Intel's Core i9-12900K beats Ryzen 9 5950X by 39% in Ashes of the Singularity

Strawman

Posts: 362   +214
In 15 years you mean. Quad core CPUs can still maintain 60 FPS in modern titles despite being "obsolete" for 5 years now. Hell a pentium 6505 can just barely manage 60, and that's a dual core with no AVX.
Nope. The 11600k can barely (and in some cases cant) maintain 60 fps. Cyberpunk, ms flight sim, new world are games that even the 5950x struggles to maintain 60 minimum.
A 4core with no ht cpu drops to the 20s in Cyberpunk. Even the mighty 7700k overclocked to hell and beyond drops to the 30s.
 

Aryassen

Posts: 175   +213
You need to check your facts and your numbers. If anything performance has stagnated now. Remember the 2011-2015 era that the high end mainstream cpu cost 300-330 euro? The prices remained static, and the performance increase was around 45% (2600k to 6700k). Numbers taken from cinebench.

Fast forward to today, and in the same msrp (around 300) in the same 4 years the performance increase was.... 33% (1700 to 5600x). If we go with overclock numbers since the 1700 was a huge overclocker that drops down to 15-20%!! So amd gave us less of a performance increase than intel did.

Funny thing is with the same money I bought my overclocked 8700k I can buy a 5600x. The difference between the two of them is around 15% when both overclocked. Wow, after 3 years, we got 15% perfornance for the same amount of money....
Interesting, but certainly a valid point: price point vs progress. However, let me augment the picture you described slightly

1) 6700K being 45% faster than 2600K: that is mainly because Intel allowed that much higher clock speeds out of the box for the latter (while at the same time the 2600K was VERY conservatively clocked, one even might say severely held back...thanks God it was dead easy to overclock to 4.5 :) ). There were a ton of clock-for-clock comparisons through 4-6 generations from Intel, and it seemed, for a long time, that there is no real innovation from them, just higher out-of-the-box clock speeds and (eventually) more cores, keeping up the illusion of progress.

2) Intel's very successful (ring bus style) cache architecture was told to be optimised for 4 cores, so understandably, they wanted to stick with it. Without any competition, they could do just that. For 6 (!) generations, they launched 4 core 8 thread CPUs as their flagship consumer models. If you wanted to have more cores, you had to go "enterprise", with astronommical prices (think $1000+ for a 6-core Xeon (yes, the same which you can buy now off Ebay for about 15 quid) plus a $3-400 motherboard)

3) Your 8700K (cracking little CPU, btw) is most probably the result of AMD's rising competition. If they didn't launch Ryzen, maybe we still would be uising 4/8 CPUs (like the 7700K, which has the same microarchitectrue as your 8700K...or the 9900K, coming to that), as "they are the best". Also, if you have a look at those above-mentioned clock for clock, core for core comparisons, I wonder what you would make of Intel's prgress over the past few years (let's skip "ancient" history, and just start with your 8700K)

4) AMD is bringing solid IPC increase with every generation so far. I wonder if they can keep it up. Either way, the good thing is that there is SOME competition at least, vs Intel monopoly for a decade with quad core processors as the pinnacle consumer technology. And just to make clear, when times were different, and AMD Athlons ruled the world while Intel's P4 was a laughing stock, it was not good either (then it was AMD's CPU prices which started to inflate disproportionally)

For one reason or another, it seems that we have turned into an era where semiconductor products are getting more expensive (instead of getting cheaper), so it may never return that for $300 you can buy a flagship model. With pandemics and the hysteria around it, with rare-metal prices going up, with manufactoring monopolies (I'm looking at you, TSMC), with China's global (manufacturer of the world) role quiestionned, and with so many other factors, it is difficult to say or see. But I don't think it is Intel's or AMD's fault.

I for one believe that innovation is there (especially in the CPU segment, more than ever before), only (sadly) it costs more, and as a result, previously established price points are being overwritten before our very eyes.
 

Avro Arrow

Posts: 1,800   +2,155
TechSpot Elite
No one said Intel is "evil" or that AMD is "good". It was only said that it is suspicious that a relatively unplayed (though hugely anticipated back in the day) game suddenly gets an update, just before the release of Intel's new processor. The wildest "accusation" (if I can call it even that) was that it has "paid optimisation" written all over it. No one said Intel is evil...apart from...you know....you...(as a sarcastic exaggeration, I presume)
The only defence for Intel is nonsense. Not really a shock is it?
Um...apparently a 10850 and 11700 are ALSO faster than a 5950x? Something sounds fishy to me.
That's because something is.
I'm thinking 24 thread enables Intel to bump up the single thread perf a bit as opposed to if they ran 16C/32T...
I think it's more that the smaller cores produce less heat and use less juice because their power efficiency levels are crap compared to AMD's. Never forget that Intel is trying to defeat an architecture that was designed by the greatest IC designer on Earth. AMD hired Jim Keller, the creator of the Athlon 64 and father of 64-bit computing.

Intel tried hiring him but he hated it there and quit. Intel has nobody anywhere near Keller's level of talent or intelligence. If Intel had treated Keller well like AMD did, Intel would have a great architecture too but they didn't and so they don't.
 
Last edited:

Avro Arrow

Posts: 1,800   +2,155
TechSpot Elite
I never said that, but you saying amd should update the benchmark is not accurate only the vendor of the software can do that. I posted that tweet so everyone can see this was an intel sponsored benchmark. Take what you want from that info.
That's the same BS defence that Intel tried using to explain why their compiler was intentionally designed to hamstring a CPU based on its CPU ID alone.
 

Strawman

Posts: 362   +214
Interesting, but certainly a valid point: price point vs progress. However, let me augment the picture you described slightly

1) 6700K being 45% faster than 2600K: that is mainly because Intel allowed that much higher clock speeds out of the box for the latter (while at the same time the 2600K was VERY conservatively clocked, one even might say severely held back...thanks God it was dead easy to overclock to 4.5 :) ). There were a ton of clock-for-clock comparisons through 4-6 generations from Intel, and it seemed, for a long time, that there is no real innovation from them, just higher out-of-the-box clock speeds and (eventually) more cores, keeping up the illusion of progress.

2) Intel's very successful (ring bus style) cache architecture was told to be optimised for 4 cores, so understandably, they wanted to stick with it. Without any competition, they could do just that. For 6 (!) generations, they launched 4 core 8 thread CPUs as their flagship consumer models. If you wanted to have more cores, you had to go "enterprise", with astronommical prices (think $1000+ for a 6-core Xeon (yes, the same which you can buy now off Ebay for about 15 quid) plus a $3-400 motherboard)

3) Your 8700K (cracking little CPU, btw) is most probably the result of AMD's rising competition. If they didn't launch Ryzen, maybe we still would be uising 4/8 CPUs (like the 7700K, which has the same microarchitectrue as your 8700K...or the 9900K, coming to that), as "they are the best". Also, if you have a look at those above-mentioned clock for clock, core for core comparisons, I wonder what you would make of Intel's prgress over the past few years (let's skip "ancient" history, and just start with your 8700K)

4) AMD is bringing solid IPC increase with every generation so far. I wonder if they can keep it up. Either way, the good thing is that there is SOME competition at least, vs Intel monopoly for a decade with quad core processors as the pinnacle consumer technology. And just to make clear, when times were different, and AMD Athlons ruled the world while Intel's P4 was a laughing stock, it was not good either (then it was AMD's CPU prices which started to inflate disproportionally)

For one reason or another, it seems that we have turned into an era where semiconductor products are getting more expensive (instead of getting cheaper), so it may never return that for $300 you can buy a flagship model. With pandemics and the hysteria around it, with rare-metal prices going up, with manufactoring monopolies (I'm looking at you, TSMC), with China's global (manufacturer of the world) role quiestionned, and with so many other factors, it is difficult to say or see. But I don't think it is Intel's or AMD's fault.

I for one believe that innovation is there (especially in the CPU segment, more than ever before), only (sadly) it costs more, and as a result, previously established price points are being overwritten before our very eyes.
1) But the same applies to Ryzen. For example the 1700 (same price point as the 5600x) could be easily overclocked 33% up, while the 5600x is pretty much maxed out. Actually overclocking the latter lowers your single thread performance.

3) I don;t have the 8700k anymore. I "upgraded" to the 10900k, which got RMA'd about 2 months ago. Now im running an 11600k temporarily waiting for Alder lake.

Sure, there isn't much progress by Intel. Nobody argues that. The problem is when people see some "huge progress" from AMD, which is far from the truth. You are complaining about Intel's 4c/8t strategy but now AMD, keeping the same prices LOWERED the number of cores you get for the same amount of money. In fact, both the 5600x and the 5800x have an MSRP that is 50% up from their predecessor.
 

Thretosix

Posts: 103   +106
Lol I love it. Marketing bull crap at it finest. Almost well played, Intel. But you gotta cover your tracks a bit better next time.
Let's be honest here though. It's not like AMD hasn't done the same. Fanboy mentalities need to take a long walk off a short pier.
 

Aryassen

Posts: 175   +213
1Sure, there isn't much progress by Intel. Nobody argues that. The problem is when people see some "huge progress" from AMD, which is far from the truth. You are complaining about Intel's 4c/8t strategy but now AMD, keeping the same prices LOWERED the number of cores you get for the same amount of money. In fact, both the 5600x and the 5800x have an MSRP that is 50% up from their predecessor.
You are right: the 5xxx series from AMD are pricey little buggers. I personally blame the pandemics madness for that, but who knows.

What is certain, that before this latest generation, AMD gave more (in some cases, much more) cores for the same (or less) money.

I personally bought my 2700 just under Ł200 (was a refurbished one, the normal "good" price was Ł240 at the time). That's 8 cores 16 threads, for less money than the oprice of my 2600k (which was Ł250, in 2011 - and I'm just going to ignore inflation). Heck, you could still buy 7700K new at the time, costing nearly Ł300, and 8700K was well over Ł350.

Then I sold my 2700 and bought a 3600 for Ł180 (as it was better in games than even the overclocked 2700), which is 6 cores 12 threads, again, so much cheaper then anything comparable from Intel (they pushed the 8700K just below Ł300 by that time).

If you compare the IPC for 1600 - 2600 - 3600 - 5600 from AMD, the progress is very well noticeable (at that's just 4 years). I know, they have had plenty to improve upon to start with, but boy, they did. Unfortunately, prices grew as well, and now that the tides have turned, Intel has some excellent "value oriented" offerings (which they didn't have before: their 8th and 9th gen prices were kept high until it was too late, they have ignored (and sniffed at) AMD's market grab, and now they are paying the price).

The good thing is that Intel is fighting since the introduction of their 10th gen product line, and there is competition, so if AMD gets unrealistic with their prices, people will have a choice at least to go Intel (though I have faith in Ms Su, to keep things in order)
 

Darth Shiv

Posts: 2,225   +799
The only defence for Intel is nonsense. Not really a shock is it.

That's because something is.

I think it's more that the smaller cores produce less heat and use less juice because their power efficiency levels are crap compared to AMD's. Never forget that Intel is trying to defeat an architecture that was designed by the greatest IC designer on Earth. AMD hired Jim Keller, the creator of the Athlon 64 and father of 64-bit computing.

Intel tried hiring him but he hated it there and quit. Intel has nobody anywhere near Keller's level of talent or intelligence. If Intel had treated Keller well like AMD did, Intel would have a great architecture too but they didn't and so they don't.
Interesting info but can you assert the perf difference would exist if they were manufactured on the same process?

Also removing hyperthreading from half the cores if it isn't productive to have them there (and lowering the thermal envelope) is a design decision. If Intel can achieve higher scores by doing so, the results are what are important. Not purist ideology stuff about half and half cores or whatever. x86/64 is largely a hack architecture with bits tacked on all over the place but has handily beaten all comers to desktop/server to date with that philosophy.

My impression is Intel's failures recently are process related not architecture. Happy to be proven wrong.
 

Avro Arrow

Posts: 1,800   +2,155
TechSpot Elite
One thing that looks like it might not change, x86 might not be able to scale down to ARM power consumption levels and maintaining the same or better performance, so they will always be limited to desktops and servers.
I don't think that x86 will be able to scale down to ARM-levels of power consumption because x86's instruction set is much larger and more capable than ARM's. ARM was chosen for mobile devices because it's very power-efficient and mobile devices don't require the same functionality as a PC.
 

