AMD Ryzen 9 7950X Geekbenched, Crushes i9-12900K, in the League of the i9-13900K

[xtreemchaos]

in 2 words "Im impressed" but a tad worried about the power draw.

 

[P4-630]

worried about the power draw.

Especially at "5.7Ghz"...

 

[lsevald]

I wonder why there is such a big jump (~10%) in single core score, comparing 13700k to13900k (2090->2314)?

 

[xtreemchaos]

yes i dont like the direction kit is going with ever increasing power demands when we are trying to save the planet, but its only a small amount of folk that would be into using 7950x.

 

[usiname]

I wonder why there is such a big jump (~10%) in single core score, comparing 13700k to13900k (2090->2314)?

The result of the 13900k was achieved with chiller

 

[Ed_1]

Some of those GB5 scores with 13xxx seem off to me as my 12600k can do 2000 ST with only max of 5.1 on two cores (5.1, 5.1, 5.0, 4.9, 4.8, 4.8). there running much higher freq and I am no DDR4 still (3867).
I guess we will know in a month or so.

Edit: I see now the 13600k only has single core freq of 5.1 so that is the reason of 20xx score, seems Intel only really pushed freq on top end models.

 

[Oberon]

The result of the 13900k was achieved with chiller

And in the unlimited 350W mode. Stock vs stock will be very close indeed.

 

[Denver]

No one will buy the poor 6-core instead of the far superior Core i5-13600K.
You are looking at performance difference +50% and lower cost!

This i5 is going to be more expensive, and honestly we don't need more toy cores heating up the house just to run games.

 

[ARF]

This i5 is going to be more expensive, and honestly we don't need more toy cores heating up the house just to run games.

More expensive than $300? How much? $350? $400?
Even if it is 50-100 more expensive, it will run with much cheaper motherboard and with DDR4.

 

[Denver]

More expensive than $300? How much? $350? $400?
Even if it is 50-100 more expensive, it will run with much cheaper motherboard and with DDR4.

Yes, about $50-100 more expensive, plus the price of a good watercooler, and don't tell me DDR5 won't make any difference, you'll lose performance using DDR4

 

[R0H1T]

Intel's gonna raise prices likely with the 13th gen or maybe slightly later, you're cuckoo if you think they can continue competing at these prices! Did you not see their latest results

I wonder why there is such a big jump (~10%) in single core score, comparing 13700k to13900k (2090->2314)?

Slightly better IPC & higher memory speeds is what I'm guessing. Oh you were talking about same 13th gen, IPC will be higher on the i9, due to more cache, but clocks as well.

 

[ThrashZone]

Both Intel and AMD will have to be competitive in pricing if they wish to lure people to buy their CPU's this generation. The performance seems to be similar, so the only factor will be performance to price ratio!

Hi,
Intel already announced price increases
AMD is always high up to 6-10 after release so competitive I see both coming out the gate high on highend stuff and intel cutthroat on mainstream chips.

 

[Valantar]

It has been discussed already that the intel platform cost will be lower..

What do you mean by this cliche "unreleased"? What is not released? We know everything already!

"Cliche"? It's a fact. The product has not been released yet. The specs are not public. That's just reality. They have been leaked, but leaks are fundamentally not trustworthy information. They might be correct, but we literally can't know that until we have official confirmation - i.e. a product launch. If this is your attitude towards reality, treating facts about real-world events as "cliches", the no wonder you get into all kinds of nonsensical arguments.

By AMD's own words and charts nothing actually changed besides a smaller node and some optimizations, it's exactly the same cpu. So i don't expect any changes there neither should you.

... what? What have you been smoking?

We have a 13% (pending independent confirmation) IPC uplift. There are major changes throughout the architecture, which their chart on the right there handily illustrates. No, it isn't a ground-up redesign, but your representation of this as "some optimizations" undersells the changes to a downright farcical degree. Calling this "exactly the same CPU" is just downright dumb - either you don't know what you're talking about, or you don't understand what you're talking about.

Also, crucially, if this was just a node shrink, how are they achieving the massive clock speed increases of this generation? Are you then saying that the node was the only thing standing in the way of Zen3 clocking higher, and that it had no architectural clock speed limitations? That nothing has been done to enable higher clocks? 'Cause what you're saying here just doesn't align with the real-world changes between these product generations.

 

[ratirt]

Both camps have CPUs with some performance. Now waiting for reviews because geekbench is not what I can call an objective or meaningful performance benchmark.

 

[Valantar]

More expensive than $300? How much? $350? $400?
Even if it is 50-100 more expensive, it will run with much cheaper motherboard and with DDR4.

