Intel Core i9-13900K Raptor Lake CPU Offers Same Performance As Core i9-12900K With “Unlimited Power” at Just 80W

TSMC is having no issues with 5nm. Besides, AMD is announcing X3D at CES. Pretty sure those chips are already in production.

Not sure it will be enough to increase sales. Time will tell.
Prices, prices, prices. That's the key. Seems the already high price of the 7XXX series will increase when they go X3D. I suspect they will be very limited release.
 
TSMC is having no issues with 5nm. Besides, AMD is announcing X3D at CES. Pretty sure those chips are already in production.

Not sure it will be enough to increase sales. Time will tell.
I'll be happy to see it. I'm still a few years away from my usual platform upgrade cycle, but a 3D cache version of Zen 4 may get me to upgrade earlier.
 
In short, a move to AMD 7xxx will cost you more and deliver less in many cases. Well done Intel, taking advantage of AMD's radical platform move. Thus, for many, if considering an upgrade, Intel becomes the better overall value which many switch AMD fans over to Intel (again?). We'll see.

With that said, and there's nothing definitive on this, if AMD does significantly lower the prices of the 7xxx series (essentially flushing/destroying gens prior in the channel though), AMD might become the value proposition again. GamersNexus suggested maybe at or before CES??

Yea its only a better value if you can drop these in. It makes zero sense to build a whole new system that is on a dead socket. This is second gen on same socket so going by intel history you got new socket coming next gen. Next time around AMD will be better value because of x670e and zen4 x3d chips.

There is value in AMD if building a new rig, there is value in raptor lake if you are dropping in. Right now raptor lake will sell more because of that. Next year you will have people buying the zen4 3d chips like hotcakes. Its a cycle. I think amd will drop prices on these when they announce some of those chips. I really doubt people on zen3 are going to go all out on zen 4 or Raptor lake right now. But next year I expect amd too move alot, as board prices normalize and they announce zen4 x3d chips.
 
Prices, prices, prices. That's the key. Seems the already high price of the 7XXX series will increase when they go X3D. I suspect they will be very limited release.
no it wont be a limited release. 5800x3d is readily available and that was their test baby. Next ones will sell like hot cakes and you can guarantee amd pumps out as much as they can.
 
Prices, prices, prices. That's the key. Seems the already high price of the 7XXX series will increase when they go X3D. I suspect they will be very limited release.
BTW I dont think X3D versions of Zen4 will see a similar relative performance uplift that 5800X -> 5800X3D saw. The "wait for X3D" mantra seems like getting hopes up beyond what can realistically be delivered.
 
BTW I dont think X3D versions of Zen4 will see a similar relative performance uplift that 5800X -> 5800X3D saw. The "wait for X3D" mantra seems like getting hopes up beyond what can realistically be delivered.
Does that feeling come the latest bunch of performance, having DDR-5 performance making a cache less important, the double of L2 cache on Ryzen 4 making the advantage of larger cache already there ? Thermal ?
 
BTW I dont think X3D versions of Zen4 will see a similar relative performance uplift that 5800X -> 5800X3D saw. The "wait for X3D" mantra seems like getting hopes up beyond what can realistically be delivered.

I don’t see why not if game is CPU limited and you are using newer GPUs. I don’t expect it to be 50% faster but faster indeed squeezing all the juice from GPU.
 
For some reason the Semianalysis site isn't responding to me so I am linking from cache, Packaging Developments
They explain that the 5800x3D is a bit of an oddity in that they relaxed the bond pitch greatly to make it work, they took it from 9 microns to 17 microns and only had it working with wafers measuring 6mm squared which was the exact size of the 5800x3d's v cache.
They also describe the very slow process in which a device physically places those two chips together to bond them and that as of June of this year was still pretty hit or miss.
The upside here is TSMC has said they have a new bonding electroplate material that has a lower resistance making it less problematic with higher voltages. The restrictions on the 5800x3d would not do well with the Zen4 chips, the clock speed decrease would surpass the performance increase of the cache by a fair bit I fear.
Well see, I hope TSMC gets it all sorted out and straightened away, but they have had some pretty smooth sailing for a long while and I wonder if they are due for their "Intel 10nm" moment in which some things just don't pan out.
 
From what I've read here there are technical difficulties with the TSMC process that they need to get around, first. Will AMD spend the time and money to get a 3D cache version into production is the big question.
Makes me wonder if the speculation about them releasing X3D versions of all their chips is off the table. I had suspected from the beginning that they may only be putting out one Gaming chip per node change due to complexity and cost to make. 7800X3D would be my guess for this generation.

Everyone is assuming AMD is going to release an entire lineup of X3D parts. I just don't see that happening. Even with an improved process, we won't be seeing the 7000 series lineup all running balls deep in 5.4 Ghz + Glory hitting max thermals. I would suspect the 3D V Cache will cause some clock adjustments just like it did for the 5000 series. Maybe not as severe, but given how hard they are to produce at the moment I wouldn't be looking at these chips as the holy grail that people are making them out to be.

My prediction of the future: This should be fun to see how well this ages... lol.
I figure the 7800X3D would be like 15-25% faster than the 5800X3D for gaming and 10% slower than the 7700X in productivity workloads.
 
Makes me wonder if the speculation about them releasing X3D versions of all their chips is off the table. I had suspected from the beginning that they may only be putting out one Gaming chip per node change due to complexity and cost to make. 7800X3D would be my guess for this generation.

Everyone is assuming AMD is going to release an entire lineup of X3D parts. I just don't see that happening. Even with an improved process, we won't be seeing the 7000 series lineup all running balls deep in 5.4 Ghz + Glory hitting max thermals. I would suspect the 3D V Cache will cause some clock adjustments just like it did for the 5000 series. Maybe not as severe, but given how hard they are to produce at the moment I wouldn't be looking at these chips as the holy grail that people are making them out to be.

My prediction of the future: This should be fun to see how well this ages... lol.
I figure the 7800X3D would be like 15-25% faster than the 5800X3D for gaming and 10% slower than the 7700X in productivity workloads.
I was thinking the reason we don't currently have a 7800x is because they are skipping it and going straight for the 7800x3D. The more I read about the current state of TSMC's stacking technology the more I think they may need to make alterations to the CCX to make it work correctly and the x3D lineup will be subtly different
 
Screenshot_20221021-152506_Samsung Internet.jpg


No real IPC difference between zen 3, zen 4, 12th gen, and 13th gen parts. Any real performance gains have been from adding e cores, slapping on more cache, and cranking the up the frequency.
 
I was thinking the reason we don't currently have a 7800x is because they are skipping it and going straight for the 7800x3D. The more I read about the current state of TSMC's stacking technology the more I think they may need to make alterations to the CCX to make it work correctly and the x3D lineup will be subtly different
Agreed
 
No real IPC difference between zen 3, zen 4, 12th gen, and 13th gen parts. Any real performance gains have been from adding e cores, slapping on more cache, and cranking the up the frequency.

The proper terminology for E-cores is actually CA-cores - Cinebench Accelerators.

How dare you not know this?
 
Last edited:
View attachment 520350

No real IPC difference between zen 3, zen 4, 12th gen, and 13th gen parts. Any real performance gains have been from adding e cores, slapping on more cache, and cranking the up the frequency.
This is what intrigues me so much about the Intel E-cores, they are so small and feature bereft but still go toe to toe with Coffee Lake, it makes me wonder what Intel may build them into in the future.
 
This is what intrigues me so much about the Intel E-cores, they are so small and feature bereft but still go toe to toe with Coffee Lake, it makes me wonder what Intel may build them into in the future.
Furnaces, water heaters, ovens, steam generators. I mean, they’re already doing space heaters, may as well expand and branch out.
 
I figure the 7800X3D would be like 15-25% faster than the 5800X3D for gaming and 10% slower than the 7700X in productivity workloads.

I was thinking the reason we don't currently have a 7800x is because they are skipping it and going straight for the 7800x3D. The more I read about the current state of TSMC's stacking technology the more I think they may need to make alterations to the CCX to make it work correctly and the x3D lineup will be subtly different

I think you guys are spot on here. A 12 or 16 core X3D Zen4 part would be glorious, and just the excuse I needed to slide an envelope of money across the table to Lisa in some Santa Clara diner, but they begin to make less for the company from a business standpoint. I guess it depends on TSMC being able to work magic again.
 
I think you guys are spot on here. A 12 or 16 core X3D Zen4 part would be glorious, and just the excuse I needed to slide an envelope of money across the table to Lisa in some Santa Clara diner, but they begin to make less for the company from a business standpoint. I guess it depends on TSMC being able to work magic again.
I see them dropping the Vcache on one CCD CPUs. Dropping it on two CCDs makes me wonder what the penalty for doing so will be. The frequency and voltage penalty for doing it to the 5800X was pretty big.

Even if the process has been advanced... There will be a cost to over all performance
 
Everyone's past CPUs were not necessarily poop. Therefore, I think that moving forward, we're going to find "value" outside of boosting something from 1,022,568 fps to 1,032,239 fps at a 2x power consumption requirement.

GPU wise, create something that gets the job done efficiently and smoothly without requiring a massive PSU and a much larger case.

CPU wise, create something that gets the job done with less power and runs much cooler (why not?).

CPU wise, if you want something better, give me at least 4 sockets, you know? Otherwise, give me something that does the "desktop" job well. I'm not interesting in running something stable for 3 seconds doing a benchmark so I can see my username in lights.

The market needs to change what it's looking for. Smooth, non-crashing experience that doesn't make me sweat. Power bills are rising. Circuits are being overloaded. Time for a change of "what is important".

(I can dream)
 
No real IPC difference between zen 3, zen 4, 12th gen, and 13th gen parts. Any real performance gains have been from adding e cores, slapping on more cache, and cranking the up the frequency.
If is this not misleading in some way, we see that it is somewhat rare to be otherwise, the 8600k to the 10900k being 2 release in a row without any change.

but it could be considering Rocket Lake finally has a jump from the 10xxxk while being one of the worst release. In the sense that it possible that they do sacrifice IPC gain to augment Cache and mhz all else being equal in a calculate way.
 
Last edited:
CPU wise, create something that gets the job done with less power and runs much cooler (why not?).
Would the 12100 and 12400 (and the upcoming 13100-13400) exactly that.

A 12100k give you a 9600k-3600-2700x performance while rarely going over 52 watt in fullthrottle perfect multithread task that make a 400w PSU overkill, for easier workload like gaming a 5800x3d draw under 60 watt.

Because of the giant Mobile-Laptop market, they always create get the job done with less power CPUs (apple, intel, amd and others), same for GPU.
 
Would the 12100 and 12400 (and the upcoming 13100-13400) exactly that.

A 12100k give you a 9600k-3600-2700x performance while rarely going over 52 watt in fullthrottle perfect multithread task that make a 400w PSU overkill, for easier workload like gaming a 5800x3d draw under 60 watt.

Because of the giant Mobile-Laptop market, they always create get the job done with less power CPUs (apple, intel, amd and others), same for GPU.
I feel like the desktop are getting the garbage. Those mobile chips are crazy good for their power draw. I mean, my i7 Mac mini 2012 is crazy high FPS playing FYL.
 
Would the 12100 and 12400 (and the upcoming 13100-13400) exactly that.

A 12100k give you a 9600k-3600-2700x performance while rarely going over 52 watt in fullthrottle perfect multithread task that make a 400w PSU overkill, for easier workload like gaming a 5800x3d draw under 60 watt.

Because of the giant Mobile-Laptop market, they always create get the job done with less power CPUs (apple, intel, amd and others), same for GPU.
This is where things like the 12600HL come in, at 65w max draw it is a little beast.
But this problem we are having now is one of the big reasons Apple moved over to ARM, their M1 and M2 lineup easily meets and exceeds even the EU's most draconian and asinine low power requirements with ease while still offering good day-to-day in any given office environment.
 
Exactly this. Server/Enterprise > Consumer/Retail
But it's "tru-ish" because enterprise wise the silicon is totally different. Epyc has more in common with TR. But even then, Eypc isn't simply better binned TR.

As someone that prefers these (even) higher end style CPUs, they're harder to get in the used market at reasonable prices vs. Intel's server end.
 
If you are buying a new system now, Zen4 all the way for future compatibility. If you are building on a budget, go Intel with ddr4 and a lower and a lower end 11th or 12th gen for a good upgrade later.
 
Exactly this. Server/Enterprise > Consumer/Retail
Well this isn't completely accurate.

It's Shareholders first, then Server/Enterprise and OEM partners, then that bacteria that lives on the inside of your refrigerator door, and then finally consumers/retail. Moore proposed it, it's yet to be falsified.
 
Well this isn't completely accurate.

It's Shareholders first, then Server/Enterprise and OEM partners, then that bacteria that lives on the inside of your refrigerator door, and then finally consumers/retail. Moore proposed it, it's yet to be falsified.
You forgot the crud growing on the underside of the toilet tank but yeah.
 
Yea its only a better value if you can drop these in. It makes zero sense to build a whole new system that is on a dead socket. This is second gen on same socket so going by intel history you got new socket coming next gen. Next time around AMD will be better value because of x670e and zen4 x3d chips.

There is value in AMD if building a new rig, there is value in raptor lake if you are dropping in. Right now raptor lake will sell more because of that. Next year you will have people buying the zen4 3d chips like hotcakes. Its a cycle. I think amd will drop prices on these when they announce some of those chips. I really doubt people on zen3 are going to go all out on zen 4 or Raptor lake right now. But next year I expect amd too move alot, as board prices normalize and they announce zen4 x3d chips.

Bingo...If you have any inkling of upgrading just your CPU and are building new now, AMD just released a platform that will be good for several years. If you are on a lower end ADL setup and want more performance, you can drop in a RKL CPU now. I have a 12600k that might turn into a 13700k after the initial rush dies down. Building a new Intel setup right now is a dead end. Some people will probably be ok with that and others won't be.
 
  • Like
Reactions: NKD
like this
Bingo...If you have any inkling of upgrading just your CPU and are building new now, AMD just released a platform that will be good for several years. If you are on a lower end ADL setup and want more performance, you can drop in a RKL CPU now. I have a 12600k that might turn into a 13700k after the initial rush dies down. Building a new Intel setup right now is a dead end. Some people will probably be ok with that and others won't be.
Yep exactly. I honestly built it for what's coming next and gonna definitely get one of those zen4 3d chips. I got the 7700x for like 350 with 10% off and some rewards I had at best buy. I will resell it and put that towards x3d, but needed to build the system now cuz I had time. You are right on about giving your system another drop in upgrade path with raptor lake. I would have done the same if I had z690.
 


The 13600k destroys the 7600x in productivity and ties it in gaming. Just as R5s were 6 core and i5s were 4 cores, AMD should have priced and named the 8 core 7700x what the 6 core 7600x is now.

Seeing AMD so outmatched in productivity in the midrange is a bit wild.
 


The 13600k destroys the 7600x in productivity and ties it in gaming. Just as R5s were 6 core and i5s were 4 cores, AMD should have priced and named the 8 core 7700x what the 6 core 7600x is now.

Seeing AMD so outmatched in productivity in the midrange is a bit wild.

Their saving grace was "keeping" boards and components, gen to gen. Intel saw an opportunity here, good timing.
 
The opposite happens next cycle, when Intel requires a new mobo but AM5 gets re-used.

Not to mention the cycle after that when the AM5 board you buy now is still good.

Platform longevity is not one of Intel's strong suits and they have been known to artificially limit platforms simply to force users to upgrade (see no CFL on Z270 boards even though der8auer showed Intel's "power issues" to be a lie).
 
Not to mention the cycle after that when the AM5 board you buy now is still good.

Platform longevity is not one of Intel's strong suits and they have been known to artificially limit platforms simply to force users to upgrade (see no CFL on Z270 boards even though der8auer showed Intel's "power issues" to be a lie).
Planned obsolescence is bad for the environment. Fuck Intel.
 
No real IPC difference between zen 3, zen 4, 12th gen, and 13th gen parts. Any real performance gains have been from adding e cores, slapping on more cache, and cranking the up the frequency.
Otherwise known as the brute force approach. Throw the 4090 in that list as well.
 
Back
Top