do you have super strict stability tests to run 1.35V ? I ran mine @ 1.26V and it was fine for gaming, handbrake or streaming (using nvenc) which is the intended use of the cpu anyway I say this because obviously then you'll get lower Watts and temperatures, maybe I just got lucky with my 9900k...or simply got a stabler memory (memory no1 cause of failure when overclocking, just downclock yours to 2133Mhz or whatever and you'll see your system is suddenly very stable) I own the H115 rgb platinum (older model) and it works great my only minus would be the pump that vibrated more than other aio and thus was hearable (maybe a defect ?) also that pump in the older H115i made Aida64 and other monitoring apps completely crash icue commander pros (when it worked fine with a H150 360mm) I hope they fixed that in this version other than that I also ran my amd 3960x on older H115i rgb platinum thing for 1 month (until I received the parts for a custom loop) and it worked "okay" but using the cpu 100% in cinebench and the like would give me 90°C (stock settings) if you have one already you can use it on threadripper (pretty much the hottest cpu you can get 250-280W only for the cpu !) but it won't be "great"

Hi Hilbert Would you say Wprime 2.10 is a good real world temperature test? So if I can get below 80c on Wprime on 5ghz with my 9900k, would you say that is safe for 24/7 use? My 9900k needs 1.32v to be completely stable at 5ghz. But it can reach 90c after an hour of AVX stress testing like Realbench etc. I've got a Corsair H150i, and my ambient temp is normally 28-29c (air conditioned room) Honestly I've been running 4.8ghz at 1.24v, the temps are like 15c better overall. I just feel a bit uncomfortable with high temps just for a 4% performance bump.

baasgene:

Would you say Wprime 2.10 is a good real world temperature test? So if I can get below 80c on Wprime on 5ghz with my 9900k, would you say that is safe for 24/7 use?

I think it is a very representable test ergo we use it, run 1024M and do that a couple of times. I am looking at different ways to stress for future articles, however, most applications tend to become a bit too viral to be called a 'real-world' test. I find wPrime to be quite representative for CPU load. What I am trying to say is just like with testing GPUs, you can run Furmark and go completely viral on the GPU, as a program that works - but is that a real-case replication of real-world temps, I tend to think that is not the case. e.g. you are not driving a Ferrari 350 kph all the time either. BTW I would not like it any proc running 80c 24/7 at 100% load. But for burst load, that's fine.

Aweseome thanks Hilbert! I just ran mine at the same 1.35v and 5ghz and got 80c package, that's at around 24c ambient and also with Kryonaut. I would've thought a H150i might do 77c roughly like the Eisbaer, but alas the Eisbaer is victorious lol! Do you do these tests with an open chassis? Or do you close the chassis as you're testing? I still rock an old school Cooler Master Haf-X lol, it was closed in my testing.

This particular review was performed by Chris, not me. He uses a closed chassis.

baasgene:

Hi Hilbert Would you say Wprime 2.10 is a good real world temperature test? So if I can get below 80c on Wprime on 5ghz with my 9900k, would you say that is safe for 24/7 use? My 9900k needs 1.32v to be completely stable at 5ghz. But it can reach 90c after an hour of AVX stress testing like Realbench etc. I've got a Corsair H150i, and my ambient temp is normally 28-29c (air conditioned room) Honestly I've been running 4.8ghz at 1.24v, the temps are like 15c better overall. I just feel a bit uncomfortable with high temps just for a 4% performance bump.

do you mean the H150i pro 360mm ? if so keep it, I can't compare to the new H115i but I had the same 9900k installed with both H115i rgb platinum (280mm) and h150i pro (360mm) the h150i pro pump made way less noise and temps were slightly cooler than h115 I highly doubt this refresh does better. H150i pro pump/usb was also way more compatible, the h115i broke most of the hardware monitoring for me (hydro x also does sadly, annoying can't have water temp in aida64) All your voltages and temps seem weird to me (guru3d, you and others) either I had a good 9900k or........you do know you're supposed to have the northbridge clock lower than your cpu clock right ? if you run cpu at 5.0ghz set it at 4.7ghz, if you check the stock "turbo" mode they all have different cpu/nb clocks and I remember reading it in an overclock guide on X99 back then, if you set them at the same speed it's near impossible to be stable for no benefit almost

kakiharaFRS:

