Core i9-10900K vs. Ryzen 9 3950X Cinebench R15 Comparison Leaked

erek

erek

[H]F Junkie
Joined
Dec 19, 2005
Messages
10,897
Impressed? It's not WCCFTech!

"In the single-threaded Cinebench R15 test, the Core i9-10900K scores 222 points, while the 3950X scores 213, which is a 4.22% lead for the new Intel flagship over AMD's. The i9-9900K is 2.81% faster than the 3950X in the same test. The landscape changes completely with multi-thread. Armed with 16 cores and 32 threads, the 3950X tests 48.61% faster than the i9-10900K, and a whopping 94.14% faster than the i9-9900K, which means the 3950X should land around 90% (±5%) faster than the i7-10700K. Core i9-10900K vs. Ryzen 9 3900X should make for a fascinating contest."

1588234062965.png


https://www.techpowerup.com/266442/core-i9-10900k-vs-ryzen-9-3950x-cinebench-r15-comparison-leaked
 
Rvenger said:
Single core 213 vs 222. That graph makes it look way better.
Click to expand...


Yeah,

If marketing departments could just get it through their thick skulls that when the axis is set at anything other than 0, it is tantamount to fraud.

There is literally never an acceptable reason to ever have an axis that does not cross at 0. They would have failed their Engineering and Science classes. Any submitted research papers would have been rejected for distorting the results.

It's literally the first thing they teach you. Always, always, always have your axes start at 0.

Marketing turds should not be allowed to make charts.


Here is what those charts should look like:

chart2.png


chart1.png
 
Zarathustra[H] said:
If marketing departments could just get it through their thick skulls that when the axis is set at anything other than 0, it is tantamount to fraud.
Click to expand...

It's tantamount to marketing. That's...that's why marketing exists.
 
Axman said:
It's tantamount to marketing. That's...that's why marketing exists.
Click to expand...

There is certainly some spin to marketing, but distorting data should be off the table.

There ought to be laws regulating marking claims for all products, like there are in the medical device and pharmaceutical industries.

Working in the medical device industry, if I want to make a marketing claim I have to have the statistical proof to support it, and I have to represent it accurately. Misrepresentation on labeling (which includes not just what is on product, but also all marketing information) can result in being shut down by FDA, and even criminal charges in some cases.

All industries should be like this.

But yes. IN most industries Marketing departments are little more than professional liars an manipulators. Trying to intentionally muddy the waters, and to get people to suspend their rational decision making processes, and make emotional decisions instead. They are literally the worst.
 
Zarathustra[H] said:
There is certainly some spin to marketing, but distorting data should be off the table.

There ought to be laws regulating marking claims for all products, like there are in the medical device and pharmaceutical industries.

Working in the medical device industry, if I want to make a marketing claim I have to have the statistical proof to support it, and I have to represent it accurately. Misrepresentation on labeling (which includes not just what is on product, but also all marketing information) can result in being shut down by FDA, and even criminal charges in some cases.

All industries should be like this.

But yes. IN most industries Marketing departments are little more than professional liars an manipulators. Trying to intentionally muddy the waters, and to get people to suspend their rational decision making processes, and make emotional decisions instead. They are literally the worst.
Click to expand...
In this case it does not bother me if the axis doesn't cross at zero, but the scale absolutely needs to be included. They do include the numbers, but people won't pay attention to that at first glance. To be fair, the ROG charts are a lot better than some of the other charts I've seen. I have seen a bar chart like this where to me it was obvious that all the bars were not on the same scale, absolutely skewing the visual representation.
 
Axman said:
No, just get better at understanding marketing. You don't appear to have a problem here, right?
Click to expand...

No.

Just because you didn't get any money when robbing a bank doesn't mean that no crime was committed.

They are not fooling those of us who have experience, technical backgrounds, and critical thinking, but it does get tiring to look at their deceitful shit day in and day out.

It is not OK to take advantage of those who are less experienced, less technical or of lesser intellect.

If they have to lie, deceive and cheat to stay in business, they shouldn't be in business at all. Marketing departments are a scourge.

Anyway, I don't need another off-topic warning, so I am going to stop here.
 
Rvenger said:
Yes, much better. The funny thing is the multicore graph looks fine... the single core graph was definitely..."fluffed"
Click to expand...
I am inclined to believe that whomever the poor schmuck was who was assigned to draft these was told that the difference is too small to see, so he needed to make it more obvious. But the whole point is that there isn't that big of a difference. I'm not a marketing person, though, so what do I know.
 
erek said:
Impressed? It's not WCCFTech!

