Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)

Page 526 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Vattila

Senior member
Oct 22, 2004
807
1,411
136
Except for the details about the improvements in the microarchitecture, we now know pretty well what to expect with Zen 3.

The leaked presentation by AMD Senior Manager Martin Hilgeman shows that EPYC 3 "Milan" will, as promised and expected, reuse the current platform (SP3), and the system architecture and packaging looks to be the same, with the same 9-die chiplet design and the same maximum core and thread-count (no SMT-4, contrary to rumour). The biggest change revealed so far is the enlargement of the compute complex from 4 cores to 8 cores, all sharing a larger L3 cache ("32+ MB", likely to double to 64 MB, I think).

Hilgeman's slides did also show that EPYC 4 "Genoa" is in the definition phase (or was at the time of the presentation in September, at least), and will come with a new platform (SP5), with new memory support (likely DDR5).

Untitled2.png


What else do you think we will see with Zen 4? PCI-Express 5 support? Increased core-count? 4-way SMT? New packaging (interposer, 2.5D, 3D)? Integrated memory on package (HBM)?

Vote in the poll and share your thoughts! :)
 
Last edited:
  • Like
Reactions: richardllewis_01

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
If you are buying a 13900k/7950X you better have some work that requires heavily threaded performance.

I'm not sure about this. Despite the amount of cores in each CPU, they are still mainstream parts. If I do upgrade this year, I'll probably get either a 13900K or 7900x, and the most multithreaded workloads I ever do are gaming and encoding.

Professional rendering or encoding though typically use Xeon, Threadrippers or Epyc CPUs.

If you only have have work that is lightly threaded, how is it going to perform better on a hybrid design than a regular design, as it is only using a few cores?

I'm talking about performance per watt. The 12900K has better performance per watt than the 5950x in gaming, browsing and other lightly threaded applications as I recall.
 

Just Benching

Banned
Sep 3, 2022
307
156
76
When NDA Lifts and Reviews come in the entire Ryzen 7000 Line Up Will dominate current Intel Skus. No Questions about it.
What do you meant current? Alderlake? Yeah sure, with the exception of the 7600x getting pummeled by the 12600k, and probably the 7700x getting its ass handed to it by the 12700k, you are right. But they don't compete with alderlake, thry compete with Raptor
 

Just Benching

Banned
Sep 3, 2022
307
156
76
IMO the energy efficiency comparision of diff. cpu-arch should be made at their respective stock power settings.
I don't think there's a reason to test at the same power cap, let alone any "iso-power"- like nonsense.
There are 2 reasons that energy efficiency testing should be done in same wattage. First of all, because a fundamental part of electronics is, the increase in wattage isnt linear with the increase in performance. That pretty much means that usually the CPU that has a higher power limit will also have worse efficiency. At that point you are not testing architectural efficiency but just stock settings efficiency. Say a 7950x at 200w is less efficient than a 7950x at 100w.

Second reason, which is pretty similar to the first one, it prevents "cheating".Intel is soon going to realize the 13900T version, which is basically power limited to 35w. That will make it the most efficient cpu on planet Earth, which might not be the case if you tested it in ISO wattage.

Point is, power draw is not an integral part of a CPU. Cores / Cache / frequency etc. are integral parts, power draw isn't. It's the one thing you can change without affecting stability. So there is no reason you shouldn't.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
Literally don't believe a 5900X is faster In doom eternal than a 7950X.

I said several pages ago that the Achilles heel of Zen 4 would be it's memory controller, and I based that statement on preliminary memory benchmarks from Aida.

Gaming stresses the CPUs memory controller and interface more than any other consumer oriented workload, so ostensibly, having a weak memory interface would hurt gaming performance.

And it's funny because that Spanish preview used slower DDR5 for the 12900K than the 7950x and the former still outperformed the latter in the memory tests, which indicates that Alder Lake's memory controller is superior to Zen 4's, and also possibly that Alder Lake may be better at hiding latency than Zen 4.
 

Bigos

Member
Jun 2, 2019
151
365
136
Can anyone even read the AnandTech review? There are so many typos it looks like a SPAM mail...

Looks better now.

