News Core i9-12900K Destroys Ryzen 9 5950X By 38% In Ashes of the Singularity

[Makaveli]

There are no facts in his/her response... sure reads like opinions based off observations to me, but who am I to let your personal fanboyism get in the way of rational thinking. (And 39% is destorying, but pinch of salt cause lack of details, remember thats in the story you obviously read).

How about before we bring up the word fact we wait until its released and independently tested. 39% would be fantastic for the market, AMD has gotten away with a "marketing" uptick in pricing on the CPU line and for good reason. Intel needs a win to push further innovation from AMD or expect that marketing uptick to continue. This generation is starting to feel eerily similar to the netburst vs Athlon leading to core architecture battle of yesteryear.

lol uh huh.

 

[helper800]

lol uh huh.

Why would intel all of a sudden be 39% faster than AMDs offering just because of this update? If anything AMDs CPU would also get a massive increase compared to the last update seeing as it also has 24+ threads... Would be interesting to see an 5900x vs 5950x here as well because the 5900x has 24 threads and is faster single core.

 

[Makaveli]

Why would intel all of a sudden be 39% faster than AMDs offering just because of this update? If anything AMDs CPU would also get a massive increase compared to the last update seeing as it also has 24+ threads... Would be interesting to see an 5900x vs 5950x here as well because the 5900x has 24 threads and is faster single core.

The update is for 24 threads there is no plus there.

Why not update it to 32 threads which is what the 5950x supports?

 

[hotaru251]

We heard this same illogical argument when Intel brought out the Core line of CPUs. No, way could a first generation of a laptop chip ported to the desktop possibly be good... Ooops! Back when that came out it was a massive leap. I'm not assuming this time will be the same, but it's just as bad to assume it won't be similar.

difference was tech improved more generationally then. Now-a-days you expect 10-15% a generation...especailly on a 1st gen new chip architecture.

again i expect they will be good chips, but not 38% better apples to apples.

 

[helper800]

The update is for 24 threads there is no plus there.

Why not update it to 32 threads which is the 5950x supports?

You seem to have a misunderstanding of how SMT and HT works. If a program supports 24 threads and a processor has 32 or 64 threads available to it both will use 24 threads with that application. The application in question updated its usable thread count from 16 > 24 which would benefit any CPUs with 24 or more threads available to it.

 

[Makaveli]

You seem to have a misunderstanding of how SMT and HT works. If a program supports 24 threads and a processor has 32 or 64 threads available to it both will use 24 threads with that application. The application in question updated its usable thread count from 16 > 24 which would benefit any CPUs with 24 or more threads available to it.

So why does the tweet say 16 to 24 and not 16 to 24+

 

[TerryLaze]

lol uh huh.

And you know that intel will release all this info together with general coding tips for alder to the general public allowing any company or coder to reap the same kind of benefits from any program that is capable of it.
The 38% mark might be off and down to high clocks but all of the code of everything will be optimized practically from day one.

 

[Makaveli]

And you know that intel will release all this info together with general coding tips for alder to the general public allowing any company or coder to reap the same kind of benefits from any program that is capable of it.
The 38% mark might be off and down to high clocks but all of the code of everything will be optimized practically from day one.

To be honest the numbers don't mean anything to me from leaks until we see 3rd party reviews.

 

[RodroX]

OMG! I wonder if I will be able to sleep tonight after reading this...

As the world, pandemic and paid job goes, this are all things I will probably never have: core i9 12xxx...., Ryzen 59xx..... and yes the all mighty and very priciey RTX 3080 and the likes.

So yes, I guess I will be able to sleep tonight just fine.

 

[punkncat]

I think it would be fair to say that over the top cooling on a high tier GPU as well as a select title were more at play here than we should assume or believe this manner of improvement over the 5xxx series. Any regular YouTuber is fully aware that some titles play better on (a) brand than the other. We should consider the above as well as this aspect for the selection of a title that is getting long in the tooth at this point and likely shouldn't even be used for gaming benchmarks any longer.

Being a person that plays it, it's very demanding and particularly as you get into large maps with lots of bots as well as large PC armies. In my own experience it seems to like high core counts and in that respect alone it is curious that a technically lesser high power core count part would outperform the other by so much. I suspect there is some untold aspect to this improvement.

As soon as release the truth will be shown in the averages.

 

[Uniblab]

New Core i9-12900K submissions flood the Ashes of the Singularity scoreboard.

