Intel Core i7-11700 or i7-11700K?

[Shou Miko]

I was recently asked which Intel CPU to buy from a friend that doesn't want AMD but Intel because he uses Intel Quick Sync and have some application that lists Intel and not AMD.

He also game so I told him that the price of the recent Intel Core i9-11900 and the i9-11900K is a joke for the price when the same 8c/16t cores.

i9-11900K is about £500 / €580
i9-11900 is about £400 / €465

I told him a i7-11700 would serve him better when he also wants PCI-E 4.0 for NVME storage but I am not sure if the extra £61 / €71 is worth it when the performance of the non-K vs K is so simular in games.

i7-11700K is about £367 / €427
i7-11700 is about £306 / €356

The 3 biggest differences are price, base clock and TDP because the 100MHz in the turbo frequency I am sure you can find on a Z590 chipset forcing a all-core boost in the bios.

I just feels that the i7-11700 with it's 65Watt TDP (I know Intel lies as always) is a good winner for the gaming performance and searching about the good old interweb and @W1zzard great review of the I7-11700KF there ain't much OC let in the Rocket Lake-S CPU's and the about 2% performance differince in CineBench there is doesn't should like £60 / €70 more performance in my brain.

 

[ThrashZone]

Hi,
I'd always do a k version because it's better to be able to oc than not be able too oc
Whether you do k or kfc version just depends on lunch preferences

I fully agree 11900k is stupidly priced lol

 

[Shou Miko]

Hi,
I'd always do a k version because it's better to be able to oc than not be able too oc
Whether you do k or kfc version just depends on lunch preferences

I fully agree 11900k is stupidly priced lol

Thanks for seeing the point and I am just not sure if the K version is worth the extra money when gaming is almost the same.

 

[ThrashZone]

Thanks for seeing the point and I am just not sure if the K version is worth the extra money when gaming is almost the same.

Hi,
Funky default intel power limits is why I'd always get k version
10900k is pretty lame without being able to get rid of intel nonsense in bios.

 

[Selaya]

Neither.
Skylake is better than Rocket Lake.
Try 10700(K), 10850K or 10900.

 

[Shou Miko]

Neither.
Skylake is better than Rocket Lake.
Try 10700(K), 10850K or 10900.

That's a no go he wants PCI-E 4.0 for NVME storage.

 

[Selaya]

Does he own an Optane P5800X?
Because that's the only PCIe 4.0 NVMe where the Gen 4 is actually more than just a decoration.

That being said, if you still are unable to sway your friend, 11700(K), obviously.

 

[oxrufiioxo]

While I agree the 10850k is the superior chip I'd go with the 11700k if i was doing a build for myself for the extra £60-70..... He can't really go wrong with either I was personally impressed by the 11500 just make sure he gets a board that is equipped with a vrm to handle it unlocked.

 

[ThrashZone]

Hi,
Only good time for a non k version would be a laptop.

 

[Bill_Bright]

Unless he has two otherwise identical machines running the same program side-by-side, it is unlikely he will be able to "see" the difference while actually gaming. On paper in benchmarks, yes - but real-world? Nope.

So it really comes down to "subjective perception" (what his brain thinks it sees), and "desire" - what does he want and will he be disappointed in himself for not spending the little bit extra? While some may not consider €70 a "little bit", when you consider the total price of the entire computer, factor in and spread out that extra cost over the life of the computer, plus add in the boost from the enjoyment value due to that "subjective perception" (ego stroking) with having the better processor, I say €70 is a tiny amount and to go for the maximum amount the budget allows - AS LONG AS that extra €70 on the CPU does NOT mean compromising in quality elsewhere, such as a quality PSU or case.

 

[Gwenved]

Unless he has two otherwise identical machines running the same program side-by-side, it is unlikely he will be able to "see" the difference while actually gaming. On paper in benchmarks, yes - but real-world? Nope.

