Confusion over RAM utilization

[John John - MVP]

Is there a utility to test for that? My neighbor's computer came to a crawl lately and I cant find any malware on it.

You would get that from the computer manufacturer.

John

 

[Paul]

Is there a utility to test for that? My neighbor's computer came to a crawl lately and I cant find any malware on it.

John

To demonstrate how many cache levels are present, you can use a tool like this.

Sisoftware Sandra Lite 2009
http://majorgeeks.com/SiSoftware_Sandra_Lite_d4664.html

Only install that, if the person doesn't have a paid for version of Sandra already,
as otherwise, the software may become confused. There is the free version,
and you can also get the pay version (which enables more of their suite
of tests).

Once it is installed, and the requests for unnecessary updates disabled
and the like, use "Cache and Memory" benchmark. That will measure bandwidth
versus block size, for memory transfers. The easiest way to view things, is in
the tabular data.

E4700 2.6GHz, VIA chipset (crappy motherboard, with low main memory bandwidth)

Block_size Bandwidth
2KB 107.93GB/sec \
4KB 113.66GB/sec \
8KB 120.38GB/sec \__ resident in L1
16KB 121.54GB/sec /
32KB 121.27GB/sec /
64KB 106.28GB/sec /
128KB 34.19GB/sec \
256KB 34.29GB/sec \__ resident in L2
512KB 34.34GB/sec /
1MB 34.36GB/sec /
4MB 7.22GB/sec \
16MB 2.38GB/sec \__ main memory bandwidth
64MB 2.35GB/sec /
256MB 2.34GB/sec /

If you had a Phenom II, it would have three break points in the result,
as a Phenom has L1, L2, L3, and main memory. My Core2 has L1 and L2
(so two break point). Many older processors have L1 and L2.

If you see the two break points, like I have in my example, then
everything is fine.

There are probably other tools that perform the same measurement
as that one, but I don't remember any names right off hand.

Paul

 

[Andy_XP_Devotee]

Here are the HD Tune results for my 3 physical drives:

http://www.file2go.net/image/db38ec6959/HD Tune Results.png

I've included Paul's benchmark results at the bottom for comparison. For
each of my 3 drives, I've also included a screen shot of the Info tab.

There are a number of interesting patterns in the data.

The first 2 graphs show patterns similar in overall shape to the benchmark,
but the 3rd graph is clearly flat. I should note that this 3rd drive is an
external drive.

My first thought was to look for info on the transfer mode, DMA or PIO. One
of the postings here said that the transfer mode would be indicated on the
Info tabs, but I'm not seeing it.

I also noticed that the yellow scatter shot points used to indicate Access
Time are much more tightly spaced in the 2nd drive than in the 1st. I am
wondering what this tells me about my drives.

Finally, in Paul's benchmark plot, I see that the SCALE of his drive is way
different than that of my drives. This is reflected in the results numbers,
too, with the benchmark having an average Transfer Rate of about 100 MB/s,
with my drives coming in at 50 MB/s or less.

I'm just guessing, but to my inexpert eye, it looks like this is telling me
that Paul's benchmark drive is working twice as fast as my own drives. And
this is a big deal to me, as my brother has often talked about how drive
access times are often the true bottleneck in any computer system.

So I guess I'd like for Paul to tell me what kind of drive he used for this
test. THIS may actually turn out to be the key to my problems.

Andy

 

[Andy_XP_Devotee]

Daave, . . . this is just great. I will work through the different tests
you suggest this evening. I will post the results back here.

 

[Daave]

My first thought was to look for info on the transfer mode, DMA or
PIO. One
of the postings here said that the transfer mode would be indicated on
the
Info tabs, but I'm not seeing it.

I'm not sure how this is done in HD Tune, but it's easy in Windows:

Open Device Manager.

Then expand the entry for IDE ATA/ATAPI controllers.

Double-click the IDE Channel entries.

Click on the Advanced Settings tab.

Ignore the ones that say Not Applicable.

Hopefully the value for current transfer mode is Ultra DMA Mode 5 (at
least, that's what mine is). If it says PIO, you need to fix it!

 

[Mike Hall - MVP]

Here are the HD Tune results for my 3 physical drives:

http://www.file2go.net/image/db38ec6959/HD Tune Results.png

I've included Paul's benchmark results at the bottom for comparison. For
each of my 3 drives, I've also included a screen shot of the Info tab.

There are a number of interesting patterns in the data.

The first 2 graphs show patterns similar in overall shape to the
benchmark,
but the 3rd graph is clearly flat. I should note that this 3rd drive is
an
external drive.

