29 minutes ago, OzWhitey said:

That's not how anyone who does overclocking defines a stable overclock. Stable overclock means putting the CPU under stress, without adverse effects occurring. Running the overclock using Prime95 or a similar program for a certain period of time would be a common definition.

The term can also be used to mean an overclock that doesn't crash in longer-term real world use.

Then let me formulate the question differently: do you call a "stable OC" something where you partially de-activate parts of your CPU to reach certain clock speeds? Again: why going for an i7 if you want to overclock to levels where HT on leads to heat issues? Or again differently: are heat issues something called "stable OC"?

5 minutes ago, OzWhitey said:

Controversial - why some guys post about this year after year, based on theory, without ever providing a single piece of data.

"For me, there is a measurable difference disabling logical core 1. "

The quote clearly presented claims a usefulness with the AM configured. The single piece of data you asked for.

 

If we are unable to see a difference consider the settings in the sim. How the sim is tested is very complicated, and not for the unwary, VSync on or off, settings levels, traffic, scenery and aircraft - aspects of the sim. Different monitors, all play a part in the outcome of testing. Reliable tests are hard to do.

Those not able to see differences as they expect have most likely some part flat-lining. Come back off all the settings and see what you get. Then see how to bring in various aspects of the sim.

1 minute ago, AnkH said:

Then let me formulate the question differently: do you call a "stable OC" something where you partially de-activate parts of your CPU to reach certain clock speeds? Again: why going for an i7 if you want to overclock to levels where HT on leads to heat issues? Or again differently: are heat issues something called "stable OC"?

Hi Ank, I'm not trying to define "stable OC" myself, it's a common term used on overclocking forums, with a meaning similar to that which I posted above. have a quick,look around some overclocking sites, and you'll,see what I mean. All the best, rob

19 minutes ago, simbol said:

Because you are talking in braille language, non blind people don't know braille code 

Edited: A mistake on last sentence

S.

Hehe, yes and around here be Pilots of Aircraft, and I think a lot of them can fly Blind. 

25 minutes ago, brucets said:

Stephen - I just don't have the confidence. Have read numerous articles over the years, but just not willing to try it. Take a look at this thread I started yesterday ---

Quote

Have to go to the dentist now

, will check back later!! Thanks all!

Bruce

So which is worse, going to the dentist or having chronic overheating problem so can't overclock? - seems like a tossup! 

4 minutes ago, SteveW said:

"For me, there is a measurable difference disabling logical core 1. "

The quote clearly presented claims a usefulness with the AM configured. The single piece of data you asked for.

 

If we are unable to see a difference consider the settings in the sim. How the sim is tested is very complicated, and not for the unwary, VSync on or off, settings levels, traffic, scenery and aircraft - aspects of the sim. Different monitors, all play a part in the outcome of testing. Reliable tests are hard to do.

Those not able to see differences as they expect have most likely some part flat-lining. Come back off all the settings and see what you get. Then see how to bring in various aspects of the sim.

Well, Ive mentioned a few times that what we need is some quantifiable data. And whilst I appreciate the pister's opinion, it's certainly not quantifiable data. Having read a bunch of these threads over the years, some people like the idea, some don't, and I know you're a vocal proponent of the concept.

i checked the LM forum, and didn't find much support for AM changes, certainly nothing that has come from the design team.

this really isn't that hard to test. Benchmark the sim, using an affinity mask as the independant variable. It'll take 10 minutes. Just like my tests of dynamic lighting earlier this evening in the other thread.

theory's always interesting, but what's important is to see if it makes a difference in the sim. I'm not against affinity masks, I just want to see some evidence. If it improves the sim experience, I'm all for it. I just don't get much out of the it works/it doesn't work/it works back and forth that most of these threads become.

i googled affinity mask for the 5820k and just got a thread that I wrote myself from 3 years ago. So remind me - for a 6 core, 12 thread CPU, what's the affinity mask that might improve performance in P3d?

 

 

30 minutes ago, OzWhitey said:

Controversial - why some guys post about this year after year, based on theory, without ever providing a single piece of data.

The mistake I see most is that HT itself cannot improve the speed of the code in the sim.

HT enabled can only accelerate the system - with its hundreds of threads doing I/O, networking, housekeeping, and the threads spawned by the addon exe apps and they might invoke simconnect clients.

The gain from HT enabled is from the underlying performance of the system on the whole. The AM prevents pairing of processes on a core where necessary, HT disabled drops all that from the worry.

Just now, SteveW said:

The mistake I see most I that that HT itself cannot improve the speed of the code in the sim.

HT enabled can only accelerate the system - with it's hundreds of threads doing I/O, networking, housekeeping, and the threads spawned by the addon exe apps and they might invoke simconnect clients.

The gain from HT enabled is from the underlying performance of the system on the whole. The AM prevents pairing of processes on a core where necessary, HT disabled drops all that from the worry.

Yes, i know that's the theory you promote, it's possibly a valid theory. But you must see that lots of experts aren't convinced by it. To the best of my knowledge, the p3d design team haven't suggested this tweak.

as per my previous question, what affinity mask do you suggest for the 5820k? I have an open mind about this issue, the page that google came up with was just me saying that this was the one tweak I'd try when I first got this cpu 3 years ago!

38 minutes ago, SteveW said:

Controversial - why some guys don't get it.

Now this should be stickied!

Greg

1 minute ago, lownslo said:

Now this should be stickied!

Greg

It sounds like you're the only guy around here with some actual data that proves all the experts who DON'T advocate setting an affinity mask wrong. No? No you don't? Oh.

Even if affinity masks help FSX or 32-bit sims - which is a highly controversial claim, as per any of the discussions you can read on the topic (and I noted the last one on the Prepar3d forum was locked by LM) - no one is providing any quantitative data to suggest that this helps in P3D v4.1. Maybe it does, maybe it doesn't, it would be great if a bunch of us could actually test this, rather than the speculation that typically accompanies an affinity mask discussion.

3 minutes ago, OzWhitey said:

Yes, i know that's the theory you promote, it's possibly a valid theory. But you must see that lots of experts aren't convinced by it. To the best of my knowledge, the p3d design team haven't suggested this tweak.

as per my previous question, what affinity mask do you suggest for the 5820k? I have an open mind about this issue, the page that google came up with was just me saying that this was the one tweak I'd try when I first got this cpu 3 years ago!

Turning off the first LP is mentioned in the P3D notes, that would require an AM, HT on or off.

[JOBSCHEDULER]

AffinityMask=14

 

However, it's more of a worry where your addon exe apps go. If they started up on the first core and you used no AM your first P3D process may be contending with those on the first core, and with HT enabled, a threaded out process from the sim would also be sharing.

16 minutes ago, SteveW said:

the threads spawned by the addon exe apps and they might invoke simconnect clients.

The gain from HT enabled is from the underlying performance of the system on the whole. 

That is the key and why I recommend to people to leave HT on, an user with ASCA, + FTX Vector, Pro-ATC X / Voice ATC or RC4, + PTA or Reshade or EVNTEX, + EZDOK or Chase plane, +Flight1 GTN Complete, etc. is having lots of add-on's trying to comunicated with the SIM and these add-on's will see a benefit on Hyper Threading.

We all use lots of tools, etc, and therefor we need to see the bigger picture, is not just about PD3 performance, it is about your whole computer performance and the slowest element in your system  will affect your overall simulation experience.

Best Regards,

Simbol

3 minutes ago, OzWhitey said:

Even if affinity masks help FSX or 32-bit sims - which is a highly controversial claim, as per any of the discussions you can read

Only because it is complicated and with that comes suspicion and doubt. Mostly it is failure to observe the problems I mention that mean some encounter irrational outcome. Mostly the sim is overloaded and flat-lined with settings which gives those poor results.

3 minutes ago, SteveW said:

Turning off the first LP is mentioned in the P3D notes, that would require an AM, HT on or off.

[JOBSCHEDULER]

AffinityMask=14

 

However, it's more of a worry where your addon exe apps go. If they started up on the first core and you used no AM your first P3D process may be contending with those on the first core, and with HT enabled, a threaded out process from the sim would also be sharing.

OK, I'm confused by the second part, I thought the whole point of this discussion was that it was hypothesized that moving P3D off the first logical processor would be good, but then you suggest that maybe this is not good?

At any rate, if you tell me that AffinityMask=14 is right for a 6 core, 12 thread processor  - with hyperthreading ON - I'll give it a try and see what happens. 

As for addon .exes - I'm not going to test with addons, that's a potential confounding factor, they can be added in later.

2 minutes ago, SteveW said:

Only because it is complicated and with that comes suspicion and doubt. Mostly it is failure to observe the problems I mention that mean some encounter irrational outcome. Mostly the sim is overloaded and flat-lined with settings which gives those poor results.

Only because, as in this thread, no one provides any useful data so it becomes a he said-she said kind of discussion. But i'm going to quickly try it out now (its after midnight here now, so it wont be a long test)