Intel Core i9-7900X 3.3 GHz

[EarthDog]

Yes, Daddy, as in moar cores and more productivity.

 

[LiveOrDie]

The "overclocking" picture you posted is marketing is what I am saying.

As I said, the mainstream supported it years ago. Its nothing remotely new. The mainstream doesn't have all unlocked processors though. The reason HEDT platform can always overclock is that the mainstream and its lesser, non-overclocking variants are there and have no place on that platform. Mainstream IS a step down from HEDT, so why cloud up the market that type of crossover????? If you stick to the primary chipset on mainstream, all the boards overclock. Even a chipset down does IIRC. THere are other cheaper variants, you are correct.

The i7 3820 used in X58 was not unlocked and required BCLK only adjustments.

In the end, we'll have to agree to disagree.

The i7 3820 takes me back pretty sure that's X79 the X58 would of been a 920 which didn't carry any K or X tag, but ether way i do get your point .

 

[EarthDog]

Oops, yes... X79... but yeah, not an "X".

X58, most all of those were not unlocked.

 

[Vego]

X is kinda better
just today i found that i still have 7740X and i oced it to 5,4GHz

1st s2066 result I am satisfied with

https://valid.x86.fr/ksr9vx

https://www.3dmark.com/3dm/23739683

 

[Vayra86]

Thanks for taking the time to run more cards than asked for through one title. I am curious solely about the 1080Ti (a card FASTER than the 1080) and all the titles though as asked last night. I wonder if the lower FPS titles will show more of an increase of if it is the same.

Regardless, I appreciate your effort.

As much as I appreciate your inquisitive stance on this, honestly, you are battling captain obvious here... Simple rules apply, you're either GPU limited or you're not, and when you're not, the next hurdle is the CPU. It all comes down to that very simple principle and a CPU still doesn't magically create additional lines of code to process as you go up in resolution - and if it does, you've got the exception to the rule, and most of the time an extremely unique game on a very uniquely built architecture or game engine (primarily isometric and 2D-engines for example with fixed, nonscaling assets - higher res will increase the amounts of assets that'll be shown and can thus be interacted with).

So we come back to the conclusion I've been presenting you numerous times: you test at low res so you gain insight on the capabilities (and limitations) of the CPU - and you do everything you possibly can to ensure that its only the CPU that you're looking at. Reviews have NEVER aimed to show 'the typical use case' - they test with high end gear that the vast majority of customers won't ever buy. Any GPU review comes with the fastest available CPU for the exact same reason; you want to push that GPU to its fullest.

Typical use cases don't exist in the PC gaming world, as what is 'typical' is always changing, very personal, and 100% customizable. As much as one person 'would never spend 1K on CPU and 200 on GPU', for the next its a typical system build because he does some light gaming alongside his professional work. Therefore: reviews should focus on the limitations, the extremes of a product, to provide insight.

 

[EarthDog]

This testing is good for what he lists it to show... high hz gaming and those that game a 720p with a high-end card. Outside of that, it really doesn't show much of anything since, again, it doesn't scale to 1080 or 4K and really doesn't show squat for those resolutions.

I'd like to see testing done at 1080p and 4K and see if the results match (they dont).

 

[bug]

This testing is good for what he lists it to show... high hz gaming and those that game a 720p with a high-end card. Outside of that, it really doesn't show much of anything since, again, it doesn't scale to 1080 or 4K and really doesn't show squat for those resolutions.

I'd like to see testing done at 1080p and 4K and see if the results match (they dont).

Neah, you're just being stubborn.

 

[Vayra86]

This testing is good for what he lists it to show... high hz gaming and those that game a 720p with a high-end card. Outside of that, it really doesn't show much of anything since, again, it doesn't scale to 1080 or 4K and really doesn't show squat for those resolutions.

I'd like to see testing done at 1080p and 4K and see if the results match (they dont).

Like all parts of a review, yes, you try to cover as wide a range of use cases as you possibly can, because again, there is no real typical one that applies everywhere. Regardless - I can tweak my GFX settings per game to once again achieve 100% CPU utilization at 1080p or perhaps even 1440p as well, I can almost guarantee you that similar GPU load will result in similar FPS on the same CPU, at different resolutions. If there is more than 2-3% deviation, I would call that an anomaly or quirk of the game.

If it satisfies your semantics - the reason we use 720p is the same reason we also include 4K and other GPU limited scenarios. Covering the product in all of its possible strengths and weaknesses.

In the end the GPU does very simple work, that scales perfectly and linearly, while the CPU always does the same job, because that's the code its executing.

 

[EarthDog]

There is little worry about most CPUs at 1080p/4K with realistic settings. Surely I can tweak a game at 1080p to be CPU limited, but, we don't game that way. So while it does show what you are purporting, that data set just isn't relevant at 1080p/4K with reasonable settings.

What is the point if CPU A shows a 10% difference at 720p with low settings when at 1080p and Ultra settings the difference is nill? The balance shifts to the GPU being the bottleneck again in the majority of cases.

Neah, you're just being stubborn.

HI, I am Earthdog, pleased to meet you.

 

[Vayra86]

There is little worry about most CPUs at 1080p/4K with realistic settings.

Really now? Again, once you go beyond 60hz mainstream monitor land and into high refresh/competitive gaming, regardless of resolution, the focus is very much or exclusively on framerate and its stability. CS:GO on your spanking new 1440p panel for example, can still be a CPU limited ordeal. Any MOBA - same story. I can play Dota 2 right now on my 1080p panel, and even on ultra, I will be CPU limited

Do note that the larger active playerbases on PC are almost all tied to low-GPU stress competitive games, MMO's (that also put massive burden on the CPU), and stuff like Minecraft. And even with a 60hz panel, there are specific titles that really do fancy running at highest possible internal FPS (CS:GO being the most well known example) to improve snappiness and smoothness.

EDIT: (or its becoming a slowchat between us here) @EarthDog below: I think you'll find there's quite a lot more of that grey area than you presume, and more importantly, the grey area becomes ever larger as it is typical for us to upgrade GPU a few times within the lifetime of a build. In addition, the idea that 'Ultra' is the only way to go is ludicrous; I personally much prefer backing down to 'High' if that's required for smooth 1080p/120hz gaming, and it'd be a shame if I'd run into my CPU limitations at that point.

 

[EarthDog]

There are exceptions, of course. What I am saying AND what you are saying are not hard and fast rules. They come with a bit of grey area. Techspot has done reviews of games showing no performance difference across a slew of CPUs in many titles. In some games, there are notable differences. It depends.

My apologies for not being more clear on that point. Again, always some grey area, but in general, 1080p/4K with Ultra settings (appropriate for the GPU used) the CPU doesn't matter nearly as much, or at all in some cases, as the datasets show at 720p.

EDIT: You are focusing on what I am agreeing with... high hz gaming it can matter. They can pile on in with the other 1% of users on 4K.

But seriously now. I see the point of it. I see it gives a valid dataset at 720p and for high hz gaming. But that doesn't mean much to those 1080p+ people with ultra settings.

 

[thebluebumblebee]

@W1zzard , I would love to see how this 3.3 GHz HEDT CPU does against the previous generations' 3.3 GHz HEDT CPU's going back to the i7-980.

 

[W1zzard]

@W1zzard , I would love to see how this 3.3 GHz HEDT CPU does against the previous generations' 3.3 GHz HEDT CPU's going back to the i7-980.

Send me those CPUs

 

[Vlada011]

I like this i9-7900X.
With 3200MHz 32GB Quad Channel... hmmmm... perfect.

It would be nice try to build him in ASRock Mini ITX with SO-DIMM.
I can't find comparison of 2666MHz SO-DIMM with 2666MHz desktop memory.
Maybe difference is so big that something like that not worth at all if you want real performance of X299 not joking.
But if difference is small, than in some ITX SFX aluminum case of weird shape.
That would be hard to find, small, attractive, SFX/Mini ITX case with 240mm radiator place.

 

[cadaveca]

I can't find comparison of 2666MHz SO-DIMM with 2666MHz desktop memory.

Where do you find these ideas?

There is ZERO difference. Desktop DDR4 does not use all pins anyway; that's how they can have RGB LEDs and offer temperature monitoring with no wires going to DIMM (4 pins were added for 12V power out for Hybrid DIMM, for example). So DDR4R SO-DIMM has 260 pins, less than the 288 of the desktop part, but those added 28 pins are primarily for things like Optane DIMMs and such, and are not used by most DDR4 modules, making there little-to-no difference between DIMM package types.

You will also see DDR4 SO-DIMMs used in a couple of my reviews in the coming weeks; reviews are done, just waiting their turn to hit the front page.

And yes, it will be using that exact board you are looking at.

 

[EarthDog]

Let me guess... asrock x299 itx and a 3800 kit?

Edit: oops he said that too i see, lol!

Ive reviewed it already, its a feature laden beast, that board. I didnt push the memory on it though.

 

[cadaveca]

Let me guess... asrock x299 itx and a 3800 kit?

No, I declined the 3800 MHz kit and opted for lower speed 2666 MHz kit, because I do not think many users will buy 3800 MHz kit when memory clocking can affect CPU OC, and you get more performance increase from CPU speed increases.

Also, because the 3800 MHz kit has much less ICs per DIMM (different ICs used, 3800 MHz kit uses specialized PCB and only 4 ICs per side), I think at lower speeds this other kit I have will be marginally faster because page size overall is smaller.

 

[EarthDog]

Makes sense.

In my case, it took us to our thermal limit w/H115i at 4.5ghz. So, it didnt affect the top end of that cooling. I dont imagine it to until custom + delid either.

 

[cadaveca]

Makes sense.

In my case, it took us to our thermal limit w/H115i at 4.5ghz. So, it didnt affect the top end of that cooling. I dont imagine it to until custom + delid either.

Seems kind of like you got a CPU like W1zz's a bit. I can hit 4.7 easily without throttle under most workloads but AVX2 on 7900X, 4.8 GHz on 7960X and 4.6 on 7980XE. The HCC CPUs then seem so amazing compared to the 7900X... all those added cores and cache, higher speeds, under the same basic thermal load is pretty damn good!

 

[EarthDog]

Yep, this one is mediocre. My retail on the other had is sub 1.2 at 4.7ghz.

I havent had a chance to play with anything greater than a 7900x yet, but those seem to scale similarly if not better. I may have a w-2195 on the way though amd to play in that realm for a bit.

 

[dalekdukesboy]

Unfortunately, that is the real world, whether you like it or not.
But even so, there seem to be few reasons to pick this over a cheaper Threadripper.

Hell, few reasons to pick this over older X79/X99 platforms! You can get those quite inexpensively and yeah few bells and whistles missing etc but...otherwise it's a joke how little Intel has moved the ball with actual performance and architecture. Hopefully with AMD at least back in the picture it won't just be die shrinks and new board platforms forcing you to upgrade your board with little benefit.

 

[LiveOrDie]

Hell, few reasons to pick this over older X79/X99 platforms! You can get those quite inexpensively and yeah few bells and whistles missing etc but...otherwise it's a joke how little Intel has moved the ball with actual performance and architecture. Hopefully with AMD at least back in the picture it won't just be die shrinks and new board platforms forcing you to upgrade your board with little benefit.

AMD CPUs are like Aids and always will be.

 

[Vya Domus]

AMD CPUs are like Aids and always will be.

A very intelligent comment.

 

[dalekdukesboy]

Yeah...I thought mine was edgy comparing older intel high end sockets to new Intel tech but calling any CPU equivalent to Aids is....well just stupid, insensitive and makes no sense to boot. My point is merely Intel has coasted for a while because yes AMD was essentially not a competitor for performance they only had good processors at low price points but couldn't compete with Intel at mid to high end whatsoever till last gen or so.

 

[dalekdukesboy]

Hey Wizzard, good review as always but what 240 AIO cooler did you use for overclocking? As someone else pointed out it does a darn good job keeping that chip cool at over 1.3 volts and 4.5 ghz.