News Ryzen 9000 Zen 5 CPU trails Core i9-14900K in leaked benchmark — Granite Ridge 5.8 GHz CPU shows Core i9-13900K-like single-threaded performance in...

[irish_adam]

What's sad is your lying. We're looking at a 2 year release cadence for AMD. Intel has never produced a 3.5% increase over a 2 year stretch. 14900k is about 25% faster than the 2 year older 12900k.

I am guessing you are very young because otherwise you are trolling right? I mean for starters where has AMD only gained 3.5% performance over a 2 year period that you are referring to? I'm pretty sure the poster you replied to was referencing the 3.5% difference between AMD and Intel in this benchmark. Also I suggest you look at the 14nm days at intel because all we got back then was a clock speed bump over many years while they tried to fix their 10nm.

Intel is hardly on a 1 year cadence currently as 13th gen and 14th gen are the same, which puts them on a 2 year cadence too.

 

[Thunder64]

Yeah, because you set the TDP while the PPT is still high.
At every single power level (TDP) ryzen uses a lot more power (ppt) than what intel uses, except for the unlimited because intel can actually go above the top setting while the 230W limit of ryzen actually is the top limit for ryzen.
https://www.anandtech.com/show/17641/lighter-touch-cpu-power-scaling-13900k-7950x/3

How convenient of you to link a screenshot of the worst case AMD scenario. Let's try another:

At 65W AMD is beating the 13900K at 125W! Or using your graph 90W vs 143W. That happens a lot. You can try to spin the data but AMD is more efficient than Intel under load.

 

[spongiemaster]

I am guessing you are very young because otherwise you are trolling right? I mean for starters where has AMD only gained 3.5% performance over a 2 year period that you are referring to? I'm pretty sure the poster you replied to was referencing the 3.5% difference between AMD and Intel in this benchmark. Also I suggest you look at the 14nm days at intel because all we got back then was a clock speed bump over many years while they tried to fix their 10nm.

Where did I say anything about AMD? I made no claims about AMD at all. Making things up is what trolls do. Guess it takes one to know one. It could be a coincidence, but the 14900k is about 3.5% faster than the 13900k, so that's what it looked like he was talking about. Even during the 14nm fiasco, Intel was delivering more than 3.5% uplift every two years. Stop with the trolling already.

 

[TerryLaze]

That power draw is only in all core, and in nearly any realistic All-core work load (longer than a single CB23 run) Intel's chip throttle way more to the point it falls to parity if not behind AMD. AMD isn't really throttling at its Tj Max of 95C, as it'll maintain an all core of 5.1-5.4. Intel will throttle much harder on any cooler that lets it hit 100C. At "stock" settings in the same scenario the 7950X hits 250w, the 14900k will be at 286w and still typically lose to the 7950x. In tasks that justify a 16+ core CPU. I'm not sure why you're referencing the 12900K since the 7950X blows it out of the water at that same power. Of course you later went on to mention power draw at lower settings which is cute, because the 7950X is still more efficient. At 35w it's using 28% more than the setting, but completes work 53%+ faster than the 13900K at the same setting. I see you going for a masters in cherry picking, as all this data is in the article you linked. Performance per watt is still on AMD's side. It can afford to pull a few more watts if its doing so for significantly less time.

We weren't even talking about efficiency so you are the one cherry picking stuff from posts just to make your narrative work.

And as to which one will throttle more, you missed the pic in the same link showing how much hotter AMD is running at every single setting with the same cooling.
And don't forget that the intel CPU is running at ~330W max here while AMD at ~215W

How convenient of you to link a screenshot of the worst case AMD scenario. Let's try another:

It's not convenient, I was talking about power and only about power and you are showing performance, these are not even the same thing....

 

[TheHerald]

So, we are now preemptively trashing another AMD product even before launching?

Ok, fine, i’ll bite, how much faster is your beloved intel?

3.5%…..aaannnd thats it?

This is sad.

You serious or what? Intel already has cpus out on the market (for 2 years nonetheless) that are faster than both the 6core and the 8core zen 5 part in MT performance. That's craaazy, amd really needs to up the core count. They been left in the dust in the mid range segment.

