Poll: Ryzen 7950X3D, 7900X3D, 7800X3D

[uscool]

Intel dead at the moment, but there are some shills in these forums constantly talking ****.

Almost like they want it to fail ? Shouldn't we want it to be the fastest so we have competition and gets intel to react and try to offer something better ? So we all win ?

The cpu space seems healthy unlike gpu

 

[delta0]

For all the operating system scheduler developers, this should be "interesting".

While not as hybrid strange as Intel's P/E core hybrids, for the dual CCD parts it will still be two different performing dies with all that involves.

And that which, IMO, made the 5800X3D such an attractive part, that is not having to go crazy with high-end memory kits and getting good performance without much if any memory tuning, might not apply to the dual CCD parts. That is a thread ends up on the CCD without the stacked 3D cache, then presumably it really needs the tuned DDR5 anyhow.

I’ve gone for DDR5 6000 CL30 EXPO RAM. I’ve not seen any info on what’s best for these new X3D CPUs

 

[Dave2150]

Where are these results from? What resolution/settings used etc? Any links so I can review please.

720P

At 4k, the 13900k is only just ahead. 7950X3D will easily beat this.

 

[Frenzy]

The higher TDP numbers on the previous chips were pointless, these TDPs are what the chips *should* have been out the gate.

GN did a video on it a while back.

7950X3D:
CCD0: 5.0Ghz/40+64MB V-Cache
CCD1: 5.7Ghz/40MB

7900X3D:
CCD0: 5.0Ghz/38+64MB V-Cache
CCD1: 5.6Ghz/38MB

7800X3D:
CCD0: 5.0Ghz/40+64MB V-Cache

So I'm guessing the lower tdp is just going to result in lower overclocks for the ccds with the extra cache, same as the 5800x3d. Probably more heat from them. I'm not biased towards intel or amd but I feel like there are a few people in this thread super over hyping themselves.

 

[stephenb]

Be cautious with AMD's slides guys

Notice how it only shows comparison to the 5800x3d and no performance comparison to ryzen 7000?

There is a good reason for that: take cs:go as an example, amd says 7800x3d is 23% faster than 5800x3d in this game. Ok, so where does that place it versus the 7700x??





Oh, so the 7800x3d is 16% slower than the 7700x in cs:go. Sounds great, so that's why AMD didn't compare x3d versus ryzen non x3d 7000... That must mean the clock speeds are significantly lower on x3d models (7700x3d must be several hundred MHz slower boost clock than the 7700x) , which makes sense given the x3d models have lower TDP versus the non x3d models

Edit: Does anyone know what the test system used was? Can't find it in the footnotes anywhere I've looked.

 

[uscool]

The 7900X3D looks like the sweet spot at first glance, cant wait to grab one!

So basically it uses 1 ccd for gaming which has the cache which run lower frequency? So wouldn't the 7900x3d be using 6 cores ?

That's why 7800x3d boost is 5ghz

 

[Robert896r1]

Some of these claims are bit strange to me. I'll take two for now.

If we look at CS:GO for example, AMD is claiming 23% gains.

Now if you look at the 7700x vs 5800x3d, without 3D, you gain that anyway:

Then when you compare their advantage over 13900k, looking at HDZ. AMD is claiming 24% with X3D:

But again, that's not really a 3D gain:

 

[Dicehunter]

Nah, if anything these new parts look uncharacteristically weak. I was hoping for much more out of them tbh. Intel still king in so many games. Can't believe I waited for this v-cache for nothing.

13900K still had an easy 20% lead over 5800X3D (which was already better than Zen 4) in Cyberpunk 2077 (and similar open world RT games). They're barely showing a 10% gain over the 5800X3D, so it's still way behind. My only hope is that I know I can't really rely on AMD benchmarks for proper results, so there might be more to eek out of them when someone competent tests them.

The worst part for me is that I'm basically going to have to keep waiting for newer and better CPUs, except Intel has dumped on the bed and is looking like they are on vacation for another 2 years so 13900K is the best they'll have for a while, and Zen 5 is largely unknown but AMD seems to focus on making better server chips and then handing them down for us on desktop, and simply slapping some more cache on it doesn't seem to be enough.

So, an actual monster of a gaming CPU that can clear all hurdles for 120 fps gaming? Seems like it's going to be a few years. Meh.

Seems logical, I'll be swapping out my 13900K for the Raptorlake Refresh in Q3 anyway which is meant to be RPL, Same socket etc... but with a clock bump and voltage optimisation, 13900K can go in my HTPC

 

[Dave2150]

Seems logical, I'll be swapping out my 13900K for the Raptorlake Refresh in Q3 anyway which is meant to be RPL, Same socket etc... but with a clock bump and voltage optimisation, 13900K can go in my HTPC

I still don't see how they can further improve on a 13900KS for the raptor lake "refresh". It'd have absolutely insane TDP.

 

[LtMatt]

Well if the multiple giant watermarks didn't give it away...

