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

[mr moose]

2 hours ago, thechinchinsong said:

Don't really agree with the power delivery argument except for the 8th and 9th gen i7/i9 processors. For the 8700k and 9900k, I can see there being power delivery problems for those but for all 4c/8t processors from since 6th gen core and even 6c/6t 8th/9th gen chips, it wouldn't be a problem.  There has been very clear and consistent TDP levels within Intel's lineup for years now and if a mobo can handle a 95w 6700k, then it should handle a 95w 7700k and all processors below TDP level. Intel themselves also rate the 8700k and 9900k at 95w tdp but we all know those processors take a better power delivery compared to 7th and 6th gen.

Intel's TDP are only for base clocks,   If the processor tries to ramp up into boost and the mobo's power delivery is inadequate then you will have issues (especially on cheaper boards).  Intel are just avoiding that as it will look bad for them regardless.

[Kisai]

16 hours ago, thechinchinsong said:

Sorry I'm not exactly sure how all of that means its not possible. There have literally been mods to support 7th and 8th gen (not sure on 9th gen) Intel CPUs on the z170 chipset. Its clear that Intel could have easily made support official.

 

 

https://www.intel.ca/content/www/ca/en/support/articles/000025694/processors/intel-core-processors.html

Quote

9th and 8th Generation Intel® Core™ Desktop Processors use an LGA1151 socket. However, they require motherboards based on Intel® 300 Series Chipset. Some Intel® 300 Series Chipset based motherboards may need a BIOS update to support 9th Generation Intel® Core™ Desktop Processors. Please check with the motherboard manufacture for details.

The 9th and 8th generation parts work on the 300 chipset.

The 6th and 7th generation parts work on the 200 chipset and 100 chipset.

 

 

Try using DDR3 memory on a 7th or 8th generation CPU on a Z170, I dare you. 

 

Notice how different chipsets have different memory support. Those older chipsets also don't support the voltages needed to have more cores on the 8th generation parts. Also note that there are different iGPU parts. The 6th gen has "HD Graphics", 7th gen has "HD Graphics 530", 8th gen has "HD Graphics 630" 

 

Like those bios hacks, people might only get away with it if they disable the iGPU and have nothing plugged into the PCH PCIe lanes.  If you go read a thread on the modding, 9th generation cpu's kill motherboards without masking off some pins on the CPU. That's a lot of effort just to save the cost of replacing the motherboard and not getting the benefit of better memory on the same board.

 

300 series (Coffee Lake Refresh)

 

100 series (kaby lake)

100 series (sky lake)

 

 

[TechyBen]

On 10/12/2019 at 11:29 PM, NumLock21 said:

Going by your post it seems you're not satisfied with what Intel is doing and, you would rather pay $500+ for a Celeron dual core?

No. just that names mean nothing anymore. I mean, if we go by Intel's logic, this is Linus:

 

[thechinchinsong]

2 hours ago, Kisai said:

 

https://www.intel.ca/content/www/ca/en/support/articles/000025694/processors/intel-core-processors.html

The 9th and 8th generation parts work on the 300 chipset.

The 6th and 7th generation parts work on the 200 chipset and 100 chipset.

 

 

Try using DDR3 memory on a 7th or 8th generation CPU on a Z170, I dare you. 

 

Notice how different chipsets have different memory support. Those older chipsets also don't support the voltages needed to have more cores on the 8th generation parts. Also note that there are different iGPU parts. The 6th gen has "HD Graphics", 7th gen has "HD Graphics 530", 8th gen has "HD Graphics 630" 

 

Like those bios hacks, people might only get away with it if they disable the iGPU and have nothing plugged into the PCH PCIe lanes.  If you go read a thread on the modding, 9th generation cpu's kill motherboards without masking off some pins on the CPU. That's a lot of effort just to save the cost of replacing the motherboard and not getting the benefit of better memory on the same board.

 

300 series (Coffee Lake Refresh)

 

100 series (kaby lake)

100 series (sky lake)

 

 

Most all z-170 through z-370 motherboards were ALL DDR4. The memory compatibility argument still doesn't make any sense. It says right there in the diagrams you posted that all three chipsets use DDR4. Furthermore, the original point of my argument was that they didn't need to change around like 5-6 ground pins just to screw over previous gen consumers. Intel could have easily kept those pins that must be modded in the same orientation as the previous gen 1151 sockets. Yes, it takes some physical modding to be able to use it from an unofficial standpoint, but my point stands that if some enthusiasts can get basic functionality through taping off a couple pins and a modded bios, Intel should have been able to easily retain functionality of their newer chips on the older motherboards.

