What to upgrade for 144hz? 4670k GTX 1060

[Glarity]

Current Build:

 

CPU: i5-4670K @ 4.2GHz

GPU: MSI 1060 6GB Armor

Motherboard: MSI Z87-G45

Memory: Vengeance 2x8GB 1600

Storage: Crucial MX300 525GB + Seagate Barracuda 2TB 7200

PSU: CXM 750W

Cooler: Hyper 212 EVO with Noctua NF-F12 fan

 

Alright so I am pretty CPU bottle necked for 144hz 1080p in AAA games. In BFV I'm getting 100% CPU usage and 40-60% GPU usage at 80-100 FPS with dips in the 50s on lowest settings. In GTA V I get 100% CPU usage and about 10% GPU usage at 80fps on high settings. This build can run about anything at 60fps at ultra or high but struggles to hit even 100fps at low in any game which is a big problem for me.

 

I was thinking about getting a 9600k but then I found this video.

 

So now I am thinking about either getting an 8700k or a 9700k but they both beat each other in different benchmarks so there is not really a clear option.

 

Any help would be very appreciated

[BigRom]

Take the 8700K, its probably cheaper and you can use the extra cash on a good CPU cooler.

[Herman Mcpootis]

8700+B360 board if you're not overclocking, best not to go 9600k since it won't last as long. 8700k if the 9700k costs quite a bit more and you wanna oc.

[Beowulff83]

upgrade to a used i7 4790k?

[BigRom]

12 minutes ago, Beowulff83 said:

upgrade to a used i7 4790k?

Performance gain probably not enough to justify a dead-end in terms of upgrade path.

[Glarity]

1 hour ago, Herman Mcpootis said:

8700+B360 board if you're not overclocking, best not to go 9600k since it won't last as long. 8700k if the 9700k costs quite a bit more and you wanna oc.

8700k for $369.99

9700k for $409.99

 

9700k then?

[jerubedo]

26 minutes ago, Glarity said:

8700k for $369.99

9700k for $409.99

 

9700k then?

If your budget allows for it I'd go with the 9700K for the $40 extra. However I will point out one thing. That video you showed us has the 8600 equivalent  at 4.9GHz staying mostly above 144 with only the occasional dip. That being said, the 9600K can usually go to 5.0GHz (if not 5.1 if you get lucky), so I think it's still the better value proposition. All this talk about it not holding well is largely unfounded. People are just looking at the older i5s and the issues they are having in a handful of new titles, where on average they still do fine and what they are forgetting is that it took 8 years to get to that point of them struggling in those handful of titles. i5s launched in 2010 and only at the end of 2017 did some titles really challenge them, but again, not on the whole.

 

The market is indeed shifting but devs are notorious for being slow to catch up, so for how long the new 6 core i5s hold up is all speculation, but I'd wager it'd be at least 4 years. Just a best guess. And even when we get there, it should still do fine on the whole.

[Herman Mcpootis]

31 minutes ago, Glarity said:

8700k for $369.99

9700k for $409.99

 

9700k then?

If you're willing to spend the extra $40, sure. But it's not that much better than the 8700k. Ignore the guy above, last thing you want is to have to spend more money in a future upgrade simply because the i5 became a bottleneck.

[BigRom]

35 minutes ago, Glarity said:

8700k for $369.99

9700k for $409.99

 

9700k then?

9700K is a lateral step from a 8700K than an upgrade tbh

Save the money, take the 8700K and enjoy having hyperthreading instead.

[jerubedo]

1 hour ago, Herman Mcpootis said:

If you're willing to spend the extra $40, sure. But it's not that much better than the 8700k. Ignore the guy above, last thing you want is to have to spend more money in a future upgrade simply because the i5 became a bottleneck.

There's no reason to be rude because I have a different view. You have no foundation to say that a 9600K will become a bottleneck in any kind of short timeframe (as in a few years). The data that we have right now shows the 9600K performing really well in averages, 1% lows, and 0.1% lows. We see things like this:

 

 

This is in a 12 threaded game. The 9600K is holding its own even with a 6 thread deficit vs the competition and giving a smooth performance. If that isn't indicative of future performance in multi-threaded games, then I don't know what is.

 

 

Again the 9600K wins in averages, 1% lows, and 0.1% lows vs the 8700K. 

[GoldenLag]

32 minutes ago, jerubedo said:

 

Again the 9600K wins in averages, 1% lows, and 0.1% lows vs the 8700K

And in the same vid fron Gamernexus

 

 

You see that in a game like Far Cry 5. Abother title that leverage a lot of threads. All the CPUs without HT suffers a lot in their 1% lows. As in the delta between average and 1% is higher than it should be. 

 

Edit: and it showed the r7 2700 at 4,2 ghz providing a overall smoother experience, even if it didn push out the same amount of frames compared to the 9700k.

[jerubedo]

10 minutes ago, GoldenLag said:

And in the same vid fron Gamernexus

 

 

You see that in a game like Far Cry 5. Abother title that leverage a lot of threads. All the CPUs without HT suffers a lot in their 1% lows. As in the delta between average and 1% is higher than it should be. 

Yes and skip to 10:02 where he says, "We'd have to run a lot more passes here to determine how much those spikes are within normal error margins." Meaning he only did 1 run. He also said earlier in the video at 08:51 that he was seeing this with all i5s and with the 9700K and that it needs further investigation. The fact that he said that means he feels it's an oddity and that it shouldn't necessarily be happening.

 

My guess is it's a bad chunk of code in the engine that assumes the game will be using HT/SMT without checking to see if it's actually available and then when it tries to allocate resources on threads that don't exist it causes a problem. IF it's not within margin of error like he stated above. Also, I have a 9700K. This is what my FCAT chart looks like, so his run could have been erroneous:

 

 

A likely difference between our two systems as well is that I've explicitly disabled HPET in Windows.

[Stefan Payne]

3 hours ago, Glarity said:

Alright so I am pretty CPU bottle necked for 144hz 1080p in AAA games. In BFV I'm getting 100% CPU usage and 40-60% GPU usage at 80-100 FPS with dips in the 50s on lowest settings. In GTA V I get 100% CPU usage and about 10% GPU usage at 80fps on high settings. This build can run about anything at 60fps at ultra or high but struggles to hit even 100fps at low in any game which is a big problem for me.

 

I was thinking about getting a 9600k but then I found this video.

 

So now I am thinking about either getting an 8700k or a 9700k but they both beat each other in different benchmarks so there is not really a clear option.

 

Any help would be very appreciated

Now a weird idea:
Throw the 144Hz Screen away and get a Freesync Screen with a wide input range.

Then its not a problem if you only play at 75fps, is it?

[GoldenLag]

2 minutes ago, jerubedo said:

Yes and skip to 10:02 where he says, "We'd have to run a lot more passes here to determine how much those spikes are within normal error margins." Meaning he only did 1 run. He also said earlier in the video at 08:51 that he was seeing this with all i5s and with the 9700K and that it needs further investigation. The fact that he said that means he feels it's an oddity and that it shouldn't necessarily be happening.

And yet every single non-ht CPU suffered. If it was a single outlier it wouldnt be a concern.

 

If it was a bugged game the great, but it isnt as if we havent seen this kind of behavior before.

 

Wait for Zen 2>8700>9700

[jerubedo]

2 minutes ago, GoldenLag said:

And yet every single non-ht CPU suffered. If it was a single outlier it wouldnt be a concern. 

On his system, see my own personal chart above from a run I did a few hours ago of the in-game bench. And again a difference could be the HPET settings. I've disabled mine explicitly. That's on a 9700K.

 

2 minutes ago, GoldenLag said:

If it was a bugged game the great, but it isnt as if we havent seen this kind of behavior before.

It could be a bug such as what I mentioned above.

[Stefan Payne]

Just now, jerubedo said:

It could be a bug such as what I mentioned above.

A very Consistent bug.

Maybe something in the CPU Architecture??
Maybe something like Spectre/Meltdown?
Or the Mitigations for that that might eventually cause Problems??

[GoldenLag]

1 minute ago, jerubedo said:

could be a bug such as what I mentioned above

And that would be great, but again. Its not as if its a unknown behavior of non-ht CPUs

[GoldenLag]

1 minute ago, Stefan Payne said:

A very Consistent bug.

Maybe something in the CPU Architecture??
Maybe something like Spectre/Meltdown?
Or the Mitigations for that that might eventually cause Problems??

Well then it would affect HT CPUs aswell but it doesnt.

[jerubedo]

2 minutes ago, Stefan Payne said:

A very Consistent bug. 

Maybe something in the CPU Architecture??
Maybe something like Spectre/Meltdown?
Or the Mitigations for that that might eventually cause Problems??

It's not consistent or necessarily wide spread. See my chart. This is with a 9700K and a 1080 Ti on my own system. Ultra preset. Frame times were measured with FRAPS and then plotted in Excel.

[jerubedo]

And again, it doesn't make sense that the 9700K would present stutter because Far Cry 5 uses 8 threads and the 9700K has 8 threads, so there's that.

[Stefan Payne]

Just now, jerubedo said:

And again, it doesn't make sense that the 9700K would present stutter because Far Cry 5 uses 8 threads and the 9700K has 8 threads, so there's that.

No, that's exactly the reason why it stutters.

Because there ain't no resources for other stuff, so CPU has to led to rest, so that it can do the other stuff, while the one with SMT isn't bothered by that.

 

But you don't seem to understand SMT and claim that its not important, so that's that.

[jerubedo]

4 minutes ago, Stefan Payne said:

No, that's exactly the reason why it stutters.

Because there ain't no resources for other stuff, so CPU has to led to rest, so that it can do the other stuff, while the one with SMT isn't bothered by that.

 

But you don't seem to understand SMT and claim that its not important, so that's that.

And what else do you propose is going on? I'm sure Steve isn't testing with other things running in the background! And I mean in my setup FRAPS was running. Windows was running. So why do you propose there's a difference? Steve admitted multiple runs were needed in order to verify as seen at 10:02, and that further investigation was needed as seen at 08:51.

 

As a senior programmer I know EXACTLY how SMT works. I have a masters in information technology and 14 years of experience utilizing multithreaded CPUs in a variety of applications written by myself and my team. 

[Glarity]

Would HT and non-HT yield different results at lower graphics compared to ultra graphics? Every benchmark I see is ultra on a 2080 or 1080ti...

[jerubedo]

2 minutes ago, Glarity said:

Would HT and non-HT yield different results at lower graphics compared to ultra graphics? Every benchmark I see is ultra on a 2080 or 1080ti...

Yes it COULD to a degree (depending on the game) when both chips are the same architecture and the same clocks and one has HT and the other doesn't. In some games it could be as much as 15% or so (that is rarer) and in other games it could be the absolutely same results.

[Glarity]

2 hours ago, jerubedo said:

If your budget allows for it I'd go with the 9700K for the $40 extra. However I will point out one thing. That video you showed us has the 8600 equivalent  at 4.9GHz staying mostly above 144 with only the occasional dip. That being said, the 9600K can usually go to 5.0GHz (if not 5.1 if you get lucky), so I think it's still the better value proposition. All this talk about it not holding well is largely unfounded. People are just looking at the older i5s and the issues they are having in a handful of new titles, where on average they still do fine and what they are forgetting is that it took 8 years to get to that point of them struggling in those handful of titles. i5s launched in 2010 and only at the end of 2017 did some titles really challenge them, but again, not on the whole.

 

The market is indeed shifting but devs are notorious for being slow to catch up, so for how long the new 6 core i5s hold up is all speculation, but I'd wager it'd be at least 4 years. Just a best guess. And even when we get there, it should still do fine on the whole.

Also, the point of the video was to show that the 8600k equivalent was maxing out at 100% usage while the 8700k was only at about 50-60% usage