This has been byhrough superior power efficiency, as Zencally a Zen 3 refinement, but on the new TSMC 5 nm process node (from TSMC 7 nm). This efficiency has allowed AMD to increase the overall TDP to 170 W from the previous 105 W but without too much penalty.
 
Last edited:

Bigos

Member
Jun 2, 2019
151
365
136
The 230W PPT setting would make sense for 24+ core SKUs, if they were to exist. Currently, 7950X @ 230W is like 5800X @ 142W - if you limit the latter to 88W (aka make it 5700X) you won't notice a thing.

Maybe AMD is planning on introducing such higher-core-count SKUs? But that might be in 2 years in the form of Zen 5...
 

zir_blazer

Golden Member
Jun 6, 2013
1,192
486
136
So AMD did the hard work even before adding the chipset connection, since the user can either connect 2x NVME PCIe 5.0 drives or 1x NVME PCIe 5.0 drive + 2x PCIe 4.0 drives hosted on an extension card. All one has to do is choose the right board for the job, since the second bundle of general purpose PCIe lanes will be used differently from one board model to another.
Using a PCIe 5.0 4x slot to get 2 PCIe 4.0 NVMe drives would require a PCIe Switch on such card and I can absolutely guarantee you than it will be too expensive to even consider it a valid option.


The chipset's purpose, or at least my interpretation in the circumstances, is to provide the needed amount of I/O that is comparatively much slower than PCIe 4.0, all those extra USB ports that may simply get filled with low bandwidth dongles and devices, some SATA drives or PCIe 3.0 drives meant for secondary storage, additional LAN or WiFi, a sound card or a capture card etc.

The only problem I can see with this approach is the user needs to plan in advance, they need to understand the strengths and weaknesses of the design and buy the motherboard configuration that fits their needs. The upside is you can get a lot more very fast storage connected straight to the CPU, and that includes some external mass storage device connected via Thunderbolt 4 / USB 4 while completely bypassing the chipset.
Modern Chipsets are pretty much I/O multiplexers because I/O is pin intensive and you would need a gargantuan socket to have all the connectivity coming from the Processor package itself. Point is, daisy chaining TWO Chipsets seems an awful idea when you could have used the new 4 lanes as another uplink. Also, consider than the downstream chiplet has a higher latency than if it was wired to a direct uplink.
I'm not expecting than anyone that actually needs that level of connectivity will not use several things simultaneously in such a way that the uplink gets bottlenecked. The B650 is fine, but the X670 as two daisy chained B650 is horrendous.

Also, while I didn't saw AM5 Motherboards specifications, I'm pretty sure than the most typical setup will be to see one 16x PCIe 5.0 Slot (Or two 8x/8x via bifurcation on high end boards), and two 4x PCIe 5.0 M.2 Slots for NVMe SSDs, but it will be rare to see one of those two ports in PCIe Slot form. So it means than most add in cards will be on the Chipset, and most likely you will need to go out of your way if you want a third Processor PCIe Slot. If the rest of the PCIe Slots are from the Chipset, more cards are adding a further bottleneck.
 

eek2121

Diamond Member
Aug 2, 2005
3,100
4,398
136
I may be wrong but, It seems DDR5 will not have significant bandwidth improvement for single CCD. It won't have double the bandwidth of DDR4 because CCD to IOD bandwidth hasn't increased much from Zen3 to Zen4. For me 230w is too much, when I buy one I would surely use it with ECO105 or ECO65. Has anyone tested how R9 7950X will perform if we simply use 125w air cooler? Will it throttle to something like 400MHz or will it throttle gradually to 125TDP and does not loose much performance? Also AVX512 is a surprise, it is performing way better than I expected. I would love to tryout some ASM coding with it.

Since the memory interfaces are on the IOD, The CCDs don't need the full bandwidth. They will never use it. IF speeds did improve, however. A lot of the 'lackluster' results with DDR5 relate to application optimization and a very slight latency regression. Desktop also wasn't really bandwidth constrained to begin with.
 
  • Like
Reactions: JustViewing

dnavas

Senior member
Feb 25, 2017
355
190
116
Threadripper 7000 'Storm Peak' leaked, looks like arrival earlier than expected?

