Speculation: Intel unlocked 2C/4T?

cbn

Lifer
Mar 27, 2009
12,968
221
106
Next year?

Optane memory enabled?

P.S. Last unlocked 2C/4T was the Core i3 7350K (4MB cache with 4.20 Ghz clock, GT2). Current Fastest Pentium is G5600 (4MB cache, 3.9 Ghz clock, GT2). Seems like we might be getting close to seeing unlocked multiplier come back to 2C/4T.
 
Last edited:

hojnikb

Senior member
Sep 18, 2014
562
45
91
I'd wish unlocked multi on the whole lineup. And maybe offer overclock options on cheaper boards as well.
 

SPBHM

Diamond Member
Sep 12, 2012
5,056
409
126
Why the hell would anyone bother overclocking a bargain basement CPU with only 2 cores in 2019? 4 cores is the baseline now.

well, a 5GHz Pentium with 2c/4t would be a fun toy
but it doesn't really make a lot of sense since an i5 8400 with an h310 board and stock cooler will demolish it for most things and the Pentium would still require a good cooler and z370

the way I see it is... why not? Intel would not lose anything with it and would gain some favor with enthusiasts IMO

but I don't see it happening,
still I also didn't expect the G3258 to exist.
 
  • Like
Reactions: VirtualLarry

cbn

Lifer
Mar 27, 2009
12,968
221
106
Remember the software codes Intel sold to unlock Cache and hyperthreading on Pentium back in 2010?

Maybe they could bring that back in some form?

Then we could unlock the 2C/4T into something better (unlocked 4C/4T, unclocked 6C/6T, etc) when it eventually runs out of steam (ie, stutters with a video card upgrade, etc).

And I actually think Intel might agree if they see potential to regain some revenue that they would otherwise lost.

--------------------------------------------------------------------------------------------------------------------------------

Swapping out unlocked 2C/4T for a used processor (either locked or unlocked) with higher core count

vs.

Buying a new motherboard or budget system along with a new budget chip with higher core count

vs.

Buying a software code that allows higher core count on the existing processor.

(The last option seems the best for both us and Intel. This assuming Intel has enough 6C dies they don't mind temporarily making 2C/4T)
 
Last edited:

DominionSeraph

Diamond Member
Jul 22, 2009
8,391
31
91
2/4 is a different die than 4/4, and they've probably removed the hyperthreading on the 4 and 6 core parts so they couldn't even make one if they wanted.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
Some information on Intel Upgrade Service from Wikipedia:

https://en.wikipedia.org/wiki/Intel_Upgrade_Service

Notice the following problem mentioned:

One reviewer noted that at the market price of the time one could actually buy the i3-530 for only $15 more than the baseline Pentium G6951, making the upgrade premium card a very questionable proposition at the official price.

......So to make an upgrade service work they have to consider price adjustments as time goes by.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126
Remember the software codes Intel sold to unlock Cache and hyperthreading on Pentium back in 2010?

Maybe they could bring that back in some form?

Then we could unlock the 2C/4T into something better (unlocked 4C/4T, unclocked 6C/6T, etc) when it eventually runs out of steam (ie, stutters with a video card upgrade, etc).

And I actually think Intel might agree if they see potential to regain some revenue that they would otherwise lost.
I've suggested this several times, "Intel Processor DLC" codes. I can see this being the thing, in the future.

OldSkool mainframe companies used to do this, as did, I think, DEC.

I mean, honestly, so few mainstream / non-enthusiast users actually upgrade their CPUs, right now, due to both technical and cost concerns.

If the procedure for a "CPU Level Up", could be made as simple as entering a code in the BIOS (or a Windows app), and voila, reboot and you're upgraded, I think that a LOT more people would do it.

I mean, if Windows' ever becomes a monthly service (well, MS Managed Desktop is going in that direction), then perhaps CPU upgrades could be "rented" along with that. Kind of like a hybrid of personal and cloud computing. You would order the "base" PC, and then pay monthly for the level of hardware that you wanted on it. Without having to physically change anything out on it.