Core i9-12900K Destroys Ryzen 9 5950X By 38% In Ashes of the Singularity : Read more

Im cool with the story but wonder if the sense of suprise and being stunned wasnt a little overkill. The AMD chip is a 2020 chip the intel hasent even come out yet. Of course it should be better but real numbers will take more time. Lets wait for suprise, if it still holds up and still remember that AMDs new chip still has a way to go so we can compare both companies newest chips. Newest flagship vs newest flagship is a more realistic "stunner" situation

 

[BadOmans]

I'd be quite disappointed with Intel if their new architecture couldn't outpreform AMD's 5000 series TBH. Though it still remains to be seen if Alder Lake will be competitive with the next gen AMD CPUs. Either way it just shows that real competition is a good thing.

 

[mdd1963]

AOS framerates are indeed interesting...!

Looking forward to BF1/BF5 comparisons, which do both still scale with both GPU and CPU performance increases....

 

[mdd1963]

I'd be quite disappointed with Intel if their new architecture couldn't outpreform AMD's 5000 series TBH. Though it still remains to be seen if Alder Lake will be competitive with the next gen AMD CPUs. Either way it just shows that real competition is a good thing.

What is the soonest in 2022 we might see some AM5 CPUs for sale? Q3 of 2022?

I hope AMD does leap again with AM5...; all these gains lead to better choices for all out shopping!

 

[cryoburner]

You seem to have a misunderstanding of how SMT and HT works. If a program supports 24 threads and a processor has 32 or 64 threads available to it both will use 24 threads with that application. The application in question updated its usable thread count from 16 > 24 which would benefit any CPUs with 24 or more threads available to it.

What they were actually pointing out is that increasing the core limit to 16 cores, but arbitrarily increasing the thread limit to just 24 threads, rather than the logical move to 32, is allowing the i9 to operate with all hardware threads utilized, while the Ryzen 9 only has 75% of its threads utilized. It's like the new update to the benchmark was tailor-made to align with the exact specifications of the 12900K, to make it look better against hardware featuring additional threads.

So, the benchmark is treating the 5950X a bit more like it were a 5900X. Which wouldn't matter for most games, but since Ashes is one of the very rare games where performance can scale significantly with the number of cores, it gives an inaccurate impression that the 12900K might perform significantly better than the 5950X at heavily-multithreaded workloads, when in reality, they will likely perform a lot closer to one another, assuming these results are even representative of what performance will be like.

The Ashes benchmark is generally used to compare heavily-multithreaded performance of CPUs, but it's not particularly useful for comparing processors featuring more threads than it supports. So, the 5950X probably shouldn't even be in this comparison, as the benchmark is not designed to fully utilize its hardware. The fact that the developers appear to have updated the benchmark specifically to fit the specifications of the 12900K while ignoring the capabilities of competing processors like the 5950X also brings into question just what other Intel-specific optimizations might have been put in place, perhaps at the request of Intel. Ashes never really was a particularly popular game, and hasn't really had much of a player-base for years, but it still sticks around as a benchmark. And since it's not really representative of performance in almost any other game, it can be thought of more as a synthetic benchmark than anything. But a benchmark that specifically tailors itself to give an advantage to one line of hardware can arguably be considered a bad benchmark.

 

[Milky_101]

What they were actually pointing out is that increasing the core limit to 16 cores, but arbitrarily increasing the thread limit to just 24 threads, rather than the logical move to 32, is allowing the i9 to operate with all hardware threads utilized, while the Ryzen 9 only has 75% of its threads utilized. It's like the new update to the benchmark was tailor-made to align with the exact specifications of the 12900K, to make it look better against hardware featuring additional threads.

So, the benchmark is treating the 5950X a bit more like it were a 5900X. Which wouldn't matter for most games, but since Ashes is one of the very rare games where performance can scale significantly with the number of cores, it gives an inaccurate impression that the 12900K might perform significantly better than the 5950X at heavily-multithreaded workloads, when in reality, they will likely perform a lot closer to one another, assuming these results are even representative of what performance will be like.

The Ashes benchmark is generally used to compare heavily-multithreaded performance of CPUs, but it's not particularly useful for comparing processors featuring more threads than it supports. So, the 5950X probably shouldn't even be in this comparison, as the benchmark is not designed to fully utilize its hardware. The fact that the developers appear to have updated the benchmark specifically to fit the specifications of the 12900K while ignoring the capabilities of competing processors like the 5950X also brings into question just what other Intel-specific optimizations might have been put in place, perhaps at the request of Intel. Ashes never really was a particularly popular game, and hasn't really had much of a player-base for years, but it still sticks around as a benchmark. And since it's not really representative of performance in almost any other game, it can be thought of more as a synthetic benchmark than anything. But a benchmark that specifically tailors itself to give an advantage to one line of hardware can arguably be considered a bad benchmark.

Erm? I thought Ashes is developed by Stardock which associated with AMD?

 

[JWNoctis]

Don't forget about the memory bandwidth difference as well - This benchmark is just as much about DDR5 vs. DDR4, and all 16 cores of that 5950X is still trapped behind that same old dual-channel DDR4 interface.

Kind of less than useful to argue about processor architectural virtues without taking that out of the equation, even though it may still take a while before any Socket AM5 part comes out.

EDIT: And that any workload above 8 threads on the same data may well face some rather severe latency penalty due to non-unified L3 cache - i.e. chiplets - on the 5950X. Though I do wonder whether that would actually apply to this specific benchmark.

 

[VforV]

Now put Alder Lake on Win10 with DDR4 and see how much of that extra performance gain is lost and how much is the actual CPU advancement in performance.

People are ignoring this time around that Win11 and DDR5 both bring extra performance to intel CPUs (Alder Lake in this case), so it's not only the CPU merit here, so to say...

And then comes Zen3+, 2 months later...

 

[VforV]

So why does the tweet say 16 to 24 and not 16 to 24+

The update is clear to support Alder Lake better, but not necessarily be better for Zen3, just like Win11 is especially better for Alder Lake too.

Also let's be real, as well as the shamefully clickbait tittle, they put the 5950x instead of the 5900x which would be the same in performance in this game, only because Alder Lake beating the 5950x sounds much grander for intel, when in fact that game does not use the 5950x to it's full potential, so actually the 5900x would be 12900k main competitor in this game.

You could limit a game to 4c/8t and say that intel's best beat's 5950x, does that make sense? I guess it does for some "kind" of individuals...

Pfft, journalism at it's best. /s Meh

 

[kaalus]

5950X maxes out at 140W. How much does 12900 use, 250W? Small wonder it will be faster.

 

[dalek1234]

I call BS. I'll believe it when I see it, and I'm betting that there won't be much to see. Intel fudges benchmarks and lies about IPC gains as a matter of course. Why should this time be any different. History will repeat itself again, I predict.

What Intel is really doing is saying: "Don't buy AMD, we have great CPUs just around the corner. Wait for them". Then the CPUs get delayed in being released, and when actually relseased, the performance claim will turn out to be another Intel lie.

Yes, I have zero facts to back my predictions up, but we have tone of Intel BSing history from the past to draw conclusions from. I'm waging my left testie that this is just another Intel smoke-and-mirrors, a smoke-and-mirrors that will require your personal power plant to fuel that CPU at max workload.

 

[sylex1]

New Core i9-12900K submissions flood the Ashes of the Singularity scoreboard.

Core i9-12900K Destroys Ryzen 9 5950X By 38% In Ashes of the Singularity : Read more

Well-well, there is a light at the end of the tunnel?

 

[larkspur]

The fact that the developers appear to have updated the benchmark specifically to fit the specifications of the 12900K while ignoring the capabilities of competing processors like the 5950X also brings into question just what other Intel-specific optimizations might have been put in place, perhaps at the request of Intel.

Stardock and AMD have a partnership/affiliation which includes optimizing their games for Ryzen. Here's what you guys are missing - Why didn't Stardock optimize Ashes for 128 threads? I mean, right now I can buy a 64c/128t Threadripper. Ok, so why didn't Stardock optimize for 32 threads? Because no one (who doesn't want to waste money) is going to buy a Threadripper OR a 5950X for gaming! These are not gaming CPUs! Sure if you actually have a use-case for 16c/32t then great get a 5950X. It'll game great and render great. But don't expect any REAL game to fully utilize it. Sheesh!

Anyway - back to the article - clearly some interesting shenanigans going on here. Most likely we're looking at some extreme OCing. But it is interesting to wonder just how much better Alder Lake will be on Win 11 vs Ryzen 3. Anyone with a brain will reserve judgement until proper reviews are released.

