Intel Core i9-10900K 10-core Processor and Z490 Chipset Arrive April 2020

[MageTank]

5 minutes ago, pas008 said:

blind faith in amd 

no

i'm looking at fact intel didnt limit themselves in the past on moving high core count cpu from their server lineup to their hedt like they are now

amd sockets might not work but they can move their eypc cpus to threadrippers like they showed us with 32 core

you think they cant move the 64 core one to tr?

 

 

 

I know how it works

 

but if intel used 2066 for mainstream in first place and then 3647 for server/hedt they wouldnt be limiting their consumers on core count either and its not like they cant offer better clocked cpus on top

 

I believe you are misunderstanding how this works. Intel is doing EXACTLY what you are suggesting. These 2066 CPU's are not their own special silicon, they are cut down from the higher end Xeons. Just because they don't share the same substrate and pinout doesn't mean they didn't come from the same wafer. If your point is that Intel is wasting money by separating the product segments based on socket size, you are grossly incorrect. AMD bringing higher core counts to their lower end product stack has less to do with their sockets, and more to do with their chiplet/CCX design.

 

I also fail to see what you mean by "limiting customers on core count". If customers want more cores, they can still buy a Xeon and the appropriate boards for them...Those that would make use of that kind of hardware likely need the additional features of their boards/chipsets anyways. Again, you want to offer the most features for your intended market without charging them more for features they don't need. 

[leadeater]

6 hours ago, pas008 said:

but if intel used 2066 for mainstream in first place and then 3647 for server/hedt they wouldnt be limiting their consumers on core count either and its not like they cant offer better clocked cpus on top

Pin count has nothing to do with the number of cores Intel offers on their CPUs for consumer platform. Even with 2066 pins Intel would only offer 10 or less cores because Ring Bus does not performance scale above that for these workloads and Mesh is worse than Ring Bus for low cores counts. Power is the other reason, you cannot have high clocks and many cores, both is impossible within normal thermal parameters.

 

Many of the pins are not used right now so increasing it to 2066 just to have more non used pins wouldn't help, all you've done is make the socket larger and more expensive to implement in to a motherboard design.

 

And it makes zero difference what HEDT and server use sockets wise, there is absolutely no cross over today for those. You might just use the same physical socket to save cost and have the pin outs totally different or only slightly different but you would only do that if you think sales volume is going to be low to not justify a dedicated physical socket design. Just because two platforms use the same physical socket doesn't actually mean they are wired the same, for all you know only 60% of them are used at all because TR does not have multi CPU support and those links use A LOT of pins.

[pas008]

19 minutes ago, MageTank said:

I believe you are misunderstanding how this works. Intel is doing EXACTLY what you are suggesting. These 2066 CPU's are not their own special silicon, they are cut down from the higher end Xeons. Just because they don't share the same substrate and pinout doesn't mean they didn't come from the same wafer. If your point is that Intel is wasting money by separating the product segments based on socket size, you are grossly incorrect. AMD bringing higher core counts to their lower end product stack has less to do with their sockets, and more to do with their chiplet/CCX design.

 

I also fail to see what you mean by "limiting customers on core count". If customers want more cores, they can still buy a Xeon and the appropriate boards for them...Those that would make use of that kind of hardware likely need the additional features of their boards/chipsets anyways. Again, you want to offer the most features for your intended market without charging them more for features they don't need. 

 

19 minutes ago, leadeater said:

Pin count has nothing to do with the number of cores Intel offers on their CPUs for consumer platform. Even with 2066 pins Intel would only offer 10 or less cores because Ring Bus does not performance scale above that for these workloads and Mesh is worse than Ring Bus for low cores counts. Power is the other reason, you cannot have high clocks and many cores, both is impossible within normal thermal parameters.

 

Many of the pins are not used right now so increasing it to 2066 just to have move non used pins wouldn't help, all you've done it make the socket larger and more expensive to implement in to a motherboard design.

 

ANd it makes zero different what HEDT and server use sockets wise, there is absolutely no cross over today for those. You might just use the same physical socket to save cost and have the pin outs totally different or only slightly different but you would only do that if you think sales volume is going to be low to not justify a dedicated physical socket design. Just because to platforms use the same physical socket doesn't actually mean the are wired the same, for all you know only 60% of them are used at all because TR does not have multi CPU support and those links you A LOT of pins.

ok you guys are right

 

intel didnt show us they could offer 4 to 18 cores on a socket or AMD showed  4 to 16 or even 8 to 32 (prolly 64 next yr)

 

x299 mobos arent that much more

i was saying if lga2066 had other chipsets like z395 and similar that only offered dual channel but you guys are right

[MageTank]

25 minutes ago, pas008 said:

 

ok you guys are right

 

intel didnt show us they could offer 4 to 18 cores on a socket or AMD showed  4 to 16 or even 8 to 32 (prolly 64 next yr)

 

x299 mobos arent that much more

i was saying if lga2066 had other chipsets like z395 and similar that only offered dual channel but you guys are right

I get your point, and I agree that everything would be easier if the entire product stack was unified from the top down, and you simply chose how many cores you want for the price you were willing to pay, but I just don't think we are there yet. Part if the issue stems from the pace that technology is evolving, and how the market is a big battle of who can offer what for the best price possible.

 

We are definitely seeing the lines blurred between the mainstream consumer platforms and the current HEDT offerings. I believe that the current HEDT will become the new mainstream, but I also believe that HEDT will evolve into something more as well. The real limit will be whether or not developers choose to take advantage of the additional resources that consumers now have. If they don't, we may see this trend reverse itself in the future. Wouldn't be the first time that AMD or Intel took some inspiration from their past designs.

 

@pas008With everything said and done, you have my respect for objectively looking at the information that was provided to you and coming to a reasonable conclusion. It's extremely rare to see that these days online, especially towards those that have not been the kindest to you. My apologies if I came off as a jerk, as I was certainly starting to get frustrated, but understand I mean no ill will towards you. Like everyone here, I get very passionate about technology and it can sometimes bring out the worst. 

[pas008]

33 minutes ago, MageTank said:

I get your point, and I agree that everything would be easier if the entire product stack was unified from the top down, and you simply chose how many cores you want for the price you were willing to pay, but I just don't think we are there yet. Part if the issue stems from the pace that technology is evolving, and how the market is a big battle of who can offer what for the best price possible.

 

We are definitely seeing the lines blurred between the mainstream consumer platforms and the current HEDT offerings. I believe that the current HEDT will become the new mainstream, but I also believe that HEDT will evolve into something more as well. The real limit will be whether or not developers choose to take advantage of the additional resources that consumers now have. If they don't, we may see this trend reverse itself in the future. Wouldn't be the first time that AMD or Intel took some inspiration from their past designs.

 

@pas008With everything said and done, you have my respect for objectively looking at the information that was provided to you and coming to a reasonable conclusion. It's extremely rare to see that these days online, especially towards those that have not been the kindest to you. My apologies if I came off as a jerk, as I was certainly starting to get frustrated, but understand I mean no ill will towards you. Like everyone here, I get very passionate about technology and it can sometimes bring out the worst. 

thx

I am too passionate about tech/electronics its too bad i didnt pursue career in it but then again i  might end up hating it

did go for electronic engineer for couple yrs(remember playing with those sockets that took p3 and others) but changed majors and then other life shit got in the way(damn kids lol)

I dislike when people say it cant happen or impossible, 20 yrs ago 2 cores seemed impossible but what do you know couple yrs later they existed and 1 company glued lol

just like when people say mulitgpu/sli/xfire/etc shit when in fact a piece of software can do it why cant you do it hardwarewise etc etc etc

 

[porina]

1 hour ago, MageTank said:

We are definitely seeing the lines blurred between the mainstream consumer platforms and the current HEDT offerings. I believe that the current HEDT will become the new mainstream, but I also believe that HEDT will evolve into something more as well.

Nice to see I'm not the only one thinking along that line. I feel performance mainstream has stagnated at dual channel ram for too long. HEDT moving into that space makes sense, leaving let's say more value sensitive mainstream on dual channel. We kinda have this situation now with Cascade Lake-X overlapping with higher end of AM4. Be interesting to see what both sides do as we go on.

[HumdrumPenguin]

Another pointless release, almost as bad as the  10980XE.

[Cribbertarian]

I cracked up so much when Linus had to say OlympicAssEaster on WAN Show.

 

Kudos to you, Sir.

[amdorintel]

On 12/13/2019 at 10:04 AM, pas008 said:

20 yrs ago 2 cores seemed impossible

yea well year after year technology advances and you are either left in the dust or you keep up with the times.

[urbanman2004]

Tsk, tsk, tsk @ Intel

[Combo Techs IT Up]

On 12/11/2019 at 11:53 AM, sowon said:

Give it up Intel, don't rush out with half-baked chips. They're going through generation numbers so fast yet there's so little 'innovation' that it means nothing but snake oil.

They'd best have decent pricing to even be at least considered alongside Zen2 CPUs.

what exactly do you mean fully by innovation? If you don't mind me asking 

[BiG StroOnZ]

Fun time junction initiated... 

 

Intel Core i9-10900K to boost up to 5.3GHz & Other Leaked Specs for 10th Gen Comet Lake:

Quote

Intel Core i9-10900K/i7-10700K/i5-10600K (K Series) 

 

The most high-end CPU in the Comet Lake-S lineup is the Core i9-10900K CPU with 10 cores and 20 threads. This part is confirmed to be 125W TDP SKU. This processor will feature clocks of 3.7 GHz to 5.2 GHz in boost. The new feature called Velocity Boost will unlock clock speeds up to 5.3 GHz, but only for two SKUs: 10900K and 10900 non-K.

 

Meanwhile, core i7-10700K is an 8-core and 16-thread CPU with clock speed up to 5.1 GHz, this part does not have Velocity Boost though.

The Core i5-10600K will boost up to 4.8 GHz with a single core. This part is a six-core and 12-thread processor.

 

 

Intel Core i9-10900 – i3-10100 (non-K Series) 

 

Only the Core i9-10900 features Velocity Boost of 5.2/4.6 GHz. The Max Turbo Boost 3.0 is also exclusive for two SKUs: 10900 and 10700 (5.1 and 4.8 GHz respectively).

 

Quote

Performance Matters:

• Up to 4.8 GHz All-Core Turbo
• Up to 5.3 / 4.0 GHz Thermal Velocity Boost Single / All-core Turbo
• Up to 5.2 GHz Intel Turbo Boost Max 3.0
• Up to 10C and 20T
• Up to DDR4-2933 MHz dual-channel
• Enhanced Core & Memory Overclocking
• Active Core Group Tuning

NEW and Featured Technologies ~

• NEW: Up to 5.3 GHz with Intel Thermal Velocity Boost
• NEW: Intel Turbo Boost Max Technology 3.0
• NEW: Intel Hyperthreading Technology across 10th Gen Core i9 and i3 processors
• NEW: Up to 10 cores 20M Intel Smart Cache
• NEW: Up to DDR4-2933 support
• NEW: Enhanced Core and & Memory Overclocking
• NEW: Intel 400 Series Chipset
• NEW: 2.5G Intel Ethernet Connection i225 (Foxville) support
• NEW: Integrated WiFi 6(AX201) Gig+ support using CNBR

 

Source: https://videocardz.com/newz/intel-core-i9-10900k-to-boost-up-to-5-3-ghz-specifications-of-10th-gen-core-comet-lake-s-leaked

Source 2: https://www.guru3d.com/news-story/core-i9-10900k-can-boost-to-5-3-ghzmore-specifications-of-10th-gen-core-comet-lake-s-leak.html

 

Side Note: Right before the story went live, Videocardz was asked not to share some of the slides, but sharing the information from them was acceptable. The slides discarded were: "Performance Matters" and "NEW and Featured Technologies". There may have been others, but they weren't mentioned in the story.

[yian88]

Pointless we need better SoC not CPU's, PC's are lacking severely in better integration, AI accelerated tasks like mobile integrated 2/3/4 and 5G better mobile laptop chips etc.

All the time we only get pure brute CPU power thats the only thing thats "new", desktop chips could be a whole lot better if they turned into SoC with smaller mobo's, maybe some integrated HBM or other type of fast high bandwidth l4 Cache, really there is nothing new to x86 since dual core was invented, on the other hand look at mobile chips integrating more and more and being so efficient while x86 becomes heavier and bulkier.

I hope for an ARM/RISC-V future for desktop/laptop/gaming computers aswell.

Even though hardware has gotten so much faster past 10 years, software has gotten slower, just browsing or using windows 10 is garbage, it used to be so fast back in 2013 with win7, an entry level SSD and a phenomii x4 every web page loaded instantly, now on a modern ryzen 3 with 8gb ddr4 it takes about 5 secons for any page to load, all web browsers are garbage, web pages with html5/css and tons of javascript are slow garbage, definately needs a faster unified front-end language to replace all 3 main front end tech. Add windows 10 telemetry and lack of optimization + bloat adds on top of that, not to mention the performance destroyer win defender and other AV's that are pure spyware on top of MS system level spyware.

If i had desktop level pro apps access and a lapdock on my SD 845 phone i would completely get rid of this x86/microsoft garbage experience im so trough with it, everyday its a pain to use, at least if i knew they are working on improving it but they are not it will only get worse with more windows bloat, more browser bloat etc.