do you have super strict stability tests to run 1.35V ? I ran mine @ 1.26V and it was fine for gaming, handbrake or streaming (using nvenc) which is the intended use of the cpu anyway I say this because obviously then you'll get lower Watts and temperatures, maybe I just got lucky with my 9900k...or simply got a stabler memory (memory no1 cause of failure when overclocking, just downclock yours to 2133Mhz or whatever and you'll see your system is suddenly very stable) I own the H115 rgb platinum (older model) and it works great my only minus would be the pump that vibrated more than other aio and thus was hearable (maybe a defect ?) also that pump in the older H115i made Aida64 and other monitoring apps completely crash icue commander pros (when it worked fine with a H150 360mm) I hope they fixed that in this version other than that I also ran my amd 3960x on older H115i rgb platinum thing for 1 month (until I received the parts for a custom loop) and it worked "okay" but using the cpu 100% in cinebench and the like would give me 90°C (stock settings) if you have one already you can use it on threadripper (pretty much the hottest cpu you can get 250-280W only for the cpu !) but it won't be "great"

Do you use the AVX offset? I don't. With -2/-3 I could lower the voltage below the 1.3 V, but I'm an old-schooler and want to do the fixed settings 😉

korn87:

9900k power consumption when performing stress tests with AVX depends on RAM performance. Therefore, it is very interesting for me to find out what power consumption the 9900k has in the author’s test. For example, my 9900kf 5GHz 1.35V at 16 threads consumes more than 220W, RAM 4270MHz cl16

In AVX tests it's more or less 200-210 W (RAM at 4133 CL17 daily)

baasgene:

Aweseome thanks Hilbert! Do you do these tests with an open chassis? Or do you close the chassis as you're testing? I still rock an old school Cooler Master Haf-X lol, it was closed in my testing.

The case is closed during the tests. As for the HAF-X is still one of the best out there in airflow department 🙂

Thanks Cap! I actually revisited my OC last night thanks to this article, and I fiddled around with Dynamic Voltage. I basically did the following : 50x core 0 AVX Offset XMP (in my case 3600mhz 16-19-19-39) VCCIO : 1.15v Agent : 1.15v Vcore : Normal (1.2v) DVID : +0.035 Internal AC/DC : Power Savings LLC : Normal All safety features like Eist, speed stepping enabled Under AVX load like Realbench it hits 1.28v, and OCCT Medium with AVX2 it hits just below 1.3v. It does spike to 1.32v on occasion but only when load is low. The strange thing is I wasn't stable at a fixed voltage of 1.3v at all before, I had to go at least 1.33v especially in the case of Monster Hunter World Iceborne, which refused to actually play more than 5 minutes on 1.32v before it crashed to desktop (not bluescreen). OCCT passed at 1.32v but gave errors on 1.31v for example (lots of them). Now it sings like a canary, my temps have dropped by 10c on Realbench (81c at 30 min compared to 91c) and everything is rock solid and happy. I never did Wprime now that I think about it, will check out the difference tonight. I finally feel comfortable actually leaving this boy on 5ghz now, I actually down clocked it before to 4.8ghz and a fixed vcore of 1.25v because I felt that these temps are just silly. My specs : 9900k Gigabyte Z390 Aorus Ultra (F10b bios) 32gb G.Skill 3600mhz Strix 1080 Ti (looking forward to the 3080 Ti) Corsair Rm1000x Corsair H150i So my advice to everybody is, fiddle with Dynamic Voltage.

@kakiharaFRS why are you using corsair software at all? i had enough H1xx and never used the software at all. pump runs off psu, fans on motherboard, and if it had leds i used the software to control it, then uninstalled it. @slicer because it is. price is not the same as value. and the cables can easily be hidden, never had any problems.

fry178:

@kakiharaFRS why are you using corsair software at all? i had enough H1xx and never used the software at all. pump runs off psu, fans on motherboard, and if it had leds i used the software to control it, then uninstalled it. @slicer because it is. price is not the same as value. and the cables can easily be hidden, never had any problems.

When you have several Corsair components its good software, I have Corsair Dominator platinum @3600 ddr4, Corsair H115i platinum, Corsair H1000i psu, Corsair K70 lux keyboard, Corsair M65 RGB Elite and finally my Corsair headset with rgb, its just a shame i can not get icue to get on with Asus Aura, if Corsair made motherboards i`d probably get one, its nice to have all my parts in one app so i can control and monitor them all.

slicer:

Arctic Freezer Liguid 280 is costing 40usd less (@95usd) and is better in every test here? Including perf. and noise. Also it only has 1 cable to connect to motherboard. This Corsair has +3 cables. How You can justify saying that Corsair H115i is a good price?

Looking at your point of view from the lowest price possible you'd be right. But you are doing just that, price, performance, period. However, if you like the iCUE ecosystem, things like warranty and more premium looks + advanced RGB setup, yeah, this is still a product that is priced pretty fair while offering more. Remember at the other side of the scope there are LCS products out there hitting 200 USD as well.

In the end, it's the consumer choice, most people will find the price quite acceptable. I certainly do. You disagree that's fine, so you won't but the product for your own (valid) reasons. Arctic has a good product at hand, absolutely, and if people feel the Artic product is, better nicer looking and can drop RGB, monitoring, unified ecosystem, and fan control, they will opt that over the Corsair products. As in the end, the consumer decides what to buy (or not). BTW: https://www.guru3d.com/articles-pages/corsair-h60-review,1.html

The Goose:

......Corsair K70 lux keyboard, Corsair M65 RGB Elite and finally my Corsair headset with rgb, its just a shame i can not get icue to get on with Asus Aura, if Corsair made keyboards i`d probably get one......

Did you mean to say something other than "if Corsair made keyboards

@slicer What other brand offers to cover your WHOLE pc (as in all parts) with a 5y warranty, in case the AIO fails? Right. For most its more important to be covered for possible damage to the rig than to save 30-50$, especially when your pc cost more than 200$. Cheaper does NOT automatically mean better value, as that is depending on the user (and his/her budget). e.g. the value of a 300.000$ two door sports car will be totally different to someone with a 100K/year income, and someone making +1M. but that doesnt have anything to do with the car itself and its quality/performance (unless its a lemon ;-)

A 5 year warranty for an AIO is probably the reason it's more expensive. Plus those ML fans aren't cheap either. I'm not a big fan of AIO coolers and went for Noctua NH-12A instead. It's proven as one of the best coolers out there especially if you aren't overclocking. Again 5 years is a long time for that pump to last.

Good cooler, but air cooling the cpu with a dedicated gpu above entry level is lowering gpu clocks noticeable. Especially when a LC is setup to exhaust heat (and dump it outside the case), the lower gpu/case temps get you better boost clocks resulting in higher fps. E.g. 20xx cards will drop clocks at 40/50/54*C and continue linear (15 mhz per 10*C), so having lower temps inside the case is crucial. I see about 30*C difference on avg. (air vs lc)

fry178:

@kakiharaFRS why are you using corsair software at all? i had enough H1xx and never used the software at all. pump runs off psu, fans on motherboard, and if it had leds i used the software to control it, then uninstalled it.

arf I lost my lenghty reply >< so I'll make it short because I have 9 different corsair components and I made profiles to tweak all of them with 1 key, 1ms change, almost instant with 1 software which is impossible to do otherwise because I rebuilt my threadripper from an empty case to custom loop+rgb 3x in 2 months because I tweak my pump and fan speeds 3x per day I tweak stuff nonstop, I'm never content with anything when it comes to my pc even now while I have 55°C running cinebenchR20 (or 78°C all cores at 4.4Ghz) I find some of my fans (the rgb ones obviously) whiney I might drop them and go "dark" @slicer lots of different people doing different jobs or not having one lots of different countries with prices hugely different lots of people wanting different things from a product you clearly mind about 40$, my motherboard alone was 700$ you can imagine I'm not at +-40 in my budget

my comments come from real life testing not watching/reading the internet and I actually bought used and tested toroughly the previous H115i "version" with both Intel and AMD so my comments are very relevant to people interested in this product especially since I test all the products I buy for several days/months/years before making statements I prefer "reality" rather than someone else's "point of view" about products but that's me

@kakiharaFRS i do tweak stuff as well (as in "optimizing"), but at some point if everything is "dialed in", i dont need to do any additional changes, and can just turn off loading at startup/ uninstall sw, as settings are usually kept in the hw and dont require sw to be running. e.g. what need is there to "tweak" things like KB/mouse/fan profiles and stuff, over and over again? as that isnt really tweaking, but more a change of settings.

slicer:

Why You are here commenting even? You are saying that You have a custom loop. You clearly just wanted to let other people know that you are buying expensive stuff, your comment does not make sense in the current conversations that we have had in this thread...

Who gave you the right to decide who's opinion is relevant to everyone else? For people that want all of the features this AIO has, it's price and performance are fine. The performance difference between this AIO and Arctic's AIO are negligible at best. If you like having RGB fans, the Arctic AIO isn't going to save you money in the long run as you'll have to buy the AIO....and then purchase RGB fans and controller separately. In the end, you will actually pay more..... That is where value comes in. For those that like RGB, and system builds with some degree of personality, there is more value in this AIO, at it's price than there is in the Arctic AIO. Personally, I like to change the appearance of my build from time to time. I have no issue paying the $140-150 this AIO costs, vs saving $40-50 on another AIO and having to source multiple sets of LED fans or a set of RGB fans and an RGB controller.....which are going to cost considerably more than the price difference.