ScienceMark thread (with score list)

[Alec§taar]

REPOSTING TO LAST PAGE WITH UPDATED RESULTS... apk

[Casheti]

Somebody sell me an Athlon for £30

[giorgos th.]

you dont need an Athlon,you need a mobo.hehe..

[Ketxxx]

lol

[Alec§taar]

Alpha0ne's score - outright wild imo.



* I knew he'd take top-spot, & said it to him once I saw his Super-Pi 1.5 mod test score... just had a feeling!

APK

P.S.=> An increase on ace80's part as well has our sum total "average" over 1400++ for the forums testers here no less, very good imo...

HOWEVER:

I do question HOW I am arriving @ the sum total group testers avg. mark (see description above)!

I.E.-> I don't think distributive properties are accurate or "possible/valid" by averaging the averages of each CPU group as I did... but, sure saves me TIME doing it, but is imo, unfortunately NOT completely accurate!

(Mainly because coming up w/ the group averages is quick & less entries, & only leaves me with 6 group averages, to average for the ENTIRE test group average - yes, I am lazy & Windows Calc.exe is a PAIN to mouse in decimal points, lol)!

So, that all said? Is me being "lazy" & doing that, erroneous/inaccurate??

I think so...

The 'average of CPU group averages' is NOT the "group average"...

I suspected this but, let it slide (to save time for me personally).

Does anyone want to recompute the group average above by doing all 31 tester's entries & see if there is a deviation from the figure I arrived @ using the 6 group CPU-type averages instead?

It would test that properly, one way or another...

E.G.-> I just got 1391.20 but, that was prior to ace80's getting 2 more points too, below, so, lol... so much for THAT figure!

EDIT PART - 1391.28 average of ALL 31 scores above, after ace's new score below.

(This is obviously not same as averaging the CPU-type group averages)

That was for the average of those 31 tests above, adding together ALL 31 tester's scores & dividing the total BY 31 (the proper way, but longer to do)

So - can anyone verify this figure I arrived @ so we can check if it is proof that averages of the CPU-type group averages is NOT the same as a TRUE average?

Thanks!

(Don't blame you if you DON'T want to take the time to do this... it IS a pain, I will warn anyone of that ahead of time... still, would be good to get a "2nd opinion" on here!)

apk

[ace80]

Another 2 pts gained, come on!!

[mikelopez]

I added all 31 scores and the avg. is 1391.277. The total amount of points is 43129.6.

[Alec§taar]

Originally Posted by mikelopez

I added all 31 scores and the avg. is 1391.277. The total amount of points is 43129.6.

Thank you Mike, I needed a second on that... I rounded up the first 7 to the right of the decimal point, based on the latter one being past 5 & all that, so it turns to .28...

Glad I checked on this... distributive properties do NOT apply as I thought they would not here... via my being blatantly/admittedly lazy & wanting to save time here.

By using the average of the group by CPU type averages, I was "off" & pretty much suspected it, lol...

Ah well! So much for 'shortcutting' things... I am surprised nobody caught this before I did actually. It's basically algebraic in nature, some stats I guess. I think it's rules I was aware of in my past actually, academically, haunting me here & I had to check on it.

* Means more work calculating it, but oh well! I'd rather do a BIT more work, than be inaccurate.

APK

[cdawall]

god dammit i moved down another stupid ddr266

[ace80]

Just found another few points and scraped through the 1100 mark, yippee!!

How many processes have all you guys managed to slim down to, while running the app?

[cdawall]

hey could we get a screen shot just hit ctrl+print scrn when you finish it would help a LOT

[ace80]

Sorry about that, forgot to attach the screenie

[Alec§taar]

Originally Posted by ace80

Just found another few points and scraped through the 1100 mark, yippee!!

How many processes have all you guys managed to slim down to, while running the app?

See above @ the bottom of the scores post below (reposted w/ your new data)...

Down to 5 here, have to restart them to get online though (but, all are plug & play that I stop so, it works out).



* I take it my "technique" is working for you then, in combination with increases in FSB etc.?

APK

[Alec§taar]

REPOSTING TO BOTTOM to account for g12rxz new score... apk

[g12rxz]

#11

[Alec§taar]

Show me your CPU-z memory timings screen please... & what type of RAM it is (DDR-400 vs. DDR-600 etc. type stuff)

I am looking for better RAM, & that Patriot stuff you have (just checked your profile)? Looks very good!

* Thanks - because when I buy? I want to see what timings to use on higher end RAM like that (that patriot brand is up there in my choices is why)

APK

[Alec§taar]

REPOSTING TO FIRST PAGE OF THREAD, PER URLYIN'S SUGGESTION TO DO SO (good one)... apk

[g12rxz]

timings are 4-4-4-12 cause my motherboard is shit and i cant loosen them. I'd like to take a moment to warn people against the ECS KA3 MVP V1.0A (extreme) (latest bios too). Very bad board to overclock/raisevoltage/tweak. I can't raise memory voltage or it wont post, cant drop htt speed, cant raise htt voltage, cant go above 1.35 volts for the processor, cant lock pci clocks, POS. Possibly returning to go for an nV5xx series single card mobo just cause i'm tired of the hassles this thing is giving me, and theres no other ati mobos out there worth buying. As for giving you a cpu-z... i've had a long day ... very long, very bad. at the moment i really dont care. sorry.

[ace80]

I take it my "technique" is working for you then, in combination with increases in FSB etc.?

Yes, the technique definatly helps, along with the various optimization guides you've posted in other threads, all been applied to this system (well the one's i feel comfortable with anyway), still reading and gaining valuable knowledge doing so.

The runs i've posted sofar have been with about 25 processes, still got to do a run with no AV and other security processes running, so that should gain some points.

[Alec§taar]

Originally Posted by g12rxz

timings are 4-4-4-12 cause my motherboard is shit and i cant loosen them. I'd like to take a moment to warn people against the ECS KA3 MVP V1.0A (extreme) (latest bios too). Very bad board to overclock/raisevoltage/tweak. I can't raise memory voltage or it wont post, cant drop htt speed, cant raise htt voltage, cant go above 1.35 volts for the processor, cant lock pci clocks, POS. Possibly returning to go for an nV5xx series single card mobo just cause i'm tired of the hassles this thing is giving me, and theres no other ati mobos out there worth buying.

It doesn't sound like it is for overclockers, or @ least not heavy o/c'ers... still, your score increase (again) is pretty impressive.

I know I'll need better RAM to beat what I got for instance, because my mobo can do the job, but my RAM can't (& I had some of the "best" here try to get me past my o/c & they almost always universally say "get better RAM" etc.)...

This is why I asked the question I did above about your Patriot RAM, because everyone that uses that stuff seems to do well on overclocks, regardless of their mobo (even restricted like you are by yours, you did very well imo @ least)!

Then - I have to "grill" Ketxxx some now as well because he knows Mushkin stuff very well also.

Originally Posted by g12rxz

As for giving you a cpu-z... i've had a long day ... very long, very bad. at the moment i really dont care. sorry.

Man, you know what? I hear/feel you, completely, in regards to yesterday... was NOT a "good day" for me either bro', it got me really DOWN!



* BUT, today is a new day, & I am 'kicking it off right', right now, w/ a NICE cup of hot coffee & some posting here!

(Thanks for info. man!)

APK

P.S.=> Have a nice day today! apk

[Alec§taar]

Originally Posted by ace80

Yes, the technique definatly helps, along with the various optimization guides you've posted in other threads, all been applied to this system (well the one's i feel comfortable with anyway), still reading and gaining valuable knowledge doing so.

Good! That's why I put them out for others to try/use...

When I see folks getting increased scores on things using them, it makes it worth it (it was many hours of work, testing, & researches over the past decade or more) & is living recorded evidence that it DOES work... the bottom-line!

I've had people from sites like 'the great arstechnica' say "cutting services is dumb/stupid" etc. & yet, proofs like yours & others in this thread show QUITE the reverse.

Same with another site called "XP Myths"...

Folks @ XP Myths (Andrew K & another called MasterTech) are not exactly my "pals" online & neither are the folks @ arstechnica!

(This ALL goes back a LONG ways, circa 2001 onwards, & a "feud online" more or less, but I don't bug them OR go to their forums, but they still (lol) try to put me down on their forums to this very day (gee, I wonder WHO got "the better" of whom))

They have, however, chased me to other forums & such starting with me (guys like MasterTech, Jeremy Reimer, Jay Little, & a few others who go online via various pseudonyms they write under & change, constantly, lol)!

To their points? Well, I simply put up VERIFIABLE evidences to the contrary, & run them off, everytime AND then they are left only with calling me names & such!

(Ad-Hominum attacks & just what I want to see when all is said & done vs. verifiable facts I put out, like this evidence of yours here in increased scores on benchmark tests).

WELL... Call me ALL the names you wish? I will just use verifiable facts as my defender... that's all!

Yes... it's pretty easy to do: Especially when you deal with people that cannot think for themselves imo (& only spit back what they read online)

The "XP Myths" site folks (Andrew K) post that to THIS VERY DAY, that cutting services is dumb/stupid, & also that ramdisk pagefile.sys placement is bad too... not on hardware SSD's like I have!

Plus, IF you have over 4gb of RAM? It is a practical use of the "excess" RAM above 4gb as well (no process in Win32 via its API can claim more than 4gb & this includes the memory mgt. of the OS itself as well).

You can use boot.ini switches to alter the 50/50 split of the 4gb total VM address space to 3gb/1gb (rather than the default 2gb OS/ 2gb apps) given to the OS (for mgt. purposes) & apps, but nothing more for THIS test.

Yet, no matter what "the great arstechnica" &/or XP Myths websites say? Heh, the results here (and elsewhere online where I tuned folks into this) show QUITE otherwise.

I.E.-> Why run things you don't use that suck up CPU cycles for things you actually DO use? Only makes sense. Do you go waving your arms around during your day, when you don't need to be? Takes up energy you could be applying to tasks you actually NEED to do is why... common sense.

Originally Posted by ace80

The runs i've posted sofar have been with about 25 processes,

Man, you CAN do better, if you cut it down... again, I can be down to 5 of them, but it took me some time to figure out which ones & such (due to the fact SOME services depend on others & some services are actually needed to, for instance, get online & be secured as well - I will note these for you).

Once you get the right "mix" down? You'll do better here & probably in ALL else... some stuff you need running though to get online for instance (DHCP client, your firewall, etc.), so keep it in mind when you stop them...

Above all:

DO NOT USE DISABLED ON A SERVICE'S STARTUP VALUE, until you KNOW you don't need it. Use MANUAL @ first, reboot, test if you can get online, & then keep testing until you 'pare it down to the bare essentials' & all that.

Takes some time & testing is all... much like doing RAM timings I guess!

Originally Posted by ace80

still got to do a run with no AV and other security processes running, so that should gain some points.

It will.



* Good luck!

APK

[ace80]

Ok, i've just done a run on a diagnostic startup via msconfig with about 10 processes running, i then went back into msconfig and choose normal startup, when it rebooted its now asking me to activate windows because of hardware changes.
I've used msconfig in the past to problem solve startup programs but this has never happened.
Correct me if i'm wrong but does it have anything to do with the BOOT.INI file, as now the modified boot.ini file is greyed out in msconfig and i think this is what i was using before.

I should be ok to just re-activate shouldn't i ?
Any help much appreciated.
Advise me to start a new thread if necessary.

[Alec§taar]

Originally Posted by ace80

Ok, i've just done a run on a diagnostic startup via msconfig with about 10 processes running,

There you go: You chopped it MORE than in 1/2 already... less processes chopping away @ CPU cycles this test could use!

And, your score went up, again (as it should have for the simple reason noted above).

Originally Posted by ace80

i then went back into msconfig and choose normal startup, when it rebooted its now asking me to activate windows because of hardware changes.
I've used msconfig in the past to problem solve startup programs but this has never happened.

This is part of WHY I recommended using services.msc instead...

Originally Posted by ace80

Correct me if i'm wrong but does it have anything to do with the BOOT.INI file, as now the modified boot.ini file is greyed out in msconfig and i think this is what i was using before.

It might - I never mess with that file, unless I am doing "debugged bootups" because that file has a TON of switches possible in it...

Originally Posted by ace80

I should be ok to just re-activate shouldn't i ?

Yes, absolutely do... you really do NOT have to be messing w/ that file @ all for the purposes of what we are doing, HERE, now.

Originally Posted by ace80

Any help much appreciated.

Show me a screenprint of your:

1.) services.msc

&

2.) Taskmgr.exe PROCESSES tab screen.

I will walk thru them with you, & cut you to a bare minimum (doing this with you can help others, it is worth doing in THIS thread because of that imo @ least).

Originally Posted by ace80

Advise me to start a new thread if necessary.

Nope, we can do this, here & now, right here... get me those screenprints.



* Like Leonardo DeCaprio said while playing the role of Howard Hughes in the film "The Aviator", after setting the world airspeed record (but crashing because he got "too into it" & ran outta gas)?

"She'll go FASTER!"

APK

P.S.=> We ARE going to get your rig "well into the 12's" when we are done I am guessing, via this method & your FSB boosts + RAM timings tweaking will take it the rest of the way... apk

[ace80]

Cheers alec, i never mess with the boot.ini file either, i guess must have selected origonal by accident. The machine was a HP origonally so when i changed the mobo it freaked, guess by reloading the origonal file it thinks i've done it again.

services and taskmanager screens attached. Thanks again.

[Alec§taar]

Originally Posted by ace80

Cheers alec, i never mess with the boot.ini file either, i guess must have selected origonal by accident.

That's OK - I cannot remember WHO said it, but it was a FAMOUS "brainiac" type:

"Genius is making mistakes. 1000's of them, & faster than others do, until you hit the RIGHT answers"

I definitely ascribe to THAT one. More than just on computers. Still, never hurts to ask others their views on things, but it is a matter of whether it applies to YOU, & how YOU USE YOUR MACHINE, really... & if it applies to that.

Originally Posted by ace80

services and taskmanager screens attached. Thanks again.

Ok then... here we go (DONE):

USING SERVICES.MSC, right-click on these processes & using the PROPERTIES submenu popup menu item, stop them (don't alter startup types until you KNOW which you need, & do NOT need - there is a DEPENDENCIES tab in services.msc right click on processes properties page, it CAN help you understand how some of them "interrelate with one another" too, do look @ it while I edit this):

THESE YOU CAN STOP (possibly permanently, depending on what they do & IF your OS or APPS need them):

AVG Email Scanner (AVG AntiVirus needs this for email virscans)
PC Tools Spyware (PC Tools program needs this for malware scans)
Print Spooler (needed for printing/faxing)
Security Center (not sure WHAT needs this, I have not seen this one before on Win2k3 Server)
Shell Hardware Detection (Explorer.exe GUI shell starts this one up)
System Event Notification
System Restore Service (If you use RESTORATION points, you need this)
Task Scheduler (boottime XP file optimization & other processes use this for PERIODIC work)
Themes (self-explanatory - especially if you use XP's "candy-coated" lifesavers candy roll look)
Windows Audio (Multimedia apps use this start it IF you need sound)
Wireless Zero Config (wireless networking uses this)

FOR A HOME LAN, these can affect that badly (keep them enabled if you do a home LAN):

Tcp/IP NetBIOS helper (If you use Client for Microsoft Networks, you need this)
Computer Browser ("")
Server
Workstation



* To get back online, after this "radical tuning"? You will have to reactivate these services (right clicking on them to start/stop them via their properties page from the popup menu you will see), in THIS order:

STOP THESE, but: You'll need these started again after test is done, to get online again:

COM+ Event System (hold off on this one when you stop others because, stopping SOME services will restart it, do it FIRST on reactivating them though)
Windows Mgt. Instrumentation
Application Layer Gateway (needed for SOME XP Firewall setups, others not)
Network Connections (needed for SOME XP Firewall setups & manipulating local connections, others not)
Network Location Awareness (you can actually set this to MANUAL startup type)
DNS Client (you can actually set this to MANUAL startup type)
DHCP Client (needed to get online)
CryptoGraphic Services (needed for online security, remember that)
Protected Storage (needed for online security, remember that)
Windows Firewall (needed for online security, remember that)
Windows Defender (needed for online security, remember that)

(Some of these Windows Server 2003 (what I use) has, some not, but that SHOULD be the correct restart order. If they "yell" @ you? Look @ that DEPENDENCIES screen I noted earlier, & start the services other services depend on, first!)

By the way?? YOU CAN ALSO STOP EXPLORER.EXE as well & use Ctrl+Alt+Del to summon taskmgr.exe again (to get your desktop GUI shell back) using taskmgr.exe's FILE menu, RUN submenu item.

That's to restart it when done doing this test (helps a LOT, explorer.exe is one busy app & eats CPU + I/O massively).

APK

P.S.=> Ok, done... stop those using services.msc, & then, we will look @ taskmgr.exe! Once we are done THERE? Run your test!

(Try to use taskmgr.exe rightclick on ScienceMark2.exe too for CPU priority (HIGH, ABOVENORMAL LOW, NORMAL, BELOWNORMAL etc. & if you have a single core/single CPU machine? STEER CLEAR OF REALTIME))

HIGH CPU priority can be good, because that DOES help for a point or two as well

(BUT, can get 'touchy' if you are at an EXTREMELY HIGH o/c near limit of what your system will take, some apps do NOT like that either - keep that in mind!) apk