Broadwell-E: Intel Core i7-6950X, 6900K, 6850K & 6800K Review

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.


Testing games with that kind of GPU power will give frame rates so high that CPU issues simply won't matter at all. You cited some numbers that show approx. only 1% difference for a framerate which is already well over 100fps, so who cares?

Besides, the modern HT was shown long ago to be beneficial. It was the old HT back in the days of the P4 which if turned off could be beneficial. Go back and read the original X58 reviews.
 


I have a better idea than empty arguing ... lets Benchmark it ! we are not living in X58 times anymore .

I wish I had 4 980 ti in SLI to test it ...it takes just 30 minutes so some one please test it !

and you are wrong about 4 GPU , you will need faster CPU for them , they bottle neck the CPU.

some one please benchmark:

1- 8 cores @ 4.4 ghz HT Turned OFF
2- SLI 1080 or 4x 980ti

and compare the result please ?
 


I want to bottleneck the 8cores cpu without HT and see if 4x SLI 980ti or 2x SLI 1080 will be faster or not ! I think it will be faster .
 


That depends on the game. How much multi-GPU testing have you done? Most games scale pretty badly beyond two GPUs anyway (how the game engine works is more important than CPU issues), and so far I've not seen any game properly gain from more than three GPUs without horrendous issues. Only synthetic benchmarks scale their performance with 3 or 4 GPUs (eg. Unigine), and they're not representative of typical games.





Sounds like you've already made up your mind.

I don't see the point in testing this when the differences are tiny and the frame rates are already enormous. What does it matter? One would never notice the difference in reality. What you're asking is for someone to spend time doing something which serves little practical purpose. By contrast, I spend time testing stuff people actually ask about, such as how newer GPUs behave with older CPUs/mbds and vice versa.


 


I want to bottleneck the 8cores cpu without HT and see if 4x SLI 980ti or 2x SLI 1080 will be faster or not ! I think it will be faster .
Click to expand...
The 1080's will be better. Its already been proven that a single 1080 OC'ec matches 2X 980's. Now given there is very little gain going from 3 to 4 cards the second 1080 will make up for the difference of the ti.
 


I think he meant testing each GPU config with HT vs. no HT, but in either case it's pointless since the frame rates will be so high anyway, nobody would notice the difference.

Samer1970, a lot of this sort of thing is very OS-dependent aswell. Plenty of people are still using Win7/8. Depends on the game, the OS, all sorts. HT is well designed these days. Who cares if one can find a corner case where turning HT off gives only a 1% improvement? Most of the time HT works well. If there's anything interesting about HT being on or off it's the power consumption and heat difference, which back in the days of P55/X58 could be considerable (turning off HT with an i7 870 dropped temps by 10C, effectively turning it into a pseudo i5 760 which could reach 4.4GHz instead of maybe just 4.2, which could help in some games, but hurts for video encoding and other tasks).

 
Its 2016 we still do not have a 5 GHZ cpu from intel, I don't care about 10 % performance per clock I want 5GHZ flat out horsepower at this stage of the game. $700 for 2 cores Intel? KMA, I hope and pray ZEN is all and more than AMD is claiming.
 

Maybe because both AMD and Intel have already tried prioritizing high frequencies in search of performance and found it to be a dead end? And frequency by itself is a meaningless number? If you don't care about IPC and just want a 5 GHz CPU, the FX-9590 already exists.
 

Clock frequency alone is meaningless: AMD's FX9xxx at 5GHz still loses many game benchmarks to Intel's 3.xGHz i3.

If you want the maximum horsepower, you need to strike the right balance of IPC and clock frequency, very similarly to striking the right balance between torque and RPM in engines. Once you sacrifice more of one than you gain on the other, you lose on overall performance.

 


I want modern re evaluation of it , now we have better threaded games coming .

I think at full load and 100% cpu usage , games will be faster with HT turned off on 8 or more cores CPU .

and I am expecting 30% improvement not 1%.

 

That proves nothing. It does indicate a lack of testing experience on your end, however. First off, a 1% variance is within the margin of error for the testing hardware and reporting software. It could be nothing more than a momentary heat rise in the system causing throttling, or any number of other explanations. You see this kind of minor fluctuation all the time when comparing nearly identical products.

Second, look at that graph again. Notice a fairly linear scaling pattern between most of the CPU configurations of 15 fps? Then notice how it tapers off after six cores? It means you're hitting a wall of some kind. It might be hardware resources, but that means it could be CPU cores, CPU efficiency, GPU limitations, RAM capacity or bandwidth limitations, or even that of the software itself. In short, it could be any number of variables, so claiming you positively know from a single chart that it's the HT overhead, and nothing else, is extremely unscientific.

Now, you could experiment on this and throw more hardware at the issue, but adding more GPUs to this requires SLI/CFX, both of which increase overhead. And, oh yeah, overhead is what you're trying to test in the first place, so adding more of it ( and even better, a different kind ) is not a good idea.

Finally, that's a synthetic test. They're great at pushing hardware in every way possible, but they're usually terrible at demonstrating real-world performance. That's because most real-world software isn't designed to run like synthetic benches. So even if in the very unlikely event you're right, if that situation doesn't occur in real-world computing, what does it matter if it's not now, nor ever will be a notable problem?
 
To those who are in authority in TH: One thing that would be very cool for y'all to do would be to overclock the 6950X (or any CPU for that matter) as far as it will go using liquid nitrogen. I have seen the article and video about the 5GHz project that occurred over 10 years ago and I think it would be very neat if that could be repeated.
 


Liquid Nitrogen overclocking is fun as its own hobby, but such overclocks are not practical for anything like gaming.

You wouldn't want to use a dragster as your daily driver for much the same reasons.
 

I know that, but it would be interesting to see how far some of these Broaswell-E CPUs will go.
 

Overclocking records where most of the chips' features get turned off for the sake of reaching the highest OC possible are pointless since nobody would run a chip that way and such a chip would lose many benchmarks against stock chips on air coolers with all features and cores enabled.

A massive OC that fails to yield meaningful real-world performance benefits on top of requiring an impractical and unsustainable setup is of extremely limited usefulness - worthless beyond bragging rights.
 
Fair enough, but Tom's isn't really about that. Just google "liquid nitrogen broadwell" and you'll find things like this:

http://www.pcworld.com/article/3077237/components-processors/overclockers-have-pushed-intels-new-broadwell-chip-to-5-7ghz.html

For more practical overclocks, I suggest you look into direct-die waterblock cooling. That probably won't work for the E-series CPUs, which typically have their IHS' soldered on. For the mainstream i7's, it seems like the best option, though I don't know about applicability to the 14 nm CPUs.
 
Hi Chris and Igor

Did you guys try running Ashes in DX12 mode with an AMD card, say a Fury or Nano? And with two, three and four AMD GPUs?

I'm asking because DX12 is supposed to allow devs to make better use of more cores - or do you think that the graphical rendering is a relatively small part of the game itself?

Thank you.
 
From my point of view, these cpu' s are not worth it both price and performance wise. From this article, it seems the i7 5960X is still the better choice when compared to the new octa core and deca- core due to easier overclocking as all these cpus hit a wall at 4.4ghz (and with performance increase that is not worth the power consumption and temps that can only be kept in check with custom water cooling or even more exotic solutions).
 
Awful tempting to upgrade.. But my 6yr old i7 980x still runs great.

STILL holding out for a Skylake 6 core HT cpu, and a 4ghz OC on air. Having the extra 2 cores is great when doing heavy multi-tasking
 
This close to Purley, you might as well wait. Broadwell-E uses Socket 2011-3, so your board won't be upgradable to Skylake-E.

But, if you get a Skylake-E, then you'll have the option of swapping in a CPU with more cores for a couple generations after.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom