cool source forge app

[ghettoside]

I've tried this freebie app that is supposed to speed up your apps.

I have it running with no problems on 2 boxes. It works pretty darn good, so I thought I'd post it. There is warning that it can cause some systems to hang- but I've had no problems, only benefits.

I'm only folding on one box. Well, after installing this app, the large WU I'm working on, the scheduled finish was April 15...as soon as I installed this app and resumed F@H, (I closed all running apps before install) the WU is scheduled to finish April 9. 6 days off the WU completion! Not bad...

On my other box, my apps are running a bit faster. I'm really pleased with this one.

7MAX info (http://www.7-max.com/)

7MAX download (http://www.7-max.com/download.html)

I'm using ver 4.01.

It was created by the developer of 7-ZIP (http://www.7-zip.org/)...which is working great for me, I do a ton of archiving.

[YARDofSTUF]

Cool, I'll give it a try on a test system.


EDIT: Looks like it just adds /3GB /NOPAE switches, then reserves some ram for 7max

[YARDofSTUF]

Well I loaded F@H into the 7max thing and so far all it did was cause an unkown gromacs error and delete my current wu.

[ghettoside]

Well I loaded F@H into the 7max thing and so far all it did was cause an unkown gromacs error and delete my current wu.

holy crap! :eek:

I have all older boxes...it's working ok on mine.

are you running console F@H? if so, need to use 7-MAXC. it's in the help file.

I should've also posted to read the help file...I seem to have operator error syndrome today.

Sorry to hear that happened to you YoS, I haven't had any probs myself.

Now I just went in and double checked my 7-MAX config...hmmm...it's showing "0M" on the apps I've attached 7Max to. Odd. I swear everything appears to be running faster. On my folding rig, same thing. Something definately sped up my folding. And i also had operator error there too...the April 15 and April9 dates- that's just for current frame end. WU scheduled to complete-get this- Oct 2! OMG.

I'll poke around on the 7MAX forum, need to do more research on this app, prolly jumped the gun a little...posted this too soon I think now. But I haven't had any probs...I absolutley wouldn't have posted this if I'd had any probs.

:o

I'll post anything useful I can find on this app.

really sorry that happened to you YoS. :(

[ghettoside]

I honestly thought this app to be useful, it's working ok for me or so it appears.

I hate when this kind of stuff happens.

:o

[YARDofSTUF]

I'm using 7max and its running fah, has 0 megs used too beacuse the memory usage is so low.

Its no biggie, the system I have it on is basically a tester/folder only.

[YARDofSTUF]

Leaving the 7max window open draws 2% cpu usage almost constant. 7max's help file says that windows xp does a bad job of managing memory? The guy writting the help file should talk when he has an unstable app that plays with it lol

[ghettoside]

Leaving the 7max window open draws 2% cpu usage almost constant. 7max's help file says that windows xp does a bad job of managing memory? The guy writting the help file should talk when he has an unstable app that plays with it lol

LOL.

This app was supposed to speed up folding- not delete the WU's.:irate:

So are you still playing with 7Max? if so I'd really appreciate it if you'd keep me informed here. I have another folding friend waiting to hear your outcome. But it sure doesn't look good. I'm using the recommended 25% of my RAM for 7MAX.

Thanks for not getting mad at me bro, I really meant to be helpful with this post. I just about crapped in my pants when I saw your post.:eek:
makes me wanna not ever post anything like this again.

I haven't been too well this weekend, if you noticed I haven't been on the boards until this morning. I dunno, I should've tested this longer and checked into it a bit more- just cuz I have no issues w/ it didn't mean anyone else wouldn't. I've been running it since weds.

i know i did post that there was a warning about possible sytem hang, but that doesn't make me feel any better about what happened to you. Damn.

I swear something sped up my folding, and it happened right after installing 7MAX-and my other box seems to be running better. :confused: ??? maybe I'm just losing my marbles. prolly am.

I will do some checking into it and post whatever I find. I have a case of barfing that's come on this afternoon, just got back from the store to get Pepto Bismol. taking a few different meds today and they don't seem to agree with me. (so what's new lol)

[YARDofSTUF]

Ya I'm using there recommended setup, try some others later. Need to snag a memory heavy WU.

[ghettoside]

I pretty much do large WU's all the time. They take at least a week for me, sometimes more.

I'm guessing? the 7Max helped by speeding up my a/v and anti-spyware apps?

Starting to wonder if this is one of those useless apps. I expect that since my boxes are all older, and 7Max is supposed to give the biggest increase on these type of boxes, I will see more improvement than you guys w/ newer equipment.

I just looked at my WU again...it went from like i think it was 28 mins per frame (that would be about right for that Oct 2 date) to now it's 57 secs a frame.
and showing april 15 for WU end. it's done a few hundred frames since i checked a couple hours ago. I'm already getting confused, gotta start writing this info down as I test this out.

it sure would be nice if this app does speed up the folding.

thanks for staying w/ this thread YoS.

[ghettoside]

YoS,
Are you using the console ver of F@H? are you using any tweaks to F@H?

I want to post in the 7Max forum, the developer is pretty good about making prompt responses. So i'd like as much info as possible, plus I have to look at F@H forum too and gather some info.

There are some postings about changing the min block size that helped in some cases.

Did find something important: May cause XP SP1 to crash! The /3Gb switch is a prob in SP1. MS KB ink (http://support.microsoft.com/?kbid=328269)

definately didn't do my homework on this one...a lesson learned when recommending an app.

from the 7Max forum, some users also claim to have gotten increases in some games- but developer says 7Max only helps apps that use physical RAM...

[YARDofSTUF]

Ya I use F@H console.

[ghettoside]

Ya I use F@H console.

Thanks YoS.

I asked about the tweaks cuz my friend is hard core folding and has done some tweaks on all her boxes for folding. Running Beta 5.04 console version on all boxes, using EM111 to track all projects, boxes programmed for best performance, all screen savers shut down, things which eat cpu usage and memory. She won't be a happy camper if she puts 7Max on and it crashes F@H. And then of course if there is a way to get more out of folding by using 7Max, anyone else can do that too. I'm giving this one a top priority for this week. hopefully my body will co-operate.

I'll gather info, do research, than post something here as soon as I have something.

Please lemme know here if anything else happens in the meanwhile. Thanks.

[ghettoside]

7maxc

command line to be used for running console apps with 7MAX

in case anybody else is brave like YoS...

and read the help file. I'm working on gathering info.

[YARDofSTUF]

Where do u see this 7maxc thing I dont see mention of it.

[ghettoside]

Where do u see this 7maxc thing I dont see mention of it.

YoS,

In the 7MAX help file, open the General Information section (under Contents tab) scroll down...it's just one tiny bit of info, and it's nowhere on the 7MAX site or on Source Forge page.

Using command line. Type in command line:
7max program_name [parameters]
For console applications you can use 7maxc instead.

The 7MAX site could really use a little work...

There's a few interesting things in the 7MAX forum that should be on the main site...take this console app thing for example. It's not even in the forum. So I expect we are the first ones using it for a console app.

The developer, Igor Pavlov, is a Russian dude I believe. I've actually "spoke' with him in email a few times, he's pretty cool, and he is prompt to respond to email and the forum.

7-ZIP (http://www.7-zip.org/) has seen much more exposure than this app, it's newer, there's a wiki (http://en.wikipedia.org/wiki/7-Zip) on it, and I think more people have more use for the archiver 7-ZIP and prolly contibuted to the 7-ZIP site...there's a whole page of 7-ZIP logos (http://www.7-zip.org/logos.html) contributed by users from across the globe.

from what I'm seeing on the 7MAX forum (http://sourceforge.net/forum/forum.php?forum_id=399008), I suspect Pavlov isn't getting enough feedback on this app. How much can one man test his own app? And without much feedback, that would seem to indicate a lack of interest by users, so that may also have an effect on how much time he spends developing this app.

If this app works to speed up Folding, either by speeding up the background services and freeing up resources for F@H, or speeding F@H itself, hell, I'll volunteer to add some info to his page.

I'm still not sure why I saw a gain in F@H ...but my Folding is so slow I'd figured what have I got to lose by trying it. And it is designed to give the biggest gains to older boxes like mine.

I'll try to put some kind of definitive post on 7MAX here today. I can shoot Igor an email but I think he's 9 or 10 hours ahead of us.

This app is primarily designed for apps that use large amounts of physical RAM...again I'm not sure why I saw an increase. Maybe there is something else he can do to it...he may have some suggestions for adjustments to the settings that could be useful to us. I know we talked about the 0Mb thing yesterday... I'm hoping there's an adjustment to the settings that can be used for us.