"In the single-threaded Cinebench R15 test, the Core i9-10900K scores 222 points, while the 3950X scores 213, which is a 4.22% lead for the new Intel flagship over AMD's. The i9-9900K is 2.81% faster than the 3950X in the same test. The landscape changes completely with multi-thread. Armed with 16 cores and 32 threads, the 3950X tests 48.61% faster than the i9-10900K, and a whopping 94.14% faster than the i9-9900K, which means the 3950X should land around 90% (±5%) faster than the i7-10700K. Core i9-10900K vs. Ryzen 9 3900X should make for a fascinating contest."

View attachment 241526

https://www.techpowerup.com/266442/core-i9-10900k-vs-ryzen-9-3950x-cinebench-r15-comparison-leaked
Click to expand...

So the takeaway is that AMD needs 60% more cores to deliver 48% more performance? :whistle:
 
  • Like
Reactions: erek
like this
Snowdog said:
So the takeaway is that AMD needs 60% more cores to deliver 48% more performance?
Click to expand...

I know nobody would actually run their computers this way but it would be interesting to see both CPUs run the test at the same clock speed, so we would know what Zen could achieve if it weren't held back by the speed deficit.
 
Vega said:
Who the heck cares about Cinebench? I bet less that 1% of users use software that routinely takes advantage of over 10 CPU cores.
Click to expand...

Cinebench has proven to be a great comparison tool over the years.

I agree that the best benchmark is one that exactly represents your workload, but there are as many different workloads as there are people who own computers, so we cant test them all. More detailed and varied benchmarks will certainly come with time, but as an early preview of what a system is capable of Cinebench isn't a bad place to start.

It is also a quick and easy benchmark to run to give you a theoretical understanding of what the single core and multicore performance really looks like.

Over the years I have found the Cinebench single core score scales just about with the average game benchmark, when you eliminate GPU bottlenecks. Yes some games will scale better or worse with different architectures, but the Cinebench tends to fall just about in the middle. Cinebench Multicore has generally scaled pretty well with rendering and encoding workloads. The only real downside IMHO is that it doesn't allow enough runtime to fully heat soak a system, so it may favor architectures with aggressive turboclocks based on core temp. There are ways to compensate for this, by running it many times in a row, or pre-heating a system with Prime95, but I'd imagine benchmarkers rarely do this.

Besides, since it was attacked by AMD fans as being an Intel biased benchmark in the FX days when Intel was ahead, and has been attacked by Intel fans as being AMD biased in the Ryzen era, I think that pretty much proves that it is a great benchmark :p

But yes. It will be better to see real world workloads. Until that time, there are a hell of a lot worse benchmarks that could be posted. Garbage like Fritz Chess or SuperPi or SiSoft Sandra? Those are usually all pretty pointless... This at least tells us something
 
  • Like
Reactions: Vega
like this
Vega said:
I bet less that 1% of users use software that routinely takes advantage of over 10 CPU cores.
Click to expand...

Most likely.

Although when I installed SQL Server on my work PC to aid development by getting rid of network speed limits, I discovered a 4th-gen i5 didn't really do all that well. But the next year I got an 8th-gen i7, and going from 4c4t to 6c12t really sped the new machine up, a lot more than the clock speed boost would account for. While I don't *need* 10 cores I could *use* them to increase parallel workloads.
 
Its good to see AMD being so competitive again and winning in multi core. Can't wait to see what Zen 3 holds in store. Thats probably where I will make the switch from my aging 6700K
 
Looks like I'm just going to be hanging on to my OC'd 8700K a little longer. I'm still one of those people that have had too many issues with AMD. This just makes me certain that going with the 10900K is also a bad idea. 11900K, I'll be waiting ;)
 
Domingo said:
Looks like I'm just going to be hanging on to my OC'd 8700K a little longer. I'm still one of those people that have had too many issues with AMD. This just makes me certain that going with the 10900K is also a bad idea. 11900K, I'll be waiting ;)
Click to expand...
I get it, I still buy AMD because there isn't a good alternative in my price range but I get it.
 
Krenum said:
Its good to see AMD being so competitive again and winning in multi core. Can't wait to see what Zen 3 holds in store. Thats probably where I will make the switch from my aging 6700K
Click to expand...


Right, this is he big takewaway from all this:

Comet Lake adds 3% clock over the 9900k. Zen 3 is expected to add 10-15% higher performance
so should erase any remaing clock speed advantage Intel has.

Oh, and they probably won't raise prices over Zen 2 launch (Intel is charging almost $100 premium for 10 core Skylake, so pretty close to the current price of 3950x ).

And the influx of HT-for-all means that all these processors will be out-of-stock for the next three months (plenty of time for Zen 3 to launch). Ever seen the sudden demand spike from Intel giving up free performance boost?
 
