Intel Core Ultra 285K, 265K, & 245K CPU Specs: Bending Fix, Power Reduction, & Prices
video description
Date: 2024-10-11
Related videos
Comments and reviews: 20
paulboyce8537
285K extreme Wattage I saw at 295W. Also the 250W is questionable as the 285K and 265K have different clock speeds.
For me this is something for new builds. I'm looking at APO and Battlemage. The new Z890 has direct 5.0 PCI lane to CPU as does X870. This is REBAR on steroids. So simply put 285K and Battlemage dual GPU performance with no restrictions link to CPU. To me Wattage is performance. In Wattage it will look 250-295W (285K) 250-275W (Battlemage) = 500-570W of combined performance. All this on paper to my view lifts the ARC from Alchemist to Battlemage more than double. We have Xe vs Xe2 that already is 50% gain. Predicted clock speed is 3000mhz if not more from 2400mhz (boost) to give a good lift also. APO another 10-50% predicted on supported games. XeSS XMX on bar with DLSS.
This is what I'm waiting to see what will this pairing do Would I dare to say beating 5080
reply
285K extreme Wattage I saw at 295W. Also the 250W is questionable as the 285K and 265K have different clock speeds.
For me this is something for new builds. I'm looking at APO and Battlemage. The new Z890 has direct 5.0 PCI lane to CPU as does X870. This is REBAR on steroids. So simply put 285K and Battlemage dual GPU performance with no restrictions link to CPU. To me Wattage is performance. In Wattage it will look 250-295W (285K) 250-275W (Battlemage) = 500-570W of combined performance. All this on paper to my view lifts the ARC from Alchemist to Battlemage more than double. We have Xe vs Xe2 that already is 50% gain. Predicted clock speed is 3000mhz if not more from 2400mhz (boost) to give a good lift also. APO another 10-50% predicted on supported games. XeSS XMX on bar with DLSS.
This is what I'm waiting to see what will this pairing do Would I dare to say beating 5080
reply
mttrashcan-bg1ro
I think Intel need to spent a few generations alternating between efficiency and performance. 11th gen and 14th gen would have been perfect for them to significantly reduce power draw, as 250w is an absolute joke when AMD can do half that on their equivalent CPUs. Regardless it's disappointing to see both companies releasing CPUs that aren't really doing much to improve over previous generations.
Games right now are becoming so insanely CPU heavy, and the 50 series isn't far off, the 5080 and 5090 are going to be pointless since no CPU will be able to push them, 7800X3D's already just barely push a 4090 best case, and worst case it's a small bottleneck. But Intel need to just make sure their CPUs work and last right now.
reply
I think Intel need to spent a few generations alternating between efficiency and performance. 11th gen and 14th gen would have been perfect for them to significantly reduce power draw, as 250w is an absolute joke when AMD can do half that on their equivalent CPUs. Regardless it's disappointing to see both companies releasing CPUs that aren't really doing much to improve over previous generations.
Games right now are becoming so insanely CPU heavy, and the 50 series isn't far off, the 5080 and 5090 are going to be pointless since no CPU will be able to push them, 7800X3D's already just barely push a 4090 best case, and worst case it's a small bottleneck. But Intel need to just make sure their CPUs work and last right now.
reply
s-x5373
Notes:
they increased prediction window, also increased rename/register pipeline depth (this is prediction window), also increased depth of instruction fetching (this also is prediction windows), all while disabling, I mean removing hyperthreading support
then they say they increased efficiency while lowering max turbo boost
they also stabilized gaming performance while doing so .. enabling APO by default
they also spite on chiplet design and say that they use Tiles instead
To me this sounds like a 14000 series cpu with disabled hyper-threading and apo enabled, that's all
reply
Notes:
they increased prediction window, also increased rename/register pipeline depth (this is prediction window), also increased depth of instruction fetching (this also is prediction windows), all while disabling, I mean removing hyperthreading support
then they say they increased efficiency while lowering max turbo boost
they also stabilized gaming performance while doing so .. enabling APO by default
they also spite on chiplet design and say that they use Tiles instead
To me this sounds like a 14000 series cpu with disabled hyper-threading and apo enabled, that's all
reply
bigbo1764
This whole launch was doomed the second raptor lake failures were made known; to try and convince consumers to buy a cpu with no performance gains simply because of efficiency gains is already crazy, but doing that while your most recent generation of CPUs is failing en masse is impossible. Idk if Intel will make it until the next generation of CPUs, I have no idea what the internal situation of the company is, but if they do they better have something like 40% gains over arrow lake or these dudes are for sure dead outside of the server and professional markets.
reply
This whole launch was doomed the second raptor lake failures were made known; to try and convince consumers to buy a cpu with no performance gains simply because of efficiency gains is already crazy, but doing that while your most recent generation of CPUs is failing en masse is impossible. Idk if Intel will make it until the next generation of CPUs, I have no idea what the internal situation of the company is, but if they do they better have something like 40% gains over arrow lake or these dudes are for sure dead outside of the server and professional markets.