I'm hearing that Zen4 is being bundled with RAM, so, something is happening with desktop sales. I like that AMD kept Zen4 prices mostly reasonable, but maybe even that wasn't enough? I guess people are looking for a deal. This shouldn't be a surprise with Intel coming out a mere month later; it's sensible for most people to just wait, but the motherboard tiering and pricing does seem to indicate that someone anticipated higher sales than I would have bet on.

Perhaps this is an attempt to move the chips to the server/Pro side? What I'm hearing is that purchasing plans for hardware are down, software is up, so I'm not sure Pro will move in quantity either. Maybe it's time for another half-hearted attempt to curry favor with non-Pro TR enthusiasts? Y'know, the market that's been starved for years now.... [tried but failed to keep the sarcasm at bay]

Could be interesting times. Oddly, I'm more interested to see if we get V3000 based products, though I'm surprised that that one doesn't seem to have an igpu (wat?) and I guess that's off-topic for here anyway.

More on-topic -- I'm surprised to see that 4x25Gbps isn't standard yet, and we're still playing with 4x10. Probably not the right forum audience, but :shrug:
 

maddie

Diamond Member
Jul 18, 2010
4,881
4,951
136
No, it's wrong because it's the opposite of all Zen behavior hither to. I'll reiterate my stance:
  1. It's the opposite of XMP which is opt-in for possible instability.
  2. It's the opposite of Zen 1-3 which were opt-in for inefficiency.
  3. Zen 4 is clearly optimized for far less power per core. Their own slides show this and its backed by benchmarks.
  4. It's the default because they didn't want to lose both 1T and MT to Raptor Lake. That's why Robert had to backtrack his 170W comment.
I don't blame AMD but it should be recognized as the Intel-like benchmark chasing it is. And I bought the 7950X because Computerbase and other competent reviewers did test at reasonable power limits.
Live a little, try something different. :)
 

A///

Diamond Member
Feb 24, 2017
4,351
3,158
136
Very interesting day! Good to see some competition. I'm already skeptical of some reviewers like TPU who have a delicate history of bias against AMD products that's been well documented online going back several years, including some bizarre hardware choices in past reviews. Outside of Intel possibly sending golden samples to reviewers which is likely given that it's Intel, Raptor Lake is what I expected it to be. It's not a bad upgrade if you're on an older Intel platform and prefer Intel. If you use a 12th gen K part with a Z690 motherboard, overclocking will be your goal here.

The 13900K is an excellent choice for those who cannot afford the AM5 platform and don't do much outside gaming and very little to no production. It's an excellent value choice and will surely earn its cred through 2024 when Intel's 14th gen comes out given the very low cost of DDR4 and DDR5 hardware compared to the premium AMD stack. Going forward, I expect future Intel platforms, such as the Z890, unless they abandon that format, to be as expensive as AM5 unless AIBs are subsidized, and Intel isn't in any financial position to undertake such a move. This is due in part to the complexity of a modern motherboard's traces and signal strength that needs to be upheld as we push new technologies as AMD has which drives up costs passed to the end consumer. There is a slight added premium with AMD's AM5 motherboards due to their longevity over the two year lifespan of a socket for Intel. I'm genuinely curious if there will be a pricing war anytime soon. I suspect Intel have little in the way of wiggle room, and any wiggle room available will be on the AMD side as they eat more into Intel's cake in data-center.

As for AMD. They're obviously still the premium product stack here save for their two overpriced lower end processors, the 7600X and 7700X. I do expect the 3D stacked processors to come out within the next 3-4 months, but alas my main gripe will be the price. The main issue with AM5 I'm seeing right now is the boot up times which is being addressed according to my industry contacts at the main motherboard vendors, but it's not the only issue being fixed or improved.

Bottom Line: If you can't afford a high end build or mid-range build on AMD, then Intel is the perfect cheap budget solution. If you only game and do very little production work, then the cheap budget 13600K and 13900K are perfect for you. If you game and do production or production work alone, the higher end AMD SKUs are better for you. Their premium status does make their prices and associated hardware eye watering. I'd hold tight if you're in no rush to see if AMD discounts them between now and post-CES.

Cooling down Intel's new budget solutions is another issue, but I expect some improvement in the near future. The issue mainly affects people who run their hardware in a case, because all these testers will have used an open bench to do their testing.

