Only using 2 cores !

[RobertR1]

Don't forget that the i5 only runs at 4.3 GHz. The i7 runs at 4.9 while the i9 has a turbo of 5.0 but often throttles down a little either due to reaching tdp limit or too high temperatures.
If you put the i5 and i7 to 5 GHz and enable the i9 to use whatever power it wants, the differences will basically vanish with the i9 being fastest, then i7, then i5 but it's like 2 fps in most games.

But there aren't many benchmarks out there which show this. You can see it from 5 GHz cinebench single thread runs though.

What this leads to though is that everybody who gets the i5 should look into overclocking it to at least 4.8 GHz all core turbo!

The 9600k is oc'd in the gamernexus review. That's why i specifically posted that one.

 

[RasmusP]

The 9600k is oc'd in the gamernexus review. That's why i specifically posted that one.

I had a look. I honestly didn't watch that one yet, so thanks for showing me!
What I see is a mix of both. Games that can make use of 8 cores will be clearly better on the 9700k, however games that don't are basically identical on both CPUs.

Not wanna start an argument, we both know what we're talking about and just stand on different sides with our opinions.
Anyway here's a summary of my knowledge. I gladly take on a teaching if you can provide facts against what I'm saying!

GTA V: (OC 9600k better than stock 9700k)
- 9600k @5.2 GHz: 87/95/126
- 9700k @5.1 GHz: 91/100/128

Assassin's Creed and other AAA games show a bigger gap.
Sadly there are no benchmarks on gamernexus with games that clearly run on only a few threads like simracing titles or cinebench single thread runs.

I tried to quickly find a few again, here are my findings:
https://www.overclock3d.net/reviews/cpu_mainboard/intel_i5_9600k_and_i7_9700k_review/7

Sadly, their i5 runs at 4.9 and the i7 at 5.2 ; if you compare it to the stock i7, you'll get 4.9 vs 4.9 though.

Single thread results: Cinebench R15 single thread:
9600k @ 4.9 GHz: 211 points
9700k @ 4.9 GHz: 204 points
9700k @ 5.2 GHz: 225 points

Basically, the clock speed alone gains the few points in the end. The 8600k max OC'ed scores 223 points compared to stock: 189 points.

Games I know that don't use many cores: PUBG and CS:GO. Benchmarks at 4:05 and 5:40.

Both CPUs Stock though!
PUBG:
- 9600k: 104/164 fps
- 9700k: 106/171 fps

CS:GO:
- 9600k: 516/666 fps
- 9700k: 521/673 fps

Basically no difference here, although being stock clocked.

Then a game that definitely CAN make use of the cores:
Assassin's Creed: Odyssey:
- 9600k: 67/90 fps
- 9700k: 78/91 fps

Yep, clear winner for this game and if someone buy a new CPU to play modern triple A games on high settings and wants 60+ fps then definitely go with the 9700k!

Then a very nice video for this case:

Cinebench single thread, stock vs 5.0 GHz:

- 9600k stock: 198 points
- 9700k stock: 214 points

- 9600k @ 5 GHz: 214 points
- 9700k @ 5 GHz: 218 points

Witcher 3, can make use of more cores:
- 9600k @ 5 GHz: 98/142 fps
- 9700k @ 5 GHz: 107/152 fps

This video is very interesting. It tests both CPUs at 5.0 GHz but the i5 with a 2080 ti vs the i7 with a 2080.
So to see a comparison close to simracing situations, we'll have to look for a comparison where the GPU is not at 100% and the CPU never goes too high either.
Only example I could quickly find is the hitman comparison at 5:45.

The whole scene basically shows identical fps. The 1% low vs the 0.1% low are a bit funny though. Probably down to the GPU spiking very shortly.

- 9600k @ 5 GHz: 1% low: 69 fps, 0.1% low: 62 fps
- 9700k @ 5 GHz: 1% low: 68 fps, 0.1% low: 16 fps


My last one, Project Cars 2 Benchmark:
https://www.tomshw.de/2018/10/19/intel-core-i9-9900k-i7-9700k-i5-9600k-im-test-review/6/

- 9600k @ stock: 74.7 / 104 fps
- 9700k @ stock: 76.2 / 114.2 fps

Add the missing 600 MHz to the 9600k and you'll probably see identical results again.

Sorry for the long post. It was a nice summary for myself though. Now I'm looking forward to add the 3600(x) and the 3700x to this and see if and if yes, what I'm gonna buy at the end of the summer

 

[RobertR1]

That's a good write up Rasmus. You're right, with clocks being even most older titles or even newer titles using older engines, you're likely not going to see a difference.

Where it'll come into play are the newer games where they can scale intelligently across more physical cores.

On a side note, I've been digging into something I noticed during the review cycles. It was really difficult to find any title that wasn't hindered by having HT enabled. Basically, the 9700k given clock parity with the 9900k would often outperform it in average but where it really pulled out ahead in a lot of titles was on the 1% frame rate. I'm gonna dig more into it and do my own testing.

For the new AMD chips, the difference was even more staggering by disabling SMT in a lot of titles.

End of summer is probably a good time. Intel should have their price drops done and AMD should have the BIOS issues sorted out by then.

 

[Case_]

and games like having 8 physical cores

People always say that about modern games, and yet there's perhaps a handful of games for which that's actually true. (And even in those cases, I'd say they don't go over 6 cores most of the time.)

 

[Bjarne Hansen]

My last one, Project Cars 2 Benchmark:
https://www.tomshw.de/2018/10/19/intel-core-i9-9900k-i7-9700k-i5-9600k-im-test-review/6/

- 9600k @ stock: 74.7 / 104 fps
- 9700k @ stock: 76.2 / 114.2 fps

Are all the German version Project Cars2 running DX12?
And a i5 8400 faster than a 9600k !

 

[RasmusP]

Are all the German version Project Cars2 running DX12?
And a i5 8400 faster than a 9600k !

Yeah there's certainly something wrong, lol. Good spot!
I wouldn't use it for a proper comparison. I wanted to show that it really doesn't matter which of the CPUs you get for that game.
I'd like to find a dx11 comparison but sadly couldn't find one quickly!

 

[Bjarne Hansen]

Only spotted it because I am interested in how the numbers of cores effect PC2.
Would really like to see a Cpu test with race sims also running wheels, motion or tactical sound and crew chief.

 

[RasmusP]

Only spotted it because I am interested in how the numbers of cores effect PC2.
Would really like to see a Cpu test with race sims also running wheels, motion or tactical sound and crew chief.

Found something
German but the charts are "international":
https://www.pcgameshardware.de/commoncfm/comparison/clickSwitch.cfm?id=141340
And English:
https://www.tomshardware.com/reviews/project-cars-2-performance-benchmarks,5430-7.html

 

[Bjarne Hansen]

Found something
German but the charts are "international":
https://www.pcgameshardware.de/commoncfm/comparison/clickSwitch.cfm?id=141340
And English:
https://www.tomshardware.com/reviews/project-cars-2-performance-benchmarks,5430-7.html

Thanks!
From the German (if we can trust the German accuracy by now? )
4 cores / 8 treads 13% slower than 8 cores/ 16 treads.
6 cores / 12 treads 3% slower than 8 cores/ 16 treads.
With other typical background programs like Teamtalk/ simhub/sim comander/ Crew Chief the 8 core Cpu's are likely to pull even more ahead.
Just not sure how much I would gain already having a i7 7700K running stable at 4.8 Ghz, not even delidded it yet.

 

[Case_]

With other typical background programs like Teamtalk/ simhub/sim comander/ Crew Chief the 8 core Cpu's are likely to pull even more ahead.

Seriously doubt that. Don't know about Sim Commander, but the other apps mentioned hardly need any CPU time at all.

 

[Bjarne Hansen]

I do not know but then again I have no idea how much "hardly" is ether?
Do know that on my 7700k in VR it's punishing to run those apps that you clam that hardly need any CPU time.
To the point that it changes from 90 hz to 20-30% reprojection time.

 

[Case_]

OK, then, some actual numbers. Running a 35 AI race in rF2 with a SimHub dashboard running on Android phone:

SimHub: one thread worth mentioning, requiring around 0.7-1 % CPU
Crew Chief: no single thread stands out, total CPU usage around 0.1 %
Teamspeak: no single thread stands out, total CPU usage around 0.1 %

I have no idea about VR, neither do I know what reprojection time is, but I find it quite suspicious such high demanding apps apparently bring your system to its knees.

 

[RasmusP]

OK, then, some actual numbers. Running a 35 AI race in rF2 with a SimHub dashboard running on Android phone:

SimHub: one thread worth mentioning, requiring around 0.7-1 % CPU
Crew Chief: no single thread stands out, total CPU usage around 0.1 %
Teamspeak: no single thread stands out, total CPU usage around 0.1 %

I have no idea about VR, neither do I know what reprojection time is, but I find it quite suspicious such high demanding apps apparently bring your system to its knees.

VR adds a lot to the big major thread. Reprojection is when the headset fills in a frame between each frame. The game drops to 45 fps vsynced with the headset filling in frames to make it 90 fps.
It adds some ghosting, input lag, likely more dizziness etc.
But only the 8700k and above are Really capable of easily running constant 90 fps in the current Sims. Only due to the single thread performance (again, urgh).
A statistic is recorded about how much percent the fps drop down to "Reprojection" so you can easily see improvements due to settings etc. Reprojection isn't always visible but sometimes, when it kicks in it might disturb you a lot. Like mid corner during a 3 wide fight.

A good benchmark to see how much the tiny apps influence the VR performance would be to run cinebench with 2 threads and then run the apps and compare the score.
Sadly the apps need a constant data feed from the game so yeah, it's not possible..

 

[Frank]

I had a look. I honestly didn't watch that one yet, so thanks for showing me!
What I see is a mix of both. Games that can make use of 8 cores will be clearly better on the 9700k, however games that don't are basically identical on both CPUs.

Not wanna start an argument, we both know what we're talking about and just stand on different sides with our opinions.
Anyway here's a summary of my knowledge. I gladly take on a teaching if you can provide facts against what I'm saying!

GTA V: (OC 9600k better than stock 9700k)
- 9600k @5.2 GHz: 87/95/126
- 9700k @5.1 GHz: 91/100/128

Assassin's Creed and other AAA games show a bigger gap.
Sadly there are no benchmarks on gamernexus with games that clearly run on only a few threads like simracing titles or cinebench single thread runs.

I tried to quickly find a few again, here are my findings:
https://www.overclock3d.net/reviews/cpu_mainboard/intel_i5_9600k_and_i7_9700k_review/7

Sadly, their i5 runs at 4.9 and the i7 at 5.2 ; if you compare it to the stock i7, you'll get 4.9 vs 4.9 though.

Single thread results: Cinebench R15 single thread:
9600k @ 4.9 GHz: 211 points
9700k @ 4.9 GHz: 204 points
9700k @ 5.2 GHz: 225 points

Basically, the clock speed alone gains the few points in the end. The 8600k max OC'ed scores 223 points compared to stock: 189 points.

Games I know that don't use many cores: PUBG and CS:GO. Benchmarks at 4:05 and 5:40.

Both CPUs Stock though!
PUBG:
- 9600k: 104/164 fps
- 9700k: 106/171 fps

CS:GO:
- 9600k: 516/666 fps
- 9700k: 521/673 fps

Basically no difference here, although being stock clocked.

Then a game that definitely CAN make use of the cores:
Assassin's Creed: Odyssey:
- 9600k: 67/90 fps
- 9700k: 78/91 fps

Yep, clear winner for this game and if someone buy a new CPU to play modern triple A games on high settings and wants 60+ fps then definitely go with the 9700k!

Then a very nice video for this case:

Cinebench single thread, stock vs 5.0 GHz:

- 9600k stock: 198 points
- 9700k stock: 214 points

- 9600k @ 5 GHz: 214 points
- 9700k @ 5 GHz: 218 points

Witcher 3, can make use of more cores:
- 9600k @ 5 GHz: 98/142 fps
- 9700k @ 5 GHz: 107/152 fps

This video is very interesting. It tests both CPUs at 5.0 GHz but the i5 with a 2080 ti vs the i7 with a 2080.
So to see a comparison close to simracing situations, we'll have to look for a comparison where the GPU is not at 100% and the CPU never goes too high either.
Only example I could quickly find is the hitman comparison at 5:45.

The whole scene basically shows identical fps. The 1% low vs the 0.1% low are a bit funny though. Probably down to the GPU spiking very shortly.

- 9600k @ 5 GHz: 1% low: 69 fps, 0.1% low: 62 fps
- 9700k @ 5 GHz: 1% low: 68 fps, 0.1% low: 16 fps


My last one, Project Cars 2 Benchmark:
https://www.tomshw.de/2018/10/19/intel-core-i9-9900k-i7-9700k-i5-9600k-im-test-review/6/

- 9600k @ stock: 74.7 / 104 fps
- 9700k @ stock: 76.2 / 114.2 fps

Add the missing 600 MHz to the 9600k and you'll probably see identical results again.

Sorry for the long post. It was a nice summary for myself though. Now I'm looking forward to add the 3600(x) and the 3700x to this and see if and if yes, what I'm gonna buy at the end of the summer

I disagreed until I read your last line :-P.

So far it looks like if you are doing anything more than just gaming, as in; running other apps in the background while gaming. AMD is going to be the king, by a mile, even more so when streaming.

 

[Kek700]

Signs are looking promising at the moment, built the computer in its own case, keeping my i5 6600k
4.6GHz and gtx1080 in the original PC.
At the moment using an old gtx760 x 2G with the 4.3Ghz i5 9600k and 2133 memory instead of 3600, admittedly with some settings dumbed down x8 instead of x16 and x2 instead of x4 i am getting 100 to 110 fps at Spa on my 3440 x 1440 120Hz monitor.

Maybe i am being a bit naive on this but for a £40 gpu and a £200 cpu that is not too shabby.

i do realise in isolation this is a bit of a meaningless statement, i am obviously interested in the out
come when i can run the i5 9600k at, i hope 5Ghz with the Gtx1080 and 4Ghz memory with it.

 

[RobertR1]

For RAM, you might want to look into running a lower bandwidth like 3600mhz and tightening the timings. The speed and reduced latency is generally what games prefer if they're memory sensitive (many aren't).

I'd take a CL14 3600mhz over CL19 4000mhz all day with reduced trfc and increase tREFI.

That's right Ernie. I'm gonna send you down the memory turning black hole so you slow down a bit in our races

 

[Kek700]

Thanks for the information Robert, will do that and see if the ram can run at CL14 @ 3666, not sure what it is at the moment.
Although out side of AC these articles may be of little use, I think they give someone who is investing in this Sim a real world scenario, making judgements on buying the best bang for the $ buy.
Often I have read an article that may very well be applicable to a general reader, when I have applied it to AC, has proved an expensive and pointless exercise.
I have read RasmusP articles on this subject for the last 12months, using his and others judgements on what to buy when I made the decision to upgrade.
So what I am saying is if it proves to be a point less exercise, it’s his fault.