 

[TheHerald]

While technically possible, 170W 8-core would be a huge departure from previous generations.
8-core 8700G or 7700 are rated at 65W.
The 7800X3D is the only one at 120W. The 7800X at 105.
170W is a ridiculous amount for 8-core and only Intel has pushed the consumption so high for low-core count CPUs, and precisely to achieve this 3.5% advantage in single core.

What's missing in this article is the power consumption indeed when running single-core bench. Intel so far has a slight advantage but burning >2X the power.

Wrong. First of all 105w tdp on amd parts means 140w actual power draw. Intel doesnt have 8 core chips at 170w tdp. Also it has nothing to do toh single core performance. Cpus don't draw 100s of watts at st workloads. They are sitting at 30-40 watts. Please, can be stop the anti intel propaganda? At least do it with facts, okay?

 

[TheHerald]

How convenient of you to link a screenshot of the worst case AMD scenario. Let's try another:

At 65W AMD is beating the 13900K at 125W! Or using your graph 90W vs 143W. That happens a lot. You can try to spin the data but AMD is more efficient than

Only problem is, you are wrong. I've tested cbr23, at 65w power draw (PPT) the 7950x scores 22k, the 13900k at 125w scores 32k. If you think a 7950x at 65w is faster than a 13900k at 125w then you have no idea what you are talking about.

 

[Thunder64]

We weren't even talking about efficiency so you are the one cherry picking stuff from posts just to make your narrative work.

And as to which one will throttle more, you missed the pic in the same link showing how much hotter AMD is running at every single setting with the same cooling.
And don't forget that the intel CPU is running at ~330W max here while AMD at ~215W

It's not convenient, I was talking about power and only about power and you are showing performance, these are not even the same thing....

I see you ignored my last post. Who cares how hot it runs if it was designed to run that way? Intel still consumes more watts which means more heat.

You serious or what? Intel already has cpus out on the market (for 2 years nonetheless) that are faster than both the 6core and the 8core zen 5 part in MT performance. That's craaazy, amd really needs to up the core count. They been left in the dust in the mid range segment.

LOL best CPU's that are even better than the unreleased Zen 5.

Only problem is, you are wrong. I've tested cbr23, at 65w power draw (PPT) the 7950x scores 22k, the 13900k at 125w scores 32k. If you think a 7950x at 65w is faster than a 13900k at 125w then you have no idea what you are talking about.

I don't care what you've tested. I care what reputable reviewers have tested. You are the one who has no idea what you are talking about.

 

[TheHerald]

I see you ignored my last post. Who cares how hot it runs if it was designed to run that way? Intel still consumes more watts which means more heat.



LOL best CPU's that are even better than the unreleased Zen 5.



I don't care what you've tested. I care what reputable reviewers have tested. You are the one who has no idea what you are talking about.

So you double down that the 7950x at 65w is faster than the 13900k at 125w. OK buddy

 

[Thunder64]

So you double down that the 7950x at 65w is faster than the 13900k at 125w. OK buddy

I mean,... that's what the data says.

 

[TheHerald]

I mean,... that's what the data says.

If that's what your data shows then your data is just flawed. Even the 12900k at 125w is faster than the 7950x at 65w. I have them all, I've actually tested them.

 

[jeremyj_83]

Even if someone doesn't want to believe the CPU-Z benchmark, when you compare AMD to AMD instead of Intel to AMD, the jump looks really good to me. Extrapolate this to the MT score and it seems like they'll increase overall metrics by a substantial margin (or I'd hope they do).

Grain of salt and all, but to me this is not a negative in the bigger picture.

Regards.

Even more impressive is when you normalize the AMD to AMD scores per GHz. When Zen 4 came out, AMD said a 13% geometric mean IPC increase over Zen 3. They included CPU-Z in that calculation and said a 1% improvement. Looking at the scores from this leak and normalized for GHz that 1% increase is true for Zen 3 (132 per GHz) and Zen 4 (134 per GHz). Zen 5 on the other hand gets 156 per GHz or a 16% increase over Zen 4. That is a pretty large gen over gen increase in a benchmark that runs purely in L1 cache.

 

[jeremyj_83]

I'm also struggling to come up with a sales pitch. Sure it's faster than Zen 3 by a couple dozen percent, but is it really worth having to also buy a new motherboard and RAM as well? Same with Zen 4, it may be 15% faster but unless you're rocking a 4080 or 4090 are you really going to notice any difference in games, if then? Similarly, if someone is rocking a pre-Pandemic system what's the argument to sway them to Socket AM5 vs AM4 or Intel LGA whatever the last generation or so was?

People who are running Zen 3 or Zen 4 aren't going to be the target of a chip like this. I'm still using my 4770k from 2013, it is showing its age, and going to be replacing it this year probably. On that system I got it at the end of the DDR3 era so I maxed out at 32GB RAM, I really need 64GB RAM to use my desktop as a virtual home lab. This is the same reason I didn't want to go Zen 3 as I didn't want to be locked on soon to be gone RAM. DDR5 now isn't much more expensive than DDR4 and you can get larger capacity DIMMs. This means that the reason to go DDR4 is quite a bit lessened for most people unless you have a VERY low budget.

 

[jeremyj_83]

I am guessing you are very young because otherwise you are trolling right? I mean for starters where has AMD only gained 3.5% performance over a 2 year period that you are referring to? I'm pretty sure the poster you replied to was referencing the 3.5% difference between AMD and Intel in this benchmark. Also I suggest you look at the 14nm days at intel because all we got back then was a clock speed bump over many years while they tried to fix their 10nm.

Intel is hardly on a 1 year cadence currently as 13th gen and 14th gen are the same, which puts them on a 2 year cadence too.

In reality 12th, 13th, and 14th Gen Intel are all the same. Raptor Cove is just Golden Cove with minor updates and higher clock speeds.

 

[blkspade]

We weren't even talking about efficiency so you are the one cherry picking stuff from posts just to make your narrative work.

And as to which one will throttle more, you missed the pic in the same link showing how much hotter AMD is running at every single setting with the same cooling.
And don't forget that the intel CPU is running at ~330W max here while AMD at ~215W


It's not convenient, I was talking about power and only about power and you are showing performance, these are not even the same thing....

I see I was wrong, you don't even understand the definition of cherry picking. If you go "look look more power", you are picking a single metric with zero context or even recognition that said power is used to do work. If the whole story is "what's happening with said power" its literally the opposite of cherry picking. What would be the actual point to talking about power without the conversation either being about cost to operate or heat being dumped into the room? Power is equated for a purpose.

 

[irish_adam]

Where did I say anything about AMD? I made no claims about AMD at all. Making things up is what trolls do. Guess it takes one to know one. It could be a coincidence, but the 14900k is about 3.5% faster than the 13900k, so that's what it looked like he was talking about. Even during the 14nm fiasco, Intel was delivering more than 3.5% uplift every two years. Stop with the trolling already.

You said:

We're looking at a 2 year release cadence for AMD. Intel has never produced a 3.5% increase over a 2 year stretch.

That to me implied that you meant AMD had produced a 3.5% over a 2 year period otherwise what were you trying to say?

If that was what he was referencing then how was he lying? you just admitted what he said was true.

Please explain how anything I said was trolling? I mean Intel really didn't add much to IPC for some years, sure they managed to add some more cores in response to AMD and they increased frequency but they did very little to actually improve their design. This is History we know this and we know why as well because all the architectural updates were tied to their 10nm process which they couldn't get working.

 

[gggplaya]

What's sad is your lying. We're looking at a 2 year release cadence for AMD. Intel has never produced a 3.5% increase over a 2 year stretch. 14900k is about 25% faster than the 2 year older 12900k.

That's a bit misleading. The 14900k has 8 more e-cores than the 12900k. When AMD moves to a 16 core ccd, are we going to consider that a 100% increase in performance?

14900k is more refined than the 12900k, so they were able to get the single core boost much higher. But if you subtract the 8 extra e-cores, and look at multi threaded performance, I bet the gains won't be that impressive. Certainly not much different than the gains AMD has had on the same zen generation.

