Just AMD teasing how they narrowed the gap a bit between the 5800x3d design and the 7000x3d design? Guess we shall see soon enough.

tty8k:

Oh and wait and it's 1080p.

Huh? Should they be testing at 480p?

tty8k:

Problem is, the bread and butter chip aka 7800X3D (still about $450) is significantly slower clock wise and less cache. That 10% gain average in 1080p by 7950X3D might turn into just a few % for 78xx. Oh and wait and it's 1080p. Yeah, meh.

Your post doesn't make sense. Specs on paper are irrelevant; real-world performance is what matters, and at least from leaked benches so far, it seems the lower clock speeds don't hurt it much. What the lower clock speeds should imply is better performance-per-watt, which means boost clocks can be maintained for longer. It also means that there's more overclocking headroom, which is another bonus over the 5800X3D, since that can't OC. One thing a lot of people don't realize is the bigger the cache, the slower it can be read or written to. The closer you get to the actual core, the more cache speed matters. Shrinking L2 is smaller might actually be the main reason why the 7800X3D is ostensibly faster, despite the lower clock speeds. Thanks to the V-cache, the L2 can be fed data more quickly. Obviously DDR5 doesn't really matter due to the hefty V-cache, and the die shrink would have a minimal impact. Lastly, when it comes to the 5800X3D, it's typically bottlenecked by the GPU, which a newer CPU isn't going to fix no matter what it is. Where the CPU is the bottleneck, you're either getting a frame rate that no display can render or the game is optimized so poorly that adding more MHz won't be enough.

tty8k:

What you're saying doesn't make sense to me, not even in theory.

So you're telling me that what's written is absolute performance, regardless of IPC, node size, ISA, etc, and that you need more than 500FPS in CS:GO? Because that's pretty much what you first post was implying.

tty8k:

@schmidtbag No, I was trying to get an overview of 7800X3D vs 7950X3D based on specs and leaked gaming gains.

Why would it be any different? The 7950X3D is the flagship so of course it'll have higher clock speeds, because how else do they convince people it's the best if it doesn't outperform a CPU with half the cores in a workload that isn't thread intensive (or CPU intensive at all), such as most games? The only way the 7800X3D will be slower is because of clock speeds. But again - it's overclockable and should retain boost clocks for longer, so what's the problem here?

schmidtbag:

Why would it be any different? The 7950X3D is the flagship so of course it'll have higher clock speeds, because how else do they convince people it's the best if it doesn't outperform a CPU with half the cores in a workload that isn't thread intensive (or CPU intensive at all), such as most games? The only way the 7800X3D will be slower is because of clock speeds. But again - it's overclockable and should retain boost clocks for longer, so what's the problem here?

Exactly the issue at hand. While many people play at 1080p how many of them have montiors over 144Hz? 200Hz? At what point does 1080p even matter as cpu's made in the previous generation can outperform monitors commonly in use? It's a performnace test that has limited validity in the pc gaming world today. PC gaming has been moving toward higher resolution with the advancement of gpu's and as we move toward 4k the cpu becomes less relevant. I made the change to 1440p a few years ago and look forward to an "affordable" 4k gpu.

schmidtbag:

Why would it be any different? The 7950X3D is the flagship so of course it'll have higher clock speeds, because how else do they convince people it's the best if it doesn't outperform a CPU with half the cores in a workload that isn't thread intensive (or CPU intensive at all), such as most games? The only way the 7800X3D will be slower is because of clock speeds. But again - it's overclockable and should retain boost clocks for longer, so what's the problem here?

Clock speed of the 7950 3d cache CCD will have the same 5.0 Ghz boost as the 7800x3d. The other CCD with the smaller cache will have the full Ghz boost that is mostly certain. Wonder how can they make the cpu to schedule gaming tasks to the correct CCD.

tty8k:

No, I meant the gain for higher resolution will be even smaller.

Naturally. What do you want them to do? Test games at higher resolutions so that we can see absolutely no difference becauee nVidia doesnt make a €10000 gaming gpu? Or should they tell us that decade old cpus are perfectly fine for 4k gaming?

tty8k:

Problem is, the bread and butter chip aka 7800X3D (still about $450) is significantly slower clock wise and less cache. That 10% gain average in 1080p by 7950X3D might turn into just a few % for 78xx. Oh and wait and it's 1080p. Yeah, meh.

I think you're forgetting that the 7800x3D only has one CCD, which is important when comparing clock speeds to the 7950x3D the 7950x3D "looks" like it has much higher clock speeds then the 7800x3D, and that isn't untrue, to some extent, but mostly for applications that will benefit from that clock speed, over the 3D cache. Anything that benefits the 3D cache over clock speed won't matter as much. Basically, the 7950x3D has a max boost speed of 5.7Ghz ONLY on the CCD WITHOUT the 3D cache, the 7800x3D has a max boost speed of 5Ghz and it only has 1 CCD and therefore that is 5Ghz on the 3D cache CCD. To my knowledge, it is unknown what the 7950x3D and 7900x3D max boost speed on the 3D cache die is, but it is expected to also be 5Ghz, potentially up to 5.2Ghz, giving them a marginal, if any performance difference to the 7800x3D in the majority of gaming scenarios. A game that can utilize more then 8/16 threads and actually get a performance jump may be marginally faster, and a game that doesn't care about the extra 3D cache but really likes Mhz may also marginally be faster (if the scheduler understands this and puts the game on the CCD without the 3D cache, which we'll just have ot wait and see) Performance benchmarks will tell the story, whatever that may be, but it is simply not expected that there will be much if any gaming performance difference between the CPUs. Oh, and you also made a statement that the 7800x3D has less cache, again, not untrue, but shouldn't affect anything from a gaming standpoint anyways, because again, the CCD with 3D cache bringing the extra performance in most gaming scenarios will all have the same cache, the missing 32MB won't really affect it since that CCD won't use it anyways. https://cdn2.xfastest.com.hk/2023/01/20230107233638.png

Yosif Videlov:

Clock speed of the 7950 3d cache CCD will have the same 5.0 Ghz boost as the 7800x3d. The other CCD with the smaller cache will have the full Ghz boost that is mostly certain. Wonder how can they make the cpu to schedule gaming tasks to the correct CCD.

This 100% so it will depend on how much more Jizzhurts ppl needs for work. Gaming wise it will be nearly the same!

Aura89:

Basically, the 7950x3D has a max boost speed of 5.7Ghz ONLY on the CCD WITHOUT the 3D cache, the 7800x3D has a max boost speed of 5Ghz and it only has 1 CCD and therefore that is 5Ghz on the 3D cache CCD.

Great explanation and I am wondering if the X3D CCD will be thermal limited like how we saw with the 5800X3D? We saw users boost higher with PBO offsets and better AIO WC kits. If it is technically thermal constrained, it would be great to find out. I really want to see that Intel mini Cryo cooling on an X3D CPU!

Anybody still gaming at 1080p needs to take a good hard look at themselves in a mirror, and conclude there is life beyond living in a basement, eating Cheetos, drinking Mountain Dew and playing Apex Legends.

pegasus1:

Anybody still gaming at 1080p needs to take a good hard look at themselves in a mirror, and conclude there is life beyond living in a basement, eating Cheetos, drinking Mountain Dew and playing Apex Legends.

oO huh?

pegasus1:

Anybody still gaming at 1080p needs to take a good hard look at themselves in a mirror, and conclude there is life beyond living in a basement, eating Cheetos, drinking Mountain Dew and playing Apex Legends.

Oh. Nooo that's me... what to do ... what to do... ? 😛

mohiuddin:

Oh. Nooo that's me... what to do ... what to do... ? 😛

Buy a 4k screen, join a dating website and buy a sports car (red ideally and if possible a convertible). :p:D:D:D

pegasus1:

Buy a 4k screen, join a dating website and buy a sports car (red ideally and if possible a convertible). :p:D:D:D

Oh you where kidding :P I totally missed that !

Yosif Videlov:

Clock speed of the 7950 3d cache CCD will have the same 5.0 Ghz boost as the 7800x3d. The other CCD with the smaller cache will have the full Ghz boost that is mostly certain. Wonder how can they make the cpu to schedule gaming tasks to the correct CCD.

AFAIK there is no unofficial/or offical confirmation anywhere for this from a reliable source. It would contradict every other dual ccd ryzen product, on existing products the boost limit is shared between both dies, silicon quality(required vid), and thermals dictate the maximum boost on the "bum" die that is present on dual ccd chips , not some arbitrary hard limit. such a thing would nullify the purpose of amd's clock/voltage/power management. They would have to have some specific reason for doing so, with the 5800x3d it was the higher voltages higher frequencies required that could damage the cache. but with zen4 the cache and the core do not share the same voltage like zen3. there is a good chance the 5ghz 7800x3d is nerfed for market segmentation purposes rather than for technological/material limitations. As always wait and see before making assumptions.

user1:

AFAIK there is no unofficial/or offical confirmation anywhere for this from a reliable source. It would contradict every other dual ccd ryzen product, on existing products the boost limit is shared between both dies, silicon quality(required vid), and thermals dictate the maximum boost on the "bum" die that is present on dual ccd chips , not some arbitrary hard limit. such a thing would nullify the purpose of amd's clock/voltage/power management. They would have to have some specific reason for doing so, with the 5800x3d it was the higher voltages higher frequencies required that could damage the cache. but with zen4 the cache and the core do not share the same voltage like zen3. there is a good chance the 5ghz 7800x3d is nerfed for market segmentation purposes rather than for technological/material limitations. As always wait and see before making assumptions.

