SCSI 1 faster than USB1.1?

28 posts / 0 new
Last post
Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
SCSI 1 faster than USB1.1?

I'm considering picking up a used Visioneer portable scanner to hook up to one of my B&W's which has an Adaptec 29160n SCSI card. The card has an external HD50 SE SCSI connector. There's two Visioneer scanners I'm looking at--the Strobe Pro and the S-10. The Strobe Pro has a USB1.1 connection. The old S-10 Mac version has a SCSI DB25 connector. I'm wondering if it's possible that connecting the S-10 to the 29160n card could possibly be faster than the USB connection. I've already got an Epson 1640su scanner which has both HD50 SCSI and USB1.1 connections, and hooked up to the 29160n, the Epson scanner is SIGNIFICANTLY faster through its SCSI connection than its USB connection. I'm wondering if it's possible the S-10's SCSI 1 connection could in any way be faster than USB too. thanks

catmistake's picture
Offline
Last seen: 2 years 4 months ago
Joined: Dec 20 2003 - 10:38
Posts: 1100
I'd say yes

I'd go further to say... I think scsi-2 is faster than usb2. I think the usb adoption by the PC industry has really muddied the waters. Think of it like RMS & "max wattage" in the car audio industry. They are lying to us.

scsi1 gets 5MB/s

usb1 gets 12Mbits/s or ~ 1.5MB/s

pretty clear

Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
Are you saying it's 1.5MB/s U

Are you saying it's 1.5MB/s USB1 compared to 5MB/s SCSI 1? I had always thought it was 12MB/s USB and 5MB/s SCSI 1. Guess I need to go back to computers 101. Applefritter is such a constant education for me.

catmistake's picture
Offline
Last seen: 2 years 4 months ago
Joined: Dec 20 2003 - 10:38
Posts: 1100
yes, that's what I'm saying

its 12 megabits per second (12Mbit/s) for usb1

scsi1 is 5 megabytes per second (5MB/s)

What you thought is actually startlingly close to usb2 speeds.

Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
And firewire's MB/s is...?

And firewire's MB/s is...?

Jon
Jon's picture
Offline
Last seen: 12 years 10 months ago
Joined: Dec 20 2003 - 10:38
Posts: 2804
400 MBit/s == 50 MByte/s, so

400 MBit/s == 50 MByte/s, so it's slightly faster than Wide 20MHz SCSI, (20x16bit)/8bit per byte == 40MByte/sec, or also narrow 40MHz SCSI.

The USB2 speed of 480Mb/s is mostly theoretical as it's CPU driven and also is not a "streamlined" as FW, so there is a lot of overhead. FW has a dedicated controller chip, so it may not hit it's theoretical peak of 400Mb/s all the time, but it gets much closer than USB2, and SCSI is the same way as it has a dedicated controller too.

catmistake's picture
Offline
Last seen: 2 years 4 months ago
Joined: Dec 20 2003 - 10:38
Posts: 1100
should be

400Mbits/s

8bits = 1byte

maxes fw400 at 50MB/s & fw800 at 100MB/s

That's the theoretical limit. USB2 claims 480Mbit/s, so you'd think it would be faster than fw400. But its totally not.

In practice, from what I've seen, fw400 hits between 15MB/s - 30MB/s. Others may have seen faster, but that's what I see. USB2, in practice, is 8-15MB/s. Again... this is my experience. YMMV.

catmistake's picture
Offline
Last seen: 2 years 4 months ago
Joined: Dec 20 2003 - 10:38
Posts: 1100
in practice...

I'd have to guess that wide scsi spanks firewire400. SCSI always got really close to its theoretical limits... but I've never seen firewire even approach its limits. But hey... firewire has many other advantages over scsi besides speed (if it doesn't live up to its theoretical expectations).

Jon
Jon's picture
Offline
Last seen: 12 years 10 months ago
Joined: Dec 20 2003 - 10:38
Posts: 2804
Yep, I'd go for an UltraWide

Yep, I'd go for an UltraWide drive of any sort for real speed over FW. FW is just so convenient for an external drive.

Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
Firewire's other advantages b

Firewire's other advantages being? Plug and play, and what else?

So what's the deal with USB2? Why has that been adopted so much more widespread than firewire? Is it a legal rights kind of thing? Is USB2 much cheaper to produce?

Jon
Jon's picture
Offline
Last seen: 12 years 10 months ago
Joined: Dec 20 2003 - 10:38
Posts: 2804
USB2 is cheaper because it us

USB2 is cheaper because it use the host CPU and not an expensive controller.