 

[PCWarrior]

The point is that results 942 points of 14900k is with unstable bios settings. In fact with intel baseline profiles single thread performance is around 8-10% lower which means 942 - 9% (85) = 857. AMD is clear winner now

1. The 14900K is not unstable in single threaded workloads.

2. Single threaded workloads are not affected by power limits as they never draw that much power as to cause the package power to go over the power limit of 253W. Also the problem doesn't seem to be the power per se but the heat transients so strictly speaking is a cooling issue.

3. The reported instabilities are not really as widespread as the media and AMD fans made them out to be. It surely is a measurable percentage, but definitely this problem is not affecting at all, at least, 80% of 14900K owners.

4. This comparison is between AMD’s next generation chip versus Intel’s sunsetting current generation chip. And the Intel chip is not even that current as it is a rerelease of a rerelease with the first release (the 13900K) in 2022. And if you count the 12900K we are even talking about 2021.

5. Intel’s upcoming next generation chip is what is going to be the competition for this AMD chip. And Intel’s next chip is built on a newer and more efficient node and with improved and more efficient architecture. So AMD’s next chip really needs to win against Intel’s current chip if it is to have any chance of competing against Intel’s next chip.

 

[jeremyj_83]

That's a bit misleading. The 14900k has 8 more e-cores than the 12900k. When AMD moves to a 16 core ccd, are we going to consider that a 100% increase in performance?

14900k is more refined than the 12900k, so they were able to get the single core boost much higher. But if you subtract the 8 extra e-cores, and look at multi threaded performance, I bet the gains won't be that impressive. Certainly not much different than the gains AMD has had on the same zen generation.

The entire performance boost from 12900k > 13900k > 14900k has been from clock speed increases in ST and added e-cores plus clock speed increases for MT. Looking at 12900k vs 14900k in ST workloads you see that the 14900k is faster but is within the 15% higher clock speed. There aren't really any IPC increases in 12th, 13th, and 14th Gen.

 

[jeremyj_83]

This comparison is between AMD’s next generation chip versus Intel’s sunsetting current generation chip. And the Intel chip is not even that current as it is a rerelease of a rerelease with the first release (the 13900K) in 2022. And if you count the 12900K we are even talking about 2021.

CPU-Z benchmark is terrible in general and really isn't usable to compare CPUs from AMD against Intel. It is really only usable to compare AMD to AMD or Intel to Intel.

Intel’s upcoming next generation chip is what is going to be the competition for this AMD chip. And Intel’s next chip is built on a newer and more efficient node and with improved and more efficient architecture. So AMD’s next chip really needs to win against Intel’s current chip if it is to have any chance of competing against Intel’s next chip.

If Zen 5 is released before 15th Gen then Zen 5 will be compared against 14th Gen. Only after 15th Gen is released will the two be compared. That said Intel is still WAY behind in terms of efficiency. There is a reason why Intel went to a hybrid approach. They knew they couldn't put 16 p-cores on a system and clock them high enough to be competitive without drawing 500W TDP.

 

[PCWarrior]

CPU-Z benchmark is terrible in general and really isn't usable to compare CPUs from AMD against Intel. It is really only usable to compare AMD to AMD or Intel to Intel.


If Zen 5 is released before 15th Gen then Zen 5 will be compared against 14th Gen. Only after 15th Gen is released will the two be compared. That said Intel is still WAY behind in terms of efficiency. There is a reason why Intel went to a hybrid approach. They knew they couldn't put 16 p-cores on a system and clock them high enough to be competitive without drawing 500W TDP.

1. Not sure if it were the other way around whether you would be saying this. There are a lot of benchmarks favouring AMD that are not being reflective of the general performance comparison between the two brands. But when it comes to those you won’t see an AMD fan dismissing them.

2. You speak as if the whatever efficiency advantage AMD has over Intel is purely a result of a superior microarchitecture, design and the such. That’s untrue though. The whatever efficiency advantage AMD cpus have is by and large, primarily thanks to being manufactured with a smaller process node. I won’t give AMD credit for something that is not their achievement. And 3D-Vcache by the way is also a TSMC achievement (TSMC's chip-on-wafer (CoW) technology).

If you really want to compare apples to apples to see whose cpu design is superior you have to compare cpus manufactured on equivalent nodes. The closest we have is TSMC 7nm/7nm+ which is equivalent to Intel 10nmSF and 10nm ESF/Intel 7. So you have to compare Intel’s 11th gen (mobile - only) and 12th- 14th gen versus AMD Zen 2 and Zen 3. In such a comparison Intel cpus have both superior raw performance and performance per watt. Especially when the Intel chips are not pushed to their limits (in order to extract another 5-10% more performance for 50-80% more power) they are remarkably efficient . In any case for this round of cpus Intel will be using TSMC’s 3nm and their own 20A process while AMD will be using TSMC's 4nm/3nm. So the playing field is either level again or in Intel’s favour.

3. Ultimately the competing Intel offering for Zen 5 cpus will be Lunar/Arrow lake as these cpus will be what Zen 5 cpus will be competing against for most of their relevant lifetime. Also the release of Zen 4 cpus a month prior to 13th gen showed that launching 1 month earlier is not an advantage as people will just wait and see how the two true competing generations compare.

 

[jeremyj_83]

1. Not sure if it were the other way around whether you would be saying this. There are a lot of benchmarks favouring AMD that are not being reflective of the general performance comparison between the two brands. But when it comes to those you won’t see an AMD fan dismissing them.

I would still say this about the CPU-Z benchmark regardless of who is ahead. I recommend you read the Chips and Cheese article about CPU-Z benchmark that @evdjj3j mentioned earlier in the thread.

2. You speak as if the whatever efficiency advantage AMD has over Intel is purely a result of a superior microarchitecture, design and the such. That’s untrue though. The whatever efficiency advantage AMD cpus have is by and large, primarily thanks to being manufactured with a smaller process node. I won’t give AMD credit for something that is not their achievement. And 3D-Vcache by the way is also a TSMC achievement (TSMC's chip-on-wafer (CoW) technology).

When they have been on similar process nodes, Zen has been more efficient. Look at Zen 1 vs 7th Gen in 2017 both were on 14nm. The 8c/16t 1800X (95W TDP) didn't draw more power than the 4c/8t 7700k (91W TDP) until ALL cores and treads were loaded on it. At 8t the 1800X drew 60W where the 7700k drew 87W. At full load it was 99 vs 91. Remember the 1800X had DOUBLE the cores/thread as the 7700k.

In the mobile arena we have seen how much more efficient Zen is since the launch of the 4000 series CPUs. At equivalent process nodes the Ryzen CPUs are getting better battery life, having higher base clocks (since the core is more efficient), similar boost clocks, AND more performance cores. The performance per watt metric also swings WAY towards AMD. IDK where you got that it is in Intel's corner as every publication has shown how much more efficient the AMD chips are in performance per watt. On the desktop space Intel needs to clock their CPUs as high as they can and throw power draw (especially in MT workloads) out the window in order to compete in benchmarks. As was already shown in the 7950X vs 13900k efficiency article from Anandtech, we reguarly see the 7950X being faster at the same TDP as Intel. Often times it take an Intel 105W TDP to comepete with the Ryzen 65W TDP. Then Intel needs to be at full unlimited power to compete with the 105W TDP from Ryzen. That article also does a very good job of showing that after the 105W TDP Zen 4 doesn't scale well with the added power. Core on the other hand keeps scaling well with the added power, however, at the same TDP Core typically loses. Overall Core was designed at a 4c/8t uArch and needed to be kept below 4.5GHz for best efficiency. However, because of Zen Intel had to just throw out that efficiency and we see how every time they added Cores and increase clocks the power draw just sky rocketed.

3. Ultimately the competing Intel offering for Zen 5 cpus will be Lunar/Arrow lake as these cpus will be what Zen 5 cpus will be competing against for most of their relevant lifetime. Also the release of Zen 4 cpus a month prior to 13th gen showed that launching 1 month earlier is not an advantage as people will just wait and see how the two true competing generations compare.

That is true. However, again we will have to see when the next Lake versions are released. Right now Arrow Lake is expeced Q4 2024 maybe Q1 2025. If these leaks for Zen 5 are true and it is out 3-4 months before Arrow Lake that will make a big difference.

 

[PCWarrior]

When they have been on similar process nodes, Zen has been more efficient. Look at Zen 1 vs 7th Gen in 2017 both were on 14nm. The 8c/16t 1800X (95W TDP) didn't draw more power than the 4c/8t 7700k (91W TDP) until ALL cores and treads were loaded on it. At 8t the 1800X drew 60W where the 7700k drew 87W. At full load it was 99 vs 91. Remember the 1800X had DOUBLE the cores/thread as the 7700k.

In the mobile arena we have seen how much more efficient Zen is since the launch of the 4000 series CPUs. At equivalent process nodes the Ryzen CPUs are getting better battery life, having higher base clocks (since the core is more efficient), similar boost clocks, AND more performance cores. The performance per watt metric also swings WAY towards AMD. IDK where you got that it is in Intel's corner as every publication has shown how much more efficient the AMD chips are in performance per watt. On the desktop space Intel needs to clock their CPUs as high as they can and throw power draw (especially in MT workloads) out the window in order to compete in benchmarks. As was already shown in the 7950X vs 13900k efficiency article from Anandtech, we regularly see the 7950X being faster at the same TDP as Intel. Often times it take an Intel 105W TDP to comepete with the Ryzen 65W TDP. Then Intel needs to be at full unlimited power to compete with the 105W TDP from Ryzen. That article also does a very good job of showing that after the 105W TDP Zen 4 doesn't scale well with the added power. Core on the other hand keeps scaling well with the added power, however, at the same TDP Core typically loses. Overall Core was designed at a 4c/8t uArch and needed to be kept below 4.5GHz for best efficiency. However, because of Zen Intel had to just throw out that efficiency and we see how every time they added Cores and increase clocks the power draw just sky rocketed.

The only truly valid comparisons about microarchitecture efficiency are with cpus manufactured on equivalent nodes, having equal core counts and running at the same clock speeds. Comparing the 7700K versus the 1800X is not a valid comparison. Adding more cores and running them at a lower all-core frequency is going to be way more efficient in MT workloads than having say half the cores and running them at double the clock speed. So if you want to go back that far and compare 7th gen to Ryzen 1000 the comparison should be the 7700K against the 1500X and that on equal clock speeds (by downclocking the 7700K). And even that wouldn’t be fair to AMD as Intel’s 14nm is superior to GloFo/Samsung 14nm. The latter when mapped on Intel’s scale is not a 14nm but a 16nm node. The former when mapped on TSMC’s/GloFo/Samsung scale is a 10nm node.

Anyway, as I said in terms of node equivalency, the closest we have is Intel 10nmSF/10nmESF/Intel 7 and TSMC 7nm/7nm+. The Intel cpus manufactured on this type of node are Intel’s 11th gen mobile and 12th-14th gen. The AMD cpus manufactured on such nodes are the Ryzen 3000, 4000 and 5000 series cpus. So the Ryzen 7000 series you mentioned are out of the comparison as there is a node advantage for AMD. Valid comparisons would be for example the following:

1. AMD R3 3100 vs Intel i3 12100F (link here). Both at 4.1GHz, both 4C/8T, both on a 7nm class node. In gaming the Intel cpu pulls 32-42W, the AMD cpu pulls 52-55W and the Intel chip delivers 40% more frames on average and 50% more 1% lows. That is Intel having 2.3x better performance per watt.

2. AMD 5600 Vs Intel 12500 (link here). Both on 7nm class, both 6C/12T. Intel is running at 4.1GHz and AMD at 4.45GHz. Intel is winning by 3-16% while AMD consumes 30% more power. So Intel is having up to 1.5X better performance per watt.

3. AMD 4900H vs Intel i9 11900H. Both 7nm, both 8C/16T, both mobile cpus. Intel is running at higher frequency. Intel wins comfortably in raw performance but AMD edges out a win in stock performance per watt. Downclocking the 11900H to match the 4900H clockspeeds and the performance per watt win goes to Intel.