Pavlov wrote this in a reply on the forum:
I can write about purposes of 7-max.
I write 7-zip compression program. And some time ago I discovered that memory access is slower than it can be. It's because windows uses 2-levels mode inside CPU for virtual memory management. But all modern CPUs allow 1-level mode for virtual memory management. It's slightly more complex for system (windows) to use such mode, since it requires some more smart algorithms. But there is some speed gain for some programs. So I rewrite that code to use 1-level mode (4 MB pages).
CPUs that have advantage in sorted order:
Duron (most advantage)
Celeron
Pentium 4
Athlon
Athlon-64 (less advantage).
When L2 cache is small advantage is larger.
If Intel will release CPU without L2, 7max will give big gain for it.

2) If it says that it is still occupying 0mb of 7-max memory, there will no any gain.

To get gain:
1) start 7-max,
2) start Application with 7-max.
3) Check that it allocates memory via 7-max.
4) close 7-max to reduce overhead of 7-max program itself.
5) use Application.

It's better if you will not use hibernation and stand-by with 7-max. Windows does not know about memory part allocated to 7-max.

So I'm still trying to figure out why I'm seeing a gain in F@H.

I thought it is interesting that Vista/2003/64-bit provides the API for 7MAX, but he's creating the API for use in Win 2K and XP SP2.

7MAX has been successfully used to get a 5% gain in encoding mpeg2 files...and that particular user mentioned testing it on Virtual Dub and xvid but then posted nothing further.

Some gamers found a few tweaks to the settings, forum (http://sourceforge.net/forum/forum.php?thread_id=1129452&forum_id=399008)
Socket A
7max Version 4.01
Windows SP2
512MB, using 128MB for 7max

Options:
Use MTRR
Minimum Block size: 128KB
Use fast working method



I'm looking for alternative progs to speed up F@H too...

[YARDofSTUF]

Ya I'll look it over pretty good. The WU that started after the failed one is on 60, and I want to see it finish, then i'll try tweaking it a bit and testing what it can do.

The idea shows potenial, but theres not much to speeding apps up with software.

If this app really does work, the gain you are, or feel you are seeing is related to the WUs with a large memory draw.

I think that 6 FPS post is BS though.

[ghettoside]

Ya I'll look it over pretty good. The WU that started after the failed one is on 60, and I want to see it finish, then i'll try tweaking it a bit and testing what it can do.

The idea shows potenial, but theres not much to speeding apps up with software.

If this app really does work, the gain you are, or feel you are seeing is related to the WUs with a large memory draw.

I think that 6 FPS post is BS though.

Thanks for your input on this YoS.

yeah, I'm just not sure what to make of the increase in F@H that I think I'm seeing. it is a large WU, but I've only done 6 WU's so far and haven't really paid that much attention to the speed of F@H so no data for me to make a conclusion. When I installed F@H, I had just kinda accepted that the box is an old piece of crap from a dumpster, and whatever I get, that's all I'm gonna get out of F@H. Just started looking last week for something to try and tweak a little more out of F@H, found this app and I thought I had something. I really could swear it is speeding up both PC's, both are dogs...I had too many probs with the other box, my main rig, to keep F@H running on it. I've done a bunch of tweaks to it to run my apps, but junk is junk. I still have that 17" box fan and the case open to control that overheating prob. I have a mobo coming from rivas and I can hardly wait, then will turn this main into a dedicated folder along with the current one, and fold on the new one as well.

That 6 FPS? I know absolutely nothing about gaming, never been into it. but I'm sure you're right!:D

[ghettoside]

YoS,
Damn I went to sleep again, just got up.

I've got a few F@H helper apps to try out, and I'm holding off on sending email to Pavlov...want to get as much info as possible. I'm going to try a few open source apps i found, just grasping at straws I think, but maybe something open source can be modified (part of reason holding off on email).

F@H core is now using 4Mb of 7MAX memory! That's an improvement, it was showing 0MB yesterday. Don't see any further change in speed, but don't have any benchmarking apps installed, on my list of to do tonite.

I tried multiple clients when I first started folding...it was dogging them all though. I'm going try an F@H manager app tonite, install 8 service clients, use with 7MAX, and see what happens. I'll try it on my main rig, not doing anything with it anyway. If it gets hosed, oh well. I moving all my data to the other hdd just in case.

Did you try that 7maxc command? I'm gonna try it tonite.

[ghettoside]

email sent to developer of 7-MAX.

I'll post info I get from him, prolly have an email reply in the morning if he is still like he used to be.

adding benchmarking tools tonite...

[ghettoside]

email sent to developer of 7-MAX.

I'll post info I get from him, prolly have an email reply in the morning if he is still like he used to be.

adding benchmarking tools tonite...

received a reply that addresses the first of 7 questions...I gave him an overview of what we're trying to do with F@H, what F@H is, & what's happened so far.

in pertinent part:

7-max can give gain only if
1) Program uses big memory amount.
2) Pattern of memory access is random.

So, at first check that you can get any gain with 7-max.
It's better to use 64-bit windows (or 32-bit Vista) + 7-max v5.00 (it's
more stable version).


That doesn't sound too hopeful. My current WU is using 4Mb of 7-MAX, so it's something, just how much that something is remains to be seen.

and in direct response to my first question: (asked for more info on usage of 7maxc)

"7maxc fh.exe ..."

which did nothing for me but give me the 7-max version and copyright, etc...
Since 7maxc at the command prompt returns the same, I suspect that fh.exe was meant to be for F@H... basically enter the console app exe after 7maxc. No switches. I hate interpreting an email like this, but I wanna give it a little more time to see if more reply will come from him.

lol.

I had to use cd command at the command prompt to set 7-max directory, 7maxc was an unrecognized command...

I'm running the graphical client still. Didn't install on my other box yet since I'm still looking at stuff and haven't been able to decide what to try first, none of it looks like what I want.

I paused my WU, stopped, then restarted it using the run with 7-max option. Well, I got 28 mins per frame again, but when the next frame started it went to 56 secs a frame again. I will say that I had been stuck at 28 mins a frame before installing 7-MAX. Just coincidence?

I dunno if I'll get more response from Pavlov or not, I noticed on the 7-max forum he seems to not want much to do with the version for win2K and XP as that version is unstable.

I thought it is interesting that Vista/2003/64-bit provides the API for 7MAX, but he's creating the API for use in Win 2K and XP SP2.

I just didn't elaborate on it there. It's detailed in the 7-MAX forum.

I'll see if anything else comes thru from him, it was around 11am his time when he sent his reply. He did address only the first question, so maybe he's going to answer piecemeal. I hope...

:(

[YARDofSTUF]

I think the best thing its doing is fooling the program that is estimating your coutner, does it actually finish in 56 secs?

I found that 7maxc does nothing, it doesnt even use teh memory that 7max allocates to itself.

[ghettoside]

I think the best thing its doing is fooling the program that is estimating your coutner, does it actually finish in 56 secs?

I found that 7maxc does nothing, it doesnt even use teh memory that 7max allocates to itself.

the frames are finishing in the ballpark of 56 secs. I checked. it appears to be in the range of -10 to +5, over the last 10 mins.

but that doesn't prove to me that it's 7MAX doing it. Maybe it was just the restart after installing 7MAX...and maybe the increase from 28 mins after the first frame completion to 56 secs would occur anyway. I'm gonna let this WU finish out as is, it'll be done Saturday. Initially, the frame completion of April 15 and Oct 2 WU end, that was running before I installed 7MAX, that was real, no mistake there. I'd like to start 2 instances of F@H at the same time, one w/ 7MAX and 1 w/o....but they'd have to both be of large variety and the same group.

I read up on the /MAXMEM /3GB /NOPAE switches added to the boot.ini. link (http://www.sysinternals.com/Information/bootini.html)

hmmm....

I don't beleive those switches are doing much for me.

when I run some of my apps w/ 7MAX, I dunno....I see photoshop cs using 4Mb sys phys, and 8 MB 7MAX (w/o opening a jpg for edit); word uses 28MB sys phys and 4 Mb 7MAX then creating a blank doc and saving it made the sys phys value 31 MB and n/c to 7MAX; pretty similar pattern among all my apps that I tested w/ 7MAX. and all are using huge swap files.

They seem to be running better...maybe that's just wishful thinking. Having no benchmarks prior to install of 7MAX, can't know anything for sure.

I dunno, looking more and more like one of those useless apps...thought I had something at first.

I'm looking for that mobo any day now, so I can look at this better then. But it does not look promising anymore. prolly just no subsitute for folding power.

I'll play w/ it a little more after the new mobo arrives and post. If that developer has anything else to say I'll post that too. I'm leaving both boxes to run under the 7MAX system for now, it doesn't seem to be hurting anything.

The program was written for memory intensive apps, so if does something, it'll prolly be noticeable then, like when archiving.

[YARDofSTUF]

Ok ur saying its not doing much, but then your saying that instaed of 28 mins a frame on F@H, now it only takes 56 secs? or are you saying 56 secs less then 28 mins?

[ghettoside]

Ok ur saying its not doing much, but then your saying that instaed of 28 mins a frame on F@H, now it only takes 56 secs? or are you saying 56 secs less then 28 mins?

lol YoS, I'm just so disgusted already...

Frames are taking 56 secs to complete- some are 46 secs, the fastest; some are 1"01'. In that range. 56 secs -10 to +5. Sorry that was unclear. That was over a 10 minute period. they were taking 28 mins per frame before I started using 7MAX.

I'm getting disgusted cuz I don't see how those switches are doing squat for me...maybe I just don't understand (as if that never happens :rolleyes: )

For one thing, I don't have nowhere near 1 Gb of RAM :rotfl: let alone 2 or 3 Gb of RAM....
If I did I prolly wouldn't have half the probs I do

I dunno, wth did my F@H speed up?

I've looked at prolly a couple thousand web pages over the past day...looking for something that prolly doesn't exist and isn't possible.

I suppose I should just stop that WU, restart into the xp sp2 partition that I normally run F@H on, restart that WU and see what happens...see if the 28 mins a frame is what I still get, and if it stays that way after it completes a few frames. That is what I should do already.

I've been researching something else for a friend that's kinda an urgency too, can't find what I need, and I'm giving myself another migraine on top of some nice frustration.

So when I made that last post, I was disappointed & disgusted in general, having a lot of doubt about this 7MAX, after reading about those switches, reading the reply I got from Pavlov (which is all I've gotten out of him) and reading your post (I trust your opinion and I know you're spending some of your time testing this too)

I so want this app to be the cause of speeding up my F@H.... it would be great if it does work to speed the folding. Everybody could benefit and the Team could rock. But I'm having my doubts. Something sped up the processing on the WU- or could it be that something had slowed down the processing...to the 28 mins/frame.

Excuse the babble, I just got out of bed again.

I just got woke up by a call from my bud, wants me to work a few days on the truck (knows I'm just loafing around and need $) so I'll have to be ready to work at 4:30 am tomorrow. I've kept all sorts of goofy hours since Jan 1, lots of no sleep for 2 days...so this should be interesting.

Anyway, I have to run to the store, need a few things like lunches, etc... then after I rest a bit I'll shut it down and restart w/o 7MAX system and see what happens. Which could mean hours before I know anything, if the frames are back to 28 mins each. I do not think I have done one of these particular WU's before, I have done all large but never saw anything like months to completion (Mark posted something on these WU's in the folding thread in general disc.) The large have been taking me like a week to complete. I'll see if anything shows what WU's I've done in the F@H certificates too when I get back.

[YARDofSTUF]

Whats your folding name and how many wus have you folded in a 24 hour period then?

Also what CPU?

[ghettoside]

I found that 7maxc does nothing, it doesnt even use teh memory that 7max allocates to itself.

as an afterthought, maybe I misunderstood you.

do you mean running F@H w/ 7maxc isn't doing anything, or that the command 7maxc did nothing?

I thought you were saying 7maxc did nothing, as in won't start the WU...

That command is to start the desired app with 7MAX. if it's already running under 7MAX it won't do anything but pop up a mesaage box saying can't do.

from the command prompt, just for grins this mroning I'd entered 7maxc FahCore_78.exe

it popped up a message box "can't do" (or soemething along those lines) I already had the core started w/ 7max...

I don't know what the command line ver looks like, but I have two processes running in the graphical version, winFAH.exe and FahCore_78.exe.
the FahCore_78.exe is the one running with 4MB of 7MAX.

[ghettoside]

Whats your folding name and how many wus have you folded in a 24 hour period then?

Also what CPU?

ghettoside

in a 24 hr period? I dunno, none i guess. ever.

pent 3 700 Mhz

I have to get my butt to the store before it rains, I'm on a bike. I'll check back as soon as I get home. i just had that aftertought on my way and didn't want to let it hang for a few hours.

thanks YoS

[YARDofSTUF]

If you havent finished a WU in 24 hours then how is it finishing frames in 56 secs. Round that up to a minute and your finishing a WU every 100 mins, or 400 mins if its a tinker. So assuming its a tinker, since that would be the slowest of the 2 to complete at a given time per frame, every 7 hours you should be finishing a WU.

Sounds like whatever is going on is playing with ya.

[ghettoside]

If you havent finished a WU in 24 hours then how is it finishing frames in 56 secs. Round that up to a minute and your finishing a WU every 100 mins, or 400 mins if its a tinker. So assuming its a tinker, since that would be the slowest of the 2 to complete at a given time per frame, every 7 hours you should be finishing a WU.

Sounds like whatever is going on is playing with ya.

I don't know how you're making those calculations. I've only got 6 WU's total and I've been folding 24/7 on that box for like 2 months.

screen shots:
http://www.websitesnservices.com/test/fah/fah1.jpg http://www.websitesnservices.com/test/fah/fah2.jpg http://www.websitesnservices.com/test/fah/fah-currentWU1.jpg
http://www.websitesnservices.com/test/fah/fah-stats1.jpg

[ghettoside]

soon it's gonna show the time to frame completion, the last 2 shots are after restarting without 7MAX
http://www.websitesnservices.com/test/fah/fah3.jpg

[ghettoside]

It's made a lot of progress since Sunday...so I believe that's why the WU finish is now Aug 1 instead of Oct 2. if I'd let it run at 28 mins per frame it'd still be an Oct 2 finish date...but the 2 days at 56 secs/frame has decreased the time to completion

I have more running to do, will see what's going on w/ this later

[YARDofSTUF]

Why not just look at the console or log and compare completed to completed? LOL

Thats saying 10,000 frames total, so then its doing a Frame, as in 1-100, or 1% of the work in about an hour and a half.

[ghettoside]

Why not just look at the console or log and compare completed to completed? LOL

Thats saying 10,000 frames total, so then its doing a Frame, as in 1-100, or 1% of the work in about an hour and a half.

what console? wth are you talking about?

compare completed to completed? what's that supposed to mean? completed what?

after restarting w/7MAX, it says 56 secs a frame again. I have to leave it like that for now, I have no more time.

[YARDofSTUF]

what console? wth are you talking about?

compare completed to completed? what's that supposed to mean? completed what?

after restarting w/7MAX, it says 56 secs a frame again. I have to leave it like that for now, I have no more time.


http://i34.photobucket.com/albums/d104/YARDofSTUF/fah.jpg

Completed to completed, shows i'm basically 20 mins per frame.

[MissTynker2]

http://i34.photobucket.com/albums/d104/YARDofSTUF/fah.jpg

Completed to completed, shows i'm basically 20 mins per frame.


If I am reading this correctly...he is running the graphical version of folding...and therefore would not see the screen you are displaying here. Now I am getting all confused!! :rotfl:

[MissTynker2]

soon it's gonna show the time to frame completion, the last 2 shots are after restarting without 7MAX
http://www.websitesnservices.com/test/fah/fah3.jpg

Maybe this one will help you some in your efforts. Its a link for the em111 monitor, relatively easy to set up...and the most accurate for measuring time per frame, per unit, etc. Its the best one I have found so far of all the choices. Thank you so much for working with this and taking all this time for the cause. Thats great!! :p

http://home.comcast.net/~wxdude1/emsite/download.html

[ghettoside]

If I am reading this correctly...he is running the graphical version of folding...and therefore would not see the screen you are displaying here. Now I am getting all confused!! :rotfl:

Yes, you read the thread correctly.

I'm running the graphical version.

I'm working on large WU's... >5Mb

The screen shots posted are of the F@H graphical display. I cropped them to only the pertinent section (WU info, time to completion, etc...) to conserve some of my bandwidth.

[ghettoside]

Maybe this one will help you some in your efforts. Its a link for the em111 monitor, relatively easy to set up...and the most accurate for measuring time per frame, per unit, etc. Its the best one I have found so far of all the choices. Thank you so much for working with this and taking all this time for the cause. Thats great!! :p

http://home.comcast.net/~wxdude1/emsite/download.html

Thanks Misty, I've added it to the others I've downloaded, and I will try it.

I have as yet not installed anything but F@H...I have a lot of choices downloaded over the past couple days and hard to decide which to try first. I think you just resolved that for me though.:)

[ghettoside]

Just one more time to make sure it's clear, the screen shots posted are of the F@H graphical display...there is no program estimating my counter. It is all F@H.

I'm not assuming the times given by F@H are accurate.

I sat at the box with a stop watch. I'm sure if I sat there an hour the average would be that 56 secs per frame.

If you assume the F@H console version log or whatever it is YoS posted are accurate, then you have to assume the F@H graphical display I posted is accurate. But it does not have to be assumed- I timed the frame completion myself, and the numbers work out for the dates.

And the 28 mins per frame I get when running F@H without 7MAX, that time is accurate too. I timed it today when I restarted thw WU w/o 7MAX. And it also does not pick up steam...it just stays at 28 mins a frame. I had previously calculated the Oct 2 date at the then current number of frames left, and it came out to Oct 2, confirming that date.

So I did not calculate the Aug 1 date...I was busy. But I'm sure at 28 mins a frame it would calculate to that Aug 1 date.

As of this time, the WU is running with 7MAX and it is constant at 56 secs per frame with a completion date of this Sat.

I don't believe 7MAX will speed up the smaller WU's. The app was designed for memory intensive programs.

If I run this P2106 WU without 7MAX, it runs at 28 mins per frame. With 7MAX, it runs at 56 secs per frame.

It looks like I won't have much time to do anything with this until Sunday.

Hopefully the new used mobo arrives by Sat. and I can have another box dedicated to F@H so I can do better testing of running F@H w/ and w/o 7MAX, as well as console versions and as a service. I'll document everything then.

I'm just going to have to test that 7maxc out myself when I have the extra folding equipment mentioned above.

I'll put something together on installation and use of 7MAX as soon as I can.

[ghettoside]

I've left F@H run w/ 7MAX for the past couple days since I haven't been home much.

It's continued to run at 56 secs/frame, as opposed to the 28 mins/frame I was getting without 7MAX.

I had a thought on how it might be possible for 7MAX to be speeding up F@H as it appears to be doing for me.


Pavlov wrote this in a reply on the forum:
...And some time ago I discovered that memory access is slower than it can be. It's because windows uses 2-levels mode inside CPU for virtual memory management. But all modern CPUs allow 1-level mode for virtual memory management. It's slightly more complex for system (windows) to use such mode, since it requires some more smart algorithms. But there is some speed gain for some programs. So I rewrite that code to use 1-level mode (4 MB pages). ...

7MAX is using different memory algorithms.

That could explain it.

I haven't had time to really research it enough, but I did already find that some special memory algorithms are used for computer science for, among other things, more efficient use of memory and for speeding up the calculations when doing those calculations.

quoting from this pdf (http://www.siam.org/pdf/news/26.pdf) "Computer Scientist Finds Small-Memory Algorithm for Fundamental Graph Problem:"
...Reingold devised an algorithm that uses, within a constant factor, the smallest possible amount of computer memory to decide whether two vertices in a given undirected graph are connected, finding a path between them if they are. ...

...The result comes at a time when enormous data sets are becoming commonplace and small-memory algorithms, though rare, are increasingly important, says Avi Wigderson, a professor of computer science at the Institute for Advanced Study in Princeton.
Among small-memory algorithms, Reingold’s result is central, Wigderson adds, because it implies the existence of small-memory algorithms for many other important problems, such as finding a minimum-weight spanning tree for a graph or determining whether a graph can be embedded in the plane.



quoting from http://g.oswego.edu/dl/html/malloc.html
... I found that they ran much more slowly and/or with much more total memory consumption than I expected them to. This was due to characteristics of the memory allocators on the systems I was running on (mainly the then-current versions of SunOs and BSD ). To counter this, at first I wrote a number of special-purpose allocators in C++, normally by overloading operator new for various classes.
However, I soon realized that building a special allocator for each new class that tended to be dynamically allocated and heavily used was not a good strategy when building kinds of general-purpose programming support classes I was writing at the time. (From 1986 to 1991, I was the the primary author of libg++ , the GNU C++ library.) A broader solution was needed -- to write an allocator that was good enough under normal C++ and C loads so that programmers would not be tempted to write special-purpose allocators except under very special conditions. ...


quoting from http://daniel.haxx.se/projects/dbestfit/thoughts.html

dbestfit is intended to be a 100% working memory allocation system. It should be capable of replacing an ordinary Operating System's own routines. It should work good in a multitasking, shared memory, non-virtual memory environment without clogging the memory. Primary aimed for small machines, CPUs and memory amounts.

I use a best-fit algorithm with a slight overhead in order to increase speed a lot. It should remain scalable and work good with very large amount of memory and free/used memory blocks too. ...


This one, although not related to memory algorithms, on usage of different algorithms, pertains specifically to Distributed Computing and should be of interest to all SG Folding Team members regardless of whether or not 7MAX can speed up F@H-
quoting Memory Efficient Pair-Wise Genome Alignment Algorithm – A Small-Scale Application with Grid Potential ( http://www.springerlink.com/openurl.asp?genre=article&issn=0302-9743&volume=3251&spage=777)
An improved genome sequence alignment algorithm is presented that significantly reduces sequence matching computation time. Yet, very long sequences can still present computation challenges. A small-scale application is implemented on a shared memory system, on a cluster, and the grid-enabled cluster. Experimental results show comparable performance of the grid-enabled version, with scalability for large sequences. The grid offers application management, enables dynamic specification of parameters, and enables a choice of distributed computation nodes.
Keywords: Grid computing, distributed computing, computational biology, biological sequence alignment.


I believe that using a "smarter' memory algorithm can speed efficiency of programs, including F@H; and one step further, "smarter" algorithms can be used within F@H and other distributed computing to speed them up.

That last link I think I'm gonna look into more...

:p

edited- link problem
LNCS 3251 - Memory Efficient Pair-Wise Genome Alignment Algorithm ...

[ghettoside]

friggin link:wth:

Nova Ahmed, Yi Pan, Art Vandenberg, Memory Efficient Pair-Wise Genome Alignment Algorithm – A Small-Scale Application with Grid Potential, Lecture Notes in Computer Science, Volume 3251, Jan 2004, Pages 777 - 782

http://www.springerlink.com/openurl.asp?genre=article&issn=0302-9743&volume=3251&spage=777

[YARDofSTUF]

I havent gotten a big memory heavy wu on the test box yet but I'm thinking 7max isnt helpful and that it may have just stopped your system from eating up all the ram and left FAH with some more than it had before.

[ghettoside]

I havent gotten a big memory heavy wu on the test box yet but I'm thinking 7max isnt helpful and that it may have just stopped your system from eating up all the ram and left FAH with some more than it had before.

Quite possible YoS.:)

Something sped up F@H and it coincided with installation and use of 7MAX. I still have my doubts about 7MAX too... but at least it is possible, I did theorize earlier in this thread that what you've just said may be the cause of F@H speed up on my box. (more or less I said that, lol :D )

It occurred to me yesterday morning on my way to work that maybe 7MAX was using a different memory algorithm and maybe that could be the cause of the speed up. First thing I did when I got home yesterday was take a nap, lol, then looked at what Pavlov had said & then started looking into the possibility of different memory algorithms and effects.

I'm also thinking about that dbestfit used in a linux OS running F@H too, but need to do more research. I think I saw something else too but not sure yet.

I know the genome project is kaput, but I'm curious to follow up on that idea in that last link, I believe something along those lines could be beneficial to F@H.

Bear in mind that 7MAX was designed to give biggest gain to older cpu's with smaller L2 cache, so it may have something to do with the box I'm running. I dunno though, just trying to pin it down one way or the other.

That mobo I'm waiting for is due today, so hopefully it arrives and I can start setting it up so that next week I can do some serious testing of this 7MAX like I want to. Actually have today off at the last minute due to the holiday, but tomorrow I have to go out of town and prolly won't return until Sunday, I'll just have to play that by ear tomorrow.

I may even try an install of a vista build I have, since Pavlov says 7MAX is stable in win2K3 and vista due to native support for the 7MAX API. Drivers are an issue for the vista I have, so maybe it'll run on the newer mobo. My curiousity is aroused.

Thanks YoS. :thumb:

[ghettoside]

I just got an email from Igor Pavlov...

He's keeping an eye on this thread! I didn't give him a link to this thread either! ... gave him a link to SG a while back when I'd posted about 7ZIP and asked permission to do a pdf demo on 7ZIP... Hey, way cool! :D


From forum:

> It's continued to run at 56 secs/frame, as opposed to the 28
mins/frame I was getting without 7MAX.

7-max can't give such gain. Gain can be from 1% to 30%.
So 7-max + your program work incorrectly.
To check gain from 7-max you can run 7-zip benchmark with and without
7-max.

Hmmm.... I wish I could determine what's causing the speed up...

definately have to benchmark, and c'mon Canada Post, need that new mobo!!!:p

[ghettoside]

It's prolly gonna take me at least 2 months to complete my testing.

Still no idea what was causing the speedup. :confused:
I needed the hdd from that box, so I'll prolly never know what was happening there.:(

I'm gonna test F@H w/ and w/o 7MAX on 3 boxes, graphical and console versions. Even if I get only a 5% increase, that'll be good enough for me. And if I don't get anything - then I'm gonna tell the developer his program sucks:D I also wanna try the "stable' version that only runs on XP64, or win2K3 or vista...

I must say though that I'm a big fan of the developer's freeware archiving utility 7ZIP, so I've got my fingers crossed the "smarter' memory algorithm will show some sort of gain in WU processing.

also gonna look into a linux distro with one of those alternate memory algorithms.... I'm thinking to try that after I'm done with 7MAX.

I'll bump this later.

I dunno if YoS is still playing with this or not.