28 minutes ago, Benjamin J said:

 

• How does this translate to the 9900K? I suppose that turbo boost on the 9900K should do a similarly good job?
• Is a boost on one single core good enough for enhanced performance in P3D v4?
• And what about v5? I read that v5 seems to spread the load over core 0, 2 and 4, rather than having it all on core 0, so that boost on just core0 seems like it would be an 'oudated' approach?
• @Noel, I see you're using the Noctua DH15, and it seems as though you were able to reach 4.9GHz with your 9900K, which is something of a relief. How does this perform in v5? As much as v4 is laoded with all the stuff I like, I do feel an 'itch' to jump to the next iteration sometime soon.

 

Ben, prior to enabling HT I ran all 8 cores at [email protected], which is hardly a radical voltage level for this CPU, and yes on the DH15.  In full summer here when house ambient was 76F-78F, in P3D v4.5 which beats on CPUs compared to what MSFS does I was seeing core temps around 60C tops w/ HT disabled.

In P3D 4.5 w/ HT enabled temps will get near or up to 70C in summer time here.   In MSFS w/ HT enabled at 4.9Ghz on all cores average core temp is around 45C.  In fact if it weren't for the fact I'm still using P3D 4.5 some I'd just dial back the 9900K to 4.2Ghz or something like.  Regardless of using 8 core or 16 logical processors I always use an affinity mask that puts everything except P3D on 0 core (or LPs 0 & 1 if HT enabled), so P3D has access to cores 1-7, or 4-15 if HT enabled, using zero as the 1st instance.  This puts competing resources on the 0 core/0-1 LP which is ample to run anything else I'm running alongside P3D v4.5.  

For now and hopefully forever (meaning, MSFS continues to improve over the next years) I won't be purchasing v5 as to me it's just another version of the same software I've used for a very many years now, and I am aware of how it differs from prior versions.  I find MSFS so compelling I'm not spending a lot of time now in P3D, though do on occasion to enjoy the lovely PMDG NGXu and 777, and my two other favs Majestic Q400 & Carenado Phenom 505 w/ GTN750.  I like P3D for dusk and night flights which looks just dandy in fact I like some of the night lighting in P3D more at times, even though when I look at photos of actual night flight views from airplanes MSFS looks much more realistic even if not as pretty to look at.  That is an area I hope is tweaked a bit in MSFS.

An i9-10900K is 3.66% faster than i9-9900K at the same clock speed FWIW.   But back to my original point:  don't r/o being able to enable hyperthreading!  It's really been the final piece as far as a CPU is concered for both P3D 4.5 & MSFS.   Right now my 9900K is half asleep while running MSFS w/ an average CPU utilization of maybe 22%.  To me this suggests there is ample headroom for the so-called study-level offerings we hope will come to MSFS.

 

@Noel, Thanks for tkaing the time to reply with such detail and with this amount of data! Makes my choice easier. Now to your points...

Your temps are encouraging! Do you have any idea how the 9700K and 9900K compare with regards to how cool they run? I ask, because I ordered a 9700K some time ago and managed to get it up to about 4.6GHz before temps starting getting in the 80+ range while running Prime95, A little below that in P3D v4. I ended up returning that processor because I didn't see much in the way of a performance improvement. As you cna imagine, I'm a little weary the 9900K will go the same way, despite your encouraging data.

That said, comparing prices it seems the cheaper option will be the 9900K by far, as I can keep using my current motherboard. A decent motherboard will cost me 150-200$, the 10700K itself is another 375$. So just getting the 9900K, at 450$, is the cheaper of the two options. That said, both are probably more expensive than I feel comfortable with right now...

So, I'm going tos tick with what your experience with regards to P3D and MSFS. I have already received my extra 16GB of RAM, so that's the only upgrade I will go with right now. I'll put the machine back together, I'll get MSFS, and when it's released P3Dv5.1, and will simply see how it works out. If it works well, great! If not, I can always decide upgrade to the 9900K at that point. Perhaps, by then, the GPU market will have become sufficiently 'loose' that I can even get a new GPU with some ease.

But if anybody ahs any data on the temps of a 9900K versus the 9700K, that'd be great! Perhaps I'm simply doing some wrong, or something isn't optimize din my setup that I couldn't get it to run at a higher clock speed... After all, this is my first self-built machine and my first OC, so...

2 minutes ago, Benjamin J said:

 

But if anybody ahs any data on the temps of a 9900K versus the 9700K, that'd be great! Perhaps I'm simply doing some wrong, or something isn't optimize din my setup that I couldn't get it to run at a higher clock speed... After all, this is my first self-built machine and my first OC, so...

Ben, my guess is something was amiss on your 9700K if you were seeing temps around 80 at 4.6Hz.  What kind of core voltage?  The first guesses would be less than optimally applied thermal paste; and 2, not setting the fan speeds correctly; and 3, perhaps contributing insufficient case ventilation, and 4 very high ambient temps.  What temp does your i5-8600K get to running Prime95?

As you may know the 9700k and 9900k are the same chip w/ the former essentially having had HT disabled and the later added more cache, and there may have been some binning to get slightly better overclocking potential for the 9900K, but that particular issue will be irrrelevant IMO.  HT disabled in P3D 4.5 which should be more CPU-limited than 5.x still runs well below 60C here on air, in summertime.  My only takehome truly is make sure IF n when you upgrade your CPU be sure it is HT-capable.  One other thing to consider is as the newest CPUs dominate the prior versions will become less available and hence more pricey for folks who want to use their current motherboard after new models have been out a while.

Thanks for getting back to me Noel!

 

5 hours ago, Noel said:

Ben, my guess is something was amiss on your 9700K if you were seeing temps around 80 at 4.6Hz.  What kind of core voltage?  The first guesses would be less than optimally applied thermal paste; and 2, not setting the fan speeds correctly; and 3, perhaps contributing insufficient case ventilation, and 4 very high ambient temps.  What temp does your i5-8600K get to running Prime95?

From what I remember it would be upper 70's or so, ~78C. Is that something that would be expected, or...? I think I'm at 1.23v or so, but I'd need to look again to be certain. The 9700K was a similar story, mind you. Perhaps I got it up to 1.21, 1.22v to get it to 4.6GHz.

When I removed the DH15 to switch back to the 8600K, my feeling was that the paste was applied to the CPU okay, it seemed spread out well enough. At least, about what I saw in the Youtube videos... Is it worthwhile buying new paste? The one I have now I bought some 2-3 years ago.

The case ventilation could be an issue. I have a cat who is very curious and loves to poke into holes and scratch and bite cables and wires. He would also routinely sit on top of the case, blocking the vents (I suppose he enjoyed the warm air). So I had to block off some of the spaces that would give access to the top and back of the case to make sure the cat doesn't get to it. I placed some cardboard barriers, which shouldn't obstruct the vents directly, but I can certainly see this making air circulation around the case inefficient. It's certainly possible this contributed, but it's hard to see what I can do about it, given that without these barriers the cat is likely to kill the computer (and himself). Mind you, I do think I have enough fans... Two in the front and one in he bottom sucking air in, and then one fan at the back and tow at the top for exhaust...

28 minutes ago, Benjamin J said:

Thanks for getting back to me Noel!

 

From what I remember it would be upper 70's or so, ~78C. Is that something that would be expected, or...? I think I'm at 1.23v or so, but I'd need to look again to be certain. The 9700K was a similar story, mind you. Perhaps I got it up to 1.21, 1.22v to get it to 4.6GHz.

When I removed the DH15 to switch back to the 8600K, my feeling was that the paste was applied to the CPU okay, it seemed spread out well enough. At least, about what I saw in the Youtube videos... Is it worthwhile buying new paste? The one I have now I bought some 2-3 years ago.

The case ventilation could be an issue. I have a cat who is very curious and loves to poke into holes and scratch and bite cables and wires. He would also routinely sit on top of the case, blocking the vents (I suppose he enjoyed the warm air). So I had to block off some of the spaces that would give access to the top and back of the case to make sure the cat doesn't get to it. I placed some cardboard barriers, which shouldn't obstruct the vents directly, but I can certainly see this making air circulation around the case inefficient. It's certainly possible this contributed, but it's hard to see what I can do about it, given that without these barriers the cat is likely to kill the computer (and himself). Mind you, I do think I have enough fans... Two in the front and one in he bottom sucking air in, and then one fan at the back and tow at the top for exhaust...

Sounds like your 9700K was on the low end of spectrum, performing poorer than average.   They are supposed to do 4900 on one core.   Look at your fan controls and make sure they're going all out when CPU core temp gets to maybe 50C so the chip doesn't heat up then take so long to cool down.  If you've looked at and followed a few different resources about how to apply paste you should be OK.  I think the tendency is to put too much on which can reduce efficacy I read.  Don't know what paste you're using but that too you can read up on and replace with the better product as indicated if you go w/ a 9900K.  From everything you've said so far it would be tempting to pick up a 9900K while they are still available and cheap.  I don't know how the Z370 does w/ that CPU v the Z390 I'm using which has been very stable.

Thanks for your input here @Noel! I suppose the turbo boost by itself would have been fine, so 4.9 on one core would be no problem. I was trying to do an all-core OC, and that's where I got stuck at 4.6GHz. So, I think I'll stick to what I have now, put my computer back together, see how it works with MSFS (and perhaps v5), and if I don't like it I might just get the 9900K. Thanks for your help!

Don't forget it was enabling HT that solved those annoyinig long frames pauses in both sims, and I see the 8600K does not support HT.  But who knows perhaps there are other factors at play that may allow you to avoid those.

Edited October 3, 20205 yr by Noel

19 hours ago, Noel said:

Don't forget it was enabling HT that solved those annoyinig long frames pauses in both sims, and I see the 8600K does not support HT.  But who knows perhaps there are other factors at play that may allow you to avoid those.

Yeah, I think I might trey P3Dv5, load it with a couple addons for NYC and see how it functions... Then I'll decide.

Meanwhile I put my computer back together with the 8600K and ran Prime95. Seems that over the course of 20 minutes of this it averages out at about 81C or so. Core #4 seems to be the hottest running core, reaching a Tmax of 86C. At this point all fans are running full speed. Does this sound 'normal', or should I re-apply the thermal paste onto the CPU? I do think I probably had similar results back when I first built this computer, mind you... However, it doesn't seem I kept any written log, sadly.

Thanks @Rob_Ainscough, good to hear. 3D Mark Time Spy seems to be a free demo at Steam, and then you need to buy it to unlock full options. Is it worth it?

4 minutes ago, Rob_Ainscough said:

No need to buy, just stress testing.

Cheers, Rob.

Little bit confused... Time Spy seems to need 3DMark in order to run, and 3DMark is 30$. Is there a free version somewhere?

5 hours ago, Rob_Ainscough said:

Hmmm ... maybe things have changed (I have the paid version but only because I do other things with it) ... 

https://store.steampowered.com/app/496100/3DMark_Time_Spy_benchmark/

https://www.guru3d.com/files-details/3dmark-download.html

Says free download, is that not accurate?

Cheers, Rob.

You're right! I went to Steam as well as their homepage and didn't see free versions - thanks for pointing me to this. The results are the following:

CPU: 6555, with an average FPS of ~22
GPU: 6022, with average FPS of ~39 and ~34 at graphics tests 1 and 2.

No idea if this is good or not... Would you mind interpreting these data, @Rob_Ainscough?

Okay, thanks @Rob_Ainscough. How many times would you recommend I run this, and what should I be looking for in terms of identifying 'instability'?

Sounds good, thanks!

Meanwhile I went ahead and installed P3Dv5, and carefully introduced new addons one at a time to test performance and associated VRAM usage. I chose NYC, which has always been something of a chokepoint for me. Just looking at the Manhattan skyline would always bring down my performance to the 10-15FPS range. Needless to say that after 'building up' the 'strain' on performance by gradually introducing AIG AI (100%), DD NYC city and all airports and the PMDG 747 I still had a flyable experience. Actually, looking at the city vs KJFK increased performance for me... I was a little shocked, to say the least. My sliders are not even particularly low. At NYC, My VRAM was hovering around 5.5-6GB out of 7.1 available, most of the time, so there is a little headroom available as well.

Mind you, there is still a lot to introduce into the equation, but this first venture was a pleasant surprise. However, while I might just go ahead and test the LA metropolitan area (most addons are already installed so I'll just go ahead and add them to the P3Dv5 library), I will stop at that point and probably remove everything again in anticipation of 5.1.

As some of you suggested, v5 was going to bring some performance boost by itself, and this has been very clear in these short sequence of tests. At the same time, my feeling is that if I were to want an all-around upgrade in both performance and graphical fidelity, i would need to upgrade both CPU and GPU, and I'm not too sure this is something I want to deal with right now. So I'll probably stick with whatever my computer can handle in P3Dv5 for the time being, knowing it's a step above what it was in P3Dv4.

2 hours ago, Benjamin J said:

So I'll probably stick with whatever my computer can handle in P3Dv5 for the time being, knowing it's a step above what it was in P3Dv4.

Try EA enabled with clear skies for some of your flights.. that will give you "improved visuals"

🙂

11 hours ago, Bert Pieke said:

Try EA enabled with clear skies for some of your flights.. that will give you "improved visuals"

🙂

Right, EA actually adds a lot! I love the atmospheric haze, it's a huge improvement to the experience. I've loaded up MSE Southern California, MSE Ultra Res LA, Nuvecta's autogen and FSDT's LAX and can't believe I'm able to fly over LA and land with an FPS averaging out at 20. I know that may not sound like much, but it's a whole lot better than what P3Dv4 use to give me...

EDIT: And then I had a BSoD... WHEA_Uncorrectable_Error, which seems to signify a hardware problem... And particualrly an unstable OC it seems. In my case though, I wonder if this relates to P3Dv5 specifically, or whether something is not quite right from rebuilding my computer. I did a memory test and downloaded an Intel utility to test the CPU, and both passed fine. I have a suspicion that this will come down to the GPU one way or another. Later on I'll do the test Rob suggested and see how that pans out. However, for now I bumped my OC down to 4.7GHz from 4.8GHz, just in case.