Last edited:
Zarathustra[H] said:
Yeah,

If marketing departments could just get it through their thick skulls that when the axis is set at anything other than 0, it is tantamount to fraud.

There is literally never an acceptable reason to ever have an axis that does not cross at 0. They would have failed their Engineering and Science classes. Any submitted research papers would have been rejected for distorting the results.

It's literally the first thing they teach you. Always, always, always have your axes start at 0.

Marketing turds should not be allowed to make charts.


Here is what those charts should look like:

View attachment 241592

View attachment 241591
Click to expand...
Now start at 212 and go up in tenths and you've got it.
 
Zarathustra[H] said:
If marketing departments could just get it through their thick skulls that when the axis is set at anything other than 0, it is tantamount to fraud.
Click to expand...

0 isn't *always* the most representative place to start. Usually, but not always.
 
Domingo said:
Looks like I'm just going to be hanging on to my OC'd 8700K a little longer. I'm still one of those people that have had too many issues with AMD. This just makes me certain that going with the 10900K is also a bad idea. 11900K, I'll be waiting ;)
Click to expand...

I'm in the same boat.
 
Domingo said:
Looks like I'm just going to be hanging on to my OC'd 8700K a little longer. I'm still one of those people that have had too many issues with AMD. This just makes me certain that going with the 10900K is also a bad idea. 11900K, I'll be waiting ;)
Click to expand...

Niner21 said:
I'm in the same boat.
Click to expand...

I'm curious. What type of issues?

Apart from the FX series disappointing performance, I've never had an issue with an AMD system that was in any way different than with Intel systems.

First gen Ryzens were a little picky about RAM. They worked great with anything on the approved list, but if you got the wrong RAM you could have some issues. That's behind us at this point. Zen+ and Zen2 have much improved memory controllers.

Living with my Threadripper 3960x has been exactly like living with my i7-3930k before it, just much faster, and with four times the cores.
 
Domingo said:
Looks like I'm just going to be hanging on to my OC'd 8700K a little longer. I'm still one of those people that have had too many issues with AMD.
Click to expand...

Ehh, I came from an ancient 4790K to my 3900X, so the jump was much bigger, but I also work more with SuperO/Xeon/ECC than Asrock or MSI, so I feel my opinions of stability and problems are valid. I just jumped to an Asus X570-E and 3900X, it was less money and more everything than the Intel build I was planning, and I have had *one* stability issue in the 2-3 months since; The DDR4 I bought over a year ago makes my machine hard-reset sometimes when it talks to Asus's garbage RGB software. Uninstalled that and zero other issues. I load my machine pretty hard, so I feel quite safe in saying it's really solid if you get decent gear.
 
Just not a fan of AMD or the way they do things, it's going to take far more than "pretty much the same, give or take" to catch my eye.

I also don't see the need for more than a 6/12 8700K at around 4.7++ Ghz right now for gaming and light web use.

If you do Cinebench or stuff like that, then I get it but I think any upper end CPU from the last 2-3 years is going to be good for gaming.

I'm actually kind of disappointed because I'd like a reason to upgrade, but I just can't see one.
 
Going to be interesting what AMD has in store with Zen 3 later this year. If the rumored 15% IPC performance increase is true then AMD may very well be in a desirable situation where they beat Intel in both single and multi-threaded applications.
 
defaultluser said:
Right, this is he big takewaway from all this:

Comet Lake adds 3% clock over the 9900k. Zen 3 is expected to add 10-15% higher performance
so should erase any remaing clock speed advantage Intel has.

Oh, and they probably won't raise prices over Zen 2 launch (Intel is charging almost $100 premium for 10 core Skylake, so pretty close to the current price of 3950x ).

And the influx of HT-for-all means that all these processors will be out-of-stock for the next three months (plenty of time for Zen 3 to launch). Ever seen the sudden demand spike from Intel giving up free performance boost?
Click to expand...

Yeah theres really no way I'm going to pay $500 for a 10 core CPU at 14nm & 125w, When I can get the same or more cores from AMD at 105w & 7nm cheaper, I'll eat the 4-5% gaming performance loss.
 
  • Like
Reactions: Meeho
like this
Derangel said:
But you know well enough that things don't scale linearly.
Click to expand...

This benchmark actually does scale linearly with cores, which is why AMD is so fond of it. It would have been more price comparable if they had used a 12 core AMD, but people never seem to do the more appropriate comparison.

Vega said:
Who the heck cares about Cinebench? I bet less that 1% of users use software that routinely takes advantage of over 10 CPU cores.
Click to expand...

AMD does. It's one of the few cases that scale linearly with cores. As you say 99% of software doesn't have this characteristic.

For most people, >6 cores is mostly an epeen contest.
 
You must log in or register to reply here.
Back
Top