Use a piece of software like photoshop CS6 and compare it with photoshop CC 2019, CC 2019 is super slow does a lot of useless backgroudn stuff, tons of telemetry and cloud nonsense in the background when all i need is to create or mod some picture.

More cores and brute power are useless without mobile style optimization, i hate the future of mobile, the more RAM they can cram into phones the more bloated the apps will get too, at least for now they are still optimized since Android/iOS doesnt allow crap to ruin your performance and battery in the background like windows does, hopefully google/apple it will continue to improve on that front so that devs dont abuse mobile aswell.

[leadeater]

12 hours ago, yian88 said:

All the time we only get pure brute CPU power thats the only thing thats "new", desktop chips could be a whole lot better if they turned into SoC with smaller mobo's, maybe some integrated HBM or other type of fast high bandwidth l4 Cache, really there is nothing new to x86 since dual core was invented, on the other hand look at mobile chips integrating more and more and being so efficient while x86 becomes heavier and bulkier.

CPUs have been integrating things for years and are SoCs. We've got embedded high-speed networking and wifi in them, GPUs with display and digital audio output, PCIe controllers with RAID and booting capability, USB controller. There isn't an x86 CPU on the market today that is not an SoC. AMD EPYC doesn't even have a chipset at all.

[ravenshrike]

On 12/13/2019 at 7:03 AM, MageTank said:

The one that stifled innovation for the sake of keeping older AM3 processors compatible,severely limiting what they could do with newer processors

This presupposes that they had a new processor design worth a damn. They had nothing in the pipeline at all. At best even with a new socket they would have sucked marginally less.

[MageTank]

3 hours ago, ravenshrike said:

This presupposes that they had a new processor design worth a damn. They had nothing in the pipeline at all. At best even with a new socket they would have sucked marginally less.

K10 was not inherently bad though, especially compared to what followed with Bulldozer. It had a pretty substantial IPC boost (10% over bulldozer if memory serves correctly). They simply needed a refined process and the ability to scale it even further. It wouldn't have been any different than Intel's current 14nm approach, where it's the same thing but slightly more refined. 

 

I imagine a refined K10 that facilitated higher clock speeds and core counts would have fared much better against Sandy/Ivy Bridge than what AMD actually delivered with Bulldozer. Not to mention we likely would have had a much better memory controller than what was delivered on bulldozer if we weren't still bound by AM3's backwards compatibility in general. 

[mr moose]

5 hours ago, MageTank said:

K10 was not inherently bad though, especially compared to what followed with Bulldozer. It had a pretty substantial IPC boost (10% over bulldozer if memory serves correctly). They simply needed a refined process and the ability to scale it even further. It wouldn't have been any different than Intel's current 14nm approach, where it's the same thing but slightly more refined. 

 

I imagine a refined K10 that facilitated higher clock speeds and core counts would have fared much better against Sandy/Ivy Bridge than what AMD actually delivered with Bulldozer. Not to mention we likely would have had a much better memory controller than what was delivered on bulldozer if we weren't still bound by AM3's backwards compatibility in general. 

 

I agree with this,  I was a phenom user on a 2+ board.  The processor was more than capable but the motherboard let it down, this is one of the reasons I do not consider platform support to be long term plus but more a short term convenience for 1%er cases.  

 

Also with people only looking at what is happening in the industry over the last few years and not taking in the wider picture with consumer demand, bread and butter products, long term strategy etc, it's easy to see why people get so strung up over Insisting companies should be doing something other than what they are.

[tech.guru]

Intel major concern should be the fact they can't even keep up the supply of their 14nm processors.

 

Enterprises are the one area where sales hasn't declined as much. Yet they are waiting long time for chips.

 

How long till these oem give up and switch to amd? Intel should be worried

[tech.guru]

This is an article  

https://www.pcworld.com/article/3448396/intels-unexpected-prolonged-processor-shortage-dampens-its-record-quarter.html

 

Many vendors are slowly considering and testing amd processors for their servers. Enterprises move slowly and intel is still king.

 

If intel loses its enterprise customers, it will really be in a world of hurt. Right now its sales are from the growing demand in cloud computing and machine learning.

 

They better be careful , EPYC outperforms and considerable cheaper than their xeon products.

Intel only has its relationships with oems and cloud providers for growth, anger them with shortages and they may just jump ship.

[BroliviaWilde]

Intel 10-Core Comet Lake-S CPU Could Suck Up To 300W

 

If this turns out to be true.... OOOFFFFF