 

[cryoburner]

Stardock and AMD have a partnership/affiliation which includes optimizing their games for Ryzen. Here's what you guys are missing - Why didn't Stardock optimize Ashes for 128 threads? I mean, right now I can buy a 64c/128t Threadripper. Ok, so why didn't Stardock optimize for 32 threads? Because no one (who doesn't want to waste money) is going to buy a Threadripper OR a 5950X for gaming! These are not gaming CPUs! Sure if you actually have a use-case for 16c/32t then great get a 5950X. It'll game great and render great. But don't expect any REAL game to fully utilize it. Sheesh!

My point was more that it makes for a poor comparison either way. A processor that only has 75% of its threads utilized compared against another that has 100% of its threads utilized does not provide a meaningful example of multithreaded performance. And since not many people are currently playing Ashes, and almost no other games are that multithreaded, it's not a relevant example of performance in games either. If the benchmark is not representative of typical game performance, and not representative of multithreaded performance on high-core-count processors, then these results are not particularly useful, and certainly don't justify the clickbait title used for this article.

As for the "partnership", it sounds like AMD worked with them to help implement FreeSync 2 and the Vulcan Graphics API along with some Ryzen optimizations for Star Control Origins a few years back. And apparently they worked with them to add DX12 benchmarking to Ashes well over five years ago, prior to the game coming out and more than a year before Ryzen launched, back when FX processors were still at the top of AMD's lineup. I don't really see much recent about them working with AMD. Intel also published an article about how their processors are great at running Ashes a few years back, so it would not be surprising if they have been working with them as well...
https://software.intel.com/content/...ing-matters-for-ashes-of-the-singularity.html

At the very least, randomly updating a more than five year old game that has mainly just been used for benchmarking in recent years to support up to the exact core and thread count of a new processor launching in a couple months seems rather suspicious. Why utilize up to 16 cores and 24 threads specifically when any processors in this range on the market today feature either 12 cores with 24 threads or 16 cores with 32 threads? From a multithreaded programming standpoint, supporting that asymmetric core count doesn't make much sense unless they were optimizing specifically for the as-yet-unreleased 12900K, the only processor using that configuration. And since Ashes is kind of a niche game at this point, and relatively few people with a 12900K-based gaming system will likely be playing it, the primary reason for that update seems like a way to make the processor look better in benchmarks, something that makes the most sense if Intel is sponsoring those optimizations.

 

[spongiemaster]

My point was more that it makes for a poor comparison either way. A processor that only has 75% of its threads utilized compared against another that has 100% of its threads utilized does not provide a meaningful example of multithreaded performance. And since not many people are currently playing Ashes, and almost no other games are that multithreaded, it's not a relevant example of performance in games either. If the benchmark is not representative of typical game performance, and not representative of multithreaded performance on high-core-count processors, then these results are not particularly useful, and certainly don't justify the clickbait title used for this article.

Agreed. This isn't really a useful comparison in any scenario. It's not representative of any other game, and even if it was an accurate representation of multi threaded performance, it would still be useless based on how rarely users need 24-32 threads

The leaked cinebench, however, give plenty of reasons for optimism.

Leaked Alder Lake Benchmark Shows Up to 21% Gain Over AMD | Digital Trends

Alder Lake benchmarks continue leaking, and the latest result shows a big boost for Team Blue over AMD's top Ryzen 9 5950X.

www.digitaltrends.com

No, the 12900k isn't beating the 5950x by 38%, but it is still winning which is pretty significant. For those saying it barely beats a year old CPU, you're missing the forest for the trees. The over 2000 single threaded cinebench score is what really matters, which is way ahead of 5000 series CPU's. 24 threads beating out the current gen 32 thread leader is a real accomplishment. Statistically zero percent of you are using a 5950x ($750 right now), and only slightly more are using a 5900x ($600+ right now). Whatever you are currently using outside of those 2, Alder Lake is going to crush. When utilizing 16 threads or less, which is overwhelmingly where most mainstream software is today, nothing is going touch Alder Lake. The winner here looks like the 12700k. The loss of 4 "efficiency" cores vs the 12900k means you're going to lose some in those rare highly threaded work loads, but lose practically nothing for everything else. If the rumored $430 is true, the 12700k is going to be a monster mainstream CPU that is going soundly beat 5000 series CPU's and crush everything else.