D865PERLL's onboard SATA 1.5 - How to achieve better performance?

[zepherzs]

Is there anything I can do with my D865PERLL MB's onboard SATA 1.5 ports to help them achieve the best (or better) performance level than its stock 1.5 spec? ..just trying if I can to take greater advantage of what my ST3750330AS Hard Drive 3.0GB/s capability can do.

In other words are there any better Drivers (or something) that I can use other than with what came stock within my D865PERLL onboard SATA chipset drivers to achive better performance? ..without having to use RAID or IAA.

Here's a Screenshot using Both HDTach & HDTune for testing. Is this the best I can do with this MB?

Thank you.

[SRTDodge05]

With ata100, sata I and sata II is all pretty close in speed. I doubt you'll see any difference going from sata I to sata II.

[zepherzs]

ATA100 ? Are you saying that because of the bottom of the screenshot with HDTunes reading that I've circled?
Why is that there? Shouldn't it be higher than that?

I realize the bus is standard PCI which max can only give 133 GB/s, but with my onboard SATA controller I was under the impression it's Not connected through the PCI bus? I thought it's connected directly to the southbridge chip on my motherboard - which the throughput on that path is much higher than 133MB/s ? That's why I was hoping I had some headroom to do something with drivers or something? Isn't there something I can do?

[Acid8000]

But even then it's unlikely that the drive itself will be able to go much faster. Sure it may support a higher speed interface, but it'll be impressive to see the drive head actually able to deliver that sort of throughput constantly.

[zepherzs]

I'm not sure why you say that though and I think maybe the opposite is true, I don't think it's the HD that's the problem but the MB, since the HD Barracuda ST3750330AS 7200.11 3Gb/s 32MB cache 7200 rpm SATA NCQ interface is rockin... sure it may be hype market rated at 3 GB/s but we know in reality it's sustained data transfer rate is surly over 100 GB/s.. ..and I would take that. However if you look at the specs from my MB link it's my MB that's the hold-up here in the expected reality dept. I'm not able to update the MB and Processor etc right now... so I just want to do what I can in the meantime with hopefully some special drivers - or am I dreamin.

And so back to my question, or hope really, with what I've been asking already using different words, is there anything I can update to get a little more out of my MB's onboard 1.5 SATA Controller to help bring the results up a bit from what it shows my screenshot.

[ctal]

I agree with the others that the hdd's internal mechanical performance is the most significant limiting factor, and speeding up the interface would not produce much improvement.

I have the 150 gB Raptor and two WD640's in my new system (described in my profile), all connected as single drives into 3.0 gbps sata2 ports. HD Tune gives the following benchmarks:

150 gB Raptor: 85/74/55 max/avg/min mB/s; 8 ms access; 116 mB/s burst
(The minimum of 55 mB/s ignores a brief downward spike to 39 that occurs near the beginning of the drive)
WD640 no. 1: 109/87/52; 12.7; 153

WD640 no. 2: 113/91/53; 12.2; 152

The WD640's I understand to have one of the fastest sustained transfer rates of any desktop drives out there, and perhaps the fastest. Yet your numbers, with a different drive through the slower interface, are not much different, aside from the burst rate parameter.

Also, as you probably realize, sata1's 1.5 mbps = 187.5 mB/s, which seems far enough above the sustained transfer capability of any desktop drive to not have a major impact on overall throughput.

Finally, as you probably realize, the downward arc of the benchmark curves (resulting in average and minimum transfer rates that are well below the maximums) is due entirely to the fact that the inner portions of the platters move at slower tangential velocities than the outer portions do. The interface thus has no relevance to that downward arc, and any benefit of a faster interface would occur mainly near the beginning (outer portions) of the platters.

Regards,
-- Al

[CK8-04]

ctal pretty much sums up everything. The limiting factor in a hard drive is its mechanical properties, not its maximum throughput. Hard drives rarely even hit the maximum transfer rate. If you want a tangible increase in speed, take a look at a Raptor. You'd see better results with a faster spindle speed than a higher maximum transfer rate.

[zepherzs]

Okay then, well I learned something today.
Thank You CTAL for your informative detailed post, it's appreciated very much, and CK80-04, Acid8000, SRTDodge05 as well.

I'm surprised really, I just logically assumed that my new SATA 3 GB/s HD would be so much faster than my previous ATA 133 drive in this same rig; The ATA 133 was 63/51/24, and my sata is 106/86/51 ...which roughly estimating is about 30%...which I guess is something I should be thankful for you're all saying.

