Core i3 is now a Core i7: Intel increases core counts

[Guest]

thats weird. I've never seen a CPU company do this before, increasing core counts while not bothering with efficiency 

[NumLock21]

2 hours ago, dizmo said:

Most people didn't really need hyper threading. I think it made sense to leave it off the chips.

HT does help with performance so having it and letting the user decided if they want to enable or disable is better than no having it at all. The lack of HT on the desktop 9th Gen Core i7 was either, they want to add the Core i9 or from some article saying HT has a security risk. It will forever be one of those weird CPUs Intel has made along with their Core X series i5 7640x and Core i7 7740x.

[QXC]

1 hour ago, floofer said:

thats weird. I've never seen a CPU company do this before, increasing core counts while not bothering with efficiency 

You don't remember the first dual core CPUs in the mid 00s, do you?

Or just generally how CPU performance went until Intel hit the clock speed wall and suddenly we had to care about how much power we threw at our CPU.

[williamcll]

Wonder how well the iGPU would be compared to the ryzen 3200G

[Fasauceome]

4 hours ago, NumLock21 said:

The pic is from a benchmarking tool that grabs the info directly from the system, it would be pretty difficult to fool it, unless that person has way too much time on their hands and edited that picture.

"Too much time" c'mon, an amateur can whip that up in like 2 minutes

[NumLock21]

2 minutes ago, Fasauceome said:

"Too much time" c'mon, an amateur can whip that up in like 2 minutes

2 minutes is a lot of time

[JoostinOnline]

3 hours ago, Opencircuit74 said:

*Looks at my i7-6700k with a measly 4 cores and 8 threads*  "You got this buddy, I don't have the budget to replace you."

Hey, I've got a 4790k. At least you don't have to replace your RAM.

[AkatsukiKun]

2 hours ago, Caroline said:

When you can make new CPUs run on the current socket but add 8 dummy ground pins just to kinda justify why you're ripping off users

How many 115x sockets have we been at now? Don't forget that Skylake, Kabylake both used 1151, same with Coffelake using the same 1151. Basically intel has been effectively using the same architecture deep down for 4 generations of cpus and the same socket. Don't get me started on tick-tock, they're running out on both by now.

[mxk]

28 minutes ago, NumLock21 said:

2 minutes is a lot of time

yeah I could poop in all that time

[mr moose]

1 hour ago, AkatsukiKun said:

How many 115x sockets have we been at now? Don't forget that Skylake, Kabylake both used 1151, same with Coffelake using the same 1151. Basically intel has been effectively using the same architecture deep down for 4 generations of cpus and the same socket. Don't get me started on tick-tock, they're running out on both by now.

if it ain't broke don't fix it.  Even though they have made what appears to bevery little revision on the surface (for more than 4 years), they're still releasing CPU's that are incrementally better and keeping up.

 

 

[Rohith_Kumar_Sp]

1 hour ago, JoostinOnline said:

Hey, I've got a 4790k. At least you don't have to replace your RAM.

i'm in the same boat. got this PC in 2014, at this point i can't upgrade anything without changing the whole PC except my graphics card. 

[leadeater]

5 hours ago, porina said:

If we're going to guess:

i3 4c8t

i5 6c12t

i7 8c16t

i9 10c20t

Seems pretty logical, this won't happen.

[GoldenLag]

7 hours ago, dizmo said:

The i3 is priced considerably cheaper than the Ryzen 5. So yes, if they keep the same pricing, it's absolutely possible, even if you have to buy a cheap cooler.

This i3 more or less killed the r5 3500. While 2 extra cores is good. The extra threads im gonna see being more useful in the long run. Ill wait for benchmarks offcourse. But it lines up to being cheaper than the 3500. With probably similar singlecore performance. 

[porina]

8 hours ago, Bananasplit_00 said:

Also Intel just can't stop making the I7 6700k huh

6700k was EOL over a year ago. If you mean "some Intel 4c8t" part, even if there isn't any significant difference in the architecture, there have been at least two refinements to process over that time gaining ball park 300+ MHz at comparable power (for a highly OC air cooled CPU).

7 hours ago, floofer said:

thats weird. I've never seen a CPU company do this before, increasing core counts while not bothering with efficiency 

There have been efficiency increases between generations. The bigger problem hit has is that 10nm, which was due around Kaby Lake, was totally broken. It isn't something you patch up and ship and they've practically missed a process generation because of that.

