @ 1:33pm 
Originally posted by Escorve:
Once more titles start using more cores, it's going to become an all-core frequency battle, and single core boosting will mean nothing. Hopefully, boosting all cores will become a standard so stock isn't useful to just single core loads.

Is this really new information tho?

https://youtu.be/vVjdhXAdKE0
r.linder Jul 17, 2020 @ 2:42pm 
Originally posted by Tiberius:
Originally posted by Escorve:
Once more titles start using more cores, it's going to become an all-core frequency battle, and single core boosting will mean nothing. Hopefully, boosting all cores will become a standard so stock isn't useful to just single core loads.

Is this really new information tho?

https://youtu.be/vVjdhXAdKE0
All core boost is, because turbo has always been single core.
xSOSxHawkens Jul 17, 2020 @ 3:21pm 
Originally posted by Tiberius:
Originally posted by Escorve:
Once more titles start using more cores, it's going to become an all-core frequency battle, and single core boosting will mean nothing. Hopefully, boosting all cores will become a standard so stock isn't useful to just single core loads.

Is this really new information tho?

https://youtu.be/vVjdhXAdKE0
Yes it is... Look at the comparison you gave, it only goes up to 12 threads, with diminishing returns past 8t...

Now look at the listed bench from the listed game. Notice how there is reasonable scaling with reasonable gains all the way up to about ~24 threads, with still noticable but marginal gains up to 32t? Thats a big change.... And a change for the good for both Intel and AMD users.

More for AMD users and less for Intel though, as anything multi-core aware will remove Intels Single core lead and put their chips *much* closer in games (or even perhaps behind in them) compared to intel, but that will be a wait and see type of thing.



Originally posted by Tiberius:
Ehh.. i was expecting a real comparison between 9700k and 3700x benchmark. This is pointless
That would be poitnless. You cant compare core scaling from AMD to Intel like that, its an apples to oranges comparison and not a valid way for one to cover core scaling. Neither would using an Intel chip with its highly volitile and large spread boosts, which is why the Intel one is locked at 4.6Ghz for the testing (Ryzen is close enough at stock from boost to base under loads to not really need this (300-400mhz varrience).

What you wanted is a comparson between the CPU's in raw game performance, not a comparison on how each one scales with core loading.
UserNotFound Jul 17, 2020 @ 7:14pm 
I recall back in the days of 2nd gen i5 4C/4T and i7 4C/8T days, I used to argue for the latter, stating that 4C/8T would be more useful as games get more and more complex (hate to bandy this phrase, but I was talking about future proofing). I agree that 4C/4T i5s were better OC'ers but no matter how good the OCs were, you can't magically get the extra threads.

Most of the time, I'd be shouted down by many who'd said that 4T was all one needed, and I simply got tired of arguing so I gave up after a while. Even now, my 4C/8T 4770K + GTX1080 (HTPC rig) does games very nicely at 4K (at mid graphics setting) and yes, I know my GPU may be holding back performance at this res, but hey, as long as gameplay is smooth, I don't care. I don't play a game and obsess over the framerate numbers displayed by MSI AB...
Tiberius Jul 17, 2020 @ 9:00pm 
Originally posted by xSOSxHawkens:
Originally posted by Tiberius:

Is this really new information tho?

https://youtu.be/vVjdhXAdKE0
Yes it is... Look at the comparison you gave, it only goes up to 12 threads, with diminishing returns past 8t...

Now look at the listed bench from the listed game. Notice how there is reasonable scaling with reasonable gains all the way up to about ~24 threads, with still noticable but marginal gains up to 32t? Thats a big change.... And a change for the good for both Intel and AMD users.

More for AMD users and less for Intel though, as anything multi-core aware will remove Intels Single core lead and put their chips *much* closer in games (or even perhaps behind in them) compared to intel, but that will be a wait and see type of thing.



Originally posted by Tiberius:
Ehh.. i was expecting a real comparison between 9700k and 3700x benchmark. This is pointless
That would be poitnless. You cant compare core scaling from AMD to Intel like that, its an apples to oranges comparison and not a valid way for one to cover core scaling. Neither would using an Intel chip with its highly volitile and large spread boosts, which is why the Intel one is locked at 4.6Ghz for the testing (Ryzen is close enough at stock from boost to base under loads to not really need this (300-400mhz varrience).

What you wanted is a comparson between the CPU's in raw game performance, not a comparison on how each one scales with core loading.


The performance jump (look at the total percentage) is even bigger on that video that what you provided on this link. :steamfacepalm:

https://mobile.twitter.com/CapFrameX/status/1283497696695857154

But then again this test is still pointless. Ppl care more if a particular processor is faster and better than the other processor
r.linder Jul 17, 2020 @ 9:05pm 
Originally posted by Tiberius:
Originally posted by xSOSxHawkens:
Yes it is... Look at the comparison you gave, it only goes up to 12 threads, with diminishing returns past 8t...

Now look at the listed bench from the listed game. Notice how there is reasonable scaling with reasonable gains all the way up to about ~24 threads, with still noticable but marginal gains up to 32t? Thats a big change.... And a change for the good for both Intel and AMD users.

More for AMD users and less for Intel though, as anything multi-core aware will remove Intels Single core lead and put their chips *much* closer in games (or even perhaps behind in them) compared to intel, but that will be a wait and see type of thing.




That would be poitnless. You cant compare core scaling from AMD to Intel like that, its an apples to oranges comparison and not a valid way for one to cover core scaling. Neither would using an Intel chip with its highly volitile and large spread boosts, which is why the Intel one is locked at 4.6Ghz for the testing (Ryzen is close enough at stock from boost to base under loads to not really need this (300-400mhz varrience).

What you wanted is a comparson between the CPU's in raw game performance, not a comparison on how each one scales with core loading.


