狂少's CDT-tweak

[massman]

eva: how 'stock' is your system when it's installed? I've seen you posting a thread which explains how to use nlite on the i4memory forums. How much tweaks are already included in your OS when you install it?

[BenchZowner]

@eva2000:

Create a 4GB file ( WinRAR, UltraISO, anything ).
Don't use maxmem.
Open up SuperPi set the calculation, and leave it at the "Now start to calculate..." window ( waiting for you to press ok )
Copy the wazza file from D:\ to C:\ and then from C:\ to D:\Temp
As soon as the last copy process finishes click OK to start the calculation.

[Gautam]

The thing is I don't. I might have a E6750 (this one is stuck at 400mhz fsb) that can do atleast 450x8 later this week.

How about that other one I posted at 8x385?

I would be joining the fun but all I have is a Core Solo 1.2GHz atm so don't despair. (I still have 6 gigs of cherry picked D9 tho...)

[BenchZowner]

I might give it a try with 1M & 32M later on, but to be honest I don't expect any changes in 1M ( still looking out for that magic trick that shove those seconds off that magic low clocks WR ), as for 32M...if it shoves off some time, and it works in combination with the "regular" copy wazza as well it might be worthy.

[Eldonko]

hmm tried 3.7GB was 2 seconds slower HA HA... will try larger file and see what happens

Its really weird waza isnt working for you. Have you tried waiting about 30 sec after the copy before the 32M run? Copying between partitions on root?

[eva2000]

Yeah tried various waiting intervals before hitting okay as well as different partitions on same drive .. super pi is located on main root partition C:\

[G H Z]

Ahhh.........I've heard of that too........and out of curiosity (maybe to get faster resaults), I have install them EVEN at 5300MHz.....NO GAIN AT ALL......Sorry.....

Exactly, complete bollocks

[elmor]

How about that other one I posted at 8x385?

I would be joining the fun but all I have is a Core Solo 1.2GHz atm so don't despair. (I still have 6 gigs of cherry picked D9 tho...)

I'll give it a try

[Gautam]

Yeah tried various waiting intervals before hitting okay as well as different partitions on same drive .. super pi is located on main root partition C:\

I start the calculation immediately after the copying completes. In fact I have the window ready to go beforehand so all I do is hit OK.

[BeardyMan]

Note how I say that it's not about which timings are the tightest . You're subs might be tighter, they will not be faster in se.

Back on the tweak, I presume testing should be done, but correctly. By the way, difference between Server03 and XP?

if they were not faster i wouldn't have pushed them like that

OFFTOPIC
What are you doing with blinds cascade?

[massman]

I start the calculation immediately after the copying completes. In fact I have the window ready to go beforehand so all I do is hit OK.

That's exactly what you should do

[spainis]

Your curve hands doesn't mean that his tweak doesn't work

[kiwi]

You don't need so large copy-waza. Maxmem +200MB or amount of RAM + 200MB should be enough

I am pretty busy right now but I will try this tweak (I hope I will understand every detail of it) myself as Hipro suggested.

[BeardyMan]

LOL

I just copied a few small zipped files from 1 map to another map on the same disk. And the result was also -0.4secs
So copying to another partition is not needed?

anyone willing to test that
I'm going to see how far you can push that

[Bigchrome]

On AMD (S754 3400+) If I run super pi WHILE moving a large file I drop from 38.063 to 37.859 in 1M and I haven't been able to replicate that every with other tweaks....anyone on conroe want to try this?

[BeardyMan]

OPB seems to have hit the nail very good




The higher i get that systemcache the better my score lol
and no need at all for copying to another partition or for zipping files

another 0.050secs by pushing that systemcache with full litterbin :p

[mrlobber]

Well, here's my report, just as Hipro wanted here:

Could someone please test it like below:

a. without any "software tweaks" (windows tweaks) - put only ramtimmings etc. (I mean, DO LargeSystemCache=1 and Priority=Very high for Super-Pi and things like that but without CopyWaza and things like that.....Only ramtimmings etc)

b. Then do what you always do up to now with only 384MB/384MB fixed Page File and Copy Waza from one disk to another (into Super-Pi folder)......Copy Waza ~768MB and more (doesn't matter if it will be 2GB)
After Copy Waza ends copying, close Explorer from task manager, close task manager and WAIT for some seconds till the HD led STOPS blinking....When it stops blinking, start the 32M

c. Do as CDT-Tweak describes

and finaly report findings?....

System:

E6600
Asus P5K-E unmodded
2x1Gb Corsair 6400C3
Asus X850XT (drivers not installed)
20 GB IDE HDD

I'm by no means ar Superpi tweaker, so the first score might look like a bit shaky... whatever.

515x7, 1:1, PL7 etc, maxmem=600, no eram, plenty of services disabled, no copy-waza, no cdt.

Result: 13m24.031s


Now the same settings plus pagefile 384 Mb, copy-waza 768 GB (in task manager I observed available memory 525Mb vs system cache 540Mb before the run):

Result: 13m14.437s


And now CDT run... done with pagefile 512Mb exactly as elmor described:

1. I created 1x 632mb file using the command "fsutil file createnew D:\myfile 647168000" (1000*1024*632) for a file of exactly 632MB.
2. Then I put it in a RAR-archive with the Store Compression method for no compression.
3. After that I renamed it to myfile2 and then myfile3 and after each rename add it to the RAR-archive with the Store option so that you get a 1896MB big RAR-archive.
4. Then I transfered it to my benchdisk with an USB flashdisk (I created the file on my laptop) and put it on SPI-partition (D:\file.rar).
5. Now you start SPI and select 32M calculation and press OK, but don't press OK to start it, let it wait there.
6. After that you copy it three times to C:\file.rar and the two other times you have to overwrite of course.
7. Next you copy D:\file.rar to any folder on D:\, for me it was D:\cw\, three times just like you did to C:\ before.
8. Now, as Kevin says, to balance it you copy the file C:\file.rar to D:\file.rar and overwrite it. I did it three times and I saw the ammount Avaliable memory raise each time.
9. Now you can close the Taskmanager if you had it open for monitoring and all the other windows not related to SPI and now you can start your calculation!

However, I couldn't observe the available memory going closer to system cache more than it was in my 2nd run pictured above. In fact, I observed the 525Mb/540Mb situation already after I did the 1st step of CDT tweak (elmor's 6th, that is, copying that big file from D:\ to C:\), and this proportion didn't change later on, while doing all the other copying.

So, since getting these two (available memory & system cache) as close as possible seem to be the key for this tweak, it's no surprise that my "CDT run" is basically the same as the "copy-waza" run, even a little worse.

Result: 13m18.422s


Gotta only 4hours left to sleep before going to work tomorrow so maybe I'll try the CDT after some 20hours once again. Things I might have done differently from elmor's run are 2 (as far as I'm aware of):
1. I calculated the "632Mb file" as 1024x1024x632 whereas elmor did 1000x1024x632
2. I used *.bat files for copying all the stuff around (you can see them on my desktop in the 3rd pic).
Elmor & other people familiar with the tweak, could you give an insight what might I have did / did not for the tweak not working for me as a result?

Edit:
3. Another subtle thing: before copying all the big files for 3 times each, I always waited a couple of seconds between the copy/replace operations. Should I have done this or should I have started the next copying immediately?

[G H Z]

The timing of this is just brilliant!

[hipro5]

Thanks mrlobber for the tests......

Anyone's willing to test it as I described?....

Tip: You can go into Registry and change the "Windows Priority" from "2" which is default for the Win XP , to 66 (it will show you 102 finaly)....
It's under "Priority"....> HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Contro l \PriorityControl
and DON'T use the Patch of Microsoft with multicore suport....It gives you NOTHING even in 3D, makes things worse...
It's under
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Contro l \Session Manager\Memory Management\Throttle......IF you have alredy install it, then put it to "0".....

[spainis]

Hipro5 maybe you can reveal your tweaks
And you think that they will work for everyone?

[hipro5]

Hipro5 maybe you can reveal your tweaks
And you think that they will work for everyone?

They are NOT MY tweaks......They are Japanise's ppl tweaks that they work like charm all these years - from the oldest till the latest of them -.....and YES they work for everyone...

[mrlobber]

I hope to be able to perform some additional tests this evening. Notice again, the idea behind CDT (at least as I see it, whichever version of it we might want to test) is to approximate available memory and system cache as close as possible. So I think it is obvious that I did something wrong in that respect since, as I already posted, I couldn't get available memory and system cache nowhere as close as others who have displayed the CDT working (and it is also obvious, that it's possible to get these two values very close, as demonstrated by various screenshots around).

Maybe I'll try what hixie posted as well

Step 1: Set MAXMEM to 600, and made a file 604MB large (600 as in the MAXMEM value and 4 because i'm running 4M Super Pi)
Step 2: Open up Super Pi, and select 4M, but stop that the start button.
Step 3: Open task manager, and keep copying the same 604MB file into THE SAME FOLDER, until System Cache and Available memory figures match each other.
Step 4: Close everything, and run Super Pi.

[hipro5]

Hipro, i would suggest you to try this tweak by yourself, and then you will be able to compare if and comment if it works for you

also, this is no =copy-waza, you can say it is based in it though there's a reason of why 3 files of 632Mb in the same folder, do the tweak properly and you will notice the improve


saludos

How much of improvement shall I see....6-7-8-9 sec lower in 3.6GHz than Copy Wazza?....

[Metallica]

How much of improvement shall I see....6-7-8-9 sec lower in 3.6GHz than Copy Wazza?....

i didn't do a deep test so i couldn't say how much of a improvement you will have

i will compare copy-waza and CDT-IV tweak for my-self this weekend and will report

[eva2000]

Note that 5.2>2.8.

@eva2000:

Create a 4GB file ( WinRAR, UltraISO, anything ).
Don't use maxmem.
Open up SuperPi set the calculation, and leave it at the "Now start to calculate..." window ( waiting for you to press ok )
Copy the wazza file from D:\ to C:\ and then from C:\ to D:\Temp
As soon as the last copy process finishes click OK to start the calculation.

but listening to both your tips i used 3.94GB file and copywaza definitely better gain of around 2 seconds versus <1 second more consistently

bumped ram to 640mhz 5-5-4-5 but same clock speed

no tweaks


just copywaza 3.94GB file copied from C: to D: and D: to C:



Going to try hipro's method as well.. didn't know about that prioritycontrol reg tweak though so let's see

Thanks mrlobber for the tests......

Anyone's willing to test it as I described?....

Tip: You can go into Registry and change the "Windows Priority" from "2" which is default for the Win XP , to 66 (it will show you 102 finaly)....
It's under "Priority"....> HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Contro l \PriorityControl
and DON'T use the Patch of Microsoft with multicore suport....It gives you NOTHING even in 3D, makes things worse...
It's under
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Contro l \Session Manager\Memory Management\Throttle......IF you have alredy install it, then put it to "0".....

but i can't see such an entry in PriorityControl ?? only entry is Win32PrioritySeparation ??

I start the calculation immediately after the copying completes. In fact I have the window ready to go beforehand so all I do is hit OK.

times vary from a gain or loss of a few ms when i do this nothing consistent heh