News AMD vs Intel 2020: Who Makes the Best CPUs?

[TheBeastFromOz]

Try this link: https://www.amazon.com/AMD-Processor-Wraith-Stealth-Cooler/dp/B07XTQZJ28

Ha, when I go to that listing it gives me an Australian $ price of $215! We really do get ripped off here...

 

[almstsobur]

Can we get some actual figures in actual benchmarks? Numbers we can compare. I'll put my 5900X up against any Intel CPU. It takes weeks of tweaking and testing to get the most out of this CPU, not just "Auto PBO". Sorry, I have very little confidence that you are using the 5900X to it's potential and I would like to run some benchmarks with mine and compare real capabilities. Not a pissing contest, real curiosity.

 

[almstsobur]

All of the online benchmarks on high end CPU kind off piss me off like this. I know you want to try to compare apples to apples, but compare the best apple to the best apple. If I'm entering an apple taste test contest, I'm not going to randomly grab a handful of apples, I'm going to look them over and take the best apples.

My point being; when you are talking the top echelon of gaming processors, you are usually talking about power users, people who will spend the time to test, read, study and setup every parameter to be the best possible. How high is the Infinity Fabric? What are all the memory timings, are they 2:1? Have you set a negative voltage offset and manual PBO? Have you raised the stupid low EDC? Have you defined the clocking parameters per core? People who spend $500+ on a CPU usually do these types of things and provide above average cooling to the CPU.

Do the same with the Intel CPU's. Show me your best, I'll show you mine. Out of the box for either CPU is doing it a disservice at this price point. It's like buying a brand new 4K TV, top of the line and leaving it on "Vivid" No one does that. No one spends the time to study and buy the best and doesn't then spend the time to get the best out of it.

 

[M42]

All of the online benchmarks on high end CPU kind off piss me off like this. I know you want to try to compare apples to apples, but compare the best apple to the best apple. If I'm entering an apple taste test contest, I'm not going to randomly grab a handful of apples, I'm going to look them over and take the best apples.

My point being; when you are talking the top echelon of gaming processors, you are usually talking about power users, people who will spend the time to test, read, study and setup every parameter to be the best possible. How high is the Infinity Fabric? What are all the memory timings, are they 2:1? Have you set a negative voltage offset and manual PBO? Have you raised the stupid low EDC? Have you defined the clocking parameters per core? People who spend $500+ on a CPU usually do these types of things and provide above average cooling to the CPU.

Do the same with the Intel CPU's. Show me your best, I'll show you mine. Out of the box for either CPU is doing it a disservice at this price point. It's like buying a brand new 4K TV, top of the line and leaving it on "Vivid" No one does that. No one spends the time to study and buy the best and doesn't then spend the time to get the best out of it.

Download PassMark's PerformanceTest benchmarking application and measure your total desktop performance against this list:
https://www.pcbenchmarks.net/fastest-desktop.html
(and please post your passmark score.)

 

[almstsobur]

Download PassMark's PerformanceTest benchmarking application and measure your total desktop performance against this list:
https://www.pcbenchmarks.net/fastest-desktop.html
(and please post your passmark score.)

I will do that, is this CPU performance only though? I unfortunately have not been able to win the GPU lottery in quite a while, my 2070 Super will obviously be a massive bottleneck in overall performance.

 

[M42]

I will do that, is this CPU performance only though? I unfortunately have not been able to win the GPU lottery in quite a while, my 2070 Super will obviously be a massive bottleneck in overall performance.

Performance is best measured as a whole system. You need to have a balance between CPU, memory, disk, and graphics. Since you mentioned "power users" in your response, it sounded like you think of yourself as a power user and do whatever it takes to get the parts to try to build/tweek the fastest system. If you are not a power user, then don't try doing the full performance test because you will likely be disappointed in your score.

 

[almstsobur]

Performance is best measured as a whole system. You need to have a balance between CPU, memory, disk, and graphics. Since you mentioned "power users" in your response, it sounded like you think of yourself as a power user and do whatever it takes to get the parts to try to build/tweek the fastest system. If you are not a power user, then don't try doing the full performance test because you will likely be disappointed in your score.

My bad, I must have posted in the wrong thread, I was certain this was AMD vs Intel CPU's. I didn't recall GPU's being part of the conversation. I said I would like to match my 5900X up against any current Intel CPU.

I will however put my 2070 Super up against any other 2070 Super. It's flashed with the highest performing BIOS for the card and running on a custom water loop. I consider a "power user" someone who goes well beyond the norm to get the absolute most out of a component. Spending $2000.00 on a GPU is another category I call "power moron"

 

[M42]

My bad, I must have posted in the wrong thread, I was certain this was AMD vs Intel CPU's. I didn't recall GPU's being part of the conversation. I said I would like to match my 5900X up against any current Intel CPU.

I will however put my 2070 Super up against any other 2070 Super. It's flashed with the highest performing BIOS for the card and running on a custom water loop. I consider a "power user" someone who goes well beyond the norm to get the absolute most out of a component. Spending $2000.00 on a GPU is another category I call "power moron"

Lack of a graphics card is a poor excuse since you did say gaming CPU here:

My point being; when you are talking the top echelon of gaming processors, you are usually talking about power users, people who will spend the time to test, read, study and setup every parameter to be the best possible.

Without a top gaming graphics card, a top gaming CPU cannot really perform its best at gaming.

Since most games are single-threaded, what's the single-threaded performance score for your system on the PerformanceTest? Here are the average scores of top-performing CPUs in single-threaded performance:

https://www.cpubenchmark.net/singleThread.html

 

[almstsobur]

Lack of a graphics card is a poor excuse since you did say gaming CPU here:

Without a top gaming graphics card, a top gaming CPU cannot really perform its best at gaming.

Since most games are single-threaded, what's the single-threaded performance score for your system on the PerformanceTest? Here are the average scores of top-performing CPUs in single-threaded performance:

https://www.cpubenchmark.net/singleThread.html

What we should do is get together in person for some benchmarks; I would like to share whatever it is you are smoking. No game designed in the last decade is single threaded. Most current games are optimized around ~8 cores to make cross-platform between current gen consoles and PC's a less complicated task. The term "gaming CPU" in the context I was using it was meant as a generic generalization for the top, non-pro workstation or server class CPU. Do what you want with your money, I personally draw the line at competing with profitable businesses for gaming hardware, it is not a fiscally responsible position to take. It is in the best interest of miners to invest in expensive GPU's, right now it pays off. I make zero profit gaming, they win for the time being.

Link to my CPU mark:

Screenshot 2021-11-29 145104.png - Google Drive

 

[M42]

What we should do is get together in person for some benchmarks; I would like to share whatever it is you are smoking. No game designed in the last decade is single threaded. Most current games are optimized around ~8 cores to make cross-platform between current gen consoles and PC's a less complicated task. The term "gaming CPU" in the context I was using it was meant as a generic generalization for the top, non-pro workstation or server class CPU. Do what you want with your money, I personally draw the line at competing with profitable businesses for gaming hardware, it is not a fiscally responsible position to take. It is in the best interest of miners to invest in expensive GPU's, right now it pays off. I make zero profit gaming, they win for the time being.

Link to my CPU mark:

Screenshot 2021-11-29 145104.png - Google Drive

Just as I thought. Even my 11900K beat your Single-Thread score with a 3816. More cores will obviously be able to do more work, but most applications benefit from higher single-threaded performance, including games.

 

[larkspur]

Most current games are optimized around ~8 cores to make cross-platform between current gen consoles and PC's a less complicated task.

There aren't any modern single core CPUs and dual-cores aren't considered viable anymore for modern gaming. 8 threads are not 8 cores. The reason a Threadripper underperforms in games vs a regular CPU is because the Threadripper has less single thread performance. Take your 5900x for example. Your ST is 3613 while the Threadripper PRO 3995WX has an ST score of 2638. The TR is the vastly superior performing chip in anything truly MT. But games simply are very bad at doing MT. They can't take advantage of all those threads. One particular process on one particular thread can bottleneck the entire performance - hence the need for high single thread performance in games.

