What to upgrade for 144hz? 4670k GTX 1060

[GoldenLag]

7 minutes ago, Glarity said:

Ok back to the question... 8700k or 9700k? And how do threads relate to high refresh rate?

HT doesnt deliver higher framerate, what it does deliver is less stutter. 

 

ill put how id rate it

 

wait for Zen 2 > 8700k > 9700k. 

 

wished intel brought a 8700k rebrand along to 9th gen, but they didnt. 

[Stefan Payne]

2 minutes ago, GoldenLag said:

same story was with the i5 of the previous era.

 

I strongly feel the deja vu with the same bullshit when the 2 Core CPUs were introduced.

They said don't get 2 Cores, nobody will ever need 2 Cores for Gaming.

They were wrong...

 

And when the 4 Cores came, they said don't get the 4 Core - also because Intel sucked at 4 Cores - because the Megahertz is important.

They were wrong again.

 

Then as you said, the i5 with SMT was the rage. And they said that nobody needs SMT. The 4 Core is more than enough.

They were wrong.

 

And now "magically" it is not wrong?! Are you serious?!

 

@Glarity

Why not just wait until "the next big thing" at the end of Next Month?? 
Because then you know what to do or not.

 

But to be honest: With the Data we have from the Past, the last ones were the i5 vs. i7 thing, we know that more threads is better.

 

Also keep in mind that the new upcoming M$+ Sony consoles will have Ryzen 3000 based CPUs wich might be a slight disadvantage for Intel because of the optimizations for the console systems.

[jerubedo]

4 minutes ago, GoldenLag said:

the 8700k was cheaper, and has more consistant performance deltas. 

What? The 9700K wins across the board in everything: Cinebench, Premiere Pro, Gaming avg FPS, 1% lows, 0.1% lows. 8 Physical cores is far superior to 6 cores and 12 threads.

 

To rephrase, it's a no brainer to get the SMT part if prices are close WHEN THE CORE COUNT IS EQUAL.

[GoldenLag]

4 minutes ago, jerubedo said:

I just noticed you said the 9700K. In general the 9700K wins across the board. 8 cores is generally better than 12 threads even in multithreaded applications like cinebench:

doesnt not show games. and Cinebench is perticularly HT sensetive. 

 

also doesnt show the historical and current issues of performance delta between average and 0,1% between HT and non-HT CPUs

[jerubedo]

Just now, GoldenLag said:

doesnt not show games. and Cinebench is perticularly HT sensetive. 

Exactly it's hugely HT sensitive and yet 8 cores beats 12 threads.

[jerubedo]

 

See the 9700K lows vs 8700K

[GoldenLag]

Just now, jerubedo said:

What? The 9700K wins across the board in everything: Cinebench, Premiere Pro, Gaming avg FPS, 1% lows, 0.1% lows. 8 Physical cores is far superior to 6 cores and 12 threads.

and neither of those are the ones im talking about............. the 9700k was also 40$ more expencive. 

 

Performance delta between 0,1% low and average. discrepency in these actually interfiere in games. they are both CPUs that push tons of frames onto the table, but only the 9700k has the thread issues. 

 

if it was premiere id probably change the tone, but games are not rendering workloads. 

1 minute ago, jerubedo said:

Exactly it's hugely HT sensitive and yet 8 cores beats 12 threads.

great for rendering workloads. games? hard to say without testing. and in testing only the non-HT CPUs have issues with performance deltas between the different  measurements of FPS. 

 

HT CPUs bring a smoother experience to the table, a more consistant expereicne between games. this isnt unknown behavior. 

[jerubedo]

6 minutes ago, Stefan Payne said:

trongly feel the deja vu with the same bullshit when the 2 Core CPUs were introduced.

They said don't get 2 Cores, nobody will ever need 2 Cores for Gaming.

They were wrong...

 

And when the 4 Cores came, they said don't get the 4 Core - also because Intel sucked at 4 Cores - because the Megahertz is important.

They were wrong again. 

 

Then as you said, the i5 with SMT was the rage. And they said that nobody needs SMT. The 4 Core is more than enough.

They were wrong.

 

And now "magically" it is not wrong?! Are you serious?!

None of them were wrong at the time. It took a long time for dual core to best single core in gaming, so the people saying to go single core at the time were correct. They only became wrong much later on. The same goes for frequency vs cores. And one can argue that even today Frequency is still pretty important in a lot of games. So when they were telling people to go with the single core or the higher frequency AT THE TIME, the people listening got better performance for a while. I like my performance up front. I'm not going to wait to see better numbers. By the time I'm done waiting, I'll be ready for my next purchase.

[jerubedo]

3 minutes ago, GoldenLag said:

and neither of those are the ones im talking about............. the 9700k was also 40$ more expencive. 

 

Performance delta between 0,1% low and average. discrepency in these actually interfiere in games. they are both CPUs that push tons of frames onto the table, but only the 9700k has the thread issues. 

 

if it was premiere id probably change the tone, but games are not rendering workloads. 

great for rendering workloads. games? hard to say without testing. and in testing only the non-HT CPUs have issues with performance deltas between the different  measurements of FPS. 

 

HT CPUs bring a smoother experience to the table, a more consistant expereicne between games. this isnt unknown behavior. 

No matter how you slice it, 8 physical cores is better than 6 with SMT/HT. The 8700K has the bigger differential as seen above. See the drops from 107 avg to 80 low? That's a bigger delta than 107 to 94. So the 9700K wins in that regard too.

[GoldenLag]

3 minutes ago, jerubedo said:

 

See the 9700K lows vs 8700K

 

thats very interesting as Gamers nexus had this review chart up. now idk the date of the either review were up. 

[Stefan Payne]

6 minutes ago, jerubedo said:

No matter how you slice it, 8 physical cores is better than 6 with SMT/HT.

That's your oppinion and proven wrong by @GoldenLag and others a couple of times.

6 minutes ago, jerubedo said:

 So the 9700K wins in that regard too.

And at what cost? 40€ more? 
That's not worth it. And also what about upcoming games that want more than 8 Threads?? Because Consoles these days only have 7 useful.

But the next gen will have 7 + SMT = 14 Threads.

You really believe your beloved 8 Core will be as good as the 12 Core??
I don't

Because its the same shit as with the i5 or the Dual Cores and everything else we've seen.

Proven wrong again and again and you still claim its better?! RLY?!

[Glarity]

Ok, then can you explain why these results happen? I don't really know anything about threads vs cores and how they work

 

[Stefan Payne]

1 minute ago, GoldenLag said:

thats very interesting as Gamers nexus had this review chart up. now idk the date of the either review were up. 

Who really trusts Toms Hardware these days?!
Especially after the "Just Buy it!" Stunt from their new Chief EDITOR??

 

I remember the Pentium D vs AMD Live Stability Test. And laughed when the AMD System didn't fail and on the Intel side the Board burned. 

 

Also its said that Intel Germany and THG were in the same building for a while...

[jerubedo]

Just now, Stefan Payne said:

That's your oppinion and proven wrong by @GoldenLag and others a couple of times.

Every person in the tech industry would tell you that having more physical cores is better than having less with SMT/HT. I've seen no such data proving that wrong

 

 

2 minutes ago, Stefan Payne said:

You really believe your beloved 8 Core will be as good as the 12 Core??

Of course 12 would be better. More cores = better. But at what core count will games stop providing meaningful results? No one knows yet.

 

3 minutes ago, Stefan Payne said:

Proven wrong again and again and you still claim its better?! RLY?!

Nothing has been proven. I've shown more data than anyone. All I have from you guys are pointing to past trends and a whole bunch of speculation on what the future might bring based on two data points.

[GoldenLag]

8 minutes ago, jerubedo said:

No matter how you slice it, 8 physical cores is better than 6 with SMT/HT. The 8700K has the bigger differential as seen above. See the drops from 107 avg to 80 low? That's a bigger delta than 107 to 94. So the 9700K wins in that regard too.

all you did was peak interest as we now have 2 benchmarks with 2 different results. 

 

now i adly couldnt find the data from that slide, what i did find was from Far Cry Primal (essentially a DLC) which over at Tom`s hardware had similar results. 

 

 

less performance delta. review from december 2018. which is rather new. 

[jerubedo]

4 minutes ago, Glarity said:

Ok, then can you explain why these results happen? I don't really know anything about threads vs cores and how they work

 

Yes, this is a case where the number of cores between both are the same, but the 7700K has more threads, so it won on lows. In the case of the 9700K there are 2 more cores and a boost in frequency. 

[jerubedo]

4 minutes ago, GoldenLag said:

all you did was peak interest as we now have 2 benchmarks with 2 different results. 

 

now i adly couldnt find the data from that slide, what i did find was from Far Cry Primal (essentially a DLC) which over at Tom`s hardware had similar results. 

 

 

less performance delta. review from december 2018. which is rather new. 

Steve is actually the outlier here. Anand has data as well. Again the 8700K has a bigger differential.

 

[Stefan Payne]

8 minutes ago, jerubedo said:

Every person in the tech industry would tell you that having more physical cores is better than having less with SMT/HT. I've seen no such data proving that wrong

We're talking about +2 Cores without SMT vs. 4 Threads on the other Hand.

 

And we already know that SMT is a net benefit, so with all those Threads, it don't look too good for your theory.

Also the only thing it proves is that Intel's SMT Implemetation is shit and don't perform too well...

 

But hey, why not get a CPU that's less affected by Spectre/Meltdown, that gets warmed up with new, similar exploits every couple of months...

 

[jerubedo]

10 minutes ago, Stefan Payne said:

We're talking about +2 Cores without SMT vs. 4 Threads on the other Hand. 

Let's get this straight. 4 threads = 2 cores. You're saying a 30% improvement on 2 cores is worth more than 100% improvement from ADDING 2 more cores. No. 

 

To put it another way, 6 cores with a 30% gain in performance = the equivalent of 7.8 cores. Verses 8 cores. 8 > 7.8. That's also why the 8700K comes close in a lot of cases, but doesn't quite beat it.

 

Of course this all assumes same architecture and same clocks.

 

This is what marketing has done to even YOU. You think you have 4 more of some "thing" but you don't.

[Stefan Payne]

Exactly, its almost as good but way cheaper.

See the Problem in your argument?!

 

And Golden Lag also proved that there are situations where SMT is really beneficial.

That's something you seem to totally ignore: The Price of the Component!

You claim that a 400€ part is better than a 200€ part, well, d'uh...

 

And now you prove that the 8700k is almost as good as the 9700k - but the 8700K is way cheaper!

[GoldenLag]

23 minutes ago, jerubedo said:

All I have from you guys are pointing to past trends and a whole bunch of speculation on what the future might bring based on two data points.

Its more than 2 if you include old i5 CPUs, but i try to keep it to modern titles. 

 

And history teaches us a lot of things. Especially when the same is essentially happening

19 minutes ago, jerubedo said:

Steve is actually the outlier here. Anand has data as well. Again the 8700K has a bigger differential.

 

Which date is this review from. Because last time i checked the CPUs having these performance characteristics were from older datasets. 

 

I had the earlier from GN  whoch showed a huge delta, and then i had the later from december which showed a smaller but similar performance characteristics 

 

Edit: maximum framerate was also higher in the other datasets

[jerubedo]

Just now, Stefan Payne said:

Exactly, its almost as good but way cheaper.

See the Problem in your argument?!

 

And Golden Lag also proved that there are situations where SMT is really beneficial.

Only when talking about the same core counts, in very limited scenarios. NOT for an 8 core vs 6 core debate.

 

And it's not WAY cheaper, it's $40 cheaper. At that price point you might as well have the more cores.

[Stefan Payne]

1 minute ago, jerubedo said:

And it's not WAY cheaper, it's $40 cheaper. At that price point you might as well have the more cores.

Yes, that's way cheaper.

And you said yourself that both are roughly equal, so why not go for the cheaper ones??
Why pay Intel more for something that's roughly the same??

 

Or why not wait for Ryzen 3000??

[jerubedo]

2 minutes ago, GoldenLag said:

Its more than 2 if you include old i5 CPUs, but i try to keep it to modern titles. 

 

And history teaches us a lot of things. Especially when the same is essentially happening

Which date is this review from. Because last time i checked the CPUs having these performance characteristics were from older datasets. 

 

I had the earlier from GN  whoch showed a huge delta, and then i had the later from december which showed a smaller but similar performance characteristics

The Anand is October 2018 and Tom's was November 2018. Not terribly long ago, and not far off of Steve's December results. There's also data from Techpowerup that agrees with Anand and Tom's, 1 sec. That would be 3 against 1. 

[jerubedo]

Just now, Stefan Payne said:

Yes, that's way cheaper.

And you said yourself that both are roughly equal, so why not go for the cheaper ones??
Why pay Intel more for something that's roughly the same??

 

Or why not wait for Ryzen 3000??

Because look at the charts above. The $40 buys quite a few more average FPS and lows.