So it really comes down to "subjective perception" (what his brain thinks it sees), and "desire" - what does he want and will he be disappointed in himself for not spending the little bit extra? While some may not consider €70 a "little bit", when you consider the total price of the entire computer, factor in and spread out that extra cost over the life of the computer, plus add in the boost from the enjoyment value due to that "subjective perception" (ego stroking) with having the better processor, I say €70 is a tiny amount and to go for the maximum amount the budget allows - AS LONG AS that extra €70 on the CPU does NOT mean compromising in quality elsewhere, such as a quality PSU or case.

I agree, I'm taking now F version as it allows me to spare some € better spent in other component. I stopped K version 10 years ago or more as overclocking wasn't necessary for me. At this time my i7-2600 paired with good memory and a good graphic card had large enough power for my use. Nowaday it is still the same or better.

 

[AusWolf]

Maybe I'm the minority here, but I always go with the non-K variant. Like said before, gaming experience is indistinguishable, and the lower power limits mean you don't necessarily need fancy cooling. Not to mention the lower price. Just be careful with the motherboard choice, don't buy or suggest anything without VRM heatsinks (it's usually a sign of bad quality).

And I agree, 11th gen Core i9 is a waste of money.

 

[Bill_Bright]

Maybe I'm the minority here

You're not.

 

[TheUn4seen]

Maybe I'm the minority here, but I always go with the non-K variant. Like said before, gaming experience is indistinguishable, and the lower power limits mean you don't necessarily need fancy cooling. Not to mention the lower price. Just be careful with the motherboard choice, don't buy or suggest anything without VRM heatsinks (it's usually a sign of bad quality).

And I agree, 11th gen Core i9 is a waste of money.

I tend to agree with this. With 11 series and for home use I wouldn't go above 11400 or at most 11600, that's what I intend to do as soon as I find a suitable motherboard. I'd say since about 8 or 9 series there's no actual point in going for high-end other than enlarging one's e-peen.

 

[AusWolf]

I tend to agree with this. With 11 series and for home use I wouldn't go above 11400 or at most 11600, that's what I intend to do as soon as I find a suitable motherboard. I'd say since about 8 or 9 series there's no actual point in going for high-end other than enlarging one's e-peen.

Well, that depends. I have a feeling the extra 2 cores might come in handy in the future, but going above Core i7 is really just for bragging rights (if that's a thing).

I myself am seriously considering retiring (or repurposing) my Ryzen 3 3100 and trying an 11700 in my thin SFF case just to see how it works under limited power options with limited airflow. The R5 3600 I tried not long ago didn't work so well, but I'm thinking that the larger monolithic die of the 11700 might dissipate heat better under similar power conditions than the smaller, more concentrated chiplets of the 3600.

 

[BarbaricSoul]

I may be bias, but what about a 10 core 20 thread 10850k? They sell for under $400 USD.

 

[The red spirit]

You can always raise TDP of chip with modifying PL values. You can safely ignore that spec. However the real hard limit is maximum all core boost. I don't know the exact number, but for i7 11700 it should be 3.8-4GHz. It means that if TDP limits aren't reached in all core load you will get no more than that. But in single threaded loads it is rated at 4.9GHz. That's solid. If you want to save some money, non k versions makes a lot of sense. Want it to be faster, raise PL values. But if you are interested in absolutely maximum performance then k version is the one you should get. Knowing how hard to cool k version is with power limits removed, there's almost no point in getting it. Considering that your friend is inexperienced, just get him non k version. Either way he/she won't utilize k features. It's just not worth it.

I'm not exactly sure what things your friend will do, but 10 core 10th gen i9 can be a better deal if workloads are heavily threaded. But whichever CPU you chose, make sure that the board you will be buying for it has good VRMs. On Intel platforms that is very important. If getting k variant means sacrificing on VRMs, then don't get k version.

 

[BarbaricSoul]

I see it's been discussed

 

[Shou Miko]

@BarbaricSoul and @The red spirit the 10th gen doesn't have PCI-E Gen4 and ruins most m.2. options on Z590 so as much of better value the i9-10850K are it not gonna be that one.

