Pagefile

wolverin0

Distinguished
Nov 21, 2007
25
0
18,540
Hi again.
I have 2 HDD:

WD Black 1Tb sata3 64mb buffer WD1002FAEX (C: Operating System HDD)
WD Caviar Blue WD5000AAKS 500GB 7200 RPM (D: Misc. HDD)

So, my question is
Should I disable the Page File @ C:\ and enable it on D:\ (technically lower speed) to gain pagefile "boost", or not?
 
Solution
Western Digital rates the WD5000AAKS at 126 MB/second (sustained)
(the number that really matters).

However, the speed at the outermost tracks will be a little faster.
The number that really matters is "buffer-to-disk".

Study these next graphs first, to see graphically why that rate drops a lot
from outermost to innermost tracks (amount of data is directly proportional
to track circumference, because modern HDDs strive to maintain the same
or similar recording density from outer to innermost tracks):

http://www.supremelaw.org/systems/io.tests/platter.transfer.crossover.graphs.2.png


So, here's what I would do:

(1) re-partition your Caviar Blue to have a relatively small primary
partition, e.g. 30GB; turn OFF the Indexing...
Generally speaking it's better to have the pagefile on a different drive than the OS. But if you've got plenty or RAM then you won't need to use the pagefile much and you won't see all that big a performance difference.

Modern systems usually have enough RAM so that the pagefile isn't the bottleneck it once was.
 

wolverin0

Distinguished
Nov 21, 2007
25
0
18,540
this are my full specs

Asus P6X58D-E
Intel Core i7 930 2.80ghz QPI 4.8GT/s 8mb Stepping D0 @ Cooler Master V8
3x2gb DDR3 Gskill 2000Mhz RIPJAWS Cas9
MSI GTX460 1Gb GDDR5 Cyclone 1GD5/OC
Samsung 23" P2370H Full HD 1920x1080 2ms HDMI
WD Black 1Tb SATA3 64MB Buffer WD1002FAEX
CoolerMaster 700W SilentProM Modular RS-700-AMBA
Cooler Master 690 II Basic + R4-S2S-124K-GP Cooler Cooler Master Silent Fan 120mm x4 + R4-L2S-122B-GP Cooler Cooler Master 120mm x2
 

wolverin0

Distinguished
Nov 21, 2007
25
0
18,540


haha i wish SSD drivers were so cheap in my country, this is the cheapest one i can find:
http://articulo.mercadolibre.com.ar/MLA-102762372-disco-rigido-ssd-gskill-phoenix-pro-60gb-para-pcnotebook-_JM

if i'd publish the price i had to pay for the whole system, you'd :eek: / :cry:

care to send me some small stuff over mail? i'd pay @ paypal haha (really)


price comparison:
USA: http://www.newegg.com/Product/Product.aspx?Item=N82E16820227550&cm_re=ssd_60gb-_-20-227-550-_-Product

vs

Argentina: http://articulo.mercadolibre.com.ar/MLA-98387618-disco-rigido-ocz-vertex-2-ssd-notebook-o-pc-35-y-25-60gb-_JM
 

XP was built when ram was expensive. It does lots of writes to the page file normally to free up ram.
Vista, and windows7 make better use of today's cheaper ram, and do not use the page file as much, if at all.
Page file is there for a reason... use it as intended.
 
I've been running my Windows 7 system without a pagefile for a year and a half without any issues. If you've got plenty of RAM to spare and no strange pagefile-dependent programs (like my old Photoshop V6) then a pagefile is superfluous.
 

wolverin0

Distinguished
Nov 21, 2007
25
0
18,540
but... is it better or worse? because i see lower values on AHCI than IDE
the only thing that changes is the 1MB block 245MBS @ Buffered Read over 256 KB block 225MB

i truly don't understand this
 


The mechanics of a hard drive do not change, regrardless of ahci or ide. It is the mechanics and the access pattern that determine performance. Do not read too much into synthetic benchmarks, they do not predict well what you will actually experience. This is true for SSD's also.
 

MRFS

Distinguished
Dec 13, 2008
1,333
0
19,360
Western Digital rates the WD5000AAKS at 126 MB/second (sustained)
(the number that really matters).

However, the speed at the outermost tracks will be a little faster.
The number that really matters is "buffer-to-disk".

Study these next graphs first, to see graphically why that rate drops a lot
from outermost to innermost tracks (amount of data is directly proportional
to track circumference, because modern HDDs strive to maintain the same
or similar recording density from outer to innermost tracks):

http://www.supremelaw.org/systems/io.tests/platter.transfer.crossover.graphs.2.png


So, here's what I would do:

(1) re-partition your Caviar Blue to have a relatively small primary
partition, e.g. 30GB; turn OFF the Indexing Service on that small partition;
re-partition the remainder as a dedicated data partition where you should
write routine drive images of your C: system partition;

(2) if you have the time and desire, reduce C: on your Caviar Black,
so the 2 primary partitions are identical in size (this, for system redundancy);
but, remember to re-locate all private data files to a dedicated data
partition on the remainder of your Caviar Black;

(3) download and run the Contig freeware, to create a perfectly
contiguous pagefile.sys on the Caviar Blue's primary partition:
http://technet.microsoft.com/en-us/sysinternals/bb897428.aspx
size that file at 1.5 TIMES the total of amount of RAM in your system
(as a general rule of thumb) e.g. 6144000000 for 4,096 MB of RAM;

(4) move pagefile.sys from C: to that primary partition on your Caviar Blue:
you'll need to re-boot, of course.


Summary: the small primary partitions will be "short-stroked"
taking advantage of the fact that there is more binary data
on the outermost tracks and, hence, their data rate is
directly proportional to their circumference (read: fastest
at the outermost tracks). Also, all "paging" will occur
more or less in parallel to I/O being done on the Caviar Black.
With AHCI enabled, that logic should effectively "burst"
paging I/O to and from pagefile.sys, because logically sequential
sectors will also be physically contiguous as well, thus minimizing
movements of the HDD servo-mechanism's READ/WRITE armature.


MRFS
 
Solution

TRENDING THREADS