The performance jump (look at the total percentage) is even bigger on that video that what you provided on this link. :steamfacepalm:

https://mobile.twitter.com/CapFrameX/status/1283497696695857154

But then again this test is still pointless. Ppl care more if a particular processor is faster and better than the other processor
When 8 threads aren't enough, the 3700X will win. Lack of HT was a bad move.
pasa Jul 18, 2020 @ 4:38am 
Originally posted by mikey:
I recall back in the days of 2nd gen i5 4C/4T and i7 4C/8T days, I used to argue for the latter, stating that 4C/8T would be more useful as games get more and more complex (hate to bandy this phrase, but I was talking about future proofing).

Still a lot of claims like that floating around. Then when it gets to asking the very simple test: disable HT and run the same things against it being enabled, crickets. Or results showing random negligible difference in any direction.

While it is true that games pick up using more threads, the workload typically requires the proper core and realized parallelism is not increased.

Certainly those who run lot of unnecessary stuff dragging the system with or without a good reason may measure some positive diff.
Originally posted by mikey:
I recall back in the days of 2nd gen i5 4C/4T and i7 4C/8T days, I used to argue for the latter, stating that 4C/8T would be more useful as games get more and more complex (hate to bandy this phrase, but I was talking about future proofing). I agree that 4C/4T i5s were better OC'ers but no matter how good the OCs were, you can't magically get the extra threads.

Most of the time, I'd be shouted down by many who'd said that 4T was all one needed, and I simply got tired of arguing so I gave up after a while. Even now, my 4C/8T 4770K + GTX1080 (HTPC rig) does games very nicely at 4K (at mid graphics setting) and yes, I know my GPU may be holding back performance at this res, but hey, as long as gameplay is smooth, I don't care. I don't play a game and obsess over the framerate numbers displayed by MSI AB...
As someone who went with the 4 core/4 thread option during that generation, and only recently replaced it, and not even because of the CPU first and foremost... it's hard to say I didn't get my money worth.

Yes, the Core i7 handled better at some games (the Battlefields being one of the prime earlier examples I remember) but those weren't games I tended to play, and I saved money for similar (or slightly better) performance for what I did play. There was no ultimate, true, better answer for that question because it depended on a few factors. Performance doesn't exist in a vacuum or we'd be saying there's no point to AMD CPUs with 8 cores/16 threads or less. Things like cost, and whether the user needs the cores/threads, are factors to be taken into equation.

As for the comment that said scaling isn't surprising because this is a game made for a high core/thread console with low IPC... that's not entirely wrong, but it's a stretch to use that fact to suggest we won't necessarily see games do this going forward. For one, and someone pointed this out, but the game scales well up to like 32 threads which is way beyond what the PlayStation 4 has. We're late in the console life cycle, yet games scaling like this aren't all over. Twice the threads but half the clock speed AND lower IPC is precisely WHY I think my Core i5 fared rather well until now. I do, however, think this will change. Next generation consoles are not going to be using parts with vastly lower IPC and clock speed. In fact, they're basically using 8 core/16 thread Ryens. They might not be clocked AS high as PC CPUs, but they not one and a half GHz anymore either. The days of quad core CPUs (hyper-threading or not) is going to soon be well past its time, and I think you're going to want 8 core/16 thread CPUs going forward (before someone throws words in my mouth, I'm not calling 6 core/12 thread, or 8 core/non SMT CPUs, like the Ryzen 5 3600, Core i7 8700K, or Core i7 9700K, dead right now or tomorrow, but I do think they will start falling towards the lower end in the coming years and you will be better off with 8 core/16 thread or more if longevity is your goal today).
SHREDDER Jul 19, 2020 @ 2:13am 
What does it means for my RYZEN 7 1700 and 16 GB DDR4 3200MHZCL15. Will it use all the cores at 100%?
Autumn_ Jul 19, 2020 @ 4:10am 
Originally posted by SHREDDER:
What does it means for my RYZEN 7 1700 and 16 GB DDR4 3200MHZCL15. Will it use all the cores at 100%?
It means 15-20 FPS, because you're using a 970.
Mr Blonde Jul 19, 2020 @ 4:29am 
Originally posted by SHREDDER:
What does it means for my RYZEN 7 1700 and 16 GB DDR4 3200MHZCL15. Will it use all the cores at 100%?

It means it finally time to upgrade
r.linder Jul 19, 2020 @ 4:42am 
Originally posted by SHREDDER:
What does it means for my RYZEN 7 1700 and 16 GB DDR4 3200MHZCL15. Will it use all the cores at 100%?
Doesn't mean anything when your GPU is a brick at 1440p. You'll never get as much FPS as you could, and the 1700 is a pretty outdated CPU at this point anyway, because a 3600 (with 2 less cores, mind you) would get better frames because of higher core frequencies and 20% higher IPC. (Zen+ was 8% higher IPC than Zen, and Zen2 is 12% higher IPC than Zen+ so it's safe to assume that Zen2 has around 20% higher IPC than Zen which is pretty significant)
Last edited by r.linder; Jul 19, 2020 @ 4:43am
SHREDDER Jul 19, 2020 @ 5:09am 
Originally posted by Mr Blonde:
Originally posted by SHREDDER:
What does it means for my RYZEN 7 1700 and 16 GB DDR4 3200MHZCL15. Will it use all the cores at 100%?

It means it finally time to upgrade
Only graphics card. Cpu and ram are ok as we see from the benchmarks. Yes i wll get RTX 3080TI if it can run ray tracing at 1440p ultra without DLSS.
< >
Showing 16-30 of 36 comments
Per page: 1530 50

Date Posted: Jul 16, 2020 @ 7:21pm
Posts: 35