Of course the choice of core count and clock speed is ultimately a design choice. And that is why Intel pivoted to the use of a hybrid architecture with many e-cores that also run at lower speeds in order to boost MT performance more efficiently. This allowed them to remain competitive for long enough despite being a full node behind. Let's wait and see what they do when they are a node ahead.

 

[blkspade]

1. Not sure if it were the other way around whether you would be saying this. There are a lot of benchmarks favouring AMD that are not being reflective of the general performance comparison between the two brands. But when it comes to those you won’t see an AMD fan dismissing them.

Its fair to evaluate if a benchmark is telling you anything relevant that would translate in to real world performance. Consumers aren't putting any effort to evaluate what the data is actually telling them. CPU-Z just doesn't correlate to anything. I would put some value in decompression performance, since like next to everything distributed online is in compressed archives. Cinebench does tie to real a world application, but that software itself isn't relevant to a whole lot of people. I would use it and Blender as an analog for something like Rhino when suggesting a build for someone that relies on that software.

In that context I would consider Intel winning single runs of CB less than relevant because they clearly tuned TAU to be more impressive in short duration benchmarks, when in the real world those workloads go beyond the TAU period. If the 13/14900k score plummet in the 10min run to point where they essentially tie a 7950X which wins outright in Blender I would consider the AMD better in MT in real world use cases.

 

[TheHerald]

The only truly valid comparisons about microarchitecture efficiency are with cpus manufactured on equivalent nodes, having equal core counts and running at the same clock speeds. Comparing the 7700K versus the 1800X is not a valid comparison. Adding more cores and running them at a lower all-core frequency is going to be way more efficient in MT workloads than having say half the cores and running them at double the clock speed. So if you want to go back that far and compare 7th gen to Ryzen 1000 the comparison should be the 7700K against the 1500X and that on equal clock speeds (by downclocking the 7700K). And even that wouldn’t be fair to AMD as Intel’s 14nm is superior to GloFo/Samsung 14nm. The latter when mapped on Intel’s scale is not a 14nm but a 16nm node. The former when mapped on TSMC’s/GloFo/Samsung scale is a 10nm node.

Anyway, as I said in terms of node equivalency, the closest we have is Intel 10nmSF/10nmESF/Intel 7 and TSMC 7nm/7nm+. The Intel cpus manufactured on this type of node are Intel’s 11th gen mobile and 12th-14th gen. The AMD cpus manufactured on such nodes are the Ryzen 3000, 4000 and 5000 series cpus. So the Ryzen 7000 series you mentioned are out of the comparison as there is a node advantage for AMD. Valid comparisons would be for example the following:

1. AMD R3 3100 vs Intel i3 12100F (link here). Both at 4.1GHz, both 4C/8T, both on a 7nm class node. In gaming the Intel cpu pulls 32-42W, the AMD cpu pulls 52-55W and the Intel chip delivers 40% more frames on average and 50% more 1% lows. That is Intel having 2.3x better performance per watt.

2. AMD 5600 Vs Intel 12500 (link here). Both on 7nm class, both 6C/12T. Intel is running at 4.1GHz and AMD at 4.45GHz. Intel is winning by 3-16% while AMD consumes 30% more power. So Intel is having up to 1.5X better performance per watt.

3. AMD 4900H vs Intel i9 11900H. Both 7nm, both 8C/16T, both mobile cpus. Intel is running at higher frequency. Intel wins comfortably in raw performance but AMD edges out a win in stock performance per watt. Downclocking the 11900H to match the 4900H clockspeeds and the performance per watt win goes to Intel.

Of course the choice of core count and clock speed is ultimately a design choice. And that is why Intel pivoted to the use of a hybrid architecture with many e-cores that also run at lower speeds in order to boost MT performance more efficiently. This allowed them to remain competitive for long enough despite being a full node behind. Let's wait and see what they do when they are a node ahead.

Why people are still arguing this is weird to me, Intel is shown to have the most efficient desktop CPUs by far