USB2.0 first experience

[Mercutio]

Had to buy a USB2 enclosure today.

Well, I didn't have to. But I got an expense authorization for one, so I did.
So I have a USB2 enclosure and a USB2 controller. I stuck an old 13GB 5400rpm Maxtor drive I had laying around in it.

Let me tell you: It's fast. Fast enough to make me forget I'm dealing with an external IDE drive (after years of using parallel, I guess its a given that external IDE generally sucks). I know the thing offers more bandwidth (50MB/sec) than that little 5400rpm drive can deliver, but honestly, it doesn't seem any worse than that same 5400rpm drive stuck inside a case.

Down on USB1.1, yup, it's pretty pokey. Slow as I'd expect. Yuck.

Still, I'm surprised by the difference. It's really nice.

 

[CougTek]

Although it might not be noticeable from a user experience, could you run something like Winbench to compare the performances of both (internal and external USB2)? I'd like to know if the access times are affected. Your drive is too slow anyway to measure possible sequential transfer rate differences. I know others have found that some implementation of USB2 are limited to ~27MBps, but a 13GB 5400rpm drive cannot reach that rate.

I think NEC has or is close to release an USB 2.0 chip with significantly better performances than their original one. At least, that's what they claim.

 

[Mercutio]

The motherboard in my game machine has the 1st-generation NEC/Via USB2 chip.

Yeah. I'm waiting for Clocker to ship his 10GB DM40+ out to me so I have a good spare drive from which to run tests.

Although, now that I think about it, I don't have a copy of Winbench, either. Never much cared to do benchmarks myself.

 

[Clocker]

The motherboard in my game machine has the 1st-generation NEC/Via USB2 chip.

Yeah. I'm waiting for Clocker to ship his 10GB DM40+ out to me so I have a good spare drive from which to run tests.

Although, now that I think about it, I don't have a copy of Winbench, either. Never much cared to do benchmarks myself.

If all goes well with my delivery (should get the burner today) I should be able to ship out your DMP40 on Thursday....

Clock

 

[Prof.Wizard]

USB 2 is indeed cool. But what if FireWire 2 arrives sometimes next year? I still love USB and I'm so glad so many companies have implemented regarding anything from joysticks to UPSs...

 

[Pradeep]

I think both USB2 and Firewire have their place in the market. Currently I have a combo card with both but only use the Firewire for an external hard drive. USB2 will be good due to backwards compatibility with USB1.1

Firewire will continue to be used for DV. We will see Firewire 2 before the end of this year. It is also one of the 2 possible connection methods between cable HDTV decoders and the HDTV monitor, the other being DVI.

 

[timwhit]

Pradeep, what card do you use that has both Firewire and USB2? Does it work well?

I want the next motherboard that I buy to have both Firewire and USB2.0 ports on it. Since I tend to keep a motherboard for at least 18 months. Are there any boards on the market right now that have both kinds of ports?

 

[time]

Good luck! I was wondering if there even are any shipping motherboards with embedded USB2.0?

 

[Pradeep]

I'm using an OrangeMicro Firewire + USB 2 combo pci card. Firewire section works fine without any drivers, the USB 2 section required the beta win2k drivers to work.

I believe Abit have a mobo (Max?) with both usb2 and firewire on board, but it gets rids of legacy ports such as serial (and parallel?) and has only 2 or 3 PCI slots.

 

[Cliptin]

Motherboards with USB2:
ASUS a7v333
MSI KT3 Ultra-ARU (MS-6380E)
MSI K7T266 Pro2-RU
Asus A7M266-D

Motherboards with USB2+FW:
ASUS a7v333-pa-raid

There are probably others!

 

[Mercutio]

I have a Gigabyte GA-7VRXP with USB2. About half the shipping KT333 boards seem to have it.

Firewire is undeniably cool; being able to do cool stuff like video editing directly between two cameras is really awesome. I'd feel a lot better about 1394 if it was an onboard standard.
The other side of that is that is that firewire represents a standard for digital encryption. Maybe it's good that it isn't widely adopted.

