8700k or 9700k?

[ilc72]

And a couple more cores.

Yes but as CPU reviews (check out GamersNexus YouTube channel) show the 9700K has poorer frame time performance when compared to 8700K, this leads to micro-stuttering in some games.

 

[IT Troll]

Yes but as CPU reviews (check out GamersNexus YouTube channel) show the 9700K has poorer frame time performance when compared to 8700K, this leads to micro-stuttering in some games.

I watch most of the GN but don’t recall that one. Have you got a link?
Their website article seems to say the opposite:
”The 9700K manages to be both lowest and most consistent, making it the most fluid courier of frames in this test.”
https://www.gamersnexus.net/hwreviews/3421-intel-i7-9700k-review-benchmark-vs-8700k-and-more

 

[ilc72]

@IT Troll

This was I think the video I remembered:

https://youtu.be/QuUwLuQGPj4

At around 10 minutes or so you’ll see some of the issues at stock frequencies in Far Cry 5 which show higher overall FPS but a less consistent frame time. It’s also showing a deficit in other productivity applications such as Blender.

I am not saying that the 9700K is a bad product, it’s just that it’s existence is purely one of product segmentation. I would take the overclocked results for the 9700K as somewhat irrelevant as most consumers wouldn’t attempt to delid a CPU with a Soldered Thermal Interface (STIM) and therefore using a 9700K overclocked to 5.3GHz is somewhat pointless.

As an overall product, the 9700K seems to be a cynical attempt to create a product that shouldn’t really exist and removing SMT or Hyperthreading is an attempt to force consumers to buy a product at a price point that really makes no sense and is more about Intel’s 14nm supply issues as a result of their botched 10nm process.

The following article from semiaccurate.com covers off the real situation with 10nm process:

https://semiaccurate.com/2019/01/25/why-semiaccurate-called-10nm-wrong/

The Intel 9900K and 9700K are good products priced terribly, at least for enthusiasts.

 

[IT Troll]

This was I think the video I remembered:

https://youtu.be/QuUwLuQGPj4

At around 10 minutes or so you’ll see some of the issues at stock frequencies in Far Cry 5 which show higher overall FPS but a less consistent frame time. It’s also showing a deficit in other productivity applications such as Blender.

That video is the companion to the article I linked. It shows three momentary spikes in frametime in Far Cry 5, which Steve himself says could be within margin of error. I think it is leap to equate this to microstuttering in some games. He goes on to say that on average the 9700K is faster overall than the 8700K and 2700.

Rendering and encoding workloads are really the only time when 6c12t will pull ahead of 8c8t and even then only by a small margin. So if that is something you do a lot of then you absolutely need to be looking at a highly threaded processor. For everything else 8 physical cores wins over 6. When it comes to games, hyperthreading offers no advantage when you have 8 cores. In some cases it can be hinderance as it can reduce boost levels and thread affinity may not be balanced optimally across the physical cores. But even in these cases the differences are small.

 

[ilc72]

That video is the companion to the article I linked. It shows three momentary spikes in frametime in Far Cry 5, which Steve himself says could be within margin of error. I think it is leap to equate this to microstuttering in some games. He goes on to say that on average the 9700K is faster overall than the 8700K and 2700.

Rendering and encoding workloads are really the only time when 6c12t will pull ahead of 8c8t and even then only by a small margin. So if that is something you do a lot of then you absolutely need to be looking at a highly threaded processor. For everything else 8 physical cores wins over 6. When it comes to games, hyperthreading offers no advantage when you have 8 cores. In some cases it can be hinderance as it can reduce boost levels and thread affinity may not be balanced optimally across the physical cores. But even in these cases the differences are small.

I think we are pretty much agreed on most things, though I’m not sure the thread affinity comment is relevant on this platform as it’s not using an architecture such as Threadripper that use an interconnect rather than a ring bus.

Though if recent news is to be believed Intel CPU shortages are set to get worse:

https://www.extremetech.com/computing/287445-intel-cpu-shortage-could-worsen-in-q2-2019-arm-amd

 

[IT Troll]

I’m not sure the thread affinity comment is relevant on this platform as it’s not using an architecture such as Threadripper that use an interconnect rather than a ring bus.

The problem is a thread may return to the same virtual hyper-threaded CPU (because of cache affinity) which is now busy doing something else. So the thread has to wait or move to an idle CPU. Without hyperthreading the threads are on separate physical cores and so don't have to compete for the same resources in the same way. This is largely down to how the code has been programmed (i.e. badly). But I know from posts in this forum that some of the VR racing sims have this exact problem when run a processor with HT and so run much smoother without.

 

[Grim5]

Might be a tricky question

But if we take a 8700k and 9700k, all core overclock both to 5ghz and using the same cooler and NO delid, how do the temps compare?

 

[IT Troll]

But if we take a 8700k and 9700k, all core overclock both to 5ghz and using the same cooler and NO delid, how do the temps compare?

With no delid on either and both under full load, I believe reviews show the 9700K is ~10 degrees cooler. But a well executed delid on either will often be even lower.

If you don’t want to delid then the 9700K runs cooler out of the box. In the case of the 9700K, the addition gains probably don’t justify the risk and loss of warranty.