World record for the fastest cpu overclocked on liquid nitrogen has finally fallen to the all mighty 13900k 😱 . Old world record was done on AMD FX-8370 eight years ago by the Stilt.
Old record was 8722.8Mhz 🙂
New world record was done by Elmor / Shamino Asus Rog r/d Team with an Asus Apex Z790 , 13900K and lot's of Ln2.
New world record is 8812.85Mhz 😱
https://hwbot.org/submission/5102721_elmor_cpu_frequency_core_i9_13900k_8812.85_mhz
closer than i expected, not a bad launch tbh, seems the e cores are paying off for now. though i will say it must hurt intel to be selling a large die for such a low price after all those years of selling fairly high margin desktop parts.
man where do you even get a memory block
Definitely too close to take AMD out of the equation. I expected those to smoke AMD in everything that was not rendering considering how meh Zen 4 is but it's within a few % most of the time. I still don't get why AMD did not increase the amount of cores with Zen 4. Clearly the e-cores are working in multi-thread workload and AMD wont be able to keep this amount of cores for long. Also shipping Zen 4 without 3d v-cache was kind of almost pointless. I get it's too expensive for cheaper cpus like dunno a potential 7600 non x or 7400 or 7200 but i feel like moving forward it's hard for gamers to take any non 3d v-cache cpus from AMD seriously. Still only thing AMD has to do is reduce price and voilà . Also the power consumption of this cpu is crazy considering the performance it doesn't look great at all.
Wow that FC6 90W benchmark is ridiculous. The 13900K is 23% faster than 7950X using the same 90W power limit.
AMD should release the 3D cache variants asap.
Yeah it's kind of crazy how quickly the power consumption goes up on these at the "last mile" so to speak.
In my eyes very disappointing HEDT performance. Imho a bit of misleading settling a dual channel ram, 20 lane CPU in this domain. AMD still has much more to offer.....
For gamers regarding cost/ performance, certainly best choice is an i5 or i7, have almost same perf. as i9@5.4Ghz. Only i9@5.8Ghz+ TVB is faster, but TVB easily doubles costs compared to i7 and triples costs compared to i5.
For OC, I think, Intel already binned CPUs quiet well, so there will be not much OC headroom without LN2. When already @5.8Ghz+, each additional mV will rocket power consumption and heat through the roof. Let´s wait for LN2 tests and see what happens....
Edit: btw, same Voltage/ Power "Problem" with Nvidia ADA.... each additional mV skyrockets heat and power consumption, looks like we are already on a very low node and well binned chips. Getting closer to the physical border is pretty sad for OCs.
2. Edit: TVB = Thermal Velocity Boost
The results on other sites (notice plural) are showing the Intel chip with far higher power use so am confused how the 13900 can use less on here GN even had it around 300w. Am not an intel fan but this is a good chip for sure, however the difference in performance advantages seems to be mostly in games, and the 5800x3d gives both new Chips from both sides a good run for, I believe this is exactly the moment AMD was waiting for and will stake money on the fact they are sitting on the newer x3d chips until the RDNA3 launch event and will casually drop a few Benchmarks on the table that will just blow all comers into the weeds for games and games related workloads
You are not considering the cost for DDR5 vs DDR4. And the cost for motherboards.
5800x3D + X570 crosshair dark hero + 3600mhz ddr4 = 150-200$ more than the z790 hero motherboard alone
for gamers there's no reason to buy the DDR5 plateforms unless...you have a 4090 that apparently the 5800x3d cannot feed fast enough (just saw this in a 13900k review don't remember which one)
also I have to add that my "only 2 sticks" problem seems to have come mostly from the 12900k cpu, raptor lake has much more 4 sticks support BUT....stilla t lower speeds than X670, 4800-5200 mostly with a few 5600..4800mhz is a no go it's like running 2133mhz ddr4 nobody does that
Wow that FC6 90W benchmark is ridiculous. The 13900K is 23% faster than 7950X using the same 90W power limit.
AMD should release the 3D cache variants asap.
The Farcry 6 engine looks to get confused by many cores or atleast not scaling very well, 7600x getting better lows then 7950x and 12400k getting better lows then 12900k.
The 13900k will win against the 7950x, because in games the 13900k is essentially a 8 core with the thread scheduler, so less confusion for the engine.
Disabling 8 cores on the 7950x probably boosts performance in Farcry 6.
The Farcry 6 engine looks to get confused by many cores or atleast not scaling very well, 7600x getting better lows then 7950x and 12400k getting better lows then 12900k.
The 13900k will win against the 7950x, because in games the 13900k is essentially a 8 core with the thread scheduler, so less confusion for the engine.
Disabling 8 cores on the 7950x probably boosts performance in Farcry 6.
Farcry 6 is horrible with all AMD from what I gather. Similar to AoE4.
Meh - seems okay but the leaked benchies made it seem like more. Rather wait for 7000 X3D. AMD should have launched with it honestly.
Intel I guess cheaper option for now but as I wrote in the other thread.. the TCCO is much lower on AMD if you plan on upgrading in the future.
I still think AMD is the better option for most people.
@Hilbert Hagedoorn Given that most of the arguments on Guru3D are that Intel is a cheaper option due to DDR4 support. Could you run a couple benches with DDR4 to see the difference in performance?
Edit: Hilbert posted below:https://www.guru3d.com/articles-pages/msi-mag-z790-tomahawk-(ddr4)-review,1.html
Given the performance loss on DDR4
Isn't the 5800X3D the better "budget" play for most people?
This review doesn't show the full picture, find a review with 4090, thats more like it
The 4K benchmarks all plus minus identical across different CPUs here, thats due to GPU bottleneck.
With 4090,m the 13900K shines even brighter and especially in 6Ghz with 4090 on waterloop,
brogadget:
Thanks for a great review.
In my eyes very disappointing HEDT performance. Imho a bit of misleading settling a dual channel ram, 20 lane CPU in this domain. AMD still has much more to offer.....
For gamers regarding cost/ performance, certainly best choice is an i5 or i7, have almost same perf. as i9@5.4Ghz. Only i9@5.8Ghz+ TVB is faster, but TVB easily doubles costs compared to i7 and triples costs compared to i5.
For OC, I think, Intel already binned CPUs quiet well, so there will be not much OC headroom without LN2. When already @5.8Ghz+, each additional mV will rocket power consumption and heat through the roof. Let´s wait for LN2 tests and see what happens....
Edit: btw, same Voltage/ Power "Problem" with Nvidia ADA.... each additional mV skyrockets heat and power consumption, looks like we are already on a very low node and well binned chips. Getting closer to the physical border is pretty sad for OCs.
This review doesn't show the full picture, find a review with 4090, thats more like it
The 4K benchmarks all plus minus identical across different CPUs here, thats due to GPU bottleneck.
With 4090,m the 13900K shines even brighter and especially in 6Ghz with 4090 on waterloop,
Testing at lower resolutions does the same thing..
I'm not saying that the 13900K isn't the fastest CPU out. It is. But I think for most people the TCCO of AMD just makes it a more valuable platform imo - unless you can afford to upgrade motherboard/cpu every gen. In which case yeah, always buy the fastest - which in this case is Intel. Then just swap to X3D when that launches because it will probably beat this. Then rinse & repeat.
Not sure what grief you are seeing. Intel had a pretty good showing but not a game changer.
Its ADL refresh and its plenty fast, + first CPU that can do 6Ghz all core with few bios changes, but it doesn't need to
Same performance as already plenty fast 12900K at 90W, you can do mini builds with this CPU and tiny heatsink and have same performance as 12900K with AIO
What more do you need? ZEN4 is not a refresh of ZEN3 and it fails to win over refresh of ADL that just got a node shrink + tad more cache
Testing at lower resolutions does the same thing..
I'm not saying that the 13900K isn't the fastest CPU out. It is. But I think for most people the TCCO of AMD just makes it a more valuable platform imo - unless you can afford to upgrade motherboard/cpu every gen. In which case yeah, always buy the fastest - which in this case is Intel. Then just swap to X3D when that launches because it will probably beat this. Then rinse & repeat.
When you read 1080p 500fps
vs 4K 160FPS is a different feeling [IMO]
I honestly dont care about keeping the same mobo for longer than 2 gens, IMHO its an overrated feature.
Lets look at AMD, X370 and X470 had PCIe gen 2.0 from chipset, so you couldn't even install Gen 3.0 M.2 [I mean more than 1]
Lack of USB 10Gbps and 20GBps, lack of 2.5Gb Ethernet and lack of otehr features
When x570 came out it was a must-have upgrade, it had 10Gbps USB, 2.5G Ethernet, everything Gen 4.0, and so on, old mobos looked sad [I had MSI Creation x570 +5950x, this mobo had more than 10 USB ports on the back, i think all USB 10 and +2 USB 20Gbps, had 10Gb Ethernet and so on]
Lets not forget about the BIOS issues AMD had, that had some CPU supported and otehrs not
Same will be with x670, new stuff will come out, stuff that will be must have like maybe full PCIe gen 5.0 from both chipset and CPU, USB 4.0, even more USB 4.0, maybe we finally move to 5gb Ethernet, maybe more PCIe lanes for M.2 and add-on cards, and so on
With Intel, im sure that next chipset will have USB 4.0, it will finally have PCIe gen 5.0 on the chipset and so on, so next year, it will be worth upgrading my z690 [right now i dont need z790, nothing new]
https://hwbot.org/submission/5102721_elmor_cpu_frequency_core_i9_13900k_8812.85_mhz
In my eyes very disappointing HEDT performance. Imho a bit of misleading settling a dual channel ram, 20 lane CPU in this domain. AMD still has much more to offer..... For gamers regarding cost/ performance, certainly best choice is an i5 or i7, have almost same perf. as i9@5.4Ghz. Only i9@5.8Ghz+ TVB is faster, but TVB easily doubles costs compared to i7 and triples costs compared to i5. For OC, I think, Intel already binned CPUs quiet well, so there will be not much OC headroom without LN2. When already @5.8Ghz+, each additional mV will rocket power consumption and heat through the roof. Let´s wait for LN2 tests and see what happens....
Edit: btw, same Voltage/ Power "Problem" with Nvidia ADA.... each additional mV skyrockets heat and power consumption, looks like we are already on a very low node and well binned chips. Getting closer to the physical border is pretty sad for OCs. 2. Edit: TVB = Thermal Velocity Boost