Going with a Z590 should allow for changing Intel's Tau values PL1 (Sustained) and PL2 (Peak) but I need to read more about this and also saw in a video from Hardware Unboxed about performance on B560 where the same cpu could performance with 30% difference depending on if manufacture was running the cpu written Intel's Tau specs for the cpu.

 

[AusWolf]

@BarbaricSoul and @The red spirit the 10th gen doesn't have PCI-E Gen4 and ruins most m.2. options on Z590 so as much of better value the i9-10850K are it not gonna be that one.

Going with a Z590 should allow for changing Intel's Tau values PL1 (Sustained) and PL2 (Peak) but I need to read more about this and also saw in a video from Hardware Unboxed about performance on B560 where the same cpu could performance with 30% difference depending on if manufacture was running the cpu written Intel's Tau specs for the cpu.

If even my rubbish H61 MSi board (in my signature) allows PL1 and PL2 modification, then every Intel motherboard should have an option for it (theoretically). The only thing is, if you want higher sustained clocks at the cost of power consumption and heat, then you'll need to choose a high quality board and decent cooling.

 

[The red spirit]

@The red spirit the 10th gen doesn't have PCI-E Gen4 and ruins most m.2. options on Z590 so as much of better value the i9-10850K are it not gonna be that one.

Does PCIe version even matter? You would be hard pressed to tell a difference between SATA SSD and NVMe SSD without looking at numbers as (1) SATA SSDs are fast enough, (2) Windows for some reason can't properly utilize all that extra speed. Making super fast SSDs somewhat pointless. The only thing I would care about in SSD is type of flash. If it's MLC it's good, if it's TLC it's tolerable and if it's QLC then it's e-waste. DRAM cache matters too. Beyond that there are almost no real differences, unless for you copying files takes up a lot of time. Then that's the only justifiable reason to shell out on fast SSD. For most people CPU will matter more than SSD, making i9 a quite reasonable buy when compared to 11700.

Going with a Z590 should allow for changing Intel's Tau values PL1 (Sustained) and PL2 (Peak) but I need to read more about this and also saw in a video from Hardware Unboxed about performance on B560 where the same cpu could performance with 30% difference depending on if manufacture was running the cpu written Intel's Tau specs for the cpu.

All Intel chipsets allow doing that. PL2 isn't exactly peak, it's sustained upper TDP for Tau duration. peak values are PL3 and PL4, that may or may not be user adjustable. BTW Intel's Z chipset is imo useless. The chips that would benefit from overclocking are locked, those that wouldn't benefit much are already clocked to almost their maximum clock speed. To be honest, AMD also has similar "problem". Nowadays TDP tweaking is much more meaningful and easier than overlocking. That HU video was mostly FUD. Sure boards now differentiate and try to hide poor VRMs by lowering PL values, but on any board with decent VRMs PL values are adjustable and default settings hardly matter. General consensus is that VRMs matter, just like they always did. There's one good thing for cheap boards, now you have more control with poor VRMs and can manage them better.

 

[AusWolf]

All Intel chipsets allow doing that. PL2 isn't exactly peak, it's sustained upper TDP for Tau duration. peak values are PL3 and PL4, that may or may not be user adjustable. BTW Intel's Z chipset is imo useless. The chips that would benefit from overclocking are locked, those that wouldn't benefit much are already clocked to almost their maximum clock speed. To be honest, AMD also has similar "problem". Nowadays TDP tweaking is much more meaningful and easier than overlocking. That HU video was mostly FUD. Sure boards now differentiate and try to hide poor VRMs by lowering PL values, but on any board with decent VRMs PL values are adjustable and default settings hardly matter. General consensus is that VRMs matter, just like they always did. There's one good thing for cheap boards, now you have more control with poor VRMs and can manage them better.

Cheap motherboards are not only cheap in price, but quality too. It's a general truth ever since the motherboard was invented. There's nothing surprising in HU's video imo.

If you want to overclock, you need a decent VRM. PL2 and/or Tau unlocking is technically overclocking, whether your motherboard manufacturer has done it for you in the 'Auto' setting or not.