FW also has direct AV capabilities. *Hot swap* plug and play, too. Wink Networking via IP over FW as well.

catmistake's picture
Offline
Last seen: 2 years 4 months ago
Joined: Dec 20 2003 - 10:38
Posts: 1100
Re: Firewire's

Plug and play, and what else?

Up to 63 peripherals on the chain... and really long cables. I just checked the wiki which says the max cable length is ~15 feet, but I have a 16ft cable I use often (great to move the cable around the room instead of the drive), and I know the bookstore sells a 25ft cable, so I'd say that, once again, wiki's info is soft.

compare that to scsi's 7 IDs (effectively 6, because the board counts as 1, er... 0), and 6 ft max cable length

Jon
Jon's picture
Offline
Last seen: 12 years 10 months ago
Joined: Dec 20 2003 - 10:38
Posts: 2804
http://en.wikipedia.org/wiki/

http://en.wikipedia.org/wiki/SCSI says 1.5-3m for regular SCSI and 12-25m for LVDs flavors.

Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
So probably, USB2 is prevalen

So probably, USB2 is prevalent moreso because there's more profit in it for the hardware manufacturers than that it keeps prices down for consumers, and the inflated high speed claims are mostly about making it sound better than firewire. It's a nasty world out there in PC land.

Jon
Jon's picture
Offline
Last seen: 12 years 10 months ago
Joined: Dec 20 2003 - 10:38
Posts: 2804
Yep, and they're already putt

Yep, and they're already putting USB in anyway (or it's a part of the mobo chipset) so they save the cost of board space and parts by not putting in FW.

Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
I also see in [url=http://en.

I also see in Wiki's USB page that Apple was originally charging $1 per unit for use of IEEE 1394, which also explains the adoption of USB2.

Jon
Jon's picture
Offline
Last seen: 12 years 10 months ago
Joined: Dec 20 2003 - 10:38
Posts: 2804
There's two good points in th

There's two good points in that link:

# USB uses a "speak-when-spoken-to" protocol; peripherals cannot communicate with the host unless the host specifically requests communication. A FireWire device can communicate with any other node at any time, subject to network conditions.
# A USB network relies on a single host at the top of the tree to control the network. In a FireWire network, any capable node can control the network.

catmistake's picture
Offline
Last seen: 2 years 4 months ago
Joined: Dec 20 2003 - 10:38
Posts: 1100
Re: I also see

oh, the irony... if it weren't for Apple, there might never have been widespread usb adoption (or is that a "post hoc, ergo propter hoc" argument?)

Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
No, a better way to put it wo

No, a better way to put it would be, if it weren't for the deceptive promotion of USB2, computer consumers everywhere would expect and demand firewire, despite Apple's desire to actually profit from their invention. A buck at wholesale kicks the retail price up $4, maybe $5? Plus the cost of including the hardware. Would consumers have payed the extra, knowing the difference in speeds and performance?

Jon
Jon's picture
Offline
Last seen: 12 years 10 months ago
Joined: Dec 20 2003 - 10:38
Posts: 2804
I think catmistake's point is

I think catmistake's point is that Apple introduced USB to the general public via the iMac (and B&W) but failed with FireWire via their strongarming of FireWire licensing, and it is ironic that USB2 was what supplanted it.

grannysmith's picture
Offline
Last seen: 12 years 1 month ago
Joined: Dec 20 2003 - 10:38
Posts: 178
Re: SCSI controller

... compare that to scsi's 7 IDs (effectively 6, because the board counts as 1, er... 0), and 6 ft max cable length

The controller uses ID=7 in both 50-pin (0-7) and 68-pin (0-15) setups, no?

de

Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
testing testing

This link was a very interesting comparison test of USB2 and Firewire. And the eMac consistently beat out the G5 with USB2! And the PC version of USB2 significantly beat out the Mac versions.

I wasn't aware that Apple was the one who gave widespread credibility to USB. Wasn't it already quite prevalent in PC's before the iMac and B&W? I thought it was just another case of Mac playing catch-up. And the B&W came with Firewire too--the first instance, I assume.

This thread has drifted far from my original topic, but that's great--I'm learning a lot here.

catmistake's picture
Offline
Last seen: 2 years 4 months ago
Joined: Dec 20 2003 - 10:38
Posts: 1100
good point

I misread the original post to mean that the external card connector was DB-25 (which, plz correct me if I'm wrong here, is always scsi1)... but I think you're right... can you confirm HC? Does the card have an external 50-pin connector and a second internal 68-pin connector?