That being said - If you are running 1440p or higher the difference between any 6+ core Zen 2/3 or CL/RL/AL CPU isn't going to be significant when running a 2070 Super. Even if you were running a RTX 3090 at 4k the difference between all those CPUs would also be negligible. Sure there's a little difference, but it is typically minimal and likely not noticeable. So when it comes to the absolute best gaming CPUs it really doesn't matter unless you are using something less than 6c/12t or something really old. Or if you are running a top-end RTX 3090 at 1080p with a high refresh rate monitor - that's the only time there is a significant difference in high-end gaming CPUs.

Since you want to discuss your 5900x - its got a great ST score like any Zen 3 CPU. And your overall scores are higher than most - looks like you've tuned it very well! But the 12700k does beat it in games. At least all the games I've seen it benchmarked. But like I said earlier - doesn't really matter unless you're at 1080p with a RTX 3090.

 

[almstsobur]

Just as I thought. Even my 11900K beat your Single-Thread score with a 3816. More cores will obviously be able to do more work, but most applications benefit from higher single-core performance, including games.

And if these were 1994, that thought would be useful.

 

[almstsobur]

I don't know where "Games" came into play so much here. I was annoyed at the otiginal article because it gave no actual performance figures for actual benchmarks; at least that I could find. It just basically said "full benchmark suite" Intel 25 points better. Well, I call <Mod Edit>. If there is nothing to hide (Hand selected, single core performance heavy, short run benchmarks) then why not post the actual benchmarks with the actual scores? That's what I was getting at. It looks to me like a paid advertisement by Intel.

Give me a Cinebench score for instance, a target I can take a comparative look at. I have a feeling Alder will throttle heavily, has less performance cores and thus will get its ass handed too it. Pi calculation? Anything another person can replicate and compare. It fishy, fishy as hell. I'm not saying Intel's best don't have a single core advantage right now, I am saying that's absolutely all they have against a well-tuned X5900+ They run hotter, use a lot more power, have less performance threads, Require yet another motherboard, seem to gain no advantage to speak of from DDR5 and are not going to come out on top in any benchmark or real world situation that requires any extended period of time under full load, or that can fully utilize a full affinity of threads.

Long story short AMD 9, Intel 1.

 

[larkspur]

Here: https://www.tomshardware.com/news/i...ryzen-9-5900x-and-5800x-face-off-intel-rising

Look in the section titled Application Performance. There are both ST and MT tests (and there are 21 different graphs for the MT tests).

None of those CPUs have been 'tuned' they are all stock. You're going to find most reviewers only use settings that are covered by warranty (hence the general lack of tuning).

 

[M42]

I don't know where "Games" came into play so much here. I was annoyed at the otiginal article because it gave no actual performance figures for actual benchmarks; at least
Long story short AMD 9, Intel 1.

How about business application performance then?

12th Gen Intel Core CPU Review Roundup

Intel has launched their new 12th Gen Intel Core desktop processors (code-named "Alder Lake") featuring support for DDR5, PCIe 5.0, as well as a completely new hybrid architecture using a mix of Performance and Efficient-cores. This is a lot of new technologies in one product which makes it hard...

www.pugetsystems.com

 

[almstsobur]

Here: https://www.tomshardware.com/news/i...ryzen-9-5900x-and-5800x-face-off-intel-rising

Look in the section titled Application Performance. There are both ST and MT tests (and there are 21 different graphs for the MT tests).

None of those CPUs have been 'tuned' they are all stock. You're going to find most reviewers only use settings that are covered by warranty (hence the general lack of tuning).

I saw that, that's more or less what I was talking about, take for instance
https://cdn.mos.cms.futurecdn.net/EYTnL9cKAffQArvVSH55pL-2560-80.png

Why is this a combined suite of benchmarks with a completely arbitrary number as a result. An even better question would be why was a Geometric mean calculation used to prove out the results, instead of just posting the actual results of these listed benchmarks? I believe the writer of this article used to work for the Federal Reserve or a Hedge fund, the best places to learn how to make numbers say anything you need them to say.

 

[almstsobur]

Then, check this out. PC Gamer, a leaked tweet pre-release says Alder Lake gets 11600 MT Cinebench score. They magically have a graph with exactly 11600 for Alder Lake, comparing among others the 5900x and 5950x. Are those also water-cooled in this benchmark? I somehow doubt it, because my 5900x tops 10,000 MT and they don't even have the 5950 breaking 10,000. I don't really care, I don't have any skin in the game, I guess it just annoys me, reminds of mainstream media and fake news. People are making numbers say whatever they need them to say, just like the headlines, then leave out the information that doesn't help the cause. It's like a world full of <Mod Edit> and liars, does anyone just shoot straight anymore?

