andrewsr@u2.rutgers.edu (Rich Andrews) (03/09/91)
Hello All! What is the difference between SCSI and SCSI-2? (My guess: SCSI is 16bit and SCSI is 32bit). Does the A3000 have a SCSI-2 port? How much better is the performance of a SCSI-2 over a SCSI? Where can I get SCSI-2 drives? (Looking for ~200MB, maybe Quantum). How much more expensive is SCSI-2 than SCSI? If there are no good (or too expensive) SCSI-2 drives around, is it worth waiting for them to come out (or drop in price)? Thanks in advance for any help! -Rich -- // Rich | "have you ever noticed that opinion without knowledge is \X/ Andrews | always a shabby sort of thing?" - Socrates
krauss@mansw1.enet.dec.com (Michael Krauss) (03/12/91)
In article <Mar.8.20.25.02.1991.5782@u2.rutgers.edu>, andrewsr@u2.rutgers.edu (Rich Andrews) writes: > >Hello All! > >What is the difference between SCSI and SCSI-2? (My guess: SCSI is >16bit and SCSI is 32bit). SCSI(-1) is 8 bit, SCSI-2 is 16 bit. > >Does the A3000 have a SCSI-2 port? No SCSI-2, it's still original SCSI. > >How much better is the performance of a SCSI-2 over a SCSI? > SCSI-2 should be around twice as fast as SCSI-1 >Where can I get SCSI-2 drives? (Looking for ~200MB, maybe Quantum). > >How much more expensive is SCSI-2 than SCSI? > >If there are no good (or too expensive) SCSI-2 drives around, is it >worth waiting for them to come out (or drop in price)? > Don't know. >Thanks in advance for any help! >-Rich >-- > // Rich | "have you ever noticed that opinion without knowledge is >\X/ Andrews | always a shabby sort of thing?" - Socrates > -- ============================================================================== Michael Krauss #include <std_disclaimer.h> DEC Mannheim, Germany "AMIGA or VAX, nothing else please !" "krauss@mansw1.enet.dec.com" or "...!decwrl!mansw1.enet!krauss"
skipper@motaus.sps.mot.com (Skipper Smith) (03/12/91)
In article <Mar.8.20.25.02.1991.5782@u2.rutgers.edu> andrewsr@u2.rutgers.edu (Rich Andrews) writes: >Hello All! > >What is the difference between SCSI and SCSI-2? (My guess: SCSI is >16bit and SCSI is 32bit). Bzzzt! SCSI-1 and SCSI-2 are both 8 bit. 16 bits and wider come under the domain of "wide" SCSI. SCSI-2 is a better definition than SCSI-1 ever was and will allow people to make compatible drives and controllers easier since they have a stricter standard to adhere to (hopefully :-). SCSI-2 also makes a few modifications to the timing diagrams for faster throughput- SCSI-1 maxed out at about 1.2 Mb/sec async and 4 Mb/sec sync whereas SCSI-2 achieves (in theory) 2 Mb/sec asynch and 5 Mb/sec synch (this is all according to a Quantum brochure that I have so blame any mistakes on them). Sure wish somebody would come out with a syncronous drive controller for SCSI-2... it would easily blow away the HardFrame (all of those reports were in MegaBYTES, not bits). > >Does the A3000 have a SCSI-2 port? No, the drive controller on the A3000 is SCSI-1, but it probably could be replaced with a SCSI-2 chip without difficulty (though I will leave it up to the designer to state whether or not that is the case). > >How much better is the performance of a SCSI-2 over a SCSI? See above. > >Where can I get SCSI-2 drives? (Looking for ~200MB, maybe Quantum). Almost all companies have switched over to making SCSI-2 specification drives since they are backwards compatible with SCSI-1 (although you only get the bonus speeds when you have a SCSI-2 compatible controller which nobody makes yet for the Amiga). Quantum, Micropolis, Maxtor, Conner... they are all making SCSI-2 compliant drives. > >How much more expensive is SCSI-2 than SCSI? No reason for them to be anymore expensive. SCSI-2 requires a tighter spec be adhered to, not that additional hardware be used (and any additional hardware used to make a drive achieve SCSI-2 compliance would mostly be very inexpensive like capacitors to reduce ringing and noise, etc...). > >If there are no good (or too expensive) SCSI-2 drives around, is it >worth waiting for them to come out (or drop in price)? > >Thanks in advance for any help! >-Rich >-- > // Rich | "have you ever noticed that opinion without knowledge is >\X/ Andrews | always a shabby sort of thing?" - Socrates -- Skipper Smith | skipper@motaus.sps.mot.com Motorola Technical Training | 8945 Guilford Rd Ste 145 All opinions are my own, not my employers | Columbia, MD 21046
daveh@cbmvax.commodore.com (Dave Haynie) (03/12/91)
In article <Mar.8.20.25.02.1991.5782@u2.rutgers.edu> andrewsr@u2.rutgers.edu (Rich Andrews) writes: >Hello All! >What is the difference between SCSI and SCSI-2? (My guess: SCSI is >16bit and SCSI is 32bit). Not quite. SCSI-1 defines an 8 bit data path. SCSI-2 generally uses the same 8 bit data path, but does define 16 and 32 bit versions. Most Amiga hard disk controllers, even modern PIO controllers, work transfers between SCSI bus and main memory in 16 bit chunks. The A3000 funnels all SCSI transfers into 32 bit chunks. >Does the A3000 have a SCSI-2 port? Yes and no. At present, the A3000 comes with a SCSI-1 port. The SCSI chip we use on it defines this "SCSI-1"-ness, and there is apparently a pin-compatible SCSI-2 part that'll drop right in and work, at least hardware-wise. The other differences, which I know less about, are software. Specifically, the SCSI command set. Since it pretty much evolved over time, and many companies that uses SCSI disk drives didn't use all the features, some of the SCSI-1 command set is effectively "optional", and other pieces aren't as standard as one would like. One of the main ideas of SCSI-2 was to very strictly define the command set, so that all SCSI-2 drives, as well as good SCSI-1 drives, would play properly on a SCSI-2 bus. >How much better is the performance of a SCSI-2 over a SCSI? There are several SCSI-1 transfer modes. Every SCSI-1 controller implements the standard asynchronous mode, which goes around 1.5 MB/s. Some have an improved asynchronous mode which goes 2.5 MB/s; I don't know much about it, other than the SCSI chip used in A3000 and A2091 supports this. Then there's the synchronous mode, which, depending on the clock used, runs up to about 5 MB/s (I think the A3000 runs it at around 4 MB/s). SCSI-2 adds a high speed synchronous mode, which kicks up to a maximum of around 10 MB/s. If you could find a 32 bit SCSI-2 drive, you could expect a theoretical maximum transfer rate of 40 MB/s. Now, none of these transfer rates really attack the question of drive speed. Good SCSI drives manage bursts of 1.25-1.50 MB/s, straight from the disk. Until any seeking is necessary. The higher transfer rates help when you have multiple devices on the SCSI bus, and also to a degree with many of the modern drives which, like the Quantum series, have some on-drive caching and won't always need to burst data straight from the disk. But you don't expect in a single drive system to double your effective disk performance going from asynchronous to synchronous or from synchronous to high speed synchronous transfers. > // Rich | "have you ever noticed that opinion without knowledge is >\X/ Andrews | always a shabby sort of thing?" - Socrates -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "What works for me might work for you" -Jimmy Buffett
erