狂少's CDT-tweak

[BeardyMan]

Maxmem IS seen on the screenshot of the Super-Pi.....

So.......kiwi.......same thing, different way right?......So why someone do ALL these things that "CDT tweak" describes and not JUST do a Copy Waza and that's it?.....

Why do you use two hdd or partitions, when all you need is one

[hipro5]

Why do you use two hdd or partitions, when all you need is one

BEST performance I ever get out of 32M is when I use TWO seperate HDs....One for windows etc and one ONLY for putting in the Page File....

Though now I tested everything into one HD with two partitions....

[sup3rfly]

i m sorry to ask which is the most correct way to perform copy wazaa???

d:\ to c:\ to d:\ or c:\ to ERAM to c:\???

kinda confuse about this as well...too many versions of copy wazaa already ...

[massman]

3oh6 has a good guide:

http://3oh6.com/forum_posted/ocx/jas...pywazza-1.html

[sup3rfly]

yup i watched that... and its c:\ to ERAM to c:\... usually ppl would say it start copy from the spi partition....but 3oh6 started from c:\... i did exactly like what 3oh6 instructed and i actually gotten 8sec improvement on the 32m spi ....i dont know if its bad or there is still room for improvement tho

[mrlobber]

yup i watched that... and its c:\ to ERAM to c:\... usually ppl would say it start copy from the spi partition....but 3oh6 started from c:\... i did exactly like what 3oh6 instructed and i actually gotten 8sec improvement on the 32m spi ....i dont know if its bad or there is still room for improvement tho

Gain of 8-10 secs from copywaza is just what you would expect

[BenchZowner]

Gain of 8-10 secs from copywaza is just what you would expect

I get 16s + , others too

[sup3rfly]

ok..... guess i m too greedy then

btw i found out 1 thing about copy wazaa 3oh6's way.... when the copy wazaa process done...the task manager seems to be very laggy...i dont know if hipro's way would have that effect...i m suspecting that a correct copy wazaa would actually make that task manager to be laggy...or was i wrong??? previously the copy wazaa way i did which i think is wrong the task manager wont be laggy at all.... just what i have found out and guys sorry for my newbie-ness

[3oh6]

I get 16s + , others too

nice...cough up the process. try to be as thorough as you can. HD setup? where everything is installed? size of copy wazza? method...etc

my method usually gets 8 seconds...depending on size of CW/maxmem/OS used it can be squeezed for 9 seconds.

a 16 second copy wazza would be a huge benefit to my 32M runs so hopefully BZ doesn't mind sharing

[BeardyMan]

BEST performance I ever get out of 32M is when I use TWO seperate HDs....One for windows etc and one ONLY for putting in the Page File....

Though now I tested everything into one HD with two partitions....

I meant don't use a partition or second HDD
Just take a new folder and put it on your desktop then coppy teh files.

[massman]

I get 16s + , others too

The gain depends on how tweaked the system is, what regestry tweaks you use for instance.

I sorta tried the copy-waza technique just a few moments ago too. Copied the program files folder to documents and settings while keeping an eye on the available memory/systemcache tab. Os was completely untweaked, so my 32M times are really bad.

System:
E6300 @ 500x7
2x512MB @ 600 5-5-5-15

Start: 14.58.9xx
Ended up with: 14.30.xxx

Like I said, untweaked or tweaked makes a big difference .

[saaya]

rather than discussing how much gain you get from it, wouldnt it make more sense to discuss what times or what performance product or whatever the number vapor calculated, you can get?

if id do the same thing on this crappy work laptop here im sure some tweaks would give me a huge boost since its a 2 year old windows install with loads of crap on it... now does that make those tweaks incredibly good or does it make my windows insteall incredibly slow and untweaked? how could you tell?
and how about the hdds? what if you have a fast hdd compared to a slow one? sata compared to ide? chipset integrated sata or external sata controller? 8mb cache or 16? or even 32mb cache?

this performance product vopor calculated seems to work very well, so why dont you compare your tweak sucess based on that. whats the product with and without this tweak, that would shed a lot more light on things i think

[BenchZowner]

The gain depends on how tweaked the system is, what regestry tweaks you use for instance.

I'm using fresh 'n' Xtra-lited Windows XP setup, and the tweak gives me the same advantage when used on its own & when combined with other tweaks as well.

[Zeus]

Here's some answers to our questions i hope:

Memory requirement:
For the maximum calculation speed, 8 MB/1 million decimal digit is favorable, but 2 MB/1 million decimal digit is completely acceptable. Working memory is automatically adjusted by the software. Anyway, available main memory size is crucial for the processing speed!

Disk storage:
As for the working disk storage, 10.5 MB per 1 million decimal digit is needed. Working disk storage is automaticall freed. As for the permanent data storage, 1 MB per 1 million decimal digit is needed. Elapsed time is very keen to the disk access time. In order to short the elapsed time, you are better to equip high speed hard disk drive!

Now if i do the math right we'll need 32x10.5mb=336mb
plus 1 mb per million decimal digit for permanent data storage, so if i understand righ we have to add another 32mb to that, 336+32=368mb.

That has got to be the size of the pagefile if the permanent data storage is done on the pagefile.

Now look at the high speed hard disk drive part, makes me wonder if a scsi setup will show some gain?

edit:

Maximum length of calculation:
33.55 million decimal digits. 340 MB of disk storage is needed for 33.55 million decimal calculation.

[saaya]

scsi... tss.... think about i-ram in raid

[BeardyMan]

rather than discussing how much gain you get from it, wouldnt it make more sense to discuss what times or what performance product or whatever the number vapor calculated, you can get?

That's just some theory talk, i would like to see that on Team Japans numbers and see how effective it is..

[hipro5]

That's just some theory talk, i would like to see that on Team Japans numbers and see how effective it is..

Theory out of 1000+ resaults?......That's NOT theory M8.....That's FACTS....

[BeardyMan]

Theory out of 1000+ resaults?......That's NOT theory M8.....That's FACTS....

When all results fall between the lines you can call it a fact yes. It's inaccurate when more then a few don't.

[BeardyMan]

Do you even understand what you're talking about?

Perhaps make yourself more clear. Are you arguing that times don't scale inversely with CPU speed?

gautam you read what i said, to my opinion the theory is not a 100% guarantee,theory talk in my eyes.
I never said you have to agree with it.

[Zeus]

Oh, and here's another one to ponder:

Memory requirement:
For the maximum calculation speed, 8 MB/1 million decimal digit is favorable, but 2 MB/1 million decimal digit is completely acceptable. Working memory is automatically adjusted by the software. Anyway, available main memory size is crucial for the processing speed!

That actually means 8x32=256mb is enough for 32M.

Has anyone ever tried 256mb+ the memory windows need maxmem for 32M?

[Zeus]

It's not enough because you've got more running than just SuperPi. I've gone down to 380ish personally...but all the numbers you've stated are correct. If you look in the SuperPi window when running 32M, you'll see that the allocated memory is exactly 256MB.

Troman, this is not theory, its fact. It's fact that you many others just haven't comprehended yet.

Yeah, realised that and edited my post soon after.

But how much mem is an undressed fully tweaked XP taking up actually at idle?

No more than ~250-300mb i guess?

[BeardyMan]

Troman, this is not theory, its fact. It's fact that you many others just haven't comprehended yet.

It's beardyman and i understand it more then enough.

12.390 @4003
= 49597
12.500 @ 3999.9
= 49998

wow talking about efficiency
let me refrain, who are you or vapor to say wich effiency number is possible at the moment and wich not?

Vapors own words,
The "ideal" line is based on "50500"....a.k.a., a VERY well-tweaked system, and what you can expect as MHz changes.

Now the grand finale:
5525 x 9.000 = 49725 (Aug 2)(wasn't this impossible?)



edit*
anyways it doesn't matter, i don't believe in it and i'll stick with that.

[kemo]

It's beardyman and i understand it more then enough.

12.390 @4003
= 49597
12.500 @ 3999.9
= 49998

wow talking about efficiency
let me refrain, who are you or vapor to say wich effiency number is possible at the moment and wich not?

Vapors own words,
The "ideal" line is based on "50500"....a.k.a., a VERY well-tweaked system, and what you can expect as MHz changes.

Now the grand finale:
5525 x 9.000 = 49725 (Aug 2)(wasn't this impossible?)



edit*
anyways it doesn't matter, i don't believe in it and i'll stick with that.

50500 valid only for runs above 5GHZ

[hipro5]

50500 valid only for runs above 5GHZ

Same here....

[Gautam]

It's beardyman and i understand it more then enough.

12.390 @4003
= 49597
12.500 @ 3999.9
= 49998

wow talking about efficiency
let me refrain, who are you or vapor to say wich effiency number is possible at the moment and wich not?

We can only state what's reasonable at a given speed range. For example the 50500 number was for 5 GHz+. At lower multipliers, you see higher efficiencies, since the FSB is proportionally higher with respect to CPU speed.

If you look at his original thread though, he himself listed several exceptional times, such as before's and kyosen's that broke the pattern. The 50500 was just meant as a ballpark guide, not an absolute number, based on over 200 results. At 4 GHz we can see most of the best times wind up around 50000...at 3.6GHz around 49800.

If we were to scale my 4 GHz pp up to 5.5GHz, I would require 550x10 2:3 with the memory at 825 4-4-4. If I could do that, then yes...50000 would be possible at 5.5GHz.