Disablin HT helped?

[bangtail 0]

I tried the -cpucount=x switch (several times) and my system never stabilized until I disabled HT while running ArmA 2. The HT setting and current drivers (191.07) have been the fix for me.

Since I reboot before/after an ArmA session anyway (process restoral) it's no problem to load a BIOS profile appropriate for what I'm doing. HT turns on as easily as it turns off.

Definitely try the -cpucount=x switch as this is the simplest solution if it works. If not run it w/o HT (applies to i7 and most i5 platforms only).

Your mileage may vary.

.

.

Yep, sadly it doesn't seem to work for everyone which is a shame. It's worked for me on 3 different MBs and it worked for 2 of the guys I game with. As long as disabling HT via the BIOS doesn't interfere with your enjoyment, no worries.

:)

Eth

[Bulldogs 10]

Basically the way it works is that when you have HT on you have 4 physical cores and 4 virtual cores. If you use -cpucount=4 then Arma 2 should only address the physical cores. But that's not always the case. If something in the background accesses more than 1 core (it's not highly likely but can happen in some setups) while loading Arma 2 then Arma 2 detect the physical cores in use and uses the virtual cores as priority instead.

It doesn't happen often but can happen, and if it does then you'll get cache thrashing, which isn't likely to damage your CPU but it will effect performance.

If you're not too worried then leave HT on, if you get bad performance then disabling HT for testing purposes can't hurt.

[IrishCoffe 10]

Yep, sadly it doesn't seem to work for everyone which is a shame. It's worked for me on 3 different MBs and it worked for 2 of the guys I game with. As long as disabling HT via the BIOS doesn't interfere with your enjoyment, no worries.

:)

Eth

As part of Nehaelm microarchitecture, SMT employ aggresive techiques to exploit the instruction-level parallelism to keep active the processing units. It is widely uncommon things to report:

unless both's of uor processor have already been subject to some rapid voltage variation and/or a wrong microcode implementation.

[kklownboy 43]

Never had any issues (except A2) and I use the cores for editing etc so disabling them is not an option.

As for heat, maybe 4 or 5 degrees but nothing to worry about.

Secondly, your statement about HT and games is incorrect. GTA 4 is 50% faster with HT enabled and FC2 29% faster with HT enabled. Conversely UT3 is slower as is WIC but it totally disproves your claim that it doesn't help in "any game, ever".

..

Oh boy... So you infer from that testing/Part4, that "One Core with HT OFF" to "Two Cores with HT OFF" gives a 142% gain..... Well there isn't any HT on...:p

The "One Core HT OFF" to the "One Core HT ON" is 50% gain...:rolleyes:. Now go back to Part three of there testing HT and you will see that GT4 with 4cores and HT on does squat... I rest my case:cool: But yeah never say say never... but it is never right now.:D

PS look at the numbers, 66fps for 4coreHT-on ,and 58fps for 2coreHT-off...

So to recap: 4cores+HT=8"cores" =66fps 2cores-HT=2cores =58fps 1core+HT=2"cores" =36fps 1core-HT=1core =24fps.

Heck by them numbers id say GT4 is really a DualCore game, since the 66fps was with turbo on!.

Edited October 27, 2009 by kklownboy

[bangtail 0]

Oh boy... So you infer from that testing/Part4, that "One Core with HT OFF" to "Two Cores with HT OFF" gives a 142% gain..... Well there isn't any HT on...:p

The "One Core HT OFF" to the "One Core HT ON" is 50% gain...:rolleyes:. Now go back to Part three of there testing HT and you will see that GT4 with 4cores and HT on does squat... I rest my case:cool: But yeah never say say never... but it is never right now.:D

PS look at the numbers, 66fps for 4coreHT-on ,and 58fps for 2coreHT-off...

So to recap: 4cores+HT=8"cores" =66fps 2cores-HT=2cores =58fps 1core+HT=2"cores" =36fps 1core-HT=1core =24fps.

Heck by them numbers id say GT4 is really a DualCore game, since the 66fps was with turbo on!.

I inferred nothing. I simply posted an article that disputes your claim (quoted below). All those assumptions (^^) are YOUR inferences, not mine.

You can rest your case if you like but I suspect that you don't see a lot of gain with all the cores enabled because the game just can't make use of all the CPU power available to it. It's fairly clear that the HT enabled core is running GTA 4 significantly faster than the non HT enabled core. As I am sure we both agree, very few games use 4 cores efficiently, let alone 8.

You will never use the extra"cores" in any regular window app, and never in anygame. ever.

In part 3 of the article, 3 of the games are still showing gains with HT on (sure, they aren't that big, but they are still gains), so again, "never" is not right now.

According to the article, HT gives you a ~10% increase as an average. While it's not earth shattering, it's significant. Some software (Maya, 3DS and Lightwave) yield up to a ~30% gain (rendering) which is very significant.

I'll concede that if all you are doing is gaming then turning HT off via the BIOS probably won't make a huge difference. By comparison, leaving it on won't cause your PC to spontaneously combust either. I know this for a fact because I have HT on across the board and I play A2 for at least 2 hours a day on average (and sometimes for significantly longer).

Another article quoting an average of ~30% increases and acknowledging a bottleneck when using all cores.

http://www.overclock.net/intel-cpus/564056-i7-ht-vs-off-benchmarks.html

See ya,

Eth

PS : We really are MILES off topic here. I think we reached a consensus and that was :

Try the -cpucount switch and if that doesn't help, disable HT via the BIOS.

I have no wish to degenerate this thread any further.

Edited October 28, 2009 by BangTail

[IrishCoffe 10]

---------- Post added at 08:03 AM ---------- Previous post was at 08:01 AM ----------

Just go away already with your fantasies.

It's not funny anymore.

Eth

As part of Nehaelm microarchitecture, SMT employ aggresive techiques to exploit the instruction-level parallelism to keep active the processing units. It is widely uncommon things to report:

unless both's of uor processor have already been subject to some rapid voltage variation and/or a wrong microcode implementation.

Resource:

http://download.intel.com/design/processor/applnots/320354.pdf

ftp://ftp.cs.wisc.edu/sohi/papers/1996/micro.trace-cache.pdf

http://en.wikipedia.org/wiki/CPU_cache

http://en.wikipedia.org/wiki/Overshoot_%28signal%29

http://en.wikipedia.org/wiki/Microcode

Edited October 27, 2009 by IrishCoffe

[bangtail 0]

Resource:

http://download.intel.com/design/processor/applnots/320354.pdf

ftp://ftp.cs.wisc.edu/sohi/papers/1996/micro.trace-cache.pdf

http://en.wikipedia.org/wiki/CPU_cache

http://en.wikipedia.org/wiki/Overshoot_%28signal%29

http://en.wikipedia.org/wiki/Microcode

Yes, we can all "copy and paste" links to things even though we don't have a clue what we are talking about.

Look, I'm an expert on subatomic particles (I must be, right, because I've copied and pasted a whole bunch of links) :

http://www.ndt-ed.org/EducationResources/HighSchool/Radiography/subatomicparticles.htm

http://science.howstuffworks.com/atom-smasher9.htm

http://www.britannica.com/EBchecked/topic/570533/subatomic-particle

http://misterguch.brinkster.net/subatomic%20particles.html

Since you can't substantiate your initial fear mongering claims about processors being destroyed by A2 etc and because you insist on trying to further derail this thread with the endless linking of white papers etc that you quite clearly don't understand, I'm afraid it's "ignore" time for you.

Bye

Eth

Edited October 27, 2009 by BangTail

[IrishCoffe 10]

Yes, we can all "copy and paste" links to things even though we don't have a clue what we are talking about.

Look, I'm an expert on subatomic particles (I must be, right, because I've copied and pasted a whole bunch of links) :

http://www.ndt-ed.org/EducationResources/HighSchool/Radiography/subatomicparticles.htm

http://science.howstuffworks.com/atom-smasher9.htm

http://www.britannica.com/EBchecked/topic/570533/subatomic-particle

http://misterguch.brinkster.net/subatomic%20particles.html

Since you can't substantiate your initial fear mongering claims about processors being destroyed by A2 etc and because you insist on trying to further derail this thread with the endless linking of white papers etc that you quite clearly don't understand, I'm afraid it's "ignore" time for you.

Bye

Eth

Mine resource are in line with the statement.

All uor "suggestions" ( not only in this thread ) are just non-sense.

Edited October 27, 2009 by IrishCoffe

[Bulldogs 10]

GTA4 actually uses virtual cores if the one of the physical cores is being stressed above 50% by another application same as Arma 2 (from personal testing on a couple of machines I have noticed this while troubleshooting some performance issues that turned out to be cache thrashing)

[W0lle 1047]

Guys can we stay on topic here? It must be possible that people here can discuss without bashing each other on disagreement. ;)

[Barso 10]

Thanks for the help Ethne, you have always been a great resource when I needed info for arma2.

You seem to be getting a hard time but I value what you have written.

I am going to disable HT in bios as I am a gamer through and through.

Can you tell me if having C1E enable in bios affects my core i7 920 OC'ed to 3.4GHz?

I ask as I noticed someone saying that the voltage changes can affect cpu stability.

I thought I might be saving the earth by saving on voltages being used but my cpu comes first, sorry eco-warriors;)

I will be OC'ing to 3.8GHz soon as I did a stability test and my max temps were only 50C although I have my cpu watercooled and also my X58 classified MB is watercooled too.

Thanks for all the input everyone but I love arma2 and love it even more after playing operation flashpoint: DR.

Please BI, never compromise the arma series by moving to consoles.

[bangtail 0]

Thanks for the help Ethne, you have always been a great resource when I needed info for arma2.

You seem to be getting a hard time but I value what you have written.

I am going to disable HT in bios as I am a gamer through and through.

Can you tell me if having C1E enable in bios affects my core i7 920 OC'ed to 3.4GHz?

I ask as I noticed someone saying that the voltage changes can affect cpu stability.

I thought I might be saving the earth by saving on voltages being used but my cpu comes first, sorry eco-warriors;)

I will be OC'ing to 3.8GHz soon as I did a stability test and my max temps were only 50C although I have my cpu watercooled and also my X58 classified MB is watercooled too.

Thanks for all the input everyone but I love arma2 and love it even more after playing operation flashpoint: DR.

Please BI, never compromise the arma series by moving to consoles.

When I OC (which is rarely), I disable EIST, C1E, C3/C6/C7 (saying that, I actually almost always have these off). Some people maintain that it is safe to leave them on and they may well be right, but because my cooling is rock solid, it's not a consideration. The voltages and clocks DO jump around when these features are enabled and obviously if you are over volting, this could be problematic in terms of stability.

Since your CPU is water cooled, it probably isn't a consideration for you either (unless you are worried about a few extra $$ on your utility bill). As always, watch your initial temps to make sure everything is nominal and run Prime for 24 hours to make sure that you are as stable as you can be.

Eth

Edited October 28, 2009 by BangTail