Core i9-13900K, i7-13700K & i5-13600K: Gaming-Könige im Test: Benchmarks in Games

Intel Raptor Lake im Test: Benchmarks in Games / Leistung in Spielen (720p, RTX 3090 Ti) / CPU-Gaming-Leistung im absoluten CPU-Limit

www.computerbase.de

I know it's computerbase but they have a lot of articles. Ah, they gimped the 7000 series by running the slowest possible memory, explains the performance differences. Will be interesting to see if X3D can still perform well running on the slowest possible memory.

EDIT - They tested Warzone not MW2 unfortunately.

 

[B1gbeard]

Well, we pretty much already know from the 5800X3D that there are games which simply don't care about cache much (or at all), so the gains won't be universal. There are also outliers the other way of course, which see truly massive gains. World of Warcraft as an example, where the 5800X3D was around 45% faster than a 5950X. That's why it's always important to check benchmarks for games that you personally care about.

For me personally, these chips tick all the boxes. The game I play most saw huge gains with the 5800X3D, and the emulator I use most sees huge gains from AVX-512. I also want a platform with longevity so I can be lazy and not have to rebuild my PC again for another five years.

My gaming is kind of the opposite at the moment where the Intel part is stronger but will wait for reviews. If it gets there abouts in the games I play will probably get a 7950x3D for the platform longevity. CPU's are coming out to fast at the moment. It almost feels like being punished for not waiting and waiting.

 

[doodah]

Take it no pricing for these 3 SKUs?

 

[Gazb]

For all the operating system scheduler developers, this should be "interesting".

While not as hybrid strange as Intel's P/E core hybrids, for the dual CCD parts it will still be two different performing dies with all that involves.

And that which, IMO, made the 5800X3D such an attractive part, that is not having to go crazy with high-end memory kits and getting good performance without much if any memory tuning, might not apply to the dual CCD parts. That is a thread ends up on the CCD without the stacked 3D cache, then presumably it really needs the tuned DDR5 anyhow.

This is my worry too, 1 CCD having stack 3D cache and the other not, and the OS unlikely to know or care which core on which CCD gets a task.
As you say, not as bad as Intel's P/E cores setup, so the performance difference between running on the different CCD's shouldn't be as drastic. (Of course some people claim this is nonsense and P or E core's isn't a concern when gaming but there are numerous accounts of having to use Process Lasso to stop game threads going onto E cores, and causing hitching/stutters, even though overall fps is very good)

I wonder if having it on only 1 of the 2 CCD's for AMD is a cost thing and/or them not wanting to lower the peak boost frequency meaning it could be slower in some tasks that don't benefit from extra cache.

 

[B1gbeard]

I know it's computerbase but they have a lot of articles. Ah, they gimped the 7000 series by running the slowest possible memory, explains the performance differences. Will be interesting to see if X3D can still perform well running on the slowest possible memory.

I have a 7950X, a 4090 and Modern Warfare 2, let's see how accurate these benchmarks are. They are using the built in benchmark and from looking at their settings, they are running 720P, 1080P and 4K max settings. I'll test 1080P with the 7950X/4090 and see what the Average FPS and 1% lows are.

Is 5200 the slowest possible? They could have used faster memory on the 13xxx series as well.

 

[delta0]

Take it no pricing for these 3 SKUs?

There’s some rumoured prices. I suspect as they aren’t coming until Feb they want to see if Intel will drop their prices and AMD will probably come in about £100 over these. Just my guess on the situation.

 

[Poneros]

Ah, they gimped the 7000 series by running the slowest possible memory

C'mon, you know this is not relevant. Let's just be honest about how weak Zen 4 is in some games, because if you really want to add memory to it you know full well how poor of an experience it is to OC mem on AM5 and how difficult (expensive) it is to get to 6000 Mhz. Memory OCing is NOT going to save Zen 4 where it is weak.

 

[Toasty]

Dual CCD parts can already have issues - now something else to the mix. Will need to determine if application is better running on lower clocked CCD with VCache or the higher clocked CCD.

I think as always wait for benchmarks and then see if the games you play are better on 13900k/s OR 7950x3D - I don't think there is 1 winner takes all here some games will still not care about cache and some will love it.

 

[uscool]

Maybe they got the guys over from the GPU department to make these

Some of these claims are bit strange to me. I'll take two for now.

If we look at CS:GO for example, AMD is claiming 23% gains.

Now if you look at the 7700x vs 5800x3d, without 3D, you gain that anyway:

Then when you compare their advantage over 13900k, looking at HDZ. AMD is claiming 24% with X3D:

But again, that's not really a 3D gain:

 

[Freddie1980]

Take it no pricing for these 3 SKUs?

Arm, leg, kidney

 

[stephenb]

AMD Unveils Three Ryzen 7000X3D V-Cache Chips, Three New 65W Non-X CPUs, Too

3D V-Cache to the masses.

www.tomshardware.com

AMD is working with Microsoft on Windows optimizations that will work in tandem with a new AMD chipset driver to identify games that prefer the increased L3 cache capacity and pin them into the CCD with the stacked cache. Other games that prefer higher frequencies more than increased L3 cache will be pinned into the bare CCD. AMD says that the bare chiplet can access the stacked L3 cache in the adjacent chiplet, but this isn’t optimal and will be rare. Yes, the chip with the extra L3 cache will run games at a slower speed, but most games don’t operate at peak clock rates, so you should still get a huge performance benefit.