Avro Arrow

Posts: 1,800   +2,155
TechSpot Elite
Let's be honest here though. It's not like AMD hasn't done the same. Fanboy mentalities need to take a long walk off a short pier.
Speaking of fanboy mentalities, making false equivalencies like what you just did there is one of the signs of an Intel or nVidia fanboy. That's why your post has zero likes after almost twenty hours in a very active thread.
Really? Right there in your post you do the very thing you're complaining about. Making a 'whataboutism' complaint about AMD.

Like a fanboy.
You beat me to it. :laughing:
Interesting info but can you assert the perf difference would exist if they were manufactured on the same process?
I don't recall saying anything about a performance difference. Intel's performance has always been good. However, their power efficiency has been terrible as of late. Why else would they come up with this 8 "smaller cores" idea? You do make a great point, it could be the process used because that does have a direct impact on power efficiency. At the same time, there's a reason why Zen won the same award Sandy Bridge did. It's a revolutionary architecture that scales linearly with core count, is extremely power efficient and is even modular in design. The modular design makes the manufacturing process far cheaper and easier by minimising the number of defective dice and therefore maximising wafer yields.

Using smaller cores for increased power efficiency is something that Intel could have done long ago but there was no need for it like there is now. In truth, I don't really care about power efficiency at the PC level and neither does Intel. Hell, I used a Piledriver CPU and it drank juice like an alcoholic with a gin mixer. The reason that Intel is concerned about their power use is the fact that AMD is just smacking them around left and right in the server space. Xeon isn't even close to being a match for EPYC and THAT is where Intel is losing a crap-tonne of profit.
Also removing hyperthreading from half the cores if it isn't productive to have them there (and lowering the thermal envelope) is a design decision. If Intel can achieve higher scores by doing so, the results are what are important.
That's just it. NOBODY can achieve higher scores with weaker cores. Intel was clearly involved in this Ashes of the Singularity v3.1 update. Why else would it be able to use 16 cores but only 24 threads? Such a configuration has never existed before and technically, still doesn't. This update was MADE SPECIFICALLY for Intel's 16-core/24-thread CPU.

AOTS' performance scales linearly with thread count so a 32-thread CPU would most likely destroy a 24-thread CPU. Remember that when maxed out, regardless of cores, a 24-thread CPU behaves more like a dodecacore CPU than a hexadecacore CPU. This version of AOTS was designed to make it appear that Intel was beating AMD in multi-thread by limiting the AMD CPU to only 75% of its available threads. In fact, what we appear to see here is a next-gen Intel product having better IPC than a current-gen AMD product. However, that's possibly not even the case because if the scheduler is using 24 threads, the Intel CPU is defintiely using 16 physical cores and 8 SMT logical cores. The 5950X could be doing the same but it's far more likely that it's using 12 physical and 12 logical cores. Physical cores always outperform SMT logical cores by a certain margin so that could be another artificial advantage for Intel.
My impression is Intel's failures recently are process related not architecture. Happy to be proven wrong.
You could be right and we're only speculating. Like most things, it's probably a little bit of this and a little bit of that.
 
Last edited:

Thretosix

Posts: 103   +106
Speaking of fanboy mentalities, making false equivalencies like what you just did there is one of the signs of an Intel or nVidia fanboy. That's why your post has zero likes after almost twenty hours in a very active thread.

You beat me to it. :laughing:

I don't recall saying anything about a performance difference. Intel's performance has always been good. However, their power efficiency has been terrible as of late. Why else would they come up with this 8 "smaller cores" idea? You do make a great point, it could be the process used because that does have a direct impact on power efficiency. At the same time, there's a reason why Zen won the same award Sandy Bridge did. It's a revolutionary architecture that scales linearly with core count, is extremely power efficient and is even modular in design. The modular design makes the manufacturing process far cheaper and easier by minimising the number of defective dice and therefore maximising wafer yields.

Using smaller cores for increased power efficiency is something that Intel could have done long ago but there was no need for it like there is now. In truth, I don't really care about power efficiency at the PC level and neither does Intel. Hell, I used a Piledriver CPU and it drank juice like an alcoholic with a gin mixer. The reason that Intel is concerned about their power use is the fact that AMD is just smacking them around left and right in the server space. Xeon isn't even close to being a match for EPYC and THAT is where Intel is losing a crap-tonne of profit.

