Hyper Threading + Process Lasso

[MountainKing]

Hi guys. Process Lasso is being given FREE today at GOTD. www.giveawayoftheday.com.

The author of this software claims that hyperthreading actually hurts CPU and that AMD implementation of same is better.

http://bitsum.com/pl_when_hyperthreading_hurts.php

What are your views?
Is it still worthwhile to use process lasso on a dual core/multi core system? If yes, I might grab it today

 

[TheReaWarmonger]

AMD Threading Model > Intel Hyperthreading

Always. Reason why is like stated in the article, AMD uses 4 extra integer cores as their "hyperthreads". Which is 4 extra full capacity threads. AMD had to market it as a 8 core, because it has 8 integer cores. But the architecture design behind bulldozer, it is meant to compete with Intel's quad cores w/ hyperthreading (hence the shared resources). When comparing AMD processors to Intel, you have to compare module count to core count. Otherwise you end up with skewed results, because Bulldozer isn't a true octa-core architecture. Comparisons can be done like below.

FX-4300 -> i3 3220
FX-6300 -> i5 3570k
FX-8350 -> i7 3770k

As for the software? Think I'll pass on that. Looks like just something else to bog down the system. Dynamic process priority won't change the performance of your machine at all. The only machines this will play a major role on is single core machines. Where all processes are fighting over a single thread. I was going to design software like this to do the same exact purpose because it works. But it only works on computers that are rocking single core processors. Plus it wont make the machine any faster, it will just make it more responsive by raising priorities on new processes temporarily to get them to actually load while the CPU is being bashed. For a multi-core machine, software like this is pointless. It may actually cause system instability by messing with process priority constantly.

 

[grimpr]

Great and useful program, what a gift to all multicore users!

 

[MisterMac]

So only useful for people with Bulldozer ++ or HT intel procs.
And as the article states win7\win8 is increasingly better at matching threads on their hardware counterparts.

In short - for power users (who most likely are the only ones with HT cores AND probably don't use XP?) the gains are minimal and normally you have to pay for it?

Isn't this just a lame clean up my pc type crap?
There are possibly factual gains but they're so bloody low it's probably not worth the price?
(and might not even be consistent as well).

EDIT:

Seems to be free tools like "Mz CPU Accelerator" and others.


How come no reputable small or big - tech website has tried testing this out and reported gains?
All report "BIG POSSIBLE GAINS" yet no one states any numbers what so ever?

That smells fishy to me?

 

[R0H1T]

So only useful for people with Bulldozer ++ or HT intel procs.
And as the article states win7\win8 is increasingly better at matching threads on their hardware counterparts.

In short - for power users (who most likely are the only ones with HT cores AND probably don't use XP?) the gains are minimal and normally you have to pay for it?

Isn't this just a lame clean up my pc type crap?
There are possibly factual gains but they're so bloody low it's probably not worth the price?
(and might not even be consistent as well).

EDIT:

Seems to be free tools like "Mz CPU Accelerator" and others.


How come no reputable small or big - tech website has tried testing this out and reported gains?
All report "BIG POSSIBLE GAINS" yet no one states any numbers what so ever?

That smells fishy to me?

Does softpedia seem reputable enough for you ~
http://www.softpedia.com/reviews/windows/Process-Lasso-Review-211367.shtml

 

[MountainKing]

Great and useful program, what a gift to all multicore users!

Is that some kind of sarcasm?? I'm asking a genuine question. I personally think dual cores can manage quite well on their own but still, you never know.

 

[R0H1T]

Is that some kind of sarcasm?? I'm asking a genuine question. I personally think dual cores can manage quite well on their own but still, you never know.

The program itself is great & if its nicely configured you'll get rid of most of those system freezes & other such problems as it particularly reduces the maximum load on the processor by dynamically adjusting the individual process & I/O priorities at any given point in time !

I'm surprised that for a tech forum like this the ignorance especially on the software side of things is seriously astounding !

 

[grimpr]

Is that some kind of sarcasm?? I'm asking a genuine question. I personally think dual cores can manage quite well on their own but still, you never know.

Nope, its a process automation program that packs multiple tools inside, especially useful for dualcore chips.

 

[MisterMac]

Does softpedia seem reputable enough for you ~
http://www.softpedia.com/reviews/windows/Process-Lasso-Review-211367.shtml

No.

Fluff review with no tests.


How hard would it be to test a Core2, BD, PD, Nehalem and a Sandy?

1 XP Round - 1 Win7 Round.

some games, some compilers, some media stuff and some general "feel"


If its free might as well try it - but it'd be nice with some numbers.

 

[MisterMac]

Nope, its a process automation program that packs multiple tools inside, especially useful for dualcore chips.

And if your on a no CMT\HT quad or dualcore - why would it benefit?

I think i'm misunderstanding how it's supposed to improve on the threading of windows to a degree that's "noticeable".

Bad code is bad code - threading and affinity can't fix that?

 

[R0H1T]

No.

Fluff review with no tests.


How hard would it be to test a Core2, BD, PD, Nehalem and a Sandy?

1 XP Round - 1 Win7 Round.

some games, some compilers, some media stuff and some general "feel"


If its free might as well try it - but it'd be nice with some numbers.

The program is for system responsiveness & last I checked there were no benchmarks for that kind of stuff ! And don't show your ignorance by doing this "no tests" routine because individual process/thread priority is what this is all about ~

The program can adjust any of the five parameters including current priority, base priority, I/O priority, page priority & affinity not to mention throttle the different processes individually ! If you don't know what these parameters do to the individual programs running or the system overall then its best you don't claim what you're doing i.e. the program being a bunch of lies because that in itself is a lie !

 

[MisterMac]

The program is for system responsiveness & last I checked there were no benchmarks for that kind of stuff ! And don't show your ignorance by doing this "no tests" routine because individual process/thread priority is what this is all about ~

The program can adjust any of the five parameters including current priority, base priority, I/O priority, page priority & affinity not to mention throttle the different processes individually ! If you don't know what these parameters do to the individual programs running or the system overall then its best you don't claim what you're doing i.e. the program being a bunch of lies because that in itself is a lie !

Jesus christ - i geuss it'd be hard to expect some nuanced words from certain fanatic people like you.

Your being a pedantic strawman ignoring what i asked for.

Anything you feel can be fairly documented on tick timings - with a various simple programming.


I never question the fact that' adjusting priorities for execution can halt and stall several bottlenecks in the processed pipeline that is the windows kernel - i questioned how exactly it's "automatism" can do it better than than windows itself on a general level.

Which is it's main selling point.


Which i hope microsoft with win8\win7 would do best themselves by now.
I do not understand why a product that would offer more smoothiness - doesn't give a copy to someone who would decently test it.

Run some hardcore resource hogging programs - and add jabbed spikes of starting other processes and timed events - while prioritizing the resource hogger to see and document a definitive better "experience" at the hands of having the lower jabs run more slowly.

While a defintive % value unit would be hard to coin out - the experience itself should be easy for a tester to explaind and document.


PS: Keep your annoying fanboi attitude towards anyone with what you think is affinity towards certain things at home

 

[R0H1T]

Jesus christ - i geuss it'd be hard to expect some nuanced words from certain fanatic people like you.

Your being a pedantic strawman ignoring what i asked for.

Anything you feel can be fairly documented on tick timings - with a various simple programming.


I never question the fact that' adjusting priorities for execution can halt and stall several bottlenecks in the processed pipeline that is the windows kernel - i questioned how exactly it's "automatism" can do it better than than windows itself on a general level.

Which is it's main selling point.


Which i hope microsoft with win8\win7 would do best themselves by now.
I do not understand why a product that would offer more smoothiness - doesn't give a copy to someone who would decently test it.

Run some hardcore resource hogging programs - and add jabbed spikes of starting other processes and timed events - while prioritizing the resource hogger to see and document a definitive better "experience" at the hands of having the lower jabs run more slowly.

While a defintive % value unit would be hard to coin out - the experience itself should be easy for a tester to explaind and document.


PS: Keep your annoying fanboi attitude towards anyone with what you think is affinity towards certain things at home

The OS does not adjust process & individual thread priorities(including I/O) like this program does, the software itself has an option for testing system responsiveness & also an option to do so with multithreading & other specific types of loads. The experience, if you read softpedia's review, is pretty good & if you don't try the program yourself, or believe someone who has been using it for the past couple of years, then you're the one not being objective !

Funny calling someone a fanatic when they can quantify the parameters on which actual performance in windows is based upon, & no hardware isn't the be all end all on a modern OS cause software is what counts for the most, & yet you ignore that one metric because assumedly you haven't tried it yourself

 

[R0H1T]

This is one of the options in the program's feature list ~

Now what the OS does is that it allots the processor to each individual process based on the priority it runs at(default/fixed) & depending on how the scheduler is set up & what this program does is that it can individually change the process priority, it can also adjust the thread & their I/O priorities separately, so do you know of any other program which can adjust these particular settings based on system load ?

 

[TheRyuu]

The author of this software claims that hyperthreading actually hurts CPU and that AMD implementation of same is better.

"Hurts CPU's", because when there's this feature that allows me to make better use of my processor resources I always thought that was a bad thing. They're different implementations, I'm not entirely sure you can directly compare the two.

It's incredibly hard to quantify and benchmark latency and system responsiveness (see the entire BFS scheduler discussion on linux). I fail to see how this software is useful in any way.

From the article:

This logical core can't do much, but it does provide a little increased parallism. It is far from being a real core. In fact, it offers 10-20% (est., likely less) the performance of a real physical core.

Does this person have any clue as to how SMT actually works, see below. On fully threaded workloads it can increase performance by up to ~25%.

Instead of completely disabling HyperThreading, you can use programs like Process Lasso (free) to set default CPU affinities for critical processes, so that their threads never get allocated to logical cores. We call this feature HyperThreaded Core Avoidance. It is better than completely disabling Hyper-Threading because it leaves the rest of the system free to take advantage of this otherwise useful feature.

Not only does windows already do this but there really isn't such a thing as a logical core (at least how they're describing it here and the even more insane description in the previous quote). You're simply seeing two logical cores per cpu core, there is no difference between the two. Whether or not you fully use one or the other doesn't make a difference.

Few people really know just how well the Windows CPU Scheduler handles logical cores, but we think it is safe to say that XP became somewhat aware of them, and they've gradually improved it since then. Again, though, the scheduler is no AI and can not make perfect decisions. It will NEVER be perfect because the OS doesn't have any knowledge of what threads are best to put on these slower logical cores.

"Slower logical cores." It hurts to read this. Also I know exactly how the windows cpu scheduler handles hyperthreading.

So much bullshit it boggles the mind.

 

[MisterMac]

The OS does not adjust process & individual thread priorities(including I/O) like this program does, the software itself has an option for testing system responsiveness & also an option to do so with multithreading & other specific types of loads. The experience, if you read softpedia's review, is pretty good & if you don't try the program yourself, or believe someone who has been using it for the past couple of years, then you're the one not being objective !

Funny calling someone a fanatic when they can quantify the parameters on which actual performance in windows is based upon, & no hardware isn't the be all end all on a modern OS cause software is what counts for the most, & yet you ignore that one metric because assumedly you haven't tried it yourself

Again - useless whining and not deflecting what i want to point out.

I will try the program - i've said that.
Ill report back with my experiences.


So... the application can test system responsiveness - but you can't quantify a metric in how it determines what to "boost affinity" to? - nor can you give a value\unit metric on how this application chooses to make it better?

Do you realize how stupid you sound (And possible are?)
There are tons of stages and handlers within our modern computer eco system. Tons of priorities, and ports that shuttle data around.
I don't really need to hear your spurt a bunch of BS about they're naming and roles to the execution of the system\OS.

..I need to hear some firsthand experience how it tangibly changes this - AND does it better on NON CMT\HT chips and those with.

I need to know how the hell it can presume to do a better job of this - than windows itself.
(Microsoft's history aside - they've bettered themselves in the technical execution last decade).


In order for this application to decide how each application needs to be assigned resources - it needs data to make decisions.

How the fuck does it get this data? - and how does it do this without a massive cloud DB that rates each program in conjunction or atleast defines types of programs?

My subjective speculation is - it seems for me impossible that someone without god knows how much information on the kernel and MANY years of testing\progrmaming could give a general tangible performance improvement (atleast on newer OS').

Can it somehow undo memory locking and thread waiting from badly programmed applications ?

Doubtful - but i geuss i'll find out in the next few days.

 

[R0H1T]

So... the application can test system responsiveness - but you can't quantify a metric in how it determines what to "boost affinity" to? - nor can you give a value\unit metric on how this application chooses to make it better?

The system doesn't hang momentarily ? How does that sound for a metric, like how aida.exe will be launched with a "high" default priority but process lasso can lower it to idle depending on the system load, also it changes affinity(depending on the option being set like that) not boost it provided you know what that means

I need to know how the hell it can presume to do a better job of this - than windows itself.
(Microsoft's history aside - they've bettered themselves in the technical execution last decade).

Because the scheduler doesn't assign process priority nor does it change it dynamically, don't you know this stuff !

 

[jkauff]

I use Process Lasso, and one of the nice things I can do with it is limit Handbrake to three cores so I can still get some work done while encoding videos.

 

[SPBHM]

I use Process Lasso, and one of the nice things I can do with it is limit Handbrake to three cores so I can still get some work done while encoding videos.

you don't need a third party software for that,
you can use the task manager..

 

[VirtualLarry]

I need to know how the hell it can presume to do a better job of this - than windows itself.
(Microsoft's history aside - they've bettered themselves in the technical execution last decade).

MS's kernel scheduler is so good... that they had to severely modify it for Windows 7 (and Vista?), by adding the multimedia scheduler, so that playing back a video and running at a high level of network utilization (high interrupt load) wouldn't bog down playback of audio/video files. Unfortunately, it can also limit network transfers to 100Mbit speeds, even on a GigE network.

 

[jkauff]

you don't need a third party software for that,
you can use the task manager..

True, but task manager doesn't save my settings.

 

[SPBHM]

True, but task manager doesn't save my settings.

you can use the shortcut for that

 

[Idontcare]

MS's kernel scheduler is so good... that they had to severely modify it for Windows 7 (and Vista?), by adding the multimedia scheduler, so that playing back a video and running at a high level of network utilization (high interrupt load) wouldn't bog down playback of audio/video files. Unfortunately, it can also limit network transfers to 100Mbit speeds, even on a GigE network.

I particularly enjoy how my 4C/8T 5GHz 2600K processor just hangs and is completely inoperable every time I open windows explorer and use it to navigate to a networked computer with network discovery enabled.

CPU utilization for the effort? Pretty damn close to zero.

Threads required for the effort? Pretty much just one, leaving 7 other processor threads available for multi-tasking.

Computer resources available while "discovering" my other computers on the router? Zero, totally hangs and is inoperable for tens of seconds, up to a minute on occasion, at a time.

But there you have it, only Microsoft can figure out a way to make a silly fast computer be completely bottlenecked by a 1Gbit Ethernet connection

 

[kbown]

I am a "power user" as I work night and day with huge Photoshop files, at the same time as working with texts, emails, sometimes watching videos, etc. Sometimes I end up working on 12GB+ files.
I maxed out my RAM to the most the machine can fit, which is 16GB. I still get to points that i am filling up the system with my work.
For me Process Lasso is the thing that makes Windows better than for example a Mac. I couldn't live without it. Before I used it, if the processes on the computer got to heavy Windows could freeze. With Process Lasso it doesn't, ever, as if a process is in danger of "taking over" the system, it just slows it down so the OS can run normally. It has saved me hours and hours of time and frustration. I cannot but recommend it, it needs little adjusting to run perfectly. It is also possible within it to immediately terminate process which are having troubles and need to be restarted. I would say aside from Photoshop, Process Lasso is the single most useful program I have ever found.

necros on old threads are not allowed except under special conditions
Markfw900

 

[cytg111]

Damn, these bots are getting good.
(hope it is not the dreaded ms-neonazi-chick)