Answer me this if you would then, on the bottom, right hand portion, of the HDTune screenshot I showed in my first post, the part I had circled in red, the "Supported" listing shows UDMA 6 ATA/133 but beneath it for "Active" shows UDMA 5 ATA/100 - my question is what/why is the Mode 5 showing there instead of both listing being Mode 6 ? Can I fix? When I test my ATA 133 drives both those listings show up as Mode 6 so that's why I'm wondering why the SATA drive shows less than.

Thanks very much

[ctal]

I don't know the explanation for the UDMA5 indication, but I wouldn't worry about it. I have 5 sata hdd's in 2 systems, with 2 very different chipsets/mobos, and I see the same thing in HDTune on all of them. Probably some Google searching would reveal the reason, but I'd expect it to be non-problematic.

You underestimate the improvements you've realized! Your maximum and average sustained transfer rates have increased over 68% (106/63 and 86/51), and the minimum has increased 112.5% (51/24)!

Regards,
-- Al

[zepherzs]

Okay thanks ctal, I won't pay attention to the UDMA 5 listing there then.

As for the % increase I underestimated, and me doing the actual math yes I did didn't I; however that does not mean it's 100% or twice as fast because the real world test I did was to zip a 1 GB file and time it with each drive, and when doing it that way is what I saw as a 30% increase faster in finish times for the new drive.. ..which is still respectable.
I'm only running 1 GB of RAM with a 3.4e processor fwiw.

Thanks again for the feedback, it's much appreciated!

[ctal]

You're welcome! I'd imagine that the reason you only saw a 30% improvement zipping a large file is that lots of other things besides the hdd are taking significant amounts of time during the process -- cpu, memory, etc., and their contributions to the overall time required would be the same regardless of which hdd was in use.

Regards,
-- Al

[zepherzs]

Hi Al,

Well since you mentioned cpu, memory, it gave me idea to try with some extra new hardware that I have on hand to try for gp.
...and so I changed from a 3.4e processor down to a 3.2c processor just to see how much I would loose, and I was surprised (but maybe not too much) that there was no discernable difference at all in the 1 GB file copy results as well as with HDTunes results - so both confirm no performance gain there; and so then I thought I would get a little difference when I tried 2GB of matched RAM instead of just 1 GB that I had... but was surprised there was No discernable difference in the tests here too?

[ctal]

Assuming you are using XP, 2 gB RAM will make no difference relative to 1 gB unless you are running a LOT of things at once.

With respect to the cpu, 3.2 is less than a 6% reduction in clock rate from 3.4, which is pretty negligible. And as far as I recall there were no significant benefits to what the Prescott ("e" series) cpu's could do on a per clock basis, either, relative to the "c" series Northwoods.

-- Al

[zepherzs]

Thanks Al,

Then no matter if I updated my MB with the latest spec'd version or not, my performance as shown in my HDTune screenshot (which you said reads similar to your readings) is the best anyone can do? ..even if they have a fancy modern performance spec'd setup?

Thank you

[ctal]

Originally Posted by zepherzs

Thanks Al,

Then no matter if I updated my MB with the latest spec'd version or not, my performance as shown in my HDTune screenshot (which you said reads similar to your readings) is the best anyone can do? ..even if they have a fancy modern performance spec'd setup?

Thank you

It depends on what you are trying to optimize.

If you are trying to optimize the time it takes to open programs, then my feeling is that access (seek) time is the most important parameter. That would mean preferring a 10,000 Raptor, with its fast access times but relatively slow sustained transfer rates.

If you are trying to optimize processes that deal with very large files, then sustained transfer rate becomes more important than access time. That would mean using a large capacity 7200 rpm drive such as my WD640's or your Seagate.

My own approach has been to optimize both, by using a 10,000 rpm Raptor as the system ("C") drive, and high capacity 7200 rpm drives (such as the WD640's) as additional drives for storage of large files. That becomes particularly important doing video editing, where it is important to keep the large video files on different drive(s) from the one containing Windows and program files.

Then there is RAID (Redundant Array of Independent [or Inexpensive] Disks), which can be used to run two or more disks as, in effect, a single disk. That produces dramatic improvement in sustained transfer rates, but many people believe that the overall improvement in system performance is small and is not worth the many complications that are involved. Do a search on "raid" with Google or in this forum, and you'll find lots of information on it.

Hope that helps,
-- Al