I'm not sure where there isn't official confirmation, it's not like all the news reporting companies just universally came up with the information on their own at the same time. Plus, if what you are saying would be true, then that'd mean the 7950x3D 5.7Ghz boost(120w TDP) is somehow, magically, more power efficient then the 7950 5.7Ghz boost(170w TDP) at the same performance (and no, the lower base doesn't account for the TDP). Plus, it has been reported that AMD is specifically working with microsoft to ensure that the scheduler will put the program/game/etc. on whatever die it will benefit the most from, higher frequency, or more cache. If you're just waiting for AMD to do a press release/graphics/pictures/etc. and choosing to not listen to the news companies that have been told that by AMD as is pretty normal then i guess you're just waiting to be disappointed lol What we know is, and isn't assumptions: 7900x3D and 7950x3D will have one CCD with higher frequency and less cache then the other CCD, which will have higher cache and lower frequency due to the 3D cache What we don't know is: What exact speed those 3D cache CCDs frequency will be, the same as 7800x3D? 5.1Ghz for 7900x3D and 5.2Ghz for 7950x3D? more? less? both faster then the 7800x3D but the same for eachother? That's the only unknown factor here.

Aura89:

I'm not sure where there isn't official confirmation, it's not like all the news reporting companies just universally came up with the information on their own at the same time. Plus, if what you are saying would be true, then that'd mean the 7950x3D 5.7Ghz boost(120w TDP) is somehow, magically, more power efficient then the 7950 5.7Ghz boost(170w TDP) at the same performance (and no, the lower base doesn't account for the TDP). Plus, it has been reported that AMD is specifically working with microsoft to ensure that the scheduler will put the program/game/etc. on whatever die it will benefit the most from, higher frequency, or more cache. If you're just waiting for AMD to do a press release/graphics/pictures/etc. and choosing to not listen to the news companies that have been told that by AMD as is pretty normal then i guess you're just waiting to be disappointed lol What we know is, and isn't assumptions: 7900x3D and 7950x3D will have one CCD with higher frequency and less cache then the other CCD, which will have higher cache and lower frequency due to the 3D cache What we don't know is: What exact speed those 3D cache CCDs frequency will be, the same as 7800x3D? 5.1Ghz for 7900x3D and 5.2Ghz for 7950x3D? more? less? both faster then the 7800x3D but the same for eachother? That's the only unknown factor here.

First off I couldn't find any primary source for this information, all I found were a couple claims on overclock.net and tweets, from randos. there are no slides or press/individual interaction describing a 5ghz fmax for the vcache die on the dual ccd parts. all there is is the 7800x3d being limited to 5ghz and people assuming that its the same for the dual ccd parts, easy jump to make , to assume that there is some "hard limit" that applies to all products, but one that ignores other factors, and knowledge of the products themselves. Second The tdp for the 7950x3d is listed at 120w, so yes it is more efficient by way of power limit, on existing products all of the cores share a frequency limit known as fmax, the power management decides how high to clock based on many factors , some of which are power draw , silicon quality , and temperature. for a 16 core cpu , most of the time many cores will be idle. so it can clock fewer cores higher and still remain within the power budget. a 7950x will still hit 5.7ghz on light workloads with a 120w power limit, its the all core frequencies that suffer. and we see this implied with the specs of the 7950x3d with a lower base frequency of 4.2ghz. And yes this actually does likely account for it since, a 120w power limited 7950x loses very little performance vs 230w limit. so it is completely possible that the vcache can make up for this (on avg), in heavily threaded workloads. [SPOILER]https://images.anandtech.com/graphs/graph17641/130507.png [/SPOILER] https://www.anandtech.com/show/17641/lighter-touch-cpu-power-scaling-13900k-7950x/2 amd is changing the preferred die via thread pinning/core prioritization presumably, so in the instance that the workload is running on the vcache die, and you're not utilizing all of the cores, why would amd limit the fmax to 5.0ghz. it doesn't make any sense since it would defeat the purpose of the power management , this isn't sandybridge with fixed ratios based on active cores. There will be 'natural limits' , like increased temps which will limit the frequency, so I'm not suggesting that the vcache die will hit 5.7ghz, but it would hasty to assume that it won't clock higher than 5.0ghz. There are some applications that do not gain anything at all from the vcache, In some rare cases running certain instructions at the same frequency will result in less performance due to increased cache latencies, so its not like such an optimization requires a huge clock deficit to be worth while. I will admit however it is possible that the rumor is true and amd is pairing silicon that is sooo bad that it won't clock higher than 5ghz, But I would think that unlikely, since even the bottom bin 7600 clocks above 5ghz easily if its fmax is raised.

Up to? https://abload.de/img/screenshot_20230220_02rinq.png

user1:

I will admit however it is possible that the rumor is true and amd is pairing silicon that is sooo bad that it won't clock higher than 5ghz, But I would think that unlikely, since even the bottom bin 7600 clocks above 5ghz easily if its fmax is raised.

7600 doesn't have 3D cache so i'm not entirely certain why the comparison. It's not about the quality of the silicon if the 3D cache is the limiting factor, won't matter how "S-Tier" the die is. but time will tell, i'd be glad to be wrong in the expectation that it won't be much higher then 5Ghz if at all, but given the TDPs and etc. it only makes sense.