And with DDR4 it will perform worse, so ... yay? No, it won't lose MT performance to match the price difference, but it's still a rather weird route to take. And, once again, if all you're doing is gaming, the extra cores/MT performance doesn't matter.

 

[Bomby569]

... what? What have you been smoking?

you show me a chart that confirms exactly what i told you

 

[ARF]

And with DDR4 it will perform worse, so ... yay? No, it won't lose MT performance to match the price difference, but it's still a rather weird route to take. And, once again, if all you're doing is gaming, the extra cores/MT performance doesn't matter.

The performance difference between a DDR4 and a DDR5 setup is in the area of the statistical error. We all know that the memory type very rarely matters and only in cherry-picked examples.

 

[Makaveli]

Its faster than the 13900k

That 2314/24396 score is an overclocked cpu at 350 watts.

This was caught on reddit and everyone reporting it seemed to have missed this.

Purported 5.7 GHz Intel Core i9-13900K Spotted at Geekbench

Raptor Lake chip boosted as high as 5.7 GHz during the tests.

www.tomshardware.com

 

[Valantar]

you show me a chart that confirms exactly what i told you

No, it confirms that the issue is that you don't understand what you're talking about. You don't get a 13% IPC improvement from "some optimizations" that are small enough for it to be "the exact same CPU". That you're actually arguing this is downright absurd. A node shrink doesn't affect IPC, and 13% is a pretty significant improvement. AMD also detailed which specific parts of the cores had seen significant changes to effect these improvements. Is that so hard to grasp?

The performance difference between a DDR4 and a DDR5 setup is in the area of the statistical error. We all know that the memory type very rarely matters and only in cherry-picked examples.

For gaming it is irrelevant. For productivity applications it absolutely isn't. Given that the 13600K's extra cores are irrelevant for gaming, I assumed that was what we were talking about - but please let me know if that wasn't the case. In productivity, that's a solid 4% performance advantage for DDR5 - well outside of error, across a broad suite of tests. Of course this depends on the application, and can vary from no difference at all to major differences, but there are plenty of common tasks in that test suite where the DDR5 shows a clear advantage. It's not huge, no, but it's there. So, you pay more for a faster CPU, you pay less for a cheaper platform, but that platform also loses you performance? Doesn't add up to me, sorry.

 

[phanbuey]

Its faster than the 13900k

That 2314/24396 score is an overclocked cpu at 350 watts.

This was caught on reddit and everyone reporting it seemed to have missed this.

Purported 5.7 GHz Intel Core i9-13900K Spotted at Geekbench

Raptor Lake chip boosted as high as 5.7 GHz during the tests.

www.tomshardware.com

View attachment 260118

That's an engineering sample - an OC'd 12900K can hit that score, so at 5.5ghz they would have to have an IPC regression across the product stack for that to hold up (they won't).

 

[R0H1T]

The performance difference between a DDR4 and a DDR5 setup is in the area of the statistical error. We all know that the memory type very rarely matters and only in cherry-picked examples.

That's not true, you can check the best tuned results on either set of memory & DDR5 will come easily on top, also if it were a statistical error then why get DDR5 support anyway? Are Intel engineers stupid to waste that extra die space?

Nope you're clearly wrong on that one, yes it is application depend but in some applications it can be a massive gap! On avg it should be high to mid single digits, talking about best DDR5 vs best DDR4 results.

 

[lilwirebrushdude]

The performance difference between a DDR4 and a DDR5 setup is in the area of the statistical error. We all know that the memory type very rarely matters and only in cherry-picked examples.

I doubt it will stay that way.

 

[junglist724]

Geekbench is highly memory sensitive. My stock 12900k gets 2050 point in st so... Yeah.

Yeah mine's slightly OC'd with 6000C30 RAM and I get 2147.

I wanna see performance in something like Metro Exodus Enhanced Edition or Crysis remastered where looking out at distant landscapes with RT on is very single thread or cache/memory bound. The first town in The Outer Worlds also performs really poorly without a good cpu.

 

[QuietBob]

The 7600X faster than the 12900K and 13700K in ST?
I wish that was true. Let's just wait for actual reviews.

 

[InVasMani]

How fair was it from AMD anyway to compare with intel CPU's from 2021.
Ok , Raptor Lake wasn't out yet, now waiting for Raptor Lake little boys.....

Yes very unfair of them to launch their product prior to Intel how dare them f*ck any competition Intel FTW and ALL HAIL QUAD CORES FROM NOW UHNTIL FOREVER AFTER!! I'll wait for the quad cores to launch...I'm sure Intel will have one as people complain 6C's isn't enough for AMD.