Leaked Intel Alder Lake numbers place it up to 22% quicker than AMD's finest Ryzen | PC Gamer

 

[M42]

An even better question would be why was a Geometric mean calculation used to prove out the results, instead of just posting the actual results of these listed benchmarks? I believe the writer of this article used to work for the Federal Reserve or a Hedge fund, the best places to learn how to make numbers say anything you need them to say.

Here's a good reason:
https://www.cse.unsw.edu.au/~cs9242/18/papers/Fleming_Wallace_86.pdf

 

[M42]

Then, check this out. PC Gamer, a leaked tweet pre-release says Alder Lake gets 11600 MT Cinebench score. They magically have a graph with exactly 11600 for Alder Lake, comparing among others the 5900x and 5950x. Are those also water-cooled in this benchmark? I somehow doubt it, because my 5900x tops 10,000 MT and they don't even have the 5950 breaking 10,000. I don't really care, I don't have any skin in the game, I guess it just annoys me, reminds of mainstream media and fake news. People are making numbers say whatever they need them to say, just like the headlines, then leave out the information that doesn't help the cause. It's like a world full of <Mod Edit> and liars, does anyone just shoot straight anymore?

The odds that an integer number will end in "00" is 1/100, so while it may seem fake to you there is a non-zero chance among thousands of benchmarks that some results will happen to have "00" in the last 2 digits. It's just basic statistics.

 

[almstsobur]

Here's a good reason:
https://www.cse.unsw.edu.au/~cs9242/18/papers/Fleming_Wallace_86.pdf

How about just posting the <Mod Edit> results?>

 

[almstsobur]

The odds that an integer number will end in "00" is 1/100, so while it may seem fake to you there is a non-zero chance among thousands of benchmarks that some results will happen to have "00" in the last 2 digits. It's just basic statistics.

The fake part is a water cooled Alder lake Vs a 5900 and 5950 without any cooler at all. How does my 5900 break 10,000 but they show a 5950 at under 10,000? That was my point. If you are going to post one CPU water cooled, better post the others as water cooled results, or you are no better than CNN or the New York Times. It's disingenuous at best, flat out <Mod Edit> is more accurate though.

<Moderator Warning: This is your second profanity violation within minutes. Next one earns you some time off.>

 

[larkspur]

The fake part is a water cooled Alder lake Vs a 5900 and 5950 without any cooler at all. How does my 5900 break 10,000 but they show a 5950 at under 10,000? That was my point. If you are going to post one CPU water cooled, better damn well post the others as water cooled results, or you are no better than CNN or the New York Times. It's disingenuous at best, flat out <Mod Edit> is more accurate though.

Yeah I don't know about PC Gamer and all that. Leaked benchies before release should be approached cautiously. Actual reviews give you a better picture and avoid such lopsided comparisons. On the link I posted you can go through each of those 21 different graphs and see the different tests with the results. The 5900x is trading blows with the 12700k throughout those tests. Nothing to get too excited about - they're very similar in MT performance. Total platform costs are also similar (assuming you use DDR4 for AL) and as such, we're in the midst of some good healthy competition. This is all good.

 

[M42]

The fake part is a water cooled Alder lake Vs a 5900 and 5950 without any cooler at all. How does my 5900 break 10,000 but they show a 5950 at under 10,000? That was my point. If you are going to post one CPU water cooled, better damn well post the others as water cooled results, or you are no better than CNN or the New York Times.

It could be LIQUID cooled (who uses water?) because that is the only LGA 1700 ready cooler that they could get.

And contrary to what some believe not all liquid coolers perform better than air coolers. (Please don't make me do a Google search to enlighten you AGAIN. Just search for yourself this time if you don't know or don't believe me about air vs liquid coolers ).

 

[ex_bubblehead]

Since this thread is over a year and a half old, AND it appears that we have a new user that doesn't want to play nice with others, this thread is now closed. However, all warnings still stand.