5 hours ago, williamcll said:

Wonder how well the iGPU would be compared to the ryzen 3200G

Probably unchanged so mostly sucks.

27 minutes ago, GoldenLag said:

This i3 more or less killed the r5 3500. While 2 extra cores is good. The extra threads im gonna see being more useful in the long run. Ill wait for benchmarks offcourse. But it lines up to being cheaper than the 3500. With probably similar singlecore performance. 

Reports are the 3500 will be a China only part. Let's see if that remains the case.

[GoldenLag]

8 minutes ago, porina said:

Reports are the 3500 will be a China only part. Let's see if that remains the case.

huh, i forgot about that. yay for making us not suffer non-ht parts. hopefully the CPUs will be mostly present in gaming cafe`s where there will be mostly esports titles

 

it also means we will probably not see many non-ht parts in the future, which is fantastic.

10 minutes ago, porina said:

There have been efficiency increases between generations. The bigger problem hit has is that 10nm, which was due around Kaby Lake, was totally broken. It isn't something you patch up and ship and they've practically missed a process generation because of that.

they have essentially been polishing the sandy bridge architecture. and atm there is a limit to how much polishing you can actually do. 

 

 

[porina]

2 minutes ago, GoldenLag said:

they have essentially been polishing the sandy bridge architecture. and atm there is a limit to how much polishing you can actually do. 

They have already designed and built the next architecture with similar IPC gains as Zen 2, by the name of Sunny Cove. The only problem for us is they seem to only want to introduce it with 10nm, so you can only get it in some mobile CPUs right now, and server CPUs expected 1H next year. Desktop hasn't even been mentioned.

 

Still, there have been big increases in IPC since Sandy Bridge. Haswell bought a big jump, and another with Skylake. Even if it kinda stalled since then (putting aside Skylake-X), their process improvements have uplifted clocks.

[GoldenLag]

Just now, porina said:

server CPUs expected 1H next year.

what sort of server CPUs? LP Servers or compute servers? because they would need to do a large chip to get any decent performance compared to their older offerings or Epyc. 

1 minute ago, porina said:

their process improvements have uplifted clocks.

their 14nm process is probably the greatest thing they have to offer atm. for the "size" of the node. its some damn good clockspeeds. suprised they dont do 14nm GPUs on it. 

[porina]

36 minutes ago, GoldenLag said:

what sort of server CPUs? LP Servers or compute servers? because they would need to do a large chip to get any decent performance compared to their older offerings or Epyc. 

I've not followed that closely since it isn't a particular interest area of mine. The statement I remember was that mobile CPUs on 10nm would be available by end of this year (done), and 10nm server 1H next year.

 

36 minutes ago, GoldenLag said:

their 14nm process is probably the greatest thing they have to offer atm. for the "size" of the node. its some damn good clockspeeds. suprised they dont do 14nm GPUs on it. 

Who's 14nm GPUs? Not Intels, since they don't currently have one, and their future one is probably designed for 10nm or newer. Given their ongoing lack of capacity to make CPUs, they're not going to offer it outside.

[leadeater]

6 minutes ago, porina said:

I've not followed that closely since it isn't a particular interest area of mine. The statement I remember was that mobile CPUs on 10nm would be available by end of this year (done), and 10nm server 1H next year.

Ice Lake-SP, it's more bringing technology features to compete on the market better like PCIe 4.0. The socket is also larger to support 8 channel memory and (assumption) more cores, going to be a rather big CPU with near as much same pins as AMD EPYC (for size reference).

[GoldenLag]

9 minutes ago, porina said:

Who's 14nm GPUs?

their own designs. 

10 minutes ago, porina said:

Given their ongoing lack of capacity to make CPUs, they're not going to offer it outside.

for now, eventually the node will be too old for their then current production, would be a shame of them to just leave such a good node without offering contract manufacture on it.

[porina]

22 minutes ago, GoldenLag said:

for now, eventually the node will be too old for their then current production, would be a shame of them to just leave such a good node without offering contract manufacture on it.

I see them continuing as they do now. Old fabs get refurbished to new nodes. 14nm wont stick around forever. It can still be useful for less important things like chipsets or other parts of their portfolio. It would also be quite a shift in their business to be a contract fab, and their design rules will be different enough others can't just drop something in without a redesign. The only way I can see Intel going this route is if they give up on x86. Leave general fabs to TSMC and Samsung.

[RobFRaschke]

On 10/12/2019 at 4:17 PM, porina said:

HT/SMT is a curious feature. It helps some things a lot (e.g. Cinebench), most things a little, some things not at all or even causes negative effects (e.g. FPU heavy compute cases). In my testing, best real world use case you get 50% more performance from HT/SMT. Cinebench R15 and R20 gain in the ball park of 30%. It's clear from AMD's direction, they want to add even more to the core, performance which might only be usefully extracted at higher thread per core counts. It remains to be seen what Intel are doing. Need a refresher on Sunny Cove changes later. I wonder what would happen if we go the other direction, take out HT/SMT. How much space does that save? Does it allow more real cores?

 

Anyway, we have to go back to Kaby lake for quad cores above i3 level.

i5-7600k, 4c4t, 3.8 base, 4.2 turbo, 

i7-7700k, 4c8t, 4.2 base, 4.8 turbo

 

Going to 8th and 9th gen (Coffee Lake) we have the x350k topping the quad core parts:

i3-8350k, 4c4t, 4.0 base, no turbo

i3-9350k, 4c4t, 4.0 base, 4.6 turbo

 

The one in OP

i3-10100, 4c8t, 3.6 base, unknown turbo

 

The model ending 100 implies it is a low end part. It might be interesting to see if they do an i3-10350k. Could be very interesting if they aim for single core 5.0 turbo.

 

The i3-10350K was my absolute first thought after seeing this leak. If they can get that in around the same price as the 9350k is right now around $180 it would be fantastic value and could actually rather upset the R5 3600. It is fewer cores and threads, but as we've seen core frequency is not nothing, and if we see relative per core increase similar to we've seen for the last two generations, it could really be a humdinger. Especially if we can do 5.0ghz all core with some extra voltage. I would actually have an intel CPU for the first time since my Pentium D 940. Because I'm stubborn.

[ImNotThere]

On 10/12/2019 at 11:13 PM, The1Dickens said:

I feel like Intel is just loosing control of their own product lines. If the i3 has HT, does that mean the i5 will? Why add HT to all the CPUs suddenly, when just last generation they removed it from all but the new i9 series? What happens to the 4c i5 when you can get a 4c i3, if both have HT? They spent that last few generations flooding the market with products for the only reason of charging obscene prices that now their own product lines are likely going to be cannibalizing each 

Well traditionally, the i3 has hyperthreading and will have the same "core count" as an i5, eg. 2c4t vs 4c4t, if this 4c8t i3 is accurate then we'll probably see an 8c8t i5 followed by presumably an 8c16t i7 and probably anywhere from 10-12c with 20-24t i9, this will be incredible for the consumer market

[The1Dickens]

2 hours ago, ImNotThere said:

Well traditionally, the i3 has hyperthreading and will have the same "core count" as an i5, eg. 2c4t vs 4c4t, if this 4c8t i3 is accurate then we'll probably see an 8c8t i5 followed by presumably an 8c16t i7 and probably anywhere from 10-12c with 20-24t i9, this will be incredible for the consumer market

So then they have no reason to not have Hyper Threading on the i5? Maybe its just me, but I'm having a hard time thinking of a person who would buy an i5 over a lower priced i3 if this turns out to be true. Or, not spend the extra few bucks and get the i7 and have Hyper Threading on all the cores you are paying for. Their pricing tree must look incredibly narrow, considering their top i9 is under 1K this generation.

[Anomnomnomaly]

On 10/12/2019 at 11:13 PM, The1Dickens said:

I feel like Intel is just loosing control of their own product lines. If the i3 has HT, does that mean the i5 will? Why add HT to all the CPUs suddenly, when just last generation they removed it from all but the new i9 series? What happens to the 4c i5 when you can get a 4c i3, if both have HT? They spent that last few generations flooding the market with products for the only reason of charging obscene prices that now their own product lines are likely going to be cannibalizing each other.

Why add HT... because AMD has HT on ALL CPU's and all are also unlocked. Intel has no real choice but to do this to stay competitive and they also need to seriously consider dropping the xxxxk varieties too.