DrBunsen's picture
Offline
Last seen: 9 years 8 months ago
Joined: Dec 20 2003 - 10:38
Posts: 946
Re: testing testing

I wasn't aware that Apple was the one who gave widespread credibility to USB. Wasn't it already quite prevalent in PC's before the iMac and B&W?
It was, but on the iMac, USB was pretty much the only expansion available. So a load of devices that previously would have been serial, parallel, even PCI cards, became available as USB.

Hawaii Cruiser's picture
Offline
Last seen: 7 years 3 weeks ago
Joined: Jan 20 2005 - 16:03
Posts: 1433
Does the PCI card make a difference?

Yeah, the Adaptec 29160n SCSI PCI card has two channels--one internal 68pin SCSI-3 LVD channel, and one SE channel with a 50pin internal and HD50 pin external connector. I don't know how high the SE channel is capable of, but like I say, the Epson scanner with its own SCSI-2 HD50 connector transfers data very fast to the external HD50 SE channel of the card. The Visioneer portable scanner has a cable with a DB25 connector, so that means it has to be SCSI-1, right?--so only capable of 5Mb/s. Of course, I could attach an HD50 adapter to the DB25 connector so I could attach the Visioneer scanner to the SCSI PCI card. Now I wonder if the SCSI controller chip makes a difference. Can data from a SCSI-1 device move faster depending on the state of the art of the SCSI controller chip? Is it possible that the SCSI-1 data will move faster through the more sophisticated Adaptec PCI card's chip than it would through an old onboard Mac SCSI chip? Or are both capable of, and limited to, 5Mb/s? And does going across the PCI bus make any performance difference?

dankephoto's picture
Offline
Last seen: 9 months 1 week ago
Joined: Dec 20 2003 - 10:38
Posts: 1899
scsi versions and connectors

Y'all are confusing the host's supported SCSI versions with the connector configuration, two different things entirely. The presence of a DB25 connector doesn't mean the device is (only) SCSI 1.

A DB25 connector can run various flavors of SCSI, including a narrow ultra SCSI data rate of 20MiB/s. I dunno how well the 25 pin connector will handle Ultra2 or better, though in theory it ought to work fine as long as the cabling itself is up to spec.

Of note is the fact that the DB25 electrical interface isn't even a 'real' SCSI connector, just a bit of Apple-hack to fit the space-limited backs of Apple computers.

So yeah, a faster host (in this case your PCI card) should equal faster data throughput, up the limits of the various devices and connecting wires. I guess you'll only find out how fast the Visioneer is by hooking it up and giving it a go.

dan k

PS: here's a nice page on SCSI IDs, especially interesting is the arbitration process, new knowledge for me.

In high-traffic settings, you will typically want to set the slower devices (scanners, tape drives) to the higher-priority IDs, to ensure that they are not crowded off the bus by the faster devices like hard disks.

Blessed Cheesemaker's picture
Offline
Last seen: 12 years 5 months ago
Joined: Mar 19 2007 - 08:52
Posts: 7
I agree with all the above and...

SCSI is high quality data transfer, but the term "SCSI Voodoo" doesn't exist without a reason...

Hooking up one or two well-behaved devices is nothing (and can be really fast), but when something doesn't work, it can be a real PIA to figure out what is at fault.

Anyway, all the above info was accurate...remember, always mind your "bits" and "bytes." Yeah, 12Mb/s (Mega-bits) sounds fast, but 5MB/s (Mega-Bytes) is faster. And that "limit" is only because that is most of the Apple on-board SCSI limit. Add a 3rd party card, and you can fly.

Other things to remember...USB and Firewire are plug and play, and hot-swappable...virtually idiot-proof, important today when there are no manuals and virtually no customer support.

And oh, yeah...a minor thing...USB was developed by Intel. Gee, I wonder why everybody adopted it?

catmistake's picture
Offline
Last seen: 2 years 4 months ago
Joined: Dec 20 2003 - 10:38
Posts: 1100
Re: I agree

And that "limit" is only because that is most of the Apple on-board SCSI limit.

well... the classic stuff is all 5MB/s, including the SE/30 and Quadra 605 and many others, but at some point (probably during the PowerMac line) the internal onboard Apple SCSI speed doubled to 10MB/s, though the external DB-25 scsi bus stayed 5MB/s. I think this is why Apple was stayed competitive using scsi instead of IDE for so long, because that is just as if not faster than ATA-3 is (in practice, if not theory).

And oh, yeah...a minor thing...USB was developed by Intel. Gee, I wonder why everybody adopted it?

Yeah, we already figured that out... it was because of the iMac.

Log in or register to post comments