[thechinchinsong]

18 hours ago, mr moose said:

Intel's TDP are only for base clocks,   If the processor tries to ramp up into boost and the mobo's power delivery is inadequate then you will have issues (especially on cheaper boards).  Intel are just avoiding that as it will look bad for them regardless.

That's true. Yes the higher core count (really only the i7-8700k and i9-9900k are relevant candidates, all others are the same core counts or similar enough), might not be able to boost to their full potential, but the core functionality (base clocks) still stands. For basically all main brand motherboards, there is adequate thermal sensing to be able to throttle back to a safe operating temperature. I can see reasons for Intel recommending updated motherboards due to the need to push partner mobos and chipsets, but if you really look into the efficiency and capability of many z-170 VRM compared to many z-370 VRM, you will find that they have similar efficiencies and capabilities at the same tiers of TDP. 

[mr moose]

5 hours ago, thechinchinsong said:

 but if you really look into the efficiency and capability of many z-170 VRM compared to many z-370 VRM, you will find that they have similar efficiencies and capabilities at the same tiers of TDP. 

Yes, but some do not, Intel can't just put out a statement saying check your motherboards power delivery for compatibility.  Who decides which boards are compatible and by what measure?   Which means Intel either have to test every single board and approve it or they say it only is guaranteed to work on chipset X where they can stipulate the minimum power delivery requirements.

[The117thCon]

On 10/13/2019 at 5:26 AM, Rohith_Kumar_Sp said:

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. 

Any Consolation to either of you I got locked into a 4790K as i was doing upgrades to an old pc, but the mobo and psu died and were proprietry parts so had to basically do a full brand new from the ground up build witk my 4790K which sucks as i know im never gonna get a return on it in the next 5-10 ish years when I upgrade.

[thechinchinsong]

On 10/17/2019 at 5:26 PM, mr moose said:

Yes, but some do not, Intel can't just put out a statement saying check your motherboards power delivery for compatibility.  Who decides which boards are compatible and by what measure?   Which means Intel either have to test every single board and approve it or they say it only is guaranteed to work on chipset X where they can stipulate the minimum power delivery requirements.

I'm saying that everything will work with no problems at least for base clock from every processor in the mainstream lineup on all Z170-Z390 chipsets for all 6th - 9th gen processors. Even the i9-9900k will work on the lowest end Z-170 mobo VRMs at base clock (which is and has been 95W for the top line Intel Core series since forever), which is the only thing Intel needs to fulfill. Everything above base clock is extra technically. Obviously that isn't optimal for the highest end, but it gives the consumer options and costs Intel nothing. If a motherboard from Z170 can handle the 95W tdp of an i7-6700k, then it can handle the 95W tdp of an i9-9900k, at least for base clocks, otherwise that's an issue on the motherboard manufacturer, not Intel. 

[mr moose]

6 hours ago, thechinchinsong said:

I'm saying that everything will work with no problems at least for base clock from every processor in the mainstream lineup on all Z170-Z390 chipsets for all 6th - 9th gen processors. Even the i9-9900k will work on the lowest end Z-170 mobo VRMs at base clock (which is and has been 95W for the top line Intel Core series since forever), which is the only thing Intel needs to fulfill. Everything above base clock is extra technically. Obviously that isn't optimal for the highest end, but it gives the consumer options and costs Intel nothing. If a motherboard from Z170 can handle the 95W tdp of an i7-6700k, then it can handle the 95W tdp of an i9-9900k, at least for base clocks, otherwise that's an issue on the motherboard manufacturer, not Intel. 

 

Intel can't just disable boost so it will work with all boards unchecked.   

[thechinchinsong]

8 hours ago, mr moose said:

 

Intel can't just disable boost so it will work with all boards unchecked.   

They don't have to, the chips themselves just won't boost as high.

[Murasaki]

Anyone else think Intel should start naming their SKUs a bit better. 2050 comes around with the realease of the i17-74007543K.

[mr moose]

3 hours ago, thechinchinsong said:

They don't have to, the chips themselves just won't boost as high.

you hope.  The are several cases documented where VRM temps have been too  high before under clocking the CPU whioch is below base and not what any consumer wants ( as well as reduced life expectancy and unpredictable future boost behavior), and again both are issues that effect Intel but that they have no control over.

[GoldenLag]

1 hour ago, comander said:

On balance moar cores increases throughout more than SMT. 

 

50% more cores > tech that on average gives a 25-30% throughout boost. 

 

Compare 8600k vs 7700k. 

For pure raw performance, yes. 

 

But when it comes to games. You are more likely to have a even experience with 4c/8t than 6c/6t

[GoldenLag]

19 minutes ago, comander said:

Same with 4C/8T vs 6C/12T

Actually its not the same with 6c and 4c as the game engines can usually take advantage of about 4+ threads. And that kinda sucks if you run out of threads.

20 minutes ago, comander said:

1) GPU performance matters ALOT more than CPU performance making hyperthreading debates mostly academic and not practical

We are talking about CPUs.....

21 minutes ago, comander said:

3) not everything is gaming. Heck most things aren't. 

Tell me a non-workstation worjload where you would benefit greatly  with the 6 core non-ht over the 4c HT.

22 minutes ago, comander said:

4) longer term peak throughout trends to dominate in even games. 

Kinda, but also not really when games start using more threads and the limiting factors are peak throughput on 1-3 cores and how many threads you have at your disposal.

 

[thechinchinsong]

On 10/19/2019 at 4:34 PM, mr moose said:

you hope.  The are several cases documented where VRM temps have been too  high before under clocking the CPU whioch is below base and not what any consumer wants ( as well as reduced life expectancy and unpredictable future boost behavior), and again both are issues that effect Intel but that they have no control over.

Like i said, that's a motherboard manufacturer problem. If a motherboard can't sustain base clocks, then that is the problem of the motherboard, not CPU maker. Yes, Intel doesn't have control over what decisions that the motherboard manufacturer makes, but if a motherboard can't handle base clock for all 95W TDP processors, Intel doesn't have to give their blessing for said motherboard then.

 

Documented cases where VRM temps are too high are even with new motherboards every generation. You have sub-par motherboards no matter how stringent you are with product segmentation. Better to let the consumer choose (to not buy those motherboards).

[mr moose]

1 hour ago, thechinchinsong said:

Like i said, that's a motherboard manufacturer problem. If a motherboard can't sustain base clocks, then that is the problem of the motherboard, not CPU maker.

But it does reflect on Intel and is a problem for them.    If I was Intel I wouldn't do it either for that reason alone.

 

1 hour ago, thechinchinsong said:

Yes, Intel doesn't have control over what decisions that the motherboard manufacturer makes, but if a motherboard can't handle base clock for all 95W TDP processors, Intel doesn't have to give their blessing for said motherboard then.

And the only way Intel can give their blessing is if they test every motherboard,  as I already said that is not going to happen because it is not worth their time or effort for the small portion of their consumer base who would need it.

 

1 hour ago, thechinchinsong said:

 

Documented cases where VRM temps are too high are even with new motherboards every generation. You have sub-par motherboards no matter how stringent you are with product segmentation. Better to let the consumer choose (to not buy those motherboards).

https://www.extremetech.com/computing/251882-overclocker-claims-x299-vrm-temperatures-disaster-limit-skylake-x-potential

https://www.techpowerup.com/234922/update-on-the-intel-x299-platform-vrm-disaster

 

These are not low end boards and highlight the exact issue that can and does occur.

 

When the CPU is throttled down to 1.2Ghz because the VRM can't keep up then that looks bad for Intel regardless who's fault it is.  That's why Intel outright doesn't list them as being compatible. 

 

[GoldenLag]

9 hours ago, comander said:

My original point was that for 99% of people that the difference in games for 4c/8T vs 6C/6T is negligible.

My point is that you are less likely to run into stuttering issues that have even started to plague 6c/6t CPUs.........

 

9 hours ago, comander said:

My original point was that for 99% of people that the difference in games for 4c/8T vs 6C/6T is negligible.

What is noticable is stuttering. Something higher threadcounts are much better at avoiding. 

[Man]

Intel increases core counts.

I blame AMD...

I'm an Intel and Nvidia fan who's probably never going to own AMD hardware ever but... God bless AMD!

[leadeater]

30 minutes ago, comander said:

The 7700 doing better is the exception, not the rule. The average case is no significant difference or a modest edge to the 9600. 

Going forward that's not going to be the case, i3's having 4 cores and HT is going to make them seriously good options on a budget and there will be no shortage of them once they go through the business workstation 4-5 year cycle then flood ebay too.

 

Today I don't think it's that good of an idea to buy a 6c/6t part unless you specifically intend to upgrade after 2 years.

[RejZoR]

Why bother with that if you can get a Ryzen 5 with 6 cores and 12 threads for relatively reasonable price that will last you for many years.

[leadeater]

2 hours ago, comander said:

I do recognize that there are a lot of 4C/8T parts on the market and that there will be more in the future but in think that argument requires an assumption that developers won't design for consoles first/factor them in significantly.

Current consoles have 8 cores, console optimization does not effect PC optimization much at all. What effects PC optimization is the general common hardware that developers then wish to target, that is primarily why 4 core CPUs do so well but a bit less so now if they don't have HT.

 

Why 4c/4t threads don't do as well now has a bit less to do with the number of threads a game uses but more the average utilization of them being high then something else, not the game, requesting CPU time. This you don't even seen in 0.1% lows that well but you do see it in frame time graphs where you get very large spikes which is much more interruptive to game play than just generally low 1% and 0.1% lows.

 

2 hours ago, comander said:

Part of the reason why titles have been relatively CPU insensitive the past few years was the mix of little progress from Intel and the fact that Jaguar based consoles (1.6 GHz with half the performance per clock of Zen 2/lake) needed to be considered.

PS4 was released in 2013, 6 years ago. This has had little effect at all or you would have seen it in a 3 year time period i.e. next iteration of games.

[otennicegg3]

I saw this coming kinda back when 1st gen Ryzen first came out. At least we see Intel trying to stay competitive unlike the pre-Ryzen days...

[GoldenLag]

4 hours ago, comander said:

provided benchmarks where the 7700k generally failed to provide better performance, including at lows, despite being overclocked, providing it a frequency advantage

Im mostly reffering to significant deltas between average and low framerate. And admiteddly it doesnt affect a whole lot of titles at this point

[leadeater]

28 minutes ago, comander said:

It's HIGHLY probably that just about every game engine put out in the last 6 years at the very least had some considerations towards the consoles. 

 

Try getting millions of dollars in funding for a project that ignores a HUGE chunk of the addressable market... It's hard. 

Yes but the optimization of them does not have to correlate so directly, it's not required. Consideration for the available hardware resources and the optimization of those are two different things. Just because one is 8 cores 8 threads and the other is 4 core 8 threads doesn't change how much CPU demand you envision for the game, that distribution of threads is purely optimization.

 

Just because you can allocate tasks out to more threads doesn't actually mean performance will improve, it also doesn't actually necessitate running the threads on their own dedicated CPU threads. If the CPU thread has enough resources to concurrently execute two game threads then you can do so. That's why a console with weak 8 cores doesn't effect the optimization of PC games because the existence of 8 cores is actually pretty negligible factor by itself.

 

28 minutes ago, comander said:

GN concluded that a Pentium G4560 wasn't significantly holding back a 1060 or 1070 depending on the title)

GN is also the one that carried out the tests to show 4c/4t and 6c/6t is being impacted by the frame time issues I talked about, this is also why 6c/6t is not a good investment if you are the type to try and carry the CPU for as long as possible. I am one of those, I still run my 4930k today and unless I try and run extremely high frame rates there is no need to upgrade now or even in the future (but I will for reasons other than CPU resources).

 

28 minutes ago, comander said:

I just don't see 4C/8T as much better than 6C/6T in the bulk of scenarios as the latter has more overall horsepower and avoids issues like cache thrashing.

I'm not saying it's better, but if you have that now it'll likely survive longer than 6c/6t when it comes to frame times. i3's coming on to the market with decent 4 cores and 8 threads is a good option at the right pricing and is much better value than what Intel is asking for to get 6c/6t, this change essentially makes i5 even more useless and i7 died when Intel removed HT from them so that doesn't change.

 

Right now Intel is that footballer that shoots for goal, hits the crossbar and rebound knocks themselves out with a ball to the face. Luckily easily shaken off.

[leadeater]

36 minutes ago, comander said:

Would you mind digging up your GN reference? The closest thing I'm finding is this - https://www.google.com/amp/s/amp.reddit.com/r/hardware/comments/a0nhns/gamers_nexus_intel_i59600k_review_vs_r7_2700_r5/

 

While it does note consistency issues relative to it's more highly threaded counterparts it doesn't make any comparisons to the 7700/6700. 

 

Currently running on 3 hours of sleep and trying to pass out while sitting down at an airport so not 100% with it right now. 

From memory it's spread all over the place and comes up in his talking points across a few videos etc.

 

But it's talked about in these:

https://www.gamersnexus.net/guides/3423-intel-i7-7700k-revisit-benchmark-vs-9700k-2700-9900k

Why 0.1% lows are not done on Far Cry 5 only I have no idea but you can see the 1% lows are better on the 7700k than the 9700k and the 0.1% are much better if you look at other sources.

 

It's a rather game specific issue, like with GTAV, but this might start cropping up more unless something is done to remedy the problem which I admit can likely be done.

 

https://www.gamersnexus.net/hwreviews/3407-intel-i5-9600k-cpu-review-vs-2700-2600-8700k

https://www.gamersnexus.net/guides/2878-gta-v-stuttering-mystery-on-i5-r5-cpus-part-2

 

And some older context

tl;dr For a long time game engines only effectively utilized 2 physical CPU cores even if 4+ threads are be utilized, more that isn't a thing the more threads start to matter.