9950X and 9950X3D is going to be terrific.

LukeTbk said:
From a quick read, people yet to update, depending on how iron out the memory controller and the little quirk (like slow post if you OC the ram) could be happy to have waited, but the x870 does not seem enough to be worth the update trouble if you do not have major issue with your current x670 kit for most people (if it allow 4x32, 4x48 to run well or 2x kit at 8000mhz maybe it will to some).

Which is part a bit of a deception but also kind of nice, that what make jumping into a new AMD platform an easy choice, faith you will not "need" (or more so want) to upgrade that board.

View attachment 657826

If we compare with TPU result of the 7950x vs the 14900k, at 720p 4090RTX (i imagine they are using a 7900xtx and different setting-drivers-game version..., so really gross comp)

7950x ->9950x performance relative to the 14900k

Boderlands3 : -23% ->.+4% (+35% boost)
Cyberpunk...: .+6% ->+13% (+20% boost)
CinebenchR24:-1.6% ->+21% (+23% boost)

Those seem good, +20/25% performance boost (+15% IPC, 7-8% higher boost)
Click to expand...

Solid bump but definitely not worth it for gamers who are able to get 7800X3D for far less money. Let's say we give the 9700X the best case scenario, a 25% boost over the 7700X. Here is the latest CPU benchmark data from TPU:

1717529703597.png


88 x 1.25 = 110 for 9700X, meanwhile
7800X3D = 105

The 9700X would barely be any faster than a 7800X3D while costing at least $400 on it's own, then factor in mobo and RAM costs would easily put you north of $600-700. A 7800X3D bundle can be bought from Microcenter right now for $480 that includes CPU/Mobo/RAM. If you do not have access to such cheap deals then I guess the argument is a lot harder to make. But also that was the best case scenario of 25% boost, if it is a 16% boost instead then the numbers would completely be in favor of the 7800X3D, cheaper AND faster.

88 x 1.16 = 102
 
MistaSparkul said:
Solid bump but definitely not worth it for gamers who are able to get 7800X3D for far less money.
Click to expand...
Almost for sure, waiting to see the 9800x3d was almost certain regardless of the regular edition would have do here.
 
Does anyone here have a 7900XTX and the latest Avatar game? If so can you provide a score using the in game benchmark at high settings at 1080p? TPU has a benchmark score for that running on a 9900X. AVG FPS is 229:

https://www.techpowerup.com/323185/hands-on-with-the-amd-ryzen-9-9950x-zen-5-desktop-processor

EDIT: Nevermind that score is completely worthless since Frame Generation is enabled. It's obvious because a 7900XTX using MEDIUM settings + FSR Quality can only get 171 fps in this game so it would be impossible to score 229fps on HIGH without the use of FG. Unless they used FSR Ultra Performance without FG to render at 360p internally then I suppose that's possible.

1717534019662.png
 
Last edited:
So we can't use the TPU data to try and gauge the performance, instead I've found that out of the 6 games AMD has chosen to compare the 14900k to the 9950x against, 4 of them have been benchmarked by Techspot against the 7800X3D.

1717539825455.png


I made some quick comparisons using the 14900K vs 7800X3D benchmark from Techspot, and this slide from AMD so we can try to see where the 9950X lands relative to the 14900K and 7800X3D.

1717539895018.png


14900K: 169fps
7800X3D: 189fps
9950X: 14900K + 13% according to AMD = 190fps, tie against the 7800X3D

1717539972716.png


14900K: 243fps
7800X3D: 272fps
9950X:14900K + 23% according to AMD = 299fps, 10% faster than 7800X3D

1717540058996.png


14900K: 230fps
7800X3D: 236fps
9950X: 14900K + 16% according to AMD = 266fps, 12.7% faster than 7800X3D

1717540140685.png


14900K: 282fps
7800X3D: 265fps
9950X: 14900K + 6% according to AMD = 307fps, 15.8% faster than 7800X3D

Of course without knowing the exact settings AMD were using, we can't say for certain that this will be the performance of the 9950X. AMD could have been using 720p lowest settings instead of 1080p ultra like Techspot for all we know.
 