My first thought was to look for info on the transfer mode, DMA or PIO.
One
of the postings here said that the transfer mode would be indicated on the
Info tabs, but I'm not seeing it.

I also noticed that the yellow scatter shot points used to indicate Access
Time are much more tightly spaced in the 2nd drive than in the 1st. I am
wondering what this tells me about my drives.

Finally, in Paul's benchmark plot, I see that the SCALE of his drive is
way
different than that of my drives. This is reflected in the results
numbers,
too, with the benchmark having an average Transfer Rate of about 100 MB/s,
with my drives coming in at 50 MB/s or less.

I'm just guessing, but to my inexpert eye, it looks like this is telling
me
that Paul's benchmark drive is working twice as fast as my own drives.
And
this is a big deal to me, as my brother has often talked about how drive
access times are often the true bottleneck in any computer system.

So I guess I'd like for Paul to tell me what kind of drive he used for
this
test. THIS may actually turn out to be the key to my problems.

Andy

Your hard drives are running hot which could affect performance, and by the
look of the numbers, something surely is..

Mount the drives further apart from each other, and if you can't do that,
change the computer case.. also look at case cooling..

 

[Andy_XP_Devotee]

These are fantastic links you've included here. I've only skimmed them, but
I am very eager to work through them all. This will take me some time,
though, as most of what I'm seeing requires that I proceed carefully. I will
post my results back here as I go.

Thank you,

Andy

 

[Paul]

Your hard drives are running hot which could affect performance, and by
the look of the numbers, something surely is..

Mount the drives further apart from each other, and if you can't do
that, change the computer case.. also look at case cooling..

I noticed that too. The drives (at 46C) are on the warm side. That
could contribute to a shorter operating life. The FDB bearing is
frictionless, until the oil evaporates. So heat isn't the best thing.

My drives currently (two 80GB IDE drives), are running 26C and 27C. There
is a fan mounted in front of them. My room temperature is 20C. In fact,
on my computer, that fan is the only fan on the case (an intake fan).
My computer case has no exhaust fan (except for the weak exhaust on
the PSU, and that doesn't really count).

The first drive in the picture, has a few scatter points that may be
indicating a sector substitution was done, or perhaps a seek to zero,
followed by a seek to the track in question. The first two drives have
transfer rates consistent with the older IDE drives I'm using.

(The sample plot I provided, is for the Velociraptor, the fastest SATA drive
you can get. There are SAS drives which are faster, but those draw
50% more power as well. The Velociraptor manages 120MB/sec near
the beginning of the disk. And at 10KRPM, the seek time will be
less than for a 7200RPM drive.)

Since we've been told the third drive is an external, the 31MB/sec
is consistent with USB2 or the like. It looks like a decent enclosure
is being used. There are some USB2 adapters which only manage half
that rate (about 16MB/sec). Being USB, you wouldn't expect stuff to
load or store quite as fast on there. Not the fault of the drive,
but the limits of USB2 performance.

One reason for the transfer rate to not show up in the Info tab, is
if a "pseudo-SCSI" driver is being used for the IDE interfaces. If
a hardware company wants to provide their own driver, the preferred
method is to make a driver that present a SCSI interface to the system.
That involves a CDB (command/data block), with the command to execute.
It is the driver's job then, to convert that SCSI CDB into an IDE specific
operation.

SCSI drivers allow the hardware manufacturer to implement caching for
the driver if they want. That would be the positive aspect of a hardware
manufacturer driver. On the minus side, there is the "information hiding"
aspect of the driver. System tools are less likely to be able to look
at features on the drive. The system can read and write just fine, but
some other details will tend to be hidden.

I don't see anything terribly wrong with the graphs. I might check the
SMART data on the first drive (Maxtor 6L300SD), to see if there
are any reports of trouble ahead. The high seek time data points
concern me a little bit, and since drives are cheap, I might swap
that one out. C: is on there, so perhaps the drive can be
retired from OS duties and used for something else.

(Speedfan offers a SMART data readout, assuming the interface
supports "passthru" of SMART commands. Select the disk and then
the SMART statistics should be shown in the window. If SMART
cannot be displayed, either the drive doesn't support SMART,
or the driver is not able to pass thru the SMART command needed.)

http://www.almico.com/speedfan437.exe

You can move the old partitions to the new drive, with a tool
provided by the manufacturer of the new disk. The download
page of each manufacturer, has some level of support, ranging
from "buzz off", all the way to providing a copy of an
Acronis tool.

http://www.seagate.com/support/discwizard/dw_ug.en.pdf

http://www.hitachigst.com/hdd/support/download.htm ("buzz off")

Make sure the old C: drive is disconnected, before booting
with the new drive the first time. After the first boot
cycle is complete, the old drive can be connected again.

I see three options to replace the C: drive

1) Purchase a new IDE drive. Simplest to install, but the
pickings are beginning to get thin for IDE. Drives with
faster (head-rate-limited) transfer rates now, are SATA.

2) Purchase a SATA drive and a SATA to IDE dongle. (Dongles
come in two flavors, so make sure the direction of adaptation
is correct.) But SATA to IDE dongles tend to suck, in terms
of getting one that works. If you're ordering online, you
might hedge your bets, by doing both (2) and (3).

3) Purchase a SATA drive and a PCI SATA adapter card. The problem
here, is finding a PCI card that you don't have to flash the BIOS
chip on it. There are a couple cards with VIA VT6421 chip on
them, and they look to work. Cards with SIL3112 or SIL3124
appear to have trouble recognizing 1TB SATA drives.

The first option is trouble free, but doesn't offer the most
performance improvement. Depending on your luck and skills
at dealing with crappy hardware, option (2) or (3) might work
out.

This one will put the old drive to shame (but it is SATA).
If you had a SATA connector on the motherboard, this would be a
winner for a C: drive. Gets somewhere between 95MB/sec and 110MB/sec
near the beginning of the disk. If you have an older computer
with only IDE, then connecting it could have its challenges.

Western Digital Caviar Black WD5001AALS 500GB 7200 RPM 32MB Cache $70
http://www.newegg.com/Product/Product.aspx?Item=N82E16822136320

Paul

 

[Andy_XP_Devotee]

Daave, . . .

Your suggestion about using the Device Manager was great. Another of the
postings had suggested that, too, but--ahem--when I'd tried to do this, I'd
gotten stuck.

You explicit note about double-clicking on the IDE Channel was what I
needed, as I hadn't previously even realized that double-clicking on these
entries would do anything! So, a whole new level of windows for me . . .

I got to the Transfer Mode info, and it's set to Ultra DMA Mode 2.

That's not PIO, which is good. But would you, by chance, be able to tell me
what it means that I'm using the DMA Mode 2 rather than the Mode 5 your
machine is using? Maybe Mode 2 just reflects the age of my drives?

Also, does this DMA transfer mode mean that *all* of my drives are working
this way? Including an external drive? (The external drive is recognized by
XP as another drive and not any kind of portable storage device.)

From what Paul said earlier, a flat-line graph from HD Tune indicates the
PIO transfer mode, and I definitely got a flat-line graph when I had HDTune
test my external drive. When I look at the Device Manager, I am not seeing a
entry that is obviously for an external drive. But I *do* find, under
Universal Serial Bus controllers, an item called USB Mass Storage device.
Double-clicking on the Mass Storage device, though, did not give me a tab
with transfer mode info.

Andy

 

[Andy_XP_Devotee]

RE: Mike Hall - MVP's posting, . . .

I wondered about the temperature numbers when I saw them in the HDTune
report. I noticed them because they were so different from Paul's numbers.

I do not know all of the details, . . . but I know that my Dell was built
with a P4 chip that ran very hot. I believe Intel only manufactured this
chip for 1 year. Because the chip runs so hot, the cooling fans often run
fast and loud. In fact, it was the noise of those fans that resulted in my
having this machine at all. This was, originally, my brother's home machine.
When the fan noise finally drove him crazy, he gave the computer to me.
(And a bit of noise was a trivial price to pay for what has been a really
nice computer.)

I have wanted to monitor the internal temps in this machine since I got it,
but, again, it's a Dell, and the usual temperature-monitoring programs don't
usually work. In fact, I think my brother said he crashed this machine
trying to get it to accept a 3rd party temperature ap.

The numbers in HD Tune were the first temperature numbers I'd ever seen for
this machine. They're for the HD and not the chip, but I was still glad to
see them.

As for cooling down this machine, . . . I will see what I can do. The
drives are not always as hot as the 46 degrees C you saw in the graph. I've
checked since then, and they start out at about 32C.

At what temperatures would you suggest I should think about shutting down
and letting them cool off? I may not be able to do anything about separating
the drives, as I believe they're already spread out as far as they'll go.
I've thought about replacing the fans with better/quieter fans, but, again,
this being a Dell, I'm not sure a replacement fan will even fit without
modding the case.

