Dual Core Comparison

[nos1eep]

It is further alleged that on or about Mon, 8 May 2006 08:25:40 -0500,
in alt.comp.hardware.amd.x86-64, the queezy keyboard of "VanShania"
<[email protected]> spewed the following:

|Well thats the difference. My times are just for the ripping part. Yours is
|for ripping and burning. Try Premiere Elements 2. It is dual core ready.

I'll check it out.

Later.cya

 

[nick]

forget raid. raid is for servers not desktop apps.

 

[Wes Newell]

forget raid. raid is for servers not desktop apps.

Really? Which of these would you rather have, hda or md0?

[root@wes2 wes]# hdparm -tT /dev/hda

/dev/hda:
Timing cached reads: 1992 MB in 2.00 seconds = 995.94 MB/sec
Timing buffered disk reads: 110 MB in 3.00 seconds = 36.62 MB/sec
[root@wes2 wes]# hdparm -tT /dev/md0

/dev/md0:
Timing cached reads: 1888 MB in 2.00 seconds = 943.94 MB/sec
Timing buffered disk reads: 276 MB in 3.00 seconds = 91.99 MB/sec
[root@wes2 wes]#

 

[nick]

transfer rate is totally meaningless.

forget raid. raid is for servers not desktop apps.

Really? Which of these would you rather have, hda or md0?

[root@wes2 wes]# hdparm -tT /dev/hda

/dev/hda:
Timing cached reads: 1992 MB in 2.00 seconds = 995.94 MB/sec
Timing buffered disk reads: 110 MB in 3.00 seconds = 36.62 MB/sec
[root@wes2 wes]# hdparm -tT /dev/md0

/dev/md0:
Timing cached reads: 1888 MB in 2.00 seconds = 943.94 MB/sec
Timing buffered disk reads: 276 MB in 3.00 seconds = 91.99 MB/sec
[root@wes2 wes]#

--
Want the ultimate in free OTA SD/HDTV Recorder? http://mythtv.org
http://mysettopbox.tv/knoppmyth.html Usenet alt.video.ptv.mythtv
My server http://wesnewell.no-ip.com/cpu.php
HD Tivo S3 compared http://wesnewell.no-ip.com/mythtivo.htm

 

[Cal Vanize]

ummmm.... this has to be a troll.

transfer rate is totally meaningless.

forget raid. raid is for servers not desktop apps.

Really? Which of these would you rather have, hda or md0?

[root@wes2 wes]# hdparm -tT /dev/hda

/dev/hda:
Timing cached reads: 1992 MB in 2.00 seconds = 995.94 MB/sec
Timing buffered disk reads: 110 MB in 3.00 seconds = 36.62 MB/sec
[root@wes2 wes]# hdparm -tT /dev/md0

/dev/md0:
Timing cached reads: 1888 MB in 2.00 seconds = 943.94 MB/sec
Timing buffered disk reads: 276 MB in 3.00 seconds = 91.99 MB/sec
[root@wes2 wes]#

--
Want the ultimate in free OTA SD/HDTV Recorder? http://mythtv.org
http://mysettopbox.tv/knoppmyth.html Usenet alt.video.ptv.mythtv
My server http://wesnewell.no-ip.com/cpu.php
HD Tivo S3 compared http://wesnewell.no-ip.com/mythtivo.htm

 

[Alan LeHun]

Nick is mostly right, transfer rate is much less important then access
time. The average read access times of the current generation of SATA
drives in in the neighborhood of 13 milliseconds. If the transfer rate
were infinite the effective transfer rate of a 256K block (assuming that
the entire block was stored contiguously) would be 20MB/second. Most file
systems use a smaller block size then that. For 64K blocks the effective
transfer rate would only be 5MBytes/second.

This is all true but it has no relevance to RAID0 which has two discs
taking the same 13ms to access two different blocks /at the same time/
and is therefore twice as fast (overheads not withstanding).

 

[General Schvantzkoph]

ummmm.... this has to be a troll.

Nick is mostly right, transfer rate is much less important then access
time. The average read access times of the current generation of SATA
drives in in the neighborhood of 13 milliseconds. If the transfer rate
were infinite the effective transfer rate of a 256K block (assuming that
the entire block was stored contiguously) would be 20MB/second. Most file
systems use a smaller block size then that. For 64K blocks the effective
transfer rate would only be 5MBytes/second. The maximum transfer rate of
the disk only matters when you are dealing with huge files that have been
laid out contiguously. If you were storing video and you used a block size
of 1M then you could start to approach the maximum transfer rate of the
disk. However using a huge block size like 1M would waste tremendous
amounts of space so it's never done except in specialized file servers
that are aimed at applications like video. For desktop applications RAID0
is useless.

 

[VanShania]

I had raid 0 set up briefly to see what it was like with my 2X320 gb WD Raid
Editions and it sure made working on the desk top more enjoyable. Whole
lotta less waiting for games everything else being loaded or transfered. I
can't wait to see how fast it will be with 4 drives.

--
Ati All-In-Wonder 9600XT
Thermaltake LanFire Midtower(4X80mm fans) with Antec 550 Watt PSU
Gigabyte GA-K8NSC-939 nForce3, A64 3500+, Stock Cooler IdleTemp 28 C
2 Gb Dual Channel PC3200 OCZ Platinum 2-3-2-5 CL2.5
Viewsonic A91f 19in Moniter
2XSATA WD 320gb Raid Edition, PATA WD 120Gb HD
Pioneer 110D+Liteon 1693S Dual Layer burner
Logitech MX 310 Optical Mouse
Microsoft Sidewinder Precision 2 Joystick
Microsoft ergonomic keyboard
Cheap computer speakers with Sennheiser HD 477 Headphones

3DMark05Free-1024X768, HighPerf Settings,5.10 drivers
Overall 1438
Cpu 4500
3Dmark2001 - 17680

Games I'm Playing- Battlezone II, IL-2 Sturmovick Series
Empire Earth 2, Need For Speed: Underground 2,
Civ IV

forget raid. raid is for servers not desktop apps.

[OK, so I cross-posted.]

Background:

I'm in the process of building a new computer for a friend who is a
rather demanding business power-user.

The computer will be used mostly for internet browsing / email, business
applications, and some light gaming. The game that would probably
present the most CPU burden would be MS Flight Simulator 2004.

User often has 4 - 6 business applications open at the same time then may
launch FS keeping the other apps in the background. He indicated that
memory usage sometimes tops 1g in his current system.

He wants a "very responsive" system. I don't want to hear him express
any concerns about stability.


System considerations:

The hard drives will be two WD SE16 250gb in RAID 0 (I have concerns
about the reliability of Raptors). Memory will be two gig (2x1gb) of
Corsair XMS Platinum CAS2 (becuase I have it).

O/S will either be W2K or XP Home.

I'll probably use a ASUS A8N-VM CSM since there is not a heavy burden on
video performance. (I run FS 2004 on a GF-6100 board without any
problems.)

The board is only capable of ~ 20% overclocking but reports indicate its
VERY stable.

I'm interested in using a dual core processor for this application and
are considering either a X2 3800 or an Opteron 165.


Question:

In this application, are there any opinions on whether the X2 3800 or
Opteron 165 would perform better?

TIA,

CV

 

[Bill]

Nick is mostly right, transfer rate is much less important then access
time.

amounts of space so it's never done except in specialized file servers
that are aimed at applications like video. For desktop applications RAID0
is useless.

I once thought as you do...but after experiencing it myself, I know your
comments are not accurate for my usage and needs.

I use Photoshop to edit large digital photos. I use a 2-disk RAID 0 and
get nearly double the transfer rates when opening and saving files. I
also do batch conversions of RAW files to JPG, and it makes a big
difference to completion time.

A friend of mine uses a 4-disk array, and it's nearly twice as fast as
my setup - about 250MB/s. But then he does DVD re-mastering and works
with 9gig files and needs the high transfer rates.

For the average joe surfing the net, writing emails, and reading text
newsgroups, yeah it won't make much difference. But if you do anything
with large files, it can make a tremendous difference.

 

[Wes Newell]

Nick is mostly right, transfer rate is much less important then access
time. The average read access times of the current generation of SATA
drives in in the neighborhood of 13 milliseconds. If the transfer rate
were infinite the effective transfer rate of a 256K block (assuming that
the entire block was stored contiguously) would be 20MB/second. Most file
systems use a smaller block size then that. For 64K blocks the effective
transfer rate would only be 5MBytes/second. The maximum transfer rate of
the disk only matters when you are dealing with huge files that have been
laid out contiguously. If you were storing video and you used a block size
of 1M then you could start to approach the maximum transfer rate of the
disk. However using a huge block size like 1M would waste tremendous
amounts of space so it's never done except in specialized file servers
that are aimed at applications like video. For desktop applications RAID0
is useless.

If you think transfer rates don't mean anything, set your drives back to
pio mode 1 and let us know how well your system performs. It's also funny
to me that you don't consider video as a desktop app. My average file size
is about 10Gig. BTW, when you set your drive back to pio mode1, let me
know how long it takes to open oo the first time, before it's cached. Now
load your system down with open apps so your swap file comes into use and
let me know how well that works for you. I think this moght change your
mind.;-)

 

[Cal Vanize]

General Schvantzkoph wrote:




I once thought as you do...but after experiencing it myself, I know your
comments are not accurate for my usage and needs.

I use Photoshop to edit large digital photos. I use a 2-disk RAID 0 and
get nearly double the transfer rates when opening and saving files. I
also do batch conversions of RAW files to JPG, and it makes a big
difference to completion time.

A friend of mine uses a 4-disk array, and it's nearly twice as fast as
my setup - about 250MB/s. But then he does DVD re-mastering and works
with 9gig files and needs the high transfer rates.

For the average joe surfing the net, writing emails, and reading text
newsgroups, yeah it won't make much difference. But if you do anything
with large files, it can make a tremendous difference.

Program start times are MUCH faster using RAID 0, especially if defrag
is run on a regulay basis.

 

[Ed Light]

I use Photoshop to edit large digital photos. I use a 2-disk RAID 0 and
get nearly double the transfer rates when opening and saving files. I
also do batch conversions of RAW files to JPG, and it makes a big
difference to completion time. ...........
For the average joe surfing the net, writing emails, and reading text
newsgroups, yeah it won't make much difference. But if you do anything
with large files, it can make a tremendous difference.

Would it make a noticeable difference in loading a whole lot of little files
1k to 15k mixed in with a few 150k ones?


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at
(e-mail address removed)
Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

Fight Spam:
http://bluesecurity.com

 

[kukkanen]

SO, with 100 drives the propability is 100%? It just don't work that
way, sorry.

 

[General Schvantzkoph]

SO, with 100 drives the propability is 100%? It just don't work that way,
sorry.

The discussion is about MTBFs. If a single drive has an MTBF or 10000
hours and you have 100 drives in the system then the MTBF of the system
would be 100 hours. That doesn't mean that the system will fail once every
100 hours like clock work. It just means that on average you would see a
drive failure every 100 hours. The consequences of a drive failure are
dependent on the RAID configuration. In a JBOD system with two drives the
likelihood of a failure that causes the loss of 50% of your data (i.e. a
single drive failure) in any one hour is 1/10000 (using a made up MTBF
number of 10000 hours). The probability of a failure that causes the loss
of 100% of your data, i.e. two drives failing in a hour, is 1/(10000 *
10000). In a RAID0 system the failure of a single drive causes the loss
of 100% of your data so the probability of a 100% data loss in any given
hour is 2/10000. In a RAID1 system the loss of a single drive doesn't
cause the loss of any data, in order to lose 100% of your data you must
have a double failure. As stated earlier the probability of two drives
failing in any give hour is 1/(10000 * 10000). However once you've
experienced a single drive failure you are left vulnerable because the
remaining good drive has the probability of failure of 1/10000. So in
figuring out what your chances are of losing your data in a RAID1 system
is you not only have to take into account the Mean Time Between Failures
but also the Mean Time to Repair. If you immediately shut the system down
and replace the bad disk then your probability of losing all your data on
a single two drive RAID1 system is close to 1/(10000 * 10000). However if
you keep the system running and it takes on average 100 hours before your
replace the bad drive then the mean time between data loses would be
100/(10000 * 10000) or 1/1000000.

 

[Ed Light]

The discussion is about MTBFs. If a single drive has an MTBF or 10000
hours and you have 100 drives in the system then the MTBF of the system
would be 100 hours. That doesn't mean that the system will fail once every
100 hours like clock work. It just means that on average you would see a
drive failure every 100 hours.

That's a pretty amazing statistic, but after the 2 or 3 duds died the other
ones would last for awhile before they started to.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at
(e-mail address removed)
Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

Fight Spam:
http://bluesecurity.com

 

[General Schvantzkoph]

That's a pretty amazing statistic, but after the 2 or 3 duds died the
other ones would last for awhile before they started to.

I'm just using 10000 hours as a nice round number to explain the math. I
don't know what the actual MTBF number for any particular drive is but I'm
pretty sure it's better than that.

 

[kukkanen]

That is very fascinating, but the probability still doesn't add up to
100% does it?