• 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’s surprise Ryzen killer

Dave2150 said:
Last few Intel CPU releases had infinitely more stock than the Ryzen 5000 series - this is fact and well known. Are you aware you literally can't find a single Ryzen 5000 CPU in stock anywhere right? Intel's last few launches were nothing like this, as they have their own fabs.

I'll be bowing out at this point, no point arguing when you struggle to see that Ryzen 5000 are completely out of stock everywhere, weeks after release.
Click to expand...

Intel can't even compete with Apple how's it gonna compete with Zen 4 lmao

https://www.reddit.com/r/hardware/c..._tiger_lake_cpu_holds_well_against_apples_m1/
 
Cenuij said:
Comedian, as I said earlier Ice Lake/Sunny Cove had problems clocking past 3.9Ghz on 10nm, it's dead on arrival on 14nm.

There will be a handfull of golden engineering samples sent out with payments for infomercial articles to show that it can "compete". Then AMD will release parts on 7nm+, then 5nm.

I take no pleasure in that, I'm on my first AMD system since Barton. I want Intel to compete better.
Click to expand...
IMO Alder Lake is the one to look out for at least, Rocket Lake may just end up being the same or slightly better per core than Zen 3 but with 8 cores max.

If Rocket Lake has been brought forward to February then maybe Alder Lake will come later next year as the roadmap shows.
Keeping hold of my 3700x for now, it's basically still the same price as when i bought it :)
 
Grim5 said:
Intel's ryzen killer is a meh burger

Apisek leaks out the 11900k Rocket lake-s top of the line cpu. Boosts to 5ghz, 8 cores and matches the 5800x in single core scores, loses in multi thread, can't touch the 5900x or 5950x

https://wccftech.com/intel-rocket-l...marks-leak-out-on-par-with-amd-ryzen-7-5800x/
Click to expand...

Thanks for this. I now feel better sitting in the overclockers queue for a 5900x. Not to mention the Ryzen chips on 7nm will draw a lot less power than intel.

By the way I just noticed your build specs - soon to be running a Strix RTX 3090 on a 750W PSU? I'm sure I've seen graphs showing they can spike to over 450W so that's gonna be tight!
 
Johno said:
...soon to be running a Strix RTX 3090 on a 750W PSU? I'm sure I've seen graphs showing they can spike to over 450W so that's gonna be tight!
Click to expand...
Given the 5950X he's also getting doesn't draw 300W I'd say it wasn't an issue. Plus the 450W spikes on the 3090 are just that, spikes. A decent PSU should be able to handle a quick spike of load that goes outside operating specs.

At 100% sustained load, I'd be surprised if the full system was over 600W at the wall.
 
Grim5 said:
Intel's ryzen killer is a meh burger

Apisek leaks out the 11900k Rocket lake-s top of the line cpu. Boosts to 5ghz, 8 cores and matches the 5800x in single core scores, loses in multi thread, can't touch the 5900x or 5950x

https://wccftech.com/intel-rocket-l...marks-leak-out-on-par-with-amd-ryzen-7-5800x/
Click to expand...

I don't get it. Why would they come out with previous cpus with extra cores to then take two away making it worse. 10900K is slower with single but gains with multi. 11900K gains with single but loses with multi.
 
Thermals, latency and process node

the 8 core 11900k has the same 125w pl1 tdp as the 10 core 10900k. Basically there isn't any power budget left for more cores.

The reason tdp is so high is partly 14nm and partly cause Rocket lake is a ring bus design and ring bus has terrible power draw scaling when adding more cores. A mesh bus design would allow for more cores at lower power draw but increase memory latency and affect gaming performance
 
V F said:
I don't get it. Why would they come out with previous cpus with extra cores to then take two away making it worse. 10900K is slower with single but gains with multi. 11900K gains with single but loses with multi.
Click to expand...
Newer core designs are faster in part because there's more transistors to provide better optimisations in for the pipelines, branch prediction etc. As the node process shrinks there's now the space & available power envelope to add this all in

Since this is a backport of a 10nm process onto 14nm it means there's less space for cores, because the cores have more technology in them. Therefore to keep the TDP within parameters Intel needs to reduce the core count.
 
Last edited:
~>Dg<~ said:
why oh why do people use this benchmark oh....i remember its a amd based benchmark lol.
Click to expand...
When I installed Modern Warfare into Windows with my new 5950x the game config now has "render core count" set to 16, the future of game code design is more concurrency, more parallelism, not less.

Here's an article from, *checks notes*, Intel explaining why higher core count is better for Ashes of the Singularity: https://software.intel.com/content/...ing-matters-for-ashes-of-the-singularity.html

So, errrm, LOL?
 
Cenuij said:
When I installed Modern Warfare into Windows with my new 5950x the game config now has "render core count" set to 16, the future of game code design is more concurrency, more parallelism, not less.

Here's an article from, *checks notes*, Intel explaining why higher core count is better for Ashes of the Singularity: https://software.intel.com/content/...ing-matters-for-ashes-of-the-singularity.html

So, errrm, LOL?
Click to expand...

AMD worked closely with them since it was made. thats why its always in AMD cpu bench marks lol.

the funny thing its always used with AMD as it favours AMD and yet guess how many play it ? a big 60 people in the world lol.
 
Cenuij said:
When I installed Modern Warfare into Windows with my new 5950x the game config now has "render core count" set to 16, the future of game code design is more concurrency, more parallelism, not less.

Here's an article from, *checks notes*, Intel explaining why higher core count is better for Ashes of the Singularity: https://software.intel.com/content/...ing-matters-for-ashes-of-the-singularity.html

So, errrm, LOL?
Click to expand...

Its not especially good calling tho, in Intel's example there the 10 core CPU is less than 5% faster than the 8 core, for 25% more cores you get under 5% more performance.

PS: i hate Ashes on the benchmark, it in no way represents anything at all, i also don't know why people keep using it. It needs to go away so people use something that actually tells us something.
 
humbug said:
Its not especially good calling tho, in Intel's example there the 10 core CPU is less than 5% faster than the 8 core, for 25% more cores you get under 5% more performance.

PS: i hate Ashes on the benchmark, it in no way represents anything at all, i also don't know why people keep using it. It needs to go away so people use something that actually tells us something.
Click to expand...
It tells us that if you write a game to take advantage of higher core counts, it will provide the user with a better experience if they buy a cpu with more cores - this works for both Intel and AMD, more cores == more revenue

It is of course a non sequitor that all games are this way or will be written this way. It's taken so long to have code maturity in games to take advantage of more cores for lots of reasons. One of them being that top tier game code having been written largely in C++ is quite hard to develop in terms of parallelism and/or concurrency. C++ is not a language that was designed with this in mind, this makes it hard for devs to reason about highly concurrent or parallel code. Same goes for the graphics API's of course.

The situation is changing though, code is like music and art - everything is built on the code that came before. Graphics API's are maturing to take advantage of higher core counts, C++ and newer high perf langauges (such as Rust) have better support and more mature libraries to assist developers in concurrent and highly threaded programming. So as I said, the direction is one way - from single core design to multi core design - because that's where the performance is.
 
You must log in or register to reply here.
Back
Top Bottom