So I may wind up just shutting down this machine when it gets too hot. A
number would be very helpful.

Thanks,

Andy

 

[Paul]

Daave, . . .

Your suggestion about using the Device Manager was great. Another of the
postings had suggested that, too, but--ahem--when I'd tried to do this, I'd
gotten stuck.

You explicit note about double-clicking on the IDE Channel was what I
needed, as I hadn't previously even realized that double-clicking on these
entries would do anything! So, a whole new level of windows for me . . .

I got to the Transfer Mode info, and it's set to Ultra DMA Mode 2.

That's not PIO, which is good. But would you, by chance, be able to tell me
what it means that I'm using the DMA Mode 2 rather than the Mode 5 your
machine is using? Maybe Mode 2 just reflects the age of my drives?

Also, does this DMA transfer mode mean that *all* of my drives are working
this way? Including an external drive? (The external drive is recognized by
XP as another drive and not any kind of portable storage device.)

From what Paul said earlier, a flat-line graph from HD Tune indicates the
PIO transfer mode, and I definitely got a flat-line graph when I had HDTune
test my external drive. When I look at the Device Manager, I am not seeing a
entry that is obviously for an external drive. But I *do* find, under
Universal Serial Bus controllers, an item called USB Mass Storage device.
Double-clicking on the Mass Storage device, though, did not give me a tab
with transfer mode info.

Andy

What a flat line in HDTune indicates, is the cabling/interface method is
limiting performance. If the cabling/interface is fast enough, then
the curved form is seen instead. The curved form tells you the head to
platter interface is the limitation.

When you have a USB2 interface on the drive, the limitation is in the
30MB/sec range. That is lower than what the head to platter is capable of.
USB2 is used for its convenience, not its transfer performance. If
you wanted performance on an external drive, you'd use ESATA. Something
which not a lot of machines are going to support out of the box. If you want
to switch to ESATA, it might mean the purchase of a different external
enclosure, and adding a card like this to a spare slot.

SYBA SD-SATA-1E1I (SIL3512, one ESATA, one internal SATA, probs with some 1TB drives) $12.99
http://www.newegg.com/Product/Product.aspx?Item=N82E16816124013

An external drive with ESATA or USB2 interface, and actual cooling $129
ESATA means no "flat line". USB2 means convenience for computers without ESATA.
Don't use both interfaces at the same time, only one or the other.

http://www.newegg.com/Product/Product.aspx?Item=N82E16822161022

Ultra DMA Mode 2 is seen on optical drives. If they don't need to transfer
data faster than that, then that interface setting is fine. If that entry
is a hard drive, then unless the drive is quite old, it may be possible to
fix it. Different interface rates can be mixed on the same cable, so
an Ultra DMA Mode 2 optical can be on the same cable as an Ultra DMA
Mode 5 hard drive.

Your first stop, with "Ultra DMA Mode 2" on a hard drive, is a cabling check.
There are two kinds of ribbon cables for IDE. The 40 wire cable has slightly
larger wires. The 40 wire cable might typically have been used in
computers, for one or two optical drives. But the 80 wire cable
(thinner wires, every second wire is a ground wire), is a preferred
cabling method. You can change the 40 wire cable to 80 wire, and then
go back to fixing Windows so that the full rate is available. The driver
for the IDE interface, can, according to the ATA/ATAPI standard, detect
most of the time, the presence of 40 versus 80 wires. The driver
can then artificially limit the transfer rate, to match the crappy
transmission quality of a 40 wire cable.

When an IDE interface "cranks down" to PIO, that is a gradual process.
The drive may start at Ultra DMA Mode 5 when the OS is first installed.
If CRC errors are detected, the OS has the ability to crank down
the interface one notch. That is an attempt to dynamically
adjust for error free operation. But once it drops to the bottom
end of the scale, you end up at 4MB/sec, with a flat line and
a sad look on the user's face. The "Workaround" procedure, is
intended to reset to the top rate - if the CRC problem
is still present, the "crank down" will happen all over again.
The driver is not designed to "crank up", when the "weather is fine".

I didn't see anything in your transfer curves, that suggested a
transfer limitation on the internal drives. Only your USB2 is
limited, which should not be a surprise due to the top end limit
of USB2.

Some pre-built computers use BTX formfactor design. They
may use one big fan, which draws air through the computer
case, and through the fins of the CPU cooler. Some of these
use simply humungous capacity fans (more than 100CFM), because
they have a hot processor to attempt to cool. They spool up
in response to internal sensors (and can sound like your vacuum
cleaner, when run at max). The hard drives in such a computer
case, may have small vents near them, intended to draw a little
cooling air over the hard drive. Examine the computer case,
to see if the hard drives are instead, packaged in a dead zone.
I've seen plenty of poorly designed solutions, so a drive
that heats up on usage is not a surprise. The fact they last so
long is the surprise. Retrofitting a bad design, is not always
easy. (In my current computer, I built a metal frame external to
the case, to hold a fan in close proximity to the hard drives.
That is why they're so cool. It required drilling four holes
in the plastic fascia, to hold the frame there.)

Paul

 

[Mike Hall - MVP]

RE: Mike Hall - MVP's posting, . . .

I wondered about the temperature numbers when I saw them in the HDTune
report. I noticed them because they were so different from Paul's
numbers.

I do not know all of the details, . . . but I know that my Dell was built
with a P4 chip that ran very hot. I believe Intel only manufactured this
chip for 1 year. Because the chip runs so hot, the cooling fans often run
fast and loud. In fact, it was the noise of those fans that resulted in
my
having this machine at all. This was, originally, my brother's home
machine.
When the fan noise finally drove him crazy, he gave the computer to me.
(And a bit of noise was a trivial price to pay for what has been a really
nice computer.)

I have wanted to monitor the internal temps in this machine since I got
it,
but, again, it's a Dell, and the usual temperature-monitoring programs
don't
usually work. In fact, I think my brother said he crashed this machine
trying to get it to accept a 3rd party temperature ap.

The numbers in HD Tune were the first temperature numbers I'd ever seen
for
this machine. They're for the HD and not the chip, but I was still glad
to
see them.

As for cooling down this machine, . . . I will see what I can do. The
drives are not always as hot as the 46 degrees C you saw in the graph.
I've
checked since then, and they start out at about 32C.

At what temperatures would you suggest I should think about shutting down
and letting them cool off? I may not be able to do anything about
separating
the drives, as I believe they're already spread out as far as they'll go.
I've thought about replacing the fans with better/quieter fans, but,
again,
this being a Dell, I'm not sure a replacement fan will even fit without
modding the case.

So I may wind up just shutting down this machine when it gets too hot. A
number would be very helpful.

Thanks,

Andy

I like to see mid 30's..

To be honest, I would look for a better case. Dells are BTX form factor so
you have to look for a BTX compatible case. Some are convertible. Don't get
a mini tower as you will have the same problem with space to spread out. If
you can't go that route for whatever reason, look to upgrade the fans..

 

[Andy_XP_Devotee]

Gerry, . . . this isn't a big deal, just a detail that's not fitting quite
right.

The HD Tune info that you copied to your posting looks somewhat similar to
what I'm seeing on the HD Tune Info tab that I'm looking at on my machine.
But it's not quite the same. I'm wondering if maybe you're using a different
version than what I downloaded yesterday. Did you look at the image file I
uploaded with the earlier posting? If you have the time to fool with it, I'd
be interested in your comments after you've looked at the file. I don't
think you'll see anything about the Mode.

I was able to find this info using the Device Manager (Daave's posting
above). My machine is currently using Ultra DMA Mode 2.

So I'm not still wondering about the transfer mode at this point. I'm just
curious as to how it is that you get this info from HD Tune and I don't.

Andy

 

[Andy_XP_Devotee]

Thank you for the info about Sandra. I wasn't looking for this, but I
really like these tests that let me see what's going on with my machine.
Maybe just for fun; I dunno! Andy.

 

[Andy_XP_Devotee]

Paul, I *really* appreciate your taking the time to type out all of this
info on the drives. I know you must have lots of other things to do. I've
gotten some really excellent help here from Daave and Gerry and the others,
but I lucked out in finding somebody so interested in drives.

And, just so everybody will know, I've copied and saved the replies I've
gotten here. Some of what I've been told I'll have to save for later, but
some of it has been so well tuned to what I needed that I'm reluctant to risk
losing track of it.

As for the specifics, . . .

After reading what everybody has said here, and after having done a few
checks on my system, . . . I'm now thinking that maybe the bottleneck in my
system is my drives. I still have a few more of the tests to do for malware
and such, but, so far, my computer has come up clean and no obvious
deficiencies have been found.

My goal when I asked here about RAM and my drives was to understand what I
needed to do with my next machine to get the performance I wanted. I'm glad
to finally know for certain that RAM isn't the issue. I'm oversimplifying
here, I know, but I'm thinking now that what I need to do next time is get
some better drives.

As it happens, I'd just recently been admiring the WD Velociraptor drive
offered with one of the Maingear PCs. I had no idea that the graph you
(Paul) posted earlier was from that drive turning at 10,000 RPMs! You really
weren't kidding when you said that the graph posted was the benchmark.

'Course, I'm not going to pay Maingear $420. for the drive; Newegg's got it
for $220.

I'm hoping to put together a new machine this fall. Between now and then, I
really won't have much time to use a new machine (or my present one), so the
delay is not a problem. My project, between now and then, is to configure
the machine I want. That might sound like a pretty easy job, but, for me, it
means reading a ton of reviews and learning a lot of new things, as I don't
begin to keep up with computer hardware. (I'm sure I'll be back here with
more questions.) Remember Tom's God Box? I think that was the name. I
found that listing of hardware really helpful, even if I didn't buy any of it.

At this point, I'd like to ask two follow-up questions.

(1) Do you have a reference for a computer model that breaks down the
computer's performance by component? That might sound like an impossible
model to put together--or maybe a useless model--considering the many
different ways that computers can be used. But I'm betting that, in fact,
people have made such models. I wouldn't be surprised to find a model like
that, for example, in an introductory Computer Science textbook. And I'm at
a point where I'd like to sit back, so to speak, and think about some of the
big picture issues before I get started. (If I've gone too far out on the
limb here, it's OK. But I wanted to ask.)

(2) Regarding hard drives in particular, . . . I had the impression that
one good-quality drive turning at 7,200 RPMs was more or less as fast as the
next good-quality drive turning at 7,200 RPMs. Is that right or not? And if
it's not, where should I look for the numbers that will let me compare the
performance of different drives?

Time, now, for a couple hours' sleep!

Andy

 

[dadiOH]

I also noticed that the yellow scatter shot points used to indicate
Access Time are much more tightly spaced in the 2nd drive than in the
1st. I am wondering what this tells me about my drives.

Finally, in Paul's benchmark plot, I see that the SCALE of his drive
is way different than that of my drives. This is reflected in the
results numbers, too, with the benchmark having an average Transfer
Rate of about 100 MB/s, with my drives coming in at 50 MB/s or less.

I'm just guessing, but to my inexpert eye, it looks like this is
telling me that Paul's benchmark drive is working twice as fast as my
own drives.

Check out the specs of his drive and you will see why.

--

dadiOH
____________________________

dadiOH's dandies v3.06...
....a help file of info about MP3s, recording from
LP/cassette and tips & tricks on this and that.
Get it at http://mysite.verizon.net/xico

 

[Gerry]

Andy

For some reason the SMART facility is not working as it should on your
computer. No information appears where it should i.e. after the words
"Supported" and "Active" in your screenshots.

Your statement "My machine is currently using Ultra DMA Mode 2." is
misinterpreting what I have read from your other replies. Ultra DMA Mode 2
applies to your external drive because of the limitation imposed by being
connected to your computer via a USB connection.

Daave's post quote "Open Device Manager. Then expand the entry for IDE
ATA/ATAPI controllers. Double-click the IDE Channel entries. Click on the
Advanced Settings tab. Ignore the ones that say Not Applicable. Hopefully
the value for current transfer mode is Ultra DMA Mode 5 (at least, that's
what mine is). If it says PIO, you need to fix it!" gave you an alternative
way to get the required information. If you are saying both the primary and
secondary ide channels both say Ultra DMA Mode 2 then this is where you have
a performance bottleneck!


--


Hope this helps.

Gerry
~~~~
FCA
Stourport, England
Enquire, plan and execute
~~~~~~~~~~~~~~~~~~~

 

[Daave]

Paul, I *really* appreciate your taking the time to type out all of
this
info on the drives. I know you must have lots of other things to do.
I've
gotten some really excellent help here from Daave and Gerry and the
others,
but I lucked out in finding somebody so interested in drives.

Glad to help you out and I am sure Gerry is, too. And it looks like Paul
is on the mark with regard to your hard drives and transfer modes (and I
believe temperatures).

I have found an excellent news group for those intereseted in building
their own PCs, which you also mentioned you'd like to do. Unfortunately,
Microsoft filters posts that contain the name of this newsgroup (at
least they used to), and I have no idea why!