reply
GR8APE69
6:10 Hopefully that NPU is implemented properly. NPUs added as coprocessors have shown pretty sizeable gains in compute power. They can be used for more than just matrix math. They work as pretty powerful hardware accelerators for things like transcoding video as well, but it's going to depend on how the CPU allocates resources. Hopefully the inclusion of an onboard NPU doesn't end up being restricted to niche processes like limiting it to matrices would do.
reply
6:10 Hopefully that NPU is implemented properly. NPUs added as coprocessors have shown pretty sizeable gains in compute power. They can be used for more than just matrix math. They work as pretty powerful hardware accelerators for things like transcoding video as well, but it's going to depend on how the CPU allocates resources. Hopefully the inclusion of an onboard NPU doesn't end up being restricted to niche processes like limiting it to matrices would do.
reply
fyinghigh1868
Seeing Intel 'humble' themselves and actually work on a different architecture that is unconventional (e.g. Hyper-Threading) and releasing chips that are somewhat competitively priced is definitely very good for us consumers. But the real competition for Intel, in my opinion, is their ability to use motherboards with the same 'type' and socket for at least more than 3 generations would be a significant contender to AMD's 'forever-lasting' motherboards.
reply
Seeing Intel 'humble' themselves and actually work on a different architecture that is unconventional (e.g. Hyper-Threading) and releasing chips that are somewhat competitively priced is definitely very good for us consumers. But the real competition for Intel, in my opinion, is their ability to use motherboards with the same 'type' and socket for at least more than 3 generations would be a significant contender to AMD's 'forever-lasting' motherboards.
reply
heyarno
I hope it's competitive.
On a surface level, AMD could just put one ccd with zen5 and one more with zen5c in a processor and match the numbers of cores.
But generally, I'm happy that chips don't come running at the absolute limit from the factory.
And the whole interconnect architecture will show it's worth when the cores become more powerful and plentiful.
Hopefully the improved competition reflects in more attractive prices for consumers.
reply
I hope it's competitive.
On a surface level, AMD could just put one ccd with zen5 and one more with zen5c in a processor and match the numbers of cores.
But generally, I'm happy that chips don't come running at the absolute limit from the factory.
And the whole interconnect architecture will show it's worth when the cores become more powerful and plentiful.
Hopefully the improved competition reflects in more attractive prices for consumers.
reply
DerKropa
The naming somehow reminds me of how the car badges were changed quite recently on BMW and Audi, at least I know of these two.
Before, the numbers meant the displacement, starting usually from 1.6 and all the way up to 6,0 or even bigger, however now they start at around 30 and go up to 70, all that looks like they afford small number as much as they can. Because bigger number better.
reply
The naming somehow reminds me of how the car badges were changed quite recently on BMW and Audi, at least I know of these two.
Before, the numbers meant the displacement, starting usually from 1.6 and all the way up to 6,0 or even bigger, however now they start at around 30 and go up to 70, all that looks like they afford small number as much as they can. Because bigger number better.
reply
advil000
OMG those product numbers. I quit. Screw it. I don't even care anymore. Not even going to try to explain to customers what they mean, and not going waste time explaining it when people ask. I'm going to send them an online link but I'm not dealing with this anymore. Who came up with this It's going to be I want an Ultra 9... whatever the hell the rest of the numbers are.
reply
OMG those product numbers. I quit. Screw it. I don't even care anymore. Not even going to try to explain to customers what they mean, and not going waste time explaining it when people ask. I'm going to send them an online link but I'm not dealing with this anymore. Who came up with this It's going to be I want an Ultra 9... whatever the hell the rest of the numbers are.
reply
lawrenceonface
Hey I will wait for it’s microcode launch who know, joking
No point to upgrade
-not compatible contact frame yet
Poor blending not fixed while they a bit PCB thickness added
-less option of cooling
-I bet the e cores with ton of heat
-no huge technically help to big improve
For the name ,it wants to confuse customers to forget 13 14 gen painful
reply
Hey I will wait for it’s microcode launch who know, joking
No point to upgrade
-not compatible contact frame yet
Poor blending not fixed while they a bit PCB thickness added
-less option of cooling
-I bet the e cores with ton of heat
-no huge technically help to big improve
For the name ,it wants to confuse customers to forget 13 14 gen painful
reply
NirreFirre
As someone with home, media and game servers and gaming needs in the same machine (space and connection based restrictions), idle power draw and temps from both default and more optimized settings would be very interesting to see comparisons around.
Comparing power draw when gaming is imho not the most important metric since most consumers use their machine at that load
reply
As someone with home, media and game servers and gaming needs in the same machine (space and connection based restrictions), idle power draw and temps from both default and more optimized settings would be very interesting to see comparisons around.
Comparing power draw when gaming is imho not the most important metric since most consumers use their machine at that load
reply
predabot__6778
The performance-improvement on the E-cores is quite insane -- 30-70%! No, no - I refuse to believe that. Now, I could see them having a really good performance-improvement though, so perhaps... 16-28 That is still a very impressive achievement of Intel, but far more realistic. What's everyone else thinking Could they really have achieved such an E-core improvement
reply
The performance-improvement on the E-cores is quite insane -- 30-70%! No, no - I refuse to believe that. Now, I could see them having a really good performance-improvement though, so perhaps... 16-28 That is still a very impressive achievement of Intel, but far more realistic. What's everyone else thinking Could they really have achieved such an E-core improvement
reply
readycheddar
So I’m assuming x86 has just hit a brick wall when it comes to performance gains I think they can’t fit in any more die shrinks so all they can do now is see how much power efficiency they can milk out of the new CPUs while squeezing the last drops of performance out of them.
Maybe I’m just talking out of my ass Please chime in if you think I’m wrong.
reply
So I’m assuming x86 has just hit a brick wall when it comes to performance gains I think they can’t fit in any more die shrinks so all they can do now is see how much power efficiency they can milk out of the new CPUs while squeezing the last drops of performance out of them.
Maybe I’m just talking out of my ass Please chime in if you think I’m wrong.
reply
ThylineTheGay
2:40 technically true, but from when SMT was first being researched, they (intel) found that it increased performance while having no significant impact on energy use, basically SMT uses the hardware far more effectively - i doubt CPUs have changed so much since then that that would be invalid
Meaning that these CPUs are likely _inherently less efficient_
reply
2:40 technically true, but from when SMT was first being researched, they (intel) found that it increased performance while having no significant impact on energy use, basically SMT uses the hardware far more effectively - i doubt CPUs have changed so much since then that that would be invalid
Meaning that these CPUs are likely _inherently less efficient_
reply
sambalsamurai9672
Please also test the IDLE power draw of 265K and 285K, not just the maximum power draw. And comapre it to 9900X and 9950X. Thanks!
Since most office business creative workstations are 90% of the time idle, waiting for that giant task like developing 1000 RAW files to JPG via favourite custom profile, as quickjly as possible before going to idle again.
reply
Please also test the IDLE power draw of 265K and 285K, not just the maximum power draw. And comapre it to 9900X and 9950X. Thanks!
Since most office business creative workstations are 90% of the time idle, waiting for that giant task like developing 1000 RAW files to JPG via favourite custom profile, as quickjly as possible before going to idle again.
reply
CrazyShiftee
So they changed the naming scheme because it was too easy to understand. Then they created a new more expensive clamping mechanism even though the current one was fine.. until now. And finally they are trying to sell you a $600 CPU that may be slower than your current one but saves you 50W worth of money.
Sucks to be poor. - Intel (undefined - 2025)
reply
So they changed the naming scheme because it was too easy to understand. Then they created a new more expensive clamping mechanism even though the current one was fine.. until now. And finally they are trying to sell you a $600 CPU that may be slower than your current one but saves you 50W worth of money.
Sucks to be poor. - Intel (undefined - 2025)
reply
MrEddieLomax
I work on chip design (FPGA) so need CPU power, these E cores need to be switched off and thus are wasted silicon. The speeds here look good and I like the power drop, but the edge seems to be a 9950x especially as there is less wasted silicon (No to GPU embedded versions too :) ), waiting for the 3D version and then probably going for it.
reply
I work on chip design (FPGA) so need CPU power, these E cores need to be switched off and thus are wasted silicon. The speeds here look good and I like the power drop, but the edge seems to be a 9950x especially as there is less wasted silicon (No to GPU embedded versions too :) ), waiting for the 3D version and then probably going for it.
reply
thejerrymobile
CPUs have been patiently biding their time, watching GPUs get massive and bloated. And finally we are starting to get Long CPU.
If there weren’t industry politics in the way, you’d think the retention mechanism would be obviated by just building the cooler into the component. But people would riot.
reply
CPUs have been patiently biding their time, watching GPUs get massive and bloated. And finally we are starting to get Long CPU.
If there weren’t industry politics in the way, you’d think the retention mechanism would be obviated by just building the cooler into the component. But people would riot.
reply
edge21str
I'm gonna be bold and claim that this name change is actually what will hurt intel the most. People who don't care about tech still know an i5 is good and an i7 is great. They don't want to learn new marketing terms and will be more receptive to alternatives if both naming schemes are unfamiliar.
reply
I'm gonna be bold and claim that this name change is actually what will hurt intel the most. People who don't care about tech still know an i5 is good and an i7 is great. They don't want to learn new marketing terms and will be more receptive to alternatives if both naming schemes are unfamiliar.
reply
bumblebeebat-te5xv
My faith in benchmarks of newly released Intel CPUs has been severely dented by the 13th/14th Gen bios power settings omnishambles.
Who's to say whether it'll all just become invalid again in 12/24 months, and by that time everyone's moved on to the next shiny new product.
reply
My faith in benchmarks of newly released Intel CPUs has been severely dented by the 13th/14th Gen bios power settings omnishambles.
Who's to say whether it'll all just become invalid again in 12/24 months, and by that time everyone's moved on to the next shiny new product.
reply
Add a review, comment
Other channel videos