This could also lead to more integrated PCs, in terms of form-factor and cooling, when upgrading becomes virtual, rather than physical.
 
  • Like
Reactions: cbn

DominionSeraph

Diamond Member
Jul 22, 2009
8,391
31
91
I've suggested this several times, "Intel Processor DLC" codes. I can see this being the thing, in the future.

I don't. Die area costs money so it doesn't make sense for any part that isn't being specifically downgraded to hit a market niche. You have a CPU that could be sold fully enabled for a higher price, and few people would be inclined to upgrade in the first place, so you'd have to charge so much for an individual upgrade to make up for the lost profits that it wouldn't make sense for anyone to upgrade, or so much more for the initial purchase that it looks really bad compared to its competitors.
A $200 Pentium with the $1000 option to upgrade it to an i9 looks pretty bad compared to an i5.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,229
9,990
126
I don't. Die area costs money so it doesn't make sense for any part that isn't being specifically downgraded to hit a market niche. You have a CPU that could be sold fully enabled for a higher price, and few people would be inclined to upgrade in the first place, so you'd have to charge so much for an individual upgrade to make up for the lost profits that it wouldn't make sense for anyone to upgrade, or so much more for the initial purchase that it looks really bad compared to its competitors.
A $200 Pentium with the $1000 option to upgrade it to an i9 looks pretty bad compared to an i5.
I disagree. Look at the bigger picture.

It costs Intel $50 to fab, test, and package a CPU, let's just say. (I've heard estimates around that.)

You might say, why does Intel sell some CPUs for less than that? It's down to market-share, and ASPs.

I don't know if Intel sells low-end at a loss, but look at it this way. If it costs Intel $50 to make, regardless, then if they sell the base model at cost, effectively, and require DLC for each tier above that, and assuming that 75% of people upgrade (admittedly a high and arbitrary number), whereas, only 50% of the people would have purchased the top-end chip outright, then, isn't that a "win" for ASPs over the lifetime of those CPUs?

It could be a product differentiator too. Instead of people telling other people, that Intel platforms are inherently "dead sockets", due to lack of upgradeability, they could instead speak of Intel CPU's "easy Level-Up capability". (Software-based microcode updates to enable more functionality.)

This might also be a somewhat savings for Intel and their distributors, and especially the branded OEMs, because they wouldn't have to stock a ton of different CPUs. Just one or so for each generation.

In fact, if there were no physical socket upgrades per platform release, Intel could do away with the socket entirely. I know that they've toyed with this idea as well.

Edit: You might say, what about overclockability, and burning out chips, and replaceability. Well, there would still be HEDT for that. I could see consumer Intel platforms, moving to BGA-only, software upgradeable, and limited packaging and test options, and HEDT for the real enthusiasts, socketed, overclockable, good VRMs/heatsinks, the works.

All while mainstream Intel platforms, would become roughly plain-vanilla, same mainboard, same CPU, etc., for all of the OEMs, mfg'd at large-scale, and soft-upgradeable, with integrated cooling and whatnot. No overclocking on consumer platforms anymore. Just Stable, Stable, Stable, for businesses and schools, and instead of replacing the whole PC after three years, simply do a "soft upgrade" on the CPUs. By them, OEM desktops should be using SSDs too, which generally-speaking, last and last and last, unless you have a power glitch or the controller somehow randomly barfs.
 
Last edited:

whm1974

Diamond Member
Jul 24, 2016
9,460
1,570
96
I'm not sure I would even bother with producing a Dual core/quad thread CPU if I was a X86-64 CPU manufacturer unless it will have considerably low TDP and power consumption with a decent amount of performance. And of course dirt cheap.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
According to Wikichip Intel doesn't make a dual core die for Coffee Lake----> https://en.wikichip.org/wiki/intel/microarchitectures/coffee_lake

Coffee Lake only comes in 4C + GT2, 6C + GT2 and 8C + GT2 variants. (So chips like G5600 are already being made from a quad core die at the very minimum).

P.S. 6C Coffee Lake vs. 8C Coffee lake:

650px-coffee_lake_die_%28hexa_core%29_%28annotated%29.png


800px-coffee_lake_die_%28octa_core%29_%28annotated%29.png
 
Last edited:

whm1974

Diamond Member
Jul 24, 2016
9,460
1,570
96

cbn

Lifer
Mar 27, 2009
12,968
221
106
I have to admit I am pretty surprised there is no dual core die.

This (Coffee Lake) is the first Intel generation to skip over offering 2C die (Cannonlake is still 2C die though)
 

PotatoWithEarsOnSide

Senior member
Feb 23, 2017
664
701
106
If their dual cores are made from quad core parts, it is because the dies are defective thus unsellable as a quad.
It's the same principle as to why AMD's 8C chiplet design is a work of art; defective dies still yield CPUs that can be sold.
Selling fully functional 8C parts as software disabled 4C parts would not be the world's best idea. Defective dies that can be repurposed, have zero cost to produce, and any revenue generated is a net positive, whereas as a functional 8C repurposed will always be a net loser.
It simply makes no financial sense to gimp an otherwise fully functional product.
 
  • Like
Reactions: whm1974

whm1974

Diamond Member
Jul 24, 2016
9,460
1,570
96
I have to admit I am pretty surprised there is no dual core die.

This (Coffee Lake) is the first Intel generation to skip over offering 2C die (Cannonlake is still 2C die though)
The reason I would even bother with designing a 2c/4t CPU is if I can get a lower then 35W TDP and less power consumption at 3000Mhz with a 1000hz iGPU that has 1030 level performance.

Is this even possible?
 

PotatoWithEarsOnSide

Senior member
Feb 23, 2017
664
701
106
I doubt there is money to be made from it even if we consider it to have been harvested from a defective die.
Intel would need AMD/Nvidia collaboration, since their own iGPU tech is severely lagging, and AMD seem to have 7nm APU as their lowest priority.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
This might also be a somewhat savings for Intel and their distributors, and especially the branded OEMs, because they wouldn't have to stock a ton of different CPUs. Just one or so for each generation.

In fact, if there were no physical socket upgrades per platform release, Intel could do away with the socket entirely. I know that they've toyed with this idea as well.

Just think about how nice that would be for laptops.

However, since some chips have defects not all processors, of course, could be "flexible SKU/Full software unlockable".....but some of them could be.

Or what Intel could do when chiplets and EMIB comes out is have a certain processor/APU made up entirely of good chiplets. This platform (possibly higher end) could be the dedicated "flexible SKU/Full software unlockable" platform while the other (possibly lesser processors) could be what we consider conventional or closer to conventional. Or maybe with chiplets and EMIB this kind of thing could extend to even the lower end as well as you appear to be writing about?
 
Last edited:

PotatoWithEarsOnSide

Senior member
Feb 23, 2017
664
701
106
As an aside, lets say that cores could be re-enabled by microcode updates. How does one go about protecting that microcode in order to be able monetize its use?
I don't doubt that you could earn more revenue from a CPU by allowing cores to be re-enabled, since all that would be required is to charge more than the difference between the initial retail prices of each variant. However, that solution impacts upon future sales, with a consequence of reduced long term revenue. It also requires that generational leaps in performance would need to be vastly increased compared to what we see now.
 

Shivansps

Diamond Member
Sep 11, 2013
3,835
1,514
136
The way i see this going i would not be suprised to see 2019 Pentiums ending as 4/4, with the posibility of Celeron being 2C/4T based either in old die or a cut down CFL die.

AMD already has a 2C/4T in the Celeron price range.
 
  • Like
Reactions: Insert_Nickname