HD Buffer Size

[Bob Simon]

Maxtor and Seagate both have 120GB 7200rpm EIDE drives with either 2MB
or 8MB buffer. (ST3120023A vs ST3120026A and 6Y120L0 vs 6Y120P0.)

Am I likely to notice a real-world performance difference from the
larger buffer with a single HD (plus CD and CR-RW) connected to an
ATA66 controller in a machine running Windows XP?

 

[Rod Speed]

Maxtor and Seagate both have 120GB 7200rpm EIDE drives with either 2MB
or 8MB buffer. (ST3120023A vs ST3120026A and 6Y120L0 vs 6Y120P0.)

Am I likely to notice a real-world performance difference from the
larger buffer with a single HD (plus CD and CR-RW) connected to
an ATA66 controller in a machine running Windows XP?

Nope. And not in a more modern PC either.

 

[Odie Ferrous]

Maxtor and Seagate both have 120GB 7200rpm EIDE drives with either 2MB
or 8MB buffer. (ST3120023A vs ST3120026A and 6Y120L0 vs 6Y120P0.)

Am I likely to notice a real-world performance difference from the
larger buffer with a single HD (plus CD and CR-RW) connected to an
ATA66 controller in a machine running Windows XP?

If you're working with smallish files, you *may* notice a difference.
Probably not.

However, the Seagate is infinitely more reliable than the Maxtor, so
that would be the better bet.


Odie

 

[dg]

Can you explain in a simple way, how is the buffer used for reading data
from the drive? I am guessing that when WRITING to the drive the buffer can
accept data as if it were written to the disk, and make small file writes
seem faster than they really are. But for reading, does the buffer
significantly help out?

THANKS!
--Dan

 

[Peter]

Maxtor and Seagate both have 120GB 7200rpm EIDE drives with either 2MB

or 8MB buffer. (ST3120023A vs ST3120026A and 6Y120L0 vs 6Y120P0.)

Am I likely to notice a real-world performance difference from the
larger buffer with a single HD (plus CD and CR-RW) connected to an
ATA66 controller in a machine running Windows XP?
Bob Simon

Well, buffer size is not the only difference.
For example comparing Seagate 26A with 23A:
- 8MB buffer vs. 2MB
- 3 heads vs. 4
- 85.4MBytes/sec internal tranfer rate vs. 570Mbits/sec
- 58MBytes/sec sustained transfer rate vs. 27-44MBytes/sec
- 8.5/8.5/9.5msec seek/read/write access time vs. 9.4/9.4/10.5msec

I would choose 26A, but real-world performance differences might be
relatively small.

 

[Rod Speed]

Can you explain in a simple way, how is the buffer used for reading data from
the drive? I am guessing that when WRITING to the drive the buffer can accept
data as if it were written to the disk, and make small file writes seem faster
than they really are. But for reading, does the buffer significantly help
out?

In theory at least the drive can read more than was asked for from
the platters on a read request, hoping that the next read request
will be for what it read from the platters without being asked for.
Then it can return the data for the second read request from the
buffer instead of having to wait for the platters to rotate the data
under the heads for the second request.

In practice its isnt that easy for the drive to work out what
might be requested next, so its all rather theoretical, particularly
with modern OSs which have multiple tasks reading much of the time.

 

[Bob Simon]

Well, buffer size is not the only difference.
For example comparing Seagate 26A with 23A:
- 8MB buffer vs. 2MB
- 3 heads vs. 4
- 85.4MBytes/sec internal tranfer rate vs. 570Mbits/sec
- 58MBytes/sec sustained transfer rate vs. 27-44MBytes/sec
- 8.5/8.5/9.5msec seek/read/write access time vs. 9.4/9.4/10.5msec

I would choose 26A, but real-world performance differences might be
relatively small.

Thanks. I also noticed those other differences when I compared the
spec sheets but I (mistakenly) thought that buffer size might be the
most significant difference since that's listed in the PriceWatch ads

I guess I should have asked which parameter is the most significant
indicator or real-world performance. Would that be sustained transfer
rate?

 

[Peter]

Well, buffer size is not the only difference.

Thanks. I also noticed those other differences when I compared the
spec sheets but I (mistakenly) thought that buffer size might be the
most significant difference since that's listed in the PriceWatch ads

I guess I should have asked which parameter is the most significant
indicator or real-world performance. Would that be sustained transfer
rate?

That would depend on real-world application, for a single user,
multimedia app - sustained transfer rate might be the most important
factor. For a multiuser database, that might be a maximum IO/sec.
In both cases 26A is better than 23A.