I'm going to try Clocker's DM40+. I might try a WD800AB if I get one in before C's drive comes (still 5400rpm, but it's within spitting distance of a Barracuda ATA IV in most of SR's tests).

Clock trials using the same PC
USB 1.1: 1.85GB to host PC = 25 minutes
USB 2.0: 1.85GB to host PC = a little under 3 minutes

 

[Pradeep]

Run ATTO on it and see how it compares to Firewire

 

[CougTek]

What we would need is an external enclosure that features both USB 2.0 and Firewire ports, as well as two controllers (one for each standards).

Nonetheless, from Pradeep's ATTO graph, it seems obvious that an external Firewire drive would be fast enough to be usable for most people.

 

[Prof.Wizard]

I might try a WD800AB if I get one in before C's drive comes (still 5400rpm, but it's within spitting distance of a Barracuda ATA IV in most of SR's tests).

That was for me to see, right Merc? :x

 

[time]

Actually, he's right.

Of course, that means it's within licking distance of a D740X.

So, Eugene wants us to believe that a drive with 33% more latency than another performs roughly the same. I guess there's really no point wasting time with 10,000 or even 15,000rpm drives then. Bring back BigFoot.

 

[Tannin]

It's difficult but not at all impossible to find a drive with 33% higher latency that can outperform a given higher RPM model, Time. All you need is to have one or other of two things:

(1) the low RPM drive has outstanding seek performance not normally expected within its class

or (2) the high-RPM drive is a slug by the standards of its class.

Consider, say, the Barracuda ATA-III (a slug) vs the Spinpoint V1020 (best in class seeks). Very, very little to choose between those two drives. Except that we sold 10 or 20 times as many Spinpoints and RMAed a half-dozen less of them. (i.e., 6 - 0 = 6).

There are probably several other similar comparisons to be made of equal validity. Was the second-last of the Quantum 7200s not an especially fast seeker, a drive that might well compare favourably with a sluggish 10,000 RPM unit?

Or, better example, Quantum Fireball LCT vs Seagate U Series slug. 4400 RPM was at least the equal of an outstandingly slow 5400.

 

[time]

Tony, go to SR and do a 'head-to-head' comparison between WD800AB, Barracuda IV and D740X, and you'll get a better understanding of what I'm saying.

They're all 40GB/platter drives, and the Seagate and WD at least have the same claimed seek times. Transfer rates are all comparable. Yet somehow the WD more or less overcomes its 1.4mS latency disadvantage.

Based on SR's bench however, the Maxtor's effective access was a blistering 8mS compared to 9.7 for Seagate. But I'm not sure that those tests are still valid with the tricks employed on current drives.

Perhaps this is the answer, from Eugene's review of the WD:

A while back in a discussion thread that has unfortunately been lost, several readers pointed out that the 60-gig WD600AB featured a "nominal" spindle speed of 5400 RPM and rotation latency that was not 5.56 milliseconds as one would expect from such a drive but rather "nominally" 5.0 milliseconds, a figure that would represent a 6000 RPM unit.

If true, this would reduce the WD's handicap to 0.8mS, maybe even less if Seagate's quoted seek time is optimistic. Doesn't explain why the Maxtor doesn't perform better though.

 

[CougTek]

Based on SR's bench however, the Maxtor's effective access was a blistering 8mS compared to 9.7 for Seagate.

This should be seek time I think.

A while back in a discussion thread that has unfortunately been lost, several readers pointed out that the 60-gig WD600AB featured a "nominal" spindle speed of 5400 RPM and rotation latency that was not 5.56 milliseconds as one would expect from such a drive but rather "nominally" 5.0 milliseconds, a figure that would represent a 6000 RPM unit.

I'm one of those who believe that the WD800AB is a 6000rpm drive instead of a 5400rpm. Impressive results and positive comments about this drive have been posted elsewhere than on SR so I believe the WD800AB is just as fast as Eugene wrote it is.

 

[Cliptin]

Run ATTO on it and see how it compares to Firewire

Touche!