 

[BarbaricSoul]

PCI-Express 4.0 NVMe SSD Performance on Ryzen 3000 & X570

AMD's new Ryzens are the first desktop processors to support PCI-Express 4.0, which doubles transfer rates over PCIe 3.0. We test real-life performance gains using the 2 TB Gigabyte Aorus Gen4 M.2 NVMe SSD, which reaches over 5 GB/s in sequential speeds.

www.techpowerup.com

 

[The red spirit]

Cheap motherboards are not only cheap in price, but quality too. It's a general truth ever since the motherboard was invented. There's nothing surprising in HU's video imo.

If you want to overclock, you need a decent VRM. PL2 and/or Tau unlocking is technically overclocking, whether your motherboard manufacturer has done it for you in the 'Auto' setting or not.

I actually wouldn't be so harsh on cheap hardware. I know it's irrational of me to defend them, but they don't usually suck. In my country cheap boards are what 95% of people have in their desktops and anything better than basic board is clearly a luxury. And it seems that cheap boards perform well and last a long time. The thing is that they usually strip down features that are marketable (yet don't really matter all that much), deliver more basic BIOS and keep vital components good enough. Obviously, sometimes they go too cheap with those too, but generally they don't.

Same thing happens with power supplies. Cheap power supplies, that YouTubers call out as being bombs are the most common units in prebuilts and they last and don't really spark or explode. They are mostly from Codegen brand and sometimes from Inter-tech. So I'm very skeptical about YouTubers claiming that cheap products are actually bad and always suspect that they overlooked some details or misunderstood products entirely. I personally saw family PC lasting 13 years with cheap "bomb" looking Codegen 300X power supply and bottom of the barrel DFI board. It worked perfectly fine, until one day capacitors gave out and RAM started to malfunction. That's pretty good, especially knowing that board was made during capacitor plaque era. During 'rona, my dad brought home work computer, which has i7 2700K and it has some cheap Inter-Tech 500W PSU and it is with some random not high end Intel board. It still works perfectly fine and was never repaired. It's now at least decade old already. Eh, people seriously underestimate cheap components.

And PL/Tau modification isn't exactly overclocking. Intel lets every OEM to freely adjust those values, except PL1 downwards. OEMs are free to adjust them as they see and Intel is cool with that. The only thing Intel doesn't give a shit about is whether your CPU will reach clock speed higher than base speed. If it fails to reach turbo speed, Intel won't RMA it. However if CPU fails to reach base speed, then Intel will RMA it. PL/Tau mods aren't overclocking, because they don't go beyond factory speed. Motherboard manufacturers are the only ones, which may claim that it's "overclocking" (aka running beyond motherboard default values) and board may not be engineered to take all that extra abuse, which will lead to excessive thermals and shorter lifespan. In rare cases, straight up BSOD. But then again, most boards come with higher than "default" (default PL values technically don't exist according to Intel) PL/Tau values anyway, making your manual adjustment unnecessary.

PCI-Express 4.0 NVMe SSD Performance on Ryzen 3000 & X570

AMD's new Ryzens are the first desktop processors to support PCI-Express 4.0, which doubles transfer rates over PCIe 3.0. We test real-life performance gains using the 2 TB Gigabyte Aorus Gen4 M.2 NVMe SSD, which reaches over 5 GB/s in sequential speeds.

www.techpowerup.com

And? All compared SSD performed pretty much the same in majority of tasks. And compared SSDs were all PCIe 4. In anything beyond synthetics, SSD speed doesn't matter (as long as its not QLC);

Unless your workloads are heavy in file transfers or reads, don't expect to see any performance gains. Also here's one trick to make existing SSD faster, disable CPU C states.

(Also that article showed that fast SSD will overheat very easily and there's a higher risk of losing your data or experiencing SSD throttling, making fast SSDs even more pointless.)

 

[BarbaricSoul]

easy Hoss, I'm promoting the 10850k also. That review is to show there is no noticeable differences between 2.0, 3.0, and 4.0