That's just it. NOBODY can achieve higher scores with weaker cores. Intel was clearly involved in this Ashes of the Singularity v3.1 update. Why else would it be able to use 16 cores but only 24 threads? Such a configuration has never existed before and technically, still doesn't. This update was MADE SPECIFICALLY for Intel's 16-core/24-thread CPU.

AOTS' performance scales linearly with thread count so a 32-thread CPU would most likely destroy a 24-thread CPU. Remember that when maxed out, regardless of cores, a 24-thread CPU behaves more like a dodecacore CPU than a hexadecacore CPU. This version of AOTS was designed to make it appear that Intel was beating AMD in multi-thread by limiting the AMD CPU to only 75% of its available threads. In fact, what we appear to see here is a next-gen Intel product having better IPC than a current-gen AMD product. However, that's possibly not even the case because if the scheduler is using 24 threads, the Intel CPU is defintiely using 16 physical cores and 8 SMT logical cores. The 5950X could be doing the same but it's far more likely that it's using 12 physical and 12 logical cores. Physical cores always outperform SMT logical cores by a certain margin so that could be another artificial advantage for Intel.

You could be right and we're only speculating. Like most things, it's probably a little bit of this and a little bit of that.
AMD has never mislead people... Interesting.
 

Thretosix

Posts: 103   +106
Really? Right there in your post you do the very thing you're complaining about. Making a 'whataboutism' complaint about AMD.

Like a fanboy.
So you are saying that AMD has never misled anyone...? I said they both do which is the opposite of fanboy. Keep drinking your cup of stupid.
 

HardReset

Posts: 1,265   +926
I don't think that x86 will be able to scale down to ARM-levels of power consumption because x86's instruction set is much larger and more capable than ARM's. ARM was chosen for mobile devices because it's very power-efficient and mobile devices don't require the same functionality as a PC.
Eh? Modern CPU's don't count x86, they count RISC like instructions. Power consumption wise about only real problem is logic that translates x86 instructions into RISC like instructions and other way around. That consumes quite little power however. It's just that modern x86 CPU's are mainly designed for higher power use. No real efforts for very low power x86 chips were ever made because ARM has so strong foothold. And why would any customer that want very low power CPU's want to use much less used x86 and lock themselves into AMD or Intel CPUs? Other choice is ARM that offers licences for virtually anyone.

ARM was "chosen" mostly because there were not much other choices tbh.
 

Lew Zealand

Posts: 1,949   +2,256
TechSpot Elite
So you are saying that AMD has never misled anyone...? I said they both do which is the opposite of fanboy. Keep drinking your cup of stupid.

No, when did I say that? LOL, now you're trying to put words in my mouth to target a corporation and then follow up that with a personal attack.

Let it go, man. They're Corps and Intel did a sleazy thing here, aside from whatever AMD's ever done. This is Intel's bad. You don't need to defend or "whatabout" a corporation that rakes in billions of dollars a year.
 
Last edited:

Daniele 00

Posts: 122   +96
Hopefully AMD cpu prices will fell down, if/when intel will be back on the scene. Good for me and whoever need to upgrade their 1" or 2" generation of Ryzen :p
 

Darth Shiv

Posts: 2,225   +799
That's just it. NOBODY can achieve higher scores with weaker cores. Intel was clearly involved in this Ashes of the Singularity v3.1 update. Why else would it be able to use 16 cores but only 24 threads? Such a configuration has never existed before and technically, still doesn't. This update was MADE SPECIFICALLY for Intel's 16-core/24-thread CPU.
If putting in smaller cores instead of fully featured ones allows you to have different thermals, that would affect how you can clock it which can allow you to achieve higher scores. 5950X vs 5900X... what are the base clock speeds? It's literally the premise that massive core count server CPUs are bound by - they trade off clock speed for massive parallelism. The point here is that the Intel hyperthread cores aren't terribly useful in such large numbers. Removing them doesn't cost you much but you can gain a lot by improving your clock speeds by doing so.