Relaying back to AMD 3D processors. There's been a wild 5800X3D processor making some headlines recently hitting higher frequencies. I suspect that processor is using the first generation cache on die and that AMD figured out their boost and voltage issue that caused the minor downclock and lockout on the 5800X3D. Though there's always the chance it's a second generation cache design that allows the processor to keep stock boost and voltage, and go even higher.

In such, I do believe Raphael X3D processors to maintain their clocks and stock voltages if not have more headroom. This would dampen whatever minuscule lead Intel has gotten with their toasty new generation. I can only guess when AMD may release their next generation Granite Ridge processors. COVID slowed production down world wide and almost made me redundant. I would not be shocked if AMD released Granite Ridge in 1H24. Is there any confirmation Granite Ridge is the next mainstream desktop processors following Raphael? I don't know of an artist named Granite Ridge. By then I suspect AMD will have anywhere from 4-7 months of sales before Intel releases their 14th gen products. AMD AM5 X770 motherboards will be out and X670 will become a prime choice for the budget conscious offering everything they need. I can't see what X770 would improve on over X670. Perhaps a fellow old fogy would be inclined to enlighten me.

As far as the lapping is concerned: if you have reliable delidding tool, delidding may be safer, assuming you can still maintain mount pressure with your chosen cooler/water block. All those metal shavings are a nightmare to deal with.
Jason (J2C) did a video yesterday or the day before on lapping. He took off less than 1 mm using a tool that Hartung made that clipped the CPU inside of a vinyl resin case. Overall using normal everyday paste (MX-4) he got it to drop down to the 88-90 at peak during Cinebench. Using Hartung's Kryonaut solution it came down more. If you're seeking lower temperatures then delid, but I'm not sure if it's worth doing so chasing after 100 Mhz more boost at stock and chancing it with overclocking.



For now I'll stick with my 10th generation Intel platform which I've come to deeply love. It does everything I want of it. I don't expect to upgrade anytime in the next 5 to 7 years. These recent releases give me nightmares back to running Pentium 4s and I'd rather not touch either chipmakers products at this time. I don't want to deal with very hot Intel systems throttling down and don't want to touch AMD due to cost and whatever bugs may come from AM5. Just the expense of current new parts makes it a tough pill to swallow. Says the person who spent a fortune on their 10th gen due to COVID.
 
Last edited:

Geddagod

Golden Member
Dec 28, 2021
1,296
1,368
106
Also the small issue of Meteor Lake very likely being delayed until mid or late 2024. No one is releasing a major processor in 2023 except minor refreshes like X3D or the KS SKUs. RPL may hold the small lead they have in production but will likely lose the gaming crown to the X3D processors. No idea what SKUs will get it. I believe that AMD will have put enough effort into them to put a large gap between RPL and those X3D SKUs that it may cause Intel users that mostly game to question their purchases.
Lmao what since when was MTL being delayed until mid or late 2024 "very likely"
I agree I doubt nothing major will release in 2023 though other than X3D
I also highly doubt AMD would take the gaming crown by a margin >10%, unless they improve x3d tech drastically, and if they do take the crown, they will most likely price that at a premium.
I doubt RPL buyers who just game will be questioning their purchases, unless they bought the 13900k for just gaming, because they prob would have gotten their platforms for much cheaper than what a potential x3d sku would have been released at. On top of not having to have waited for a x3d launch. And if you bought a 13900k for just gaming, I have no sympathy for you regretting your purchase regardless hahah
 

LightningZ71

Golden Member
Mar 10, 2017
1,792
2,151
136
The 7600x and 7700x are in the stack to give gamers something approachable cost wise for AM5 to help with platform volume. Looking at the benchmarks, there isn't a game out there that they can't give you a good framerate on at any resolution with the appropriate video card. Lets not kid ourselves, with consoles having 8 cores, with one or two of them likely reserved for console housekeeping, there isn't going to be a game produced anytime soon that is going to absolutely require more than 6 physical cores to run well enough. They may run slightly better with a few more, but nothing will be unplayable with a 7600x.

What you get when you buy that 7600x is a platform that is advertised byvthe manufacturer to see at least two more full generation upgrades available, taking you through a good 4 years or more of being relevant. That's not what Intel is offering for just a little less money overall for low end boards.
 

Hans Gruber

Platinum Member
Dec 23, 2006
2,302
1,216
136
I think it's time for AMD to start making Zen 4 CPU's on the AM4 platform. When people say that is not possible. AMD promised that AM4 would be supported for 5 years. Then they said Zen 3 wasn't compatible with B350/x370 boards and then suddenly it was.

It's the one thing AMD can do to save their slowly sinking ship. A drop in processor option would greatly help Zen 4 sales. Intel's next CPU will be on a true 7nm process. 5nm has not done much for AMD performance. I don't know if 3nm will bring more performance other than energy efficiency for Zen 5.

Zen 4 is looking more like a Zen 3 on 5nm with a much higher TDP than Zen 3. I wonder how a Zen 3 would perform against Zen 4 with the same TDP. What Zen 4 lacks in IPC they gain in higher power consumption than Zen 3. People can argue the power efficiency gains going from 7nm to 5nm offsets the increased power of Zen4. I know that is the Intel way but AMD should have done a better job with Zen 4.

Obviously Zen 4 will continue to dominate the server market. I am wondering if that will change when Intel has 7nm silicon. Will AMD still hold a server market lead after Zen 4.

I thought 5nm would mean AMD would have a 24 or 32 core Zen 4 consumer CPU.
 
Jul 27, 2020
19,823
13,588
146
Whoever is or isn't "desperate" has little to do with when Raphael-X will launch.
You don't find the pairing of 7900 XTX and 7950X ridiculous? AMD is telling gamers that this is the premium gaming experience. Typical gamer sees the marketing and ads and buys it. Few months later, AMD releases X3D. Now gamer sees his "premium" gaming machine getting beaten by relatively cheaper DIY desktops with X3D. Do you really expect the gamer to not utter an expletive at that point? It's just a really bad move by AMD.
 
Jul 27, 2020
19,823
13,588
146
Zen 4 gets slightly faster with AMD's optimized compiler: https://www.phoronix.com/review/amd-aocc-4

Additionally, I personally continue to be a bit perplexed why AMD waits until the later EPYC launch before posting AOCC 4.0 rather than striking initially when the Ryzen 7000 series first debuted.

Heck why not ship it earlier as AOCC "beta" for Ryzen testing? I imagine this too comes down to engineering resources and needing all the time they can get to work out the compiler optimizations.

Yes, AMD. Why are you so tardy? GIve your software people more resources.
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
Holy cherry picking 😂
As if a 1/54th outlier would change the outcome that much (HU already covered it anyway).

As I said above, HZD is also another one.

How are you going to spin the 5950x beating a 13900K+DDR5? I realize that this is an AMD thread and even the entire forum leans AMD, but even you have to admit that makes no sense.

Both CPUs are x86-64 so you can't pin it down to optimization issues either.

Edit: BFV and BF2042 may use the same family of engines, but they're not the same are they? Codemaster's F1 and DIRT games also use the same family of engines between them, that doesn't mean we don't see wild variance in results between AMD, Intel and nVidia GPUs and CPUs. This is a non-argument.

BF2042 is much more CPU demanding and has a later version of the Frostbite engine. BF5 can achieve much higher FPS than BF2042 due to being less demanding.

What's next? Factorio and MMOs are banned in V-Cache testing? E-Sports titles banned on AMD testing?

Factorio has an explanation in that it runs primarily out of the cache, which is why the 5800x3D is so dominant.

TPU just did a E-core on v off big round-up, and Raptor Lake performs virtually the same in games in both situations.

The E cores is a possibility, but such a big discrepancy looks to me more like a configuration problem or a bug in the game itself.

And if those "arguments" fail, guess you'll just rant on how if you use $400+ 7600MT kits, 13600K would be faster or something😊

I realize you're being flippant, but this is just an observation by me because it obviously doesn't add up when you think about it. In HZD, a 5950x with less IPC, DDR4 and lower clock speeds is beating the fastest gaming CPU that is currently available.

While ignoring that the kit they did the 13600K 54 game tests on would make the 7600X even faster in the first place 😁

I don't own a Zen 4 CPU, but even I know that Zen 4 runs best in a 1:1 gear mode, which makes DDR5-7600 useless for Zen 4. :rolleyes: