CPU or GPU? my upgrade suggestion

[MrFuzzy 2,118]

On 28/7/2017 at 5:10 AM, dommel1234 said:

I said essentialy, of cource it is, if you delid it and loose your guarantee can reach some 5 Ghz and beyond. The point is that they dont represent the technological revolution to make worth the investment especially when the 2XXX performance is comparable or a bit inferior

Mine is not delidded and rock stable at 4,9 GHz - 1,25 V. 

Besides, between 2xxx generation (Sandy Bridge) and 7xxx (Kaby Lake) there is a 30% average IPC improvement.

https://www.pcper.com/reviews/Processors/Intel-Core-i7-7700K-Review-Kaby-Lake-and-14nm/Clock-Clock-Kaby-Lake-Skylake-Broad

Therefore a 5 GHz 7700K is about 50% faster than a 2600K @ 4,4 GHz.

 

[Noel 4,492]

6 hours ago, enek0id said:

Therefore a 5 GHz 7700K is about 50% faster than a 2600K @ 4,4 GHz.

That's baloney in terms of real world results:  http://cpu.userbenchmark.com/Compare/Intel-Core-i7-7700-vs-Intel-Core-i7-3930K/3887vs1487

This doesn't accurately reflect the overclocked result but even when you do you're lucky to get 25% increase in single core performance.

[MrFuzzy 2,118]

OMG

That website is well known to be rubbish, it's just self-generated numbers based on an algorithm, please! 

Check the link I provided and the following:

http://www.gamersnexus.net/guides/2867-intel-i7-2600k-2017-benchmark-vs-7700k-1700-more/page-2

https://www.hardocp.com/article/2017/01/13/kaby_lake_7700k_vs_sandy_bridge_2600k_ipc_review/4

 

[vortex681 806]

10 hours ago, enek0id said:

Check the link I provided and the following:

http://www.gamersnexus.net/guides/2867-intel-i7-2600k-2017-benchmark-vs-7700k-1700-more/page-2

https://www.hardocp.com/article/2017/01/13/kaby_lake_7700k_vs_sandy_bridge_2600k_ipc_review/4

On the last page of the second link you posted, it says:

On the desktop gaming front in terms of benchmarks, we have seen Kaby Lake pull a general 20% advantage. Keep in mind these are comparing Kaby Lake and Sandy Bridge at identical clocks. But I will suggest that when we go back and test real-world gaming, that delta will get a lot more narrow...

If you own a highly clocked 2600K/2500K Sandy Bridge processor and it is still giving you stable performance, it is hard for me to make the argument that it is time for you to upgrade

The comparison was with both CPUs running at the same clock speed of 4.5GHz. If you assume that increasing the clock speed will have a linear effect on performance (as the IPC remains the same), then increasing the 7700K to 5Ghz would result in an extra 10% improvement, which gets you nowhere near the total of 50% you claimed!

If you look at page 2, the single thread performance of the two CPUs is very similar. Bearing in mind that P3D still tends to use one core more than the others, this would result in an even smaller performance gap between the 2 CPUs

Take a look at https://www.youtube.com/watch?v=4sx1kLGVAF0&feature=youtu.be. Although it uses a 6700K, there's virtually no difference in IPC between the 6700K and the 7700K. As you go further through the video (in different games), the 2600K gets much closer to the 6700K.

There's no doubt that the 7700K will be the faster CPU and will overclock higher, but the difference for gaming will not be as massive as you claim.

[MrFuzzy 2,118]

I was talking about the overall IPC increment indeed, which is highlighted by synthetic benchmarks.

Of course +50% * IPC is very different from +50% * fps in gaming!

From a 2600K at 4.4 GHz to a 7700K at 5 GHz I would expect +10-15 fps in Prepar3D. If it's worth the upgrade, it only depends on the OP pockets :)

 

[Guest]

Please guys, stay closer to P3D CPU behaviour and P3D fps, not just CPU abstract reviews.

[Noel 4,492]

6 hours ago, enek0id said:

From a 2600K at 4.4 GHz to a 7700K at 5 GHz I would expect +10-15 fps in Prepar3D. 

I think you have to express change as a percentage.  IOW, if you're flying thru ultra dense scenery in the NGX and have gone down to a frame rate of 18 on the 2600K at 4.4 you are going to see 28-33 frames per second on your 7700K.   You might if baseline is 60 frames and you go to 70-75.  In the first example you're seeing a frame rate increase of 55%, whereas in the latter only 14%.  It would be useful to have frames jump to 30 from 18, but I don't think you're going to see that whatsoever, it will be more like the 14% increase, or moving from 18 to 21, which is absolutely hardly worth noting.  Therein lies the issue w/ paltry improvements in single-core performance.
 

[MrFuzzy 2,118]

Yes you are correct, my estimation of 10-15 fps is based on his configuration (i7 2600K @ 4,4 GHz + GTX 1070) at 1080p though. If he has an average of 40-45 fps now, with a 7700K @ 4,8 GHz (easily achievable even with an air cooler) he would get 50-55. CPU is the bottleneck at 1080p, and +40-50% IPC would give +20-25% fps IMHO.

There are too many parameters, including P3D internal settings, it's very difficult even to give a rough estimation actually. 

[Noel 4,492]

On 7/30/2017 at 10:07 PM, enek0id said:

OMG

That website is well known to be rubbish, it's just self-generated numbers based on an algorithm, please! 

Check the link I provided and the following:

http://www.gamersnexus.net/guides/2867-intel-i7-2600k-2017-benchmark-vs-7700k-1700-more/page-2

https://www.hardocp.com/article/2017/01/13/kaby_lake_7700k_vs_sandy_bridge_2600k_ipc_review/4

 

OMG, from your second link!  Please!  Here, from old Sandy Bridge to new Kaby Lake, both overclocked to 4.5Ghz:

1. In this gaming engine we see a 19.868% increase with the 7700K
2. Heaven comes in very close to what we saw above with the Kaby Lake providing a 20.755% increase.
3. The 7700K drops a bit with a 17.927% increase in frame rate.
4. Metro is certainly our most graphic-intensive benchmark, and we see the 7700K provide a 14.907% increase in frames.
5. Back on track with AotS we see the 7700K give us an increase of 20.588% increase in frame rates.

I rest my case!  Wake me up when LM makes P3D use all of those sleepy cores!

[BlueStar 323]

On ‎7‎/‎27‎/‎2017 at 6:54 PM, dommel1234 said:

The improvement is less in higher resolution when the GPU takes charge. But even this is not worth the money to buy an essentialy not overclockable overheating chip like the 7700k. Surely there is much better cpus to come soon

I ran my 7700k up to 4.8 on air and the max. temp was 80.  I've backed it off to 4.5 and the max temp is 70C.

blaustern

[Jiri Kocman 20]

On 28. 7. 2017 at 2:18 AM, lownslo said:

Not overclockable?  Hmmm, lots of folks have pushed that chip to 5Ghz and beyond.  It's proven itself to be overclockable.

Greg

Two major retailers in my country selling i7-7700k pretested for 5 and 5.1GHz OC by "PRO" overclocker, they give you guarantee that you can OC that CPU at this clocks, sure those picked ones have higher price tag, but better than buy it, then test it to OC and if it is not able to achieve higher clocks, you will return it and pick another one... Not good for you, not good for retailer (I live in EU where law allows you to return stuff and get 100% money back with no ask why during 14 day period after purchase)