 

[Rod Speed]

Thanks. I also noticed those other differences when I compared the
spec sheets but I (mistakenly) thought that buffer size might be the
most significant difference since that's listed in the PriceWatch ads

I guess I should have asked which parameter is the most significant
indicator or real-world performance. Would that be sustained transfer
rate?

Nope, the rate the sectors move under the heads.

Unfortunately that isnt a tidy number usually listed on the manufacturer's web
site.

 

[Derek Baker]

Nope. And not in a more modern PC either.

http://faq.storagereview.com/tiki-index.php?page=BufferSizes

 

[Bob Simon]

Nope, the rate the sectors move under the heads.

Unfortunately that isnt a tidy number usually listed on the manufacturer's web
site.

Wouldn't that imply that for a given rpm, larger drives will give
better performance? I hope so because I just ordered an ST3200822A
from Fry. With the $50 rebate (that ends today) it was less than the
ST3120026A.

(BTW, I have read about problems with Seagate rebate fullfillment but
I am patient and persistant so I expect that this will work out ok for
me in 2 or 3 months.)

 

[Rod Speed]

http://faq.storagereview.com/tiki-index.php?page=BufferSizes

Dont believe it. Bet you wouldnt be able to pick it in a double
blind trial without being allowed to use a benchmark.

 

[Rod Speed]

Wouldn't that imply that for a given rpm,
larger drives will give better performance?

Its more complicated than that because the large drive may
have more platters and/or more tracks per platter surface.

I hope so because I just ordered an ST3200822A from Fry. With
the $50 rebate (that ends today) it was less than the ST3120026A.

The ST3200822A would be marginally faster, its got 50GB
per surface compared with 40GB for the ST3120026A.

(BTW, I have read about problems with Seagate rebate
fullfillment but I am patient and persistant so I expect that
this will work out ok for me in 2 or 3 months.)

Yeah, with an operation of that size, it should be possible to
do them over on the rebate if you follow the detail required
precisely and keep demanding your rebate if it doesnt show up.

 

[Derek Baker]

Dont believe it. Bet you wouldnt be able to pick it in a double
blind trial without being allowed to use a benchmark.

I do believe they got those results in their benchmarks: I trust
storagereview.com.
http://www.storagereview.com/php/be...&numDrives=1&devID_0=218&devID_1=219&devCnt=2

I agree about the subjective performance though: I didn't notice that my
ATA-100 drive with a 40MB/sec average transfer rate was running at 15MB/sec
recently.

 

[Rod Speed]

I do believe they got those results in their benchmarks: I trust
storagereview.com.
http://www.storagereview.com/php/be...&numDrives=1&devID_0=218&devID_1=219&devCnt=2

Sure, but what matters is how relevant the
benchmark is to the real world use of the drives.

I agree about the subjective performance though: I didn't notice that my
ATA-100 drive with a 40MB/sec average transfer rate was running at 15MB/sec
recently.

Yeah, thats what I meant.

 

[Arno Wagner]

Maxtor and Seagate both have 120GB 7200rpm EIDE drives with either 2MB
or 8MB buffer. (ST3120023A vs ST3120026A and 6Y120L0 vs 6Y120P0.)

Am I likely to notice a real-world performance difference from the
larger buffer with a single HD (plus CD and CR-RW) connected to an
ATA66 controller in a machine running Windows XP?

No. But you will notice the warranty difference on the Maxtor
drives: 3 years for the 8MB model, 1 year for the other one.

Arno

 

[Chuck U. Farley]

http://faq.storagereview.com/tiki-index.php?page=BufferSizes

I do believe they got those results in their benchmarks: I trust
storagereview.com.
http://www.storagereview.com/php/be...&numDrives=1&devID_0=218&devID_1=219&devCnt=2

I agree about the subjective performance though: I didn't notice that my
ATA-100 drive with a 40MB/sec average transfer rate was running at 15MB/sec
recently.

It all depends on what applications are being used as well. Would someone
using the box for websurfing, email and the occasional Word document notice
a difference? Probably not. Would someone using disk intensive video apps
that routinely use 50-100 gig files notice a difference? You bet.

 

[chrisv]

Maxtor and Seagate both have 120GB 7200rpm EIDE drives with either 2MB
or 8MB buffer. (ST3120023A vs ST3120026A and 6Y120L0 vs 6Y120P0.)

Am I likely to notice a real-world performance difference from the
larger buffer with a single HD (plus CD and CR-RW) connected to an
ATA66 controller in a machine running Windows XP?

Gosh, you're the first person to ever ask that question!

Ever hear of Google, moron?

 

[Rod Speed]

It all depends on what applications are being used as well. Would someone
using the box for websurfing, email and the occasional Word document notice
a difference? Probably not. Would someone using disk intensive video apps
that routinely use 50-100 gig files notice a difference? You bet.

I doubt it on that last.