Intel Core i9-7900X 3.3 GHz

[EarthDog]

Yea it does. Look at the core usage.

See if you can push some of your cores abit higher. You might as well just get a 8700k if you want to lock all your cores at the same speed.

Im good where i am at, for 24/7 really. When benching i manipulate as you do depending on the benchmark and its core needs.

Though, seems like the 4.9/5ghz cores arent doing the work from your ss..core 9 isnt loaded as core 6 is. Core 7 has more load than 9 and runs at 4.3.

 

[LiveOrDie]

Looks like you got a lemon there W1zzard, my chip does 4.7Ghz on 1.2v.

Yea it does. Look at the core usage.

See if you can push some of your cores abit higher. You might as well just get a 8700k if you want to lock all your cores at the same speed.

HWinNFO i wouldn't trust for clock speeds.

 

[EarthDog]

Or .976 as you posted a pic of CPUz without a load.

 

[Vayra86]

Testing at 720p wouldnt remotely tell you that

Rog realbench runs 3 things at once... all real world testing... at least 2 of three iirc.

But how often do your real world scenarios reflect the actual benchmark runs you see in ANY review, anyway? How often do you focus on that one task and actually watch the progress bar and how it differed from your last CPU?

You don't.

If you cannot or will not read benchmarks as exactly what they are meant to be, ie. AN INDICATOR of performance that is reproducible and works the same way every single time, same duration, same type of loads, free of as many other bottlenecking as you can, then no single 'real world scenario' is going to give you the right information. 1080p testing is just as inaccurate as 720p in terms of a real world scenario, and at the same time, you do introduce other side effects into the benchmark, such as the GPU performance. For historical purposes too - if you start jumping to 1080p, all your 720p benches have been rendered pointless for comparison.

The argument of 'but nobody runs 720p Low' is just as useless in this discussion as saying 'the majority runs 1080p'. Sure they do, but not on Low! The 1080p bench is not indicative of a real world scenario either.

You already know my stance on the 720p testing, but again, trying to 'mimic the real performance' is not and was never the purpose of benchmarking. The purpose of benchmarking is being able to measure RELATIVE performance between products. If you want your real world scenario information, there are other articles for that, most specifically the 'can I run' type articles, where a range of hardware is tested on a specific game. For a product review the information you want is see how it stacks up against the rest and whether that justifies the price, and you want to see the product in a wide range of different situations with similar conditions.

The bottom line with all this discussion around CPU testing for games to me, is a simple one: do NOT dumb down the hardware review to the level of the uninformed, misguided readers. I'm sure its good for your page hits, but in the longer term, you create a false image of actual performance and get all sorts of wrong ideas into people's heads. When people are dumb, educate them. Show them why you do what you do. Thát to me constitutes a good review and a good reviewer.

 

[btarunr]

Looks like you got a lemon there W1zzard, my chip does 4.7Ghz on 1.2v.

HWinNFO i wouldn't trust for clock speeds.

You're on liquid cooling. Our cooling is potato NH-U14S.

 

[EarthDog]

That potato is also trying to cool 1.35V at a lower clockspeed. I'd also like to know how that potato is managing to cool a 4.5 GHz CPU at 1.35V......

A dud is a dud though bt.

But how often do your real world scenarios reflect the actual benchmark runs you see in ANY review, anyway? How often do you focus on that one task and actually watch the progress bar and how it differed from your last CPU?

This depends on the test, doesn't it? Synthetics... not so much, handbrake, POVRay, Cinebench, and many others are real world tests.

If you cannot or will not read benchmarks as exactly what they are meant to be, ie. AN INDICATOR of performance

But that is what I am trying to say it ISN'T an indicator of performance outside of 720p as it doesn't scale. 1080p and the CPU load in gaming is a completely different beast.

1080p testing is just as inaccurate as 720p in terms of a real world scenario, and at the same time, you do introduce other side effects into the benchmark, such as the GPU performance. For historical purposes too - if you start jumping to 1080p, all your 720p benches have been rendered pointless for comparison.

1080p testing is more real world as the majority of people use it according to steam. To put simply, 720p a 'fake' environment intentionally used to exaggerate a result so a difference can be seen. I was saying add 1080p results to keep those who believe the 720p are worthwhile happy. Not to compare.

The argument of 'but nobody runs 720p Low' is just as useless in this discussion as saying 'the majority runs 1080p'. Sure they do, but not on Low! The 1080p bench is not indicative of a real-world scenario either.

Sure it is. it will show how little the CPU matters at a res most play at. Real world res, real-world settings (ultra)...

The purpose of benchmarking is being able to measure RELATIVE performance between products.

Indeed, but how does an unrealistic environment show relative performance where it matters? It is exaggerating an environment which few use. Again, who runs low settings and 720p with a high-end gpu? It's more realistic to show 1080p with settings appropriate for the test system than to exaggerate differences to me.

The bottom line with all this discussion around CPU testing for games to me, is a simple one: do NOT dumb down the hardware review to the level of the uninformed, misguided readers.

But that is EXACTLY what its doing!!!!!!! Its misleading to those not in the know to see these results and think hmm, this CPU will give me 10% more FPS if I use it.... not knowing it won't come close to that at 1080p or not at all when running a higher res.

Its impossible to please everyone, of course, but to run a high-end gpu at 720p and low settings to show CPU results is just making a mountain out of a molehill and to the detriment of those reading it.

 

[cadaveca]

A dud is a dud though bt.

W1zz could likely get the voltage a bit lower. That's using the pre-programmed VID for 4.5 GHz. He did say, all he did was change the multi... highlighting that Intel tests and programs these CPU for far more than just stock. If you are a user that doesn't know how to OC, you don't need to know much, not even temperatures, as the CPU will throttle when it gets too hot.

That's also an early ES CPU. I have a QS myself.

 

[Vayra86]

That potato is also trying to cool 1.35V at a lower clockspeed. I'd also like to know how that potato is managing to cool a 4.5 GHz CPU at 1.35V......

A dud is a dud though bt.





This depends on the test, doesn't it? Synthetics... not so much, handbrake, POVRay, Cinebench, and many others are real world tests.

But that is what I am trying to say it ISN'T an indicator of performance outside of 720p as it doesn't scale. 1080p and the CPU load in gaming is a completely different beast.

1080p testing is more real world as the majority of people use it according to steam. To put simply, 720p a 'fake' environment intentionally used to exaggerate a result so a difference can be seen. I was saying add 1080p results to keep those who believe the 720p are worthwhile happy. Not to compare.

Sure it is. it will show how little the CPU matters at a res most play at. Real world res, real-world settings (ultra)...

Indeed, but how does an unrealistic environment show relative performance where it matters? It is exaggerating an environment which few use. Again, who runs low settings and 720p with a high-end gpu? It's more realistic to show 1080p with settings appropriate for the test system than to exaggerate differences to me.

But that is EXACTLY what its doing!!!!!!! Its misleading to those not in the know to see these results and think hmm, this CPU will give me 10% more FPS if I use it.... not knowing it won't come close to that at 1080p or not at all when running a higher res.

Its impossible to please everyone, of course, but to run a high-end gpu at 720p and low settings to show CPU results is just making a mountain out of a molehill and to the detriment of those reading it.

Like a few others have said, I'll respectfully agree to disagree Now you're even advocating we use 1080p ultra... That's just painful.

 

[EarthDog]

W1zz could likely get the voltage a bit lower. That's using the pre-programmed VID for 4.5 GHz. He did say, all he did was change the multi... highlighting that Intel tests and programs these CPU for far more than just stock. If you are a user that doesn't know how to OC, you don't need to know much, not even temperatures, as the CPU will throttle when it gets too hot.

That's also an early ES CPU. I have a QS myself.

Auto overclocking... ick. Gotcha... and good to know. 1.35V is a lot for 4.5 GHz. Is each CPUs VID for a given clock different I assume? My sample sets 1.30V when using auto across several X299 boards.

EDIT: I take that back, both my 7900Xs, one early ES and one Retail, both set to 1.30V on auto at 4.5 GHz. Though both need relatively different voltages at 4.5 Ghz and its stock VID.

I'd still like to know why the overclocking CPUz images show 8c/16t, likely just a pic snafu.

I'd also like to know how that air cooler can work better than a 2x120mm AIO with .12V more... lol. Though POVR and OCCT are pretty stressful so that may be it.

I have an ES as well, from day0.

Now you're even advocating we use 1080p ultra... That's just painful.

Yes, I simply like unexaggerated results, that's all.

I also have no idea how you can say this:

The bottom line with all this discussion around CPU testing for games to me, is a simple one: do NOT dumb down the hardware review to the level of the uninformed, misguided readers.

...when that is exactly what it is doing. If your concern is not to mislead readers than testing should be completed in a more realistic environment for the test system.

But, ok, agree to disagree. I digress.

 

[cadaveca]

I have an ES as well, from day0.

Yeah, I'm pretty sure that we all got our CPUs at roughly the same time. You know how THAT all went down... I do have a much later chip that was binned for me by ASRock, and it does 4.6 GHz @ 1.225, but is amazingly hot, and draws roughly 25W more than the one W1zz has.

THE THERMAL LIMITS ARE REAL!!!

Screenshots I can't comment on at all.

And yes, CPUs all seem to have different VIDs for different multis as per the norm for Intel now; as such vInput, Vcore, vRing, VCCIO and VCCSA are still chip-dependent. My chip requests 1.295V for 4.5 GHz. If you are not seeing this, the boards you are using are screwing with you.

That's another thing to consider as well when it comes to OC; these chips are so huge and have so many different varying voltage domains that raw clockspeed isn't always the best indicator of what's going on... board matters quite a bit from my own testing on OC with this platform.

 

[LiveOrDie]

Or .976 as you posted a pic of CPUz without a load.

Your point ?

You're on liquid cooling. Our cooling is potato NH-U14S.

Haha yeah i also delidded my chip it dropped 15c off my temps.


I say its just CPU-z they uses to read the 7900x ES as i7s at some point as well i guess they didn't fix the core counts lol or it thinks its a 7800x.

 

[dicktracy]

Faster RAM is more important for Skylake-X:

 

[Al Chafai]

i didn't like this review at all,seems hummm ,honestly it's not what i've seen in lot of Youtubers reviews,the 1950x kicks the 7900x in the teeth when it comes to efficiency and MT workloads (also price).but here it says that it's on par,so idk seems kinda biased.
correct me if i am wrong,or check these youtubers (Paul's hardware,Linus tech tips,hardware unboxed ect)

 

[Vayra86]

i didn't like this review at all,seems hummm ,honestly it's not what i've seen in lot of Youtubers reviews,the 1950x kicks the 7900x in the teeth when it comes to efficiency and MT workloads (also price).but here it says that it's on par,so idk seems kinda biased.
correct me if i am wrong,or check these youtubers (Paul's hardware,Linus tech tips,hardware unboxed ect)

Well, there is the overclockable part where TR loses its edge. 4.5 Ghz is nothing to sneeze at, and this isn't even a great chip.

But as an overall product yes, I agree with you, I think TR has much better positioning and a lower price tag... These points are mentioned in the review though, even with a sarcastic undertone towards Intel and their weird platform. Sorry, W1zzard is on the money for me

 

[Vego]

1st - who buys 1k$ CPU and games at 720p... ???
2nd - I had like 12 different 7900X, all of them did 100% stable 4,7GHz @1,25v max - and in the beginning there was no APEX so I tested it with cheapest MSI board, SLI+
After delid it is a perfect CPU in this budget, very good for SLI setups

 

[EarthDog]

Your point ?

Haha yeah i also delidded my chip it dropped 15c off my temps.


I say its just CPU-z they uses to read the 7900x ES as i7s at some point as well i guess they didn't fix the core counts lol or it thinks its a 7800x.

I still don't see that voltage anywhere... lol! AISuite/DIP5 will read it and so does Coretemp. You can thank Asus' custom ICs for this.

It would just be nice when someone says I ran X at Y that their screenshot actually supported what was said. Not that you are lying, but... we see it all the time. Hilarious.

 

[efikkan]

So how much Blender rendering and 7-Zip decompression do you have to do to justify buying Threadripper 1950X over i9-7900X? I'll take 10 good cores over 16 lesser cores any day.

That is just an indication about how well the CPU is doing when pushing game engines. It's not meant to be translated into overall gaming performance, but you're right, noobs won't read it that way. At the same time, I wouldn't get rid of those benchmarks, but I don't know how to better convey what they represent either. Maybe a disclaimer on the respective page(s)?

Imposing artificial bottlenecks on benchmarks is pointless, and benchmarking scenarios with no relation to actual usage is useless.

Finally, I think real world usage is not even the target of benchmarks. Benchmarks simply highlight the basics, the "building blocks" of real world usage. We're simply supposed to pick the "building blocks" that make up our own usage and decide on what to buy based on that.

Real world benchmarks should always be the standard. What you call "building blocks" of real world usage is the problem; it has caused people to cherry-pick benchmarks and build their anecdote based on edge cases, resulting in numerous false claims about performance, and ridiculous claims like "this CPU is better because it has more cores" or "this CPU is better because of {insert random tech spec here}". Real world tests is the only thing which should matter to buyers.

 

[cadaveca]

1st - who buys 1k$ CPU and games at 720p... ???
2nd - I had like 12 different 7900X, all of them did 100% stable 4,7GHz @1,25v max - and in the beginning there was no APEX so I tested it with cheapest MSI board, SLI+
After delid it is a perfect CPU in this budget, very good for SLI setups

I have that same board direct from MSI as review sample and the review has been posted.

https://www.techpowerup.com/reviews/MSI/X299_SLI_PLUS/

I still have mine in use, with my 7900X sample, and my CPU overheats @ 4.7.

You got lucky on your CPUs.

now if you delidded them all, then that's totally believable to me...

everyone asked for 720P benchmarks as this removes GPU limitations...? You can never please everyone.

 

[W1zzard]

who buys 1k$ CPU and games at 720p... ???

You are aware that 720p tells you the limits for 144/240 Hz gaming?

 

[dicktracy]

i didn't like this review at all,seems hummm ,honestly it's not what i've seen in lot of Youtubers reviews,the 1950x kicks the 7900x in the teeth when it comes to efficiency and MT workloads (also price).but here it says that it's on par,so idk seems kinda biased.
correct me if i am wrong,or check these youtubers (Paul's hardware,Linus tech tips,hardware unboxed ect)

I don’t think anyone does realworld productivity testing better than Puget Systems. They sell all kinds of brands so they are practically free from bias.
https://www.pugetsystems.com/all_articles.php

In many tests (they have crap ton of articles) 7900x can hold its own, and the 7920x is more than competitive. 7960x and 7980XE is in its own league in real world multithreaded workloads. These findings are worlds apart from BS sites like Hardware Unboxed.

 

[EarthDog]

@W1zzard

You are aware that 720p tells you the limits for 144/240 Hz gaming?

Does it? Cool... (links?)! Even though its pumping out a ton of frames, I wonder, with the graphics turned up as it normally would be, if the CPU is really holding things back.

That is the first worthwhile argument I've heard at this forum to come out of 720p testing.



Was the 8c/16t CPUz ss a simple error or did we all miss something?

I am also curious to hear what your temps were at that god awful voltage and air cooling.........as I mentioned for mine, at 1.23V and 4.5 Ghz all cores, the AIO is at its end already....

 

[cadaveca]

i didn't like this review at all,seems hummm ,honestly it's not what i've seen in lot of Youtubers reviews,the 1950x kicks the 7900x in the teeth when it comes to efficiency and MT workloads (also price).but here it says that it's on par,so idk seems kinda biased.
correct me if i am wrong,or check these youtubers (Paul's hardware,Linus tech tips,hardware unboxed ect)

We don't do YouTube reviews for a reason...

Those aren't reviews, and those people aren't product reviewers. They are INFLUENCERS, and they choose to show things a ways that INFLUENCE your opinion. Now that you've watched them, you've been influenced to see things the way THEY wanted you to. Them doing this is how and why they get paid. We offer something different, so at times, what we present is different. Paul came from Newegg, Linus from NCIX. W1zz writes GPU-Z, and we all joined him from that... from something that has NOTHING to do with selling you a product. It's a far different perspective when money isn't involved.

 

[W1zzard]

Does it? Cool... (links?)! Even though its pumping out a ton of frames, I wonder, with the graphics turned up as it normally would be, if the CPU is really holding things back.

That is the first worthwhile argument I've heard at this forum to come out of 720p testing.

It clearly identifies the FPS rate at which the CPU becomes the bottleneck. If you can't get 120 FPS at 720p, then no way you're getting it at a higher resolution.

MT workloads (also price).but here it says that it's on par,so idk seems kinda biased.

If I understand you correctly you want to see a test suite of only highly multi-threaded tests to show that TR beats 7900X there? Sounds like bias to me.

I tried to pick a well rounded suite of typical applications to cover a wide field of usages. If you only run Blender all day, sure, just look at Blender and buy Threadripper, it's great for that, as I mentioned in my conclusion.

If you have suggestions for additional tests, let me know, I'm happy to look into adding them.

I like the encode+gaming scenario mentioned further above and will try to figure out how to reliably test it, for the next complete rebench.

 

[Xzibit]

We don't do YouTube reviews for a reason...

Those aren't reviews, and those people aren't product reviewers. They are INFLUENCERS, and they choose to show things a ways that INFLUENCE your opinion. Now that you've watched them, you've been influenced to see things the way THEY wanted you to. Them doing this is how and why they get paid. We offer something different, so at times, what we present is different. Paul came from Newegg, Linus from NCIX. W1zz writes GPU-Z, and we all joined him from that... from something that has NOTHING to do with selling you a product. It's a far different perspective when money isn't involved.

I don't frequent Youtube or their reviews but I wouldn't go tin foil hat on them.

Same thing about written form. One relies on others perspective and the way that information is delivered be it written or visual. Different targeted audiences with-in the same tech community but saying their transition disqualifies them is out there (Well maybe Linus j/k). That's akin to having a _ mentality.

Money is always involved.

 

[LiveOrDie]

I still don't see that voltage anywhere... lol! AISuite/DIP5 will read it and so does Coretemp. You can thank Asus' custom ICs for this.

It would just be nice when someone says I ran X at Y that their screenshot actually supported what was said. Not that you are lying, but... we see it all the time. Hilarious.

I did say 1.2v