X3D as well, 8 core first then 16+ as systems get built
 
MistaSparkul said:
Of course without knowing the exact settings AMD were using, we can't say for certain that this will be the performance of the 9950X. AMD could have been using 720p lowest settings instead of 1080p ultra like Techspot for all we know.
Click to expand...
Indeed. They have fudged things a bit, in the past. Like running their Intel numbers with JEDEC RAM settings for Intel stock speeds-----but, using faster than stock, for their own Ryzen numbers.

That said, I'm pretty optimistic about Zen 5. The rumors and leaks have been touting some pretty nice IPC gains, latency improvements, AVX improvements, etc, for months. And this aligns pretty well with that. The productivity slide has gigantic gains for handbrake and blender, for example.
 
So did AMD use the newer Intel settings for 14000 series or the motherboard toaster settings? I would recommend actual reviews on these processors.

As for putting VCache on both two CCD processors, in many if not most multi-threaded applications there is a performance loss due to reduced frequencies with VCache, unless Zen 5 VCache chips can hit the same frequencies, it would be a performance loss in general for many applications and little if any improvement for games.
 
noko said:
So did AMD use the newer Intel settings for 14000 series or the motherboard toaster settings? I would recommend actual reviews on these processors.

As for putting VCache on both two CCD processors, in many if not most multi-threaded applications there is a performance loss due to reduced frequencies with VCache, unless Zen 5 VCache chips can hit the same frequencies, it would be a performance loss in general for many applications and little if any improvement for games.
Click to expand...
Review embargoes would have to end first. These are just preliminary marketing announcements.

The vcache on both CCDs has been discussed ad nauseam in this thread and other threads.
 
chameleoneel said:
Indeed. They have fudged things a bit, in the past. Like running their Intel numbers with JEDEC RAM settings for Intel stock speeds-----but, using faster than stock, for their own Ryzen numbers.

That said, I'm pretty optimistic about Zen 5. The rumors and leaks have been touting some pretty nice IPC gains, latency improvements, AVX improvements, etc, for months. And this aligns pretty well with that. The productivity slide has gigantic gains for handbrake and blender, for example.
Click to expand...

They've done that before? Yikes. The last time I remember AMD misleading people pretty hard was when they publicly stated that the 7900XTX was going to be some crazy 60 or 70% faster than a 6950XT when in reality it ended up being only like 35% faster.
 
MistaSparkul said:
They've done that before? Yikes. The last time I remember AMD misleading people pretty hard was when they publicly stated that the 7900XTX was going to be some crazy 60 or 70% faster than a 6950XT when in reality it ended up being only like 35% faster.
Click to expand...
Yes, they have. They compared a 12900k with DDR5 5200, to a 5800X3d with DDR4 3600 (JEDEC/stock speed is 3200).
And then they compared a 12900k with DDR5 6000 to pre-production Zen 4 with DDR5 6400 (most Zen 4 have trouble being stable at 6400. especially before the big bios update for memory compatibility around July 2023).
You can see it here in the "footnotes" of their previous slides:
https://cdn.mos.cms.futurecdn.net/XKtx5sguyCfx3BnHpZC9PV.jpg

I finally found a site which posted the "endnotes" for AMD's Computex 2024 slides and actually.....this comparison seems pretty equal: DDR5 6000 for both the 9950X and the 14900K. and most everything else seems equal, as well.

https://cdn.mos.cms.futurecdn.net/gfZtZrdBfk92wvvDz5Kdm9.jpg

*top right paragraph for the gaming performance notes. and bottom left for the productivity notes.
 
Last edited:
I'm interested to see where these go and I'm glad for the announcement - Zen5 seems a step forward nicely, but until we see the 3DCache chips and their release window, it will be hard to assess the entire line vs both the previous AMD CPUs and Intel's current and upcoming. I can only hope they're not going to wait until either late in the year or next year to debut the 3D V-Cache high ends, especially if they're going to stick with the lower frequency, heterogeneous CCD configuration from Zen4. Though ideally, they've finally come up with an enhanced process that allows full frequency/thermals for the Vcache bonding and can apply it on multiple CCDs.

Also, as an aside... I do wish they woud have updated a chipset like X870E to allow for quad channel memory and more PCI-E lanes! Sure, let the A and B chipsets stick to dual, allow X to go either way at manufacturer discretion, and put X-E's highest end to all be quad channel RAM capable. These new Zen5 chips are using more memory bandwidth than ever and AMD seems to get an uplift from doing so thus, especially given how Threadripper/PRO has become "Jr Server, highly paralleized,expensive" rather than "HEDT multi use enthusiast" the way it was in the 1000-3000 era, it seems beneficial. For PCI-E lanes it seems like more users than ever are making use of more lanes thanks to ubiquitious PCI-E storage and other content, so why not do like old HEDT and give some extra lanes, perhaps varying in both the chipset and/or the CPU itself (ie the way that the Intel 5820K had fewer lane than the 5930K and 5960X)? We've been stuck on dual channel and relatively limited PCI-E lanes for ages and both upgrades would directy improve AMD' performance and feature set. AMD should remember how their road to Ryzen success came in breaking the near decade long Intel precedent of "4 cores, 8 threads if you're lucky" , they could take another step forward here.
 
RanceJustice said:
Also, as an aside... I do wish they woud have updated a chipset like X870E to allow for quad channel memory and more PCI-E lanes! Sure, let the A and B chipsets stick to dual, allow X to go either way at manufacturer discretion, and put X-E's highest end to all be quad channel RAM capable.
Click to expand...

The memory controller is on die and is a limitation of the CPU not the chipset.
 
ND40oz said:
The memory controller is on die and is a limitation of the CPU not the chipset.
Click to expand...
Oh I know, that would have to be updated as well. I meant that the board chipset families would be a way to delineate support for certain features v price , the way that for Intel's Alder Lake (I think) you had some DDR4 and other DDR5 supporting boards, or how there were ITX boards years ago on X99 chipsets that only supported dual channel because they only had 2 physical RAM slots etc
 
RanceJustice said:
I do wish they woud have updated a chipset like X870E to allow for quad channel memory and more PCI-E lanes!
Click to expand...
I'd like to see quad channel memory too, but it would require more pins, and thus a new CPU socket and even more expensive motherboards. I think moving to CAMM might be more practical, as it should allow for higher bandwidth while maintaining socket compatibility.

I doubt we'll see more PCIe lanes, as the number of people who can make use of them is very small.
 
Mr Evil said:
I'd like to see quad channel memory too, but it would require more pins, and thus a new CPU socket and even more expensive motherboards. I think moving to CAMM might be more practical, as it should allow for higher bandwidth while maintaining socket compatibility.

I doubt we'll see more PCIe lanes, as the number of people who can make use of them is very small.
Click to expand...

Same with additional ram bandwidth, really. There are extreme diminishing returns over ddr5 6000 in almost all apps, so really another niche use case that Threadripper or Epyc solve.
 
chameleoneel said:
Yes, they have. They compared a 12900k with DDR5 5200, to a 5800X3d with DDR4 3600 (JEDEC/stock speed is 3200).
And then they compared a 12900k with DDR5 6000 to pre-production Zen 4 with DDR5 6400 (most Zen 4 have trouble being stable at 6400. especially before the big bios update for memory compatibility around July 2023).
You can see it here in the "footnotes" of their previous slides:
https://cdn.mos.cms.futurecdn.net/XKtx5sguyCfx3BnHpZC9PV.jpg

I finally found a site which posted the "endnotes" for AMD's Computex 2024 slides and actually.....this comparison seems pretty equal: DDR5 6000 for both the 9950X and the 14900K. and most everything else seems equal, as well.

https://cdn.mos.cms.futurecdn.net/gfZtZrdBfk92wvvDz5Kdm9.jpg

*top right paragraph for the gaming performance notes. and bottom left for the productivity notes.
Click to expand...

Hmm DDR5 6000 for the 14900K eh? Techspot was using DDR5 7200 so in this case AMD would be lowballing the 14900K's performance by a few %. From Techspot's comparison, the 7800X3D is average 5% faster than the 14900K when it is paired to 7200Mhz RAM so if it was paired to 6000 MHz RAM instead then I would imagine the lead would probably stretch out to maybe 7 or 8%?

1717607156358.png


So for the 9700X to match the 7800X3D it would need to be at least 8% faster than the 14900K on average. My final guess is it's going to end up around 5% faster than a 7800X3D at the end of the day.
 
MistaSparkul said:
Hmm DDR5 6000 for the 14900K eh? Techspot was using DDR5 7200 so in this case AMD would be lowballing the 14900K's performance by a few %. From Techspot's comparison, the 7800X3D is average 5% faster than the 14900K when it is paired to 7200Mhz RAM so if it was paired to 6000 MHz RAM instead then I would imagine the lead would probably stretch out to maybe 7 or 8%?

View attachment 658055

So for the 9700X to match the 7800X3D it would need to be at least 8% faster than the 14900K on average. My final guess is it's going to end up around 5% faster than a 7800X3D at the end of the day.
Click to expand...
Yes, Hardware Unboxed/Techspot, use timings/sub-timings provided by Buildzoid. DDR5 6000 for Zen 4 and DDR5 7200, for Intel 13th and 14th gen. They also use a 4090. AMD used a 7900 XTX.
 
bobzdar said:
Same with additional ram bandwidth, really. There are extreme diminishing returns over ddr5 6000 in almost all apps, so really another niche use case that Threadripper or Epyc solve.
Click to expand...
I grant that heavy users of both would be in the minority, but especially given how Threadripper and HEDT is not what it once was in term of affordability, features,or overall use case, having certain features (just like core counts above 8) peeled off into Ryzen seems to be helpful. Its my understanding that memory bandwidth (and other parameters) can have real benefit for overclocking the CPU and overall performance, so even if the hardcore edge case tweakers are in the minority I'd think going quad channel would help. As far as diminishing returns over 6000mhz is that still accurate? My understandig wass that it was the "sweet spot" for the 7000 series especially as attempting to go over 6400mhz vastly increased the chance of problems. However, for 9000 series and with the X870E and related chipsets, I thought one of the bigger features was support for faster RAM timings going up to DDR5 8000 (or even higher, from Gskill's Computex pics), but admittedly I'm unsure where the line of diminishing returns is and how it varies on Zen5 vs earlier. In any event, I'd certainly like the option to support up to quad channel.

As far as PCI-E, I imagine its less niche than ever now, isn't it? Thanks to proliferation of PCI-E M.2 SSDs, and how higher end boards (ie X670E or upcoming X870E) differentiate themselves in part by having more lanes, PCI-E 5.0 supported lanes, and increassing onboard support for more PCI-E M.2 devices, sometimes in exceess of 4-6 of them on high end boards. Of course, even on X670E boards if you actualy make use of more of these M.2 sockets, there's usually a tradeoff. Oh, if you use M.2_1, then PCI-E x16_2 will onyl work at x8 maximum and so on. Not all slots are PCI-E 5.0 compatible too. You'd think with al this in mind, having more lanes of the latest generation would be a benefit so that users could use a full compliment of drives M.2 without hobling their limited PCI-E slots or other functionality. In the old days, if you were using many lanes it was pretty much a CrossFireX/SLI setup or a handful of other slot-heavy expansion cards, but today its a lot less niche to make use of PCI-E devices aside from just a video card
 
Nevermind I take back my prediction after watching this video:
View: https://www.youtube.com/watch?v=MpM53KgVNT0

If you remove the Avatar result since it's GPU limited and not CPU limited, the 7800X3D is on average a whopping 32% faster than the 7700X. There is just no way the 9700X is going to be 32%+ faster than a 7700X so it will end up slower than the 7800X3D overall.
 
RanceJustice said:
I grant that heavy users of both would be in the minority, but especially given how Threadripper and HEDT is not what it once was in term of affordability, features,or overall use case, having certain features (just like core counts above 8) peeled off into Ryzen seems to be helpful. Its my understanding that memory bandwidth (and other parameters) can have real benefit for overclocking the CPU and overall performance, so even if the hardcore edge case tweakers are in the minority I'd think going quad channel would help. As far as diminishing returns over 6000mhz is that still accurate? My understandig wass that it was the "sweet spot" for the 7000 series especially as attempting to go over 6400mhz vastly increased the chance of problems. However, for 9000 series and with the X870E and related chipsets, I thought one of the bigger features was support for faster RAM timings going up to DDR5 8000 (or even higher, from Gskill's Computex pics), but admittedly I'm unsure where the line of diminishing returns is and how it varies on Zen5 vs earlier. In any event, I'd certainly like the option to support up to quad channel.

As far as PCI-E, I imagine its less niche than ever now, isn't it? Thanks to proliferation of PCI-E M.2 SSDs, and how higher end boards (ie X670E or upcoming X870E) differentiate themselves in part by having more lanes, PCI-E 5.0 supported lanes, and increassing onboard support for more PCI-E M.2 devices, sometimes in exceess of 4-6 of them on high end boards. Of course, even on X670E boards if you actualy make use of more of these M.2 sockets, there's usually a tradeoff. Oh, if you use M.2_1, then PCI-E x16_2 will onyl work at x8 maximum and so on. Not all slots are PCI-E 5.0 compatible too. You'd think with al this in mind, having more lanes of the latest generation would be a benefit so that users could use a full compliment of drives M.2 without hobling their limited PCI-E slots or other functionality. In the old days, if you were using many lanes it was pretty much a CrossFireX/SLI setup or a handful of other slot-heavy expansion cards, but today its a lot less niche to make use of PCI-E devices aside from just a video card
Click to expand...

I don't think so because that kind of massive high speed storage is, in itself, a niche use case. I always ask in here when people say they need all that high speed storage, why they need it. It's a lot of 'hosting my media collection' stuff which is really really niche, spending $5k on hardware to store and host movies to save a few bucks each in streaming rental costs makes no sense to me. Most users are fine with a single high speed m.2 SSD and then slower (but still fast) SATA SSD's for additional storage. Even then, you can usually run 2 high speed m.2 drives even in B series chipsets. Having an array of 4-8 m.2 drives makes zero sense for 99+% of users, that's why that stuff is all in HEDT. I think HEDT is actually better than it was before in terms of features, the TR 7000 PRO series is kind of insane in what it supports. 96 cores, 8 channel ram, 144 pcie lanes...it's nuts. TR 7000 is a little less nuts with 'only' 64c, quad channel and 80 lanes, but still probably hits 99% of the 1% niche use cases. Sure, you have to pay for it, but for those that make their money off of it it's not hard to justify. For hobby use sure, that's hard to justify, and honestly there's not much an x870E with twin high speed m.2's, 16 cores and a single 4090 (or 5090) can't do amazingly well.
 
bobzdar said:
I don't think so because that kind of massive high speed storage is, in itself, a niche use case. I always ask in here when people say they need all that high speed storage, why they need it. It's a lot of 'hosting my media collection' stuff which is really really niche, spending $5k on hardware to store and host movies to save a few bucks each in streaming rental costs makes no sense to me. Most users are fine with a single high speed m.2 SSD and then slower (but still fast) SATA SSD's for additional storage. Even then, you can usually run 2 high speed m.2 drives even in B series chipsets. Having an array of 4-8 m.2 drives makes zero sense for 99+% of users, that's why that stuff is all in HEDT. I think HEDT is actually better than it was before in terms of features, the TR 7000 PRO series is kind of insane in what it supports. 96 cores, 8 channel ram, 144 pcie lanes...it's nuts. TR 7000 is a little less nuts with 'only' 64c, quad channel and 80 lanes, but still probably hits 99% of the 1% niche use cases. Sure, you have to pay for it, but for those that make their money off of it it's not hard to justify. For hobby use sure, that's hard to justify, and honestly there's not much an x870E with twin high speed m.2's, 16 cores and a single 4090 (or 5090) can't do amazingly well.
Click to expand...
Personally I am disappointed by the limited number of lanes, which will translate to boards with relatively few expansion slots. Even my now aging TR3 setup had plenty of lanes, yet surprisingly few slots. I typically have a GPU, a 25GbE or 40GbE card, a video capture card (think HD SDI) and typically a high accuracy sound input as well. It's not unusual that a 5th random thing like a MIDI output ends up in there as well. Doing these sorts of things was trivial 15 years ago but almost impossible now, as MBs increasingly assume one GPU and the rest of the board space gets allocated to an ever increasing number of M.2 slots. Further complicating this is the sheer size of modern GPUs that often block one or more slots.

I'd love a 9950X that had 48 or 60 lanes allocated across 6 or 7 slots, and just one or two M.2 slots.
 
sphinx99 said:
Personally I am disappointed by the limited number of lanes, which will translate to boards with relatively few expansion slots. Even my now aging TR3 setup had plenty of lanes, yet surprisingly few slots. I typically have a GPU, a 25GbE or 40GbE card, a video capture card (think HD SDI) and typically a high accuracy sound input as well. It's not unusual that a 5th random thing like a MIDI output ends up in there as well. Doing these sorts of things was trivial 15 years ago but almost impossible now, as MBs increasingly assume one GPU and the rest of the board space gets allocated to an ever increasing number of M.2 slots. Further complicating this is the sheer size of modern GPUs that often block one or more slots.

I'd love a 9950X that had 48 or 60 lanes allocated across 6 or 7 slots, and just one or two M.2 slots.
Click to expand...

That's TRX50 man. You can get the Asus SAGE that will do what you want, you might need to use a single riser cable if you have a 3 slot GPU to get the 5th PCIE slot you need but that's fairly trivial to do.
 
HEDT is dead and doesn't have any sign of coming back soon. You're either consumer or workstation.

Multi-GPU has been dead for years now, so of course motherboard manufacturers are going to cater to what benefits the majority of consumers. You can get a M.2 to PCI-E adapter (doesn't really work for rear M.2 slots).
 
Tsumi said:
HEDT is dead and doesn't have any sign of coming back soon. You're either consumer or workstation.

Multi-GPU has been dead for years now, so of course motherboard manufacturers are going to cater to what benefits the majority of consumers. You can get a M.2 to PCI-E adapter (doesn't really work for rear M.2 slots).
Click to expand...

TRX50 = HEDT, no? WRX80 (ie PRO) is Workstation.
 
bobzdar said:
TRX50 = HEDT, no? WRX80 (ie PRO) is Workstation.
Click to expand...
There would not be some clear definition of what people mean by high end desktop.

For some HEDT mean high single thread performance money can buy with extras, think the -XE series intel cpu, X299 chipset, you had more cores than a 10900x but clocked has high-higher, quad channel memory support, way more lanes, etc.... outside money you pay zero cost for those extra.

TRX50 and WRX80 (or sapphire lake) would be more classed workstation and more expensive workstation to some than HEDT vs workstation, both aiming at the workstation market.
 
LukeTbk said:
There would not be some clear definition of what people mean by high end desktop.

For some HEDT mean high single thread performance money can buy with extras, think the -XE series intel cpu, X299 chipset, you had more cores than a 10900x but clocked has high-higher, quad channel memory support, way more lanes, etc.... outside money you pay zero cost for those extra.

TRX50 and WRX80 (or sapphire lake) would be more classed workstation and more expensive workstation to some than HEDT vs workstation, both aiming at the workstation market.
Click to expand...

Yeah, I guess the line is a bit blurry. I think if AMD released a 7000 (or 9000) series threadripper with a vcache ccd, that would be pretty clearly an HEDT offering as it wouldn't be catering to just productivity/work.
 
bobzdar said:
TRX50 = HEDT, no? WRX80 (ie PRO) is Workstation.
Click to expand...
I consider them both to be workstation because the pricing is so far removed from consumer pricing. HEDT would be like in the X58 and X79 days where getting into the platform was only $200-400 more. Getting into TRX50 is at least $1200 more.
 
I currently have a 5800X...I mostly use my PC for gaming...I've been waiting for Zen 5 to upgrade but am wondering if the 9800X (or whatever the 5800X equivalent will be named) is worth getting or is it best to get the 7800X3D?...I don't have the patience to wait until February 2025 for the 9000 series 3D V-Cache models
 
polonyc2 said:
I currently have a 5800X...I mostly use my PC for gaming...I've been waiting for Zen 5 to upgrade but am wondering if the 9800X (or whatever the 5800X equivalent will be named) is worth getting or is it best to get the 7800X3D?...I don't have the patience to wait until February 2025 for the 9000 series 3D V-Cache models
Click to expand...

You already waited this long but don't want to wait for the 9800X3D?
 
MistaSparkul said:
Well based on the video that I posted earlier:
View: https://www.youtube.com/watch?v=MpM53KgVNT0

I really don't think Zen 5 is going to beat the 7800X3D in gaming. Sure it will win in titles where VCache doesn't do much, but a lot of titles do seem to benefit from VCache and some benefit HUGELY. I would just get the 7800X3D.
Click to expand...


I think the 9800X (or whatever the 7800X equivalent is named) will come really close or be slightly faster...but it still makes the case for getting a 7800X3D...seems like the significant Zen 5 feature is USB 4 support and I'm not sure that really matters much to me at this point
 
Last edited:
polonyc2 said:
I think the 9800X (or whatever the 5800X equivalent is named) will come really close or be slightly faster...but it still makes the case for getting a 7800X3D...seems like the significant Zen 5 feature is USB 4 support and I'm not sure that really matters much to me at this point
Click to expand...

It'll be close while costing quite a bit more. The 7800X3D has been out for over a year already and now have some good discounts on it, while Zen 5 would easily cost $400 just for the CPU alone.
 
sphinx99 said:
Personally I am disappointed by the limited number of lanes, which will translate to boards with relatively few expansion slots. Even my now aging TR3 setup had plenty of lanes, yet surprisingly few slots. I typically have a GPU, a 25GbE or 40GbE card, a video capture card (think HD SDI) and typically a high accuracy sound input as well. It's not unusual that a 5th random thing like a MIDI output ends up in there as well. Doing these sorts of things was trivial 15 years ago but almost impossible now, as MBs increasingly assume one GPU and the rest of the board space gets allocated to an ever increasing number of M.2 slots. Further complicating this is the sheer size of modern GPUs that often block one or more slots.

I'd love a 9950X that had 48 or 60 lanes allocated across 6 or 7 slots, and just one or two M.2 slots.
Click to expand...
Yup this is why I chose the ROG X670E Extreme board, one of the few with three PCIe slots, I'm using a Marian Dante sound card and a Pro Tools HD Native card, plus I'm probably getting a Sonnet Thunderbolt 3 PCIe chassis to put those two cards in so I can free the slots up for other things and it has the 10 GbE built in. It's pricey but I don't think there is a single other X670E mobo out there with all that connectivity and it still allows for 4 M.2 SSDs to be connected. Every generation is seeing less and less connectivity thinking people are only going to connect a GPU and that's it.
 
polyh3dron said:
Yup this is why I chose the ROG X670E Extreme board, one of the few with three PCIe slots, I'm using a Marian Dante sound card and a Pro Tools HD Native card, plus I'm probably getting a Sonnet Thunderbolt 3 PCIe chassis to put those two cards in so I can free the slots up for other things and it has the 10 GbE built in. It's pricey but I don't think there is a single other X670E mobo out there with all that connectivity and it still allows for 4 M.2 SSDs to be connected. Every generation is seeing less and less connectivity thinking people are only going to connect a GPU and that's it.
Click to expand...
Well, I am glad to know that I’m not the only person using more than one PCIE slot in the 2020s.
 
I'll want to use all three M.2 Slots. Then the x16 for the GPU and then another x4 whatever for my P4800X Optane drive thats mounted in a PCIe adapter.

Should be fine. Would be nice to be able to use another M.2 to PCIe adaprter as well. Even a PCIe 4 x1 slot would do.
 
daglesj said:
I'll want to use all three M.2 Slots. Then the x16 for the GPU and then another x4 whatever for my P4800X Optane drive thats mounted in a PCIe adapter.

Should be fine. Would be nice to be able to use another M.2 to PCIe adaprter as well. Even a PCIe 4 x1 slot would do.
Click to expand...
How sturdy are those?
 
You must log in or register to reply here.
Back
Top