Anyway, if you'd like to search the archives, here's a Google Groups
link:

http://tinyurl.com/cqcym5

Since you are using Microsoft's Web interface, that group is unavailable
there. You will either need to use Google (or other Web interface) or a
news reader (like Outlook Express, Thunderbird, Agent, etc.), which is
always the preferred method to access newsgroups. For more information,
see:

http://www.elephantboycomputers.com/page2.html#Tech_Support

and scroll down to C. Usenet.

Also see this page:

http://michaelstevenstech.com/outlookexpressnewreader.htm

Then it will be easy to access the particular newgroup:

@lt.comp.hardware.pc-homebuilt

(a changed to @ to avoid the filtering!)

 

[Paul]

Paul, I *really* appreciate your taking the time to type out all of this
info on the drives. I know you must have lots of other things to do. I've
gotten some really excellent help here from Daave and Gerry and the others,
but I lucked out in finding somebody so interested in drives.

And, just so everybody will know, I've copied and saved the replies I've
gotten here. Some of what I've been told I'll have to save for later, but
some of it has been so well tuned to what I needed that I'm reluctant to risk
losing track of it.

As for the specifics, . . .

After reading what everybody has said here, and after having done a few
checks on my system, . . . I'm now thinking that maybe the bottleneck in my
system is my drives. I still have a few more of the tests to do for malware
and such, but, so far, my computer has come up clean and no obvious
deficiencies have been found.

My goal when I asked here about RAM and my drives was to understand what I
needed to do with my next machine to get the performance I wanted. I'm glad
to finally know for certain that RAM isn't the issue. I'm oversimplifying
here, I know, but I'm thinking now that what I need to do next time is get
some better drives.

As it happens, I'd just recently been admiring the WD Velociraptor drive
offered with one of the Maingear PCs. I had no idea that the graph you
(Paul) posted earlier was from that drive turning at 10,000 RPMs! You really
weren't kidding when you said that the graph posted was the benchmark.

'Course, I'm not going to pay Maingear $420. for the drive; Newegg's got it
for $220.

I'm hoping to put together a new machine this fall. Between now and then, I
really won't have much time to use a new machine (or my present one), so the
delay is not a problem. My project, between now and then, is to configure
the machine I want. That might sound like a pretty easy job, but, for me, it
means reading a ton of reviews and learning a lot of new things, as I don't
begin to keep up with computer hardware. (I'm sure I'll be back here with
more questions.) Remember Tom's God Box? I think that was the name. I
found that listing of hardware really helpful, even if I didn't buy any of it.

At this point, I'd like to ask two follow-up questions.

(1) Do you have a reference for a computer model that breaks down the
computer's performance by component? That might sound like an impossible
model to put together--or maybe a useless model--considering the many
different ways that computers can be used. But I'm betting that, in fact,
people have made such models. I wouldn't be surprised to find a model like
that, for example, in an introductory Computer Science textbook. And I'm at
a point where I'd like to sit back, so to speak, and think about some of the
big picture issues before I get started. (If I've gone too far out on the
limb here, it's OK. But I wanted to ask.)

(2) Regarding hard drives in particular, . . . I had the impression that
one good-quality drive turning at 7,200 RPMs was more or less as fast as the
next good-quality drive turning at 7,200 RPMs. Is that right or not? And if
it's not, where should I look for the numbers that will let me compare the
performance of different drives?

Time, now, for a couple hours' sleep!

Andy

In my wanderings yesterday, I ran into an article with an exhaustive
examination of a hard drive. This is a little too much info for
making purchasing decisions. But it does show all the nuances you
can find, if you look hard enough.

http://www.xbitlabs.com/articles/storage/display/1tb-14hdd-roundup.html

*******
The following is a simplified view of a computer. Sure, there are
pathological cases you can make up, to refute every statement in
the following. This represents what I see in everyday usage.

For your typical program, once the program is loaded, the CPU and memory
determine the performance. If you examine the impact of memory speed
and latency, the effects there are relatively small. That leaves the
CPU as the single determinant of performance while the program is
running. For a given family (i.e. comparing similar Core2 processors),
comparing the clock speed would be enough. For comparing between families,
then you need benchmarks. Core2, for example, can retire more instructions
per clock tick, than the P4. Which means it can do more work per cycle
than the P4. The enormous caches on the processor, mean not a lot of
instruction memory requests have to go to main memory. (Thus, I don't
get alarmed, when my old P4 ran at 3.1GHz and my new Core2 runs at
2.6GHz. The Core2 is much faster than the P4, even comparing single
threaded benchmarks. Clock rates don't compare between different
generations, which is why benchmarks are necessary.)

( 24.05 seconds on Core2, 45 to 50 seconds on my P4 or AthlonXP systems, SuperPI benchmark.
Notice the difference the clock rate can make. The world's record for SuperPI with
the settings used here, is somewhere in the 8 second range.)
http://groups.google.ca/group/alt.comp.hardware.overclocking/msg/5f98cd237d534dba?dmode=source

The Core i7 adds to that, a memory interface right on the processor,
the same way that AMD does it. That eliminates some of the latency
getting to main memory. But is nothing compared to the bandwidth
and latency of the internal caches. (The Sandra benchmark shows
how effective those can be, compared to main memory.)

In terms of price performance, these are some possibilities.

*******
Fastest dual core. You would buy this, if you don't do a lot of multimedia
that has multithreading capability. I use a dual core, because I don't own
a lot of modern software, and the older software tends to use one core.

Intel Core 2 Duo E8600 Wolfdale 3.33GHz 6MB L2 Cache LGA 775 65W Dual-Core $270
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115054

*******
Buying this one, means getting decent pricing on motherboard and RAM (compared
to Core i7). You're not getting gouged on the infrastructure. This is equivalent
to two silicon dies like the E8600, packaged inside the CPU (that is where the
12MB of cache comes from - two chips with 6MB each). The clock rate isn't as
high as the E8600. So if you had the older software, single threaded stuff,
then those apps might run a little slower on this one compared to the E8600.
But if you're rendering a movie in a video editor, this is the one to own.
If a program runs on all four cores at once, performance screams.
There are faster ones than this, but they cost too much. This
was outside my budget range, when I upgraded.

Intel Core 2 Quad Q9550 Yorkfield 2.83GHz 12MB L2 Cache LGA 775 95W Quad-Core $275
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115041

*******
Finally, the Core i7 with memory controller, in the LGA1366 package, is
the best there is. Again, I've selected the "cheap one", to get most
of the benefit, without the big price. Salt to taste (whatever you can
afford). Motherboard and RAM are part of the overall system expense.
The cache structure on this is three level (similar to AMD Phenom II).
The silicon is a single die (they don't clamp two together to make these).

Intel Core i7 920 Nehalem 2.66GHz 4 x 256KB L2 Cache 8MB L3 Cache LGA 1366 130W Quad-Core
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115202

For charts, try this. Virtually every test here is multithreaded, so
most of the time you're seeing the benefit of multiple cores. There
is hardly anything left here any more, which compares single core
performance for older software. SuperPI would have been nice, for
example. But these guys have an agenda, so don't expect them to be
dispassionate scientists like the Xbitlabs people.

http://www.tomshardware.com/charts/desktop-cpu-charts-q3-2008/benchmarks,31.html

HTH,
Paul

 

[Andy_XP_Devotee]

Gerry, hi.

No surprise about why SMART is not working on my machine. I'm using a Dell,
and this is one of the dumb Dell quirks. NONE of the temperature-sensing
programs will work on my machine. I have no idea why not, but this is one of
the known oddities for my particular Dell. (I don't know if this is true for
all Dells, many Dells, or just a few.)

Gerry, I looked again at the Device Manager after reading your note. Going
to the IDE ATA/ATAPI controlers, I found 3 listings, one of which was Primary
IDE Channel. This was the only listing that gave a properties box with an
Advanced Settings tab, so I'm doubly-sure I've gotten the right spot.

On the Advanced Settings tab, I've Device 0 and Device 1 (which must be my
two physical local drives). Both of them show a Current Transfer Mode of
"Ultra DMA Mode 2."

I still need to work on some of the material Paul gave me, but Paul pretty
much said (nicely) that I needed a new drive! Maybe you would agree?

What would be of great interest to me would be a general sense of just how
much this DMA Mode 2 is slowing me down.

There was discussion earlier about how a drive might slow down over time,
eventually winding up at PIO. When I later found this info about the DMA
Mode 2, I didn't think about applying a fix for it as had been discussed
earlier (Paul, again).

Do you happen to know if this DMA Mode 2 is something I can/should fix? Or
do you vote with Paul that I need to toss this drive and get a new one?!

I'm asking to get a better feel for the situation. I'm not actually going
to replace this one (not if don't have to, anyway), because I'm hoping it
won't be too long before I have a completely new machine.

What do you think?

Andy

PS: I have other postings here that I'd really like to answer, but the
Microsoft Site is running soooooo slowly tonight that it's wearing me out.
Maybe the site is busy because it's Saturday. But I will be back later for
sure. A.