• 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.

Intel Core i9-12900K allegedly scores 30K points in Cinebench R23

Caporegime
Joined
17 Mar 2012
Posts
47,382
Location
ARC-L1, Stanton System
Don't be daft. You words were "Single core performance is NOT important" which is obviously and demonstrably wrong.

As it stands single core performance is key for gaming, that may change in the future, maybe, but it hasn't really done so in the 15 years since multiform CPUs come out and won't be in the immediate future.

We're talking about CPUs that will be released imminently or in the near future not in far off fantasy land where games are beautifully multithreaded and money is free and unicorns prance across the plains.

It is one of many factors, it is not "Key"

Rocket Lake


Your Welcome :)
 
Caporegime
Joined
8 Jan 2004
Posts
31,994
Location
Rutland
It is one of many factors, it is not "Key"

Rocket Lake

No single threaded performance (I shouldn't call it single core - no one wants a single core CPU) is absolutely key still even in this day. Very few engines will perform well if single core performance is garbage. Games need good SC performance, low latency and "enough" threads.

Rocketlake is a garbage CPU with strong SC performance hampered by poor cache latency.
 
Last edited:
Caporegime
Joined
17 Mar 2012
Posts
47,382
Location
ARC-L1, Stanton System
No it absolutely is key still even in this day. Very few engines will perform well if single core performance is garbage.

Rocketlake is a garbage CPU with strong SC performance hampered by poor cache latency.

Using that measure would suggest that core to core latency is more important than IPC and clock speed in games where single core performance matters.

tkxJJoE.png
GSVcwoc

GSVcwoc.png
 
Soldato
Joined
6 Feb 2019
Posts
17,466
Perhaps Intel will insure it does....

Apparently ADL review NDA lifts on the 27'th, AMD said Microsofts performance issues on Win 11 will also be fixed in this month, i think it will be fixed before reviews go out.

it's already fixed in the latest beta version of w11
 
Soldato
Joined
6 Aug 2009
Posts
7,070
Windows 11 has something else to say about that and will make your 3700x 15% slower. :eek:

Lol, I saw that. No plans to upgrade to W11 for some time, hoping it will get patched. Still on F22 BIOS so no compatibility for the 5XXX series CPUs yet let alone W11. I'll update it when I eventually switch to W11.
 
Caporegime
Joined
17 Feb 2006
Posts
29,263
Location
Cornwall
You are wrong. What you describe is poor software optimisations, lazy game developers but things will improve over time and yet again you will be proven wrong.
More cores is always better, no matter what arguments you bring in the discussion..
Those phrases roll off the tongue but what do they actually mean... "poor optimisation," "lazy devs," etc. They could mean anything, it's just a baseless accusation a lot of the time.

From my admittedly very rough understanding, there is a cost to making things parallel. And it's entirely possible that for a certain algorithm or game function, the cost of (the overheads) of parallel processing actually outweigh the benefits.

Collision detection is one I always think about. You can update the position of every game object on its own thread (for the lulz), but the cost of synchronising all the data to do collision detection would probably (maybe) be higher than processing all those moving objects ST-style in sequence.
 
Permabanned
Joined
2 Sep 2017
Posts
10,490
Those phrases roll off the tongue but what do they actually mean... "poor optimisation," "lazy devs," etc. They could mean anything, it's just a baseless accusation a lot of the time.

From my admittedly very rough understanding, there is a cost to making things parallel. And it's entirely possible that for a certain algorithm or game function, the cost of (the overheads) of parallel processing actually outweigh the benefits.

Collision detection is one I always think about. You can update the position of every game object on its own thread (for the lulz), but the cost of synchronising all the data to do collision detection would probably (maybe) be higher than processing all those moving objects ST-style in sequence.

I think the best would be if the game developers begin to do this:
- optimise for 12-core CPU for top ultra-high settings;
- optimise for 8-core for high settings;
- optimise for 6-core for medium settings;
- optimise for 4-core for low-medium settings;

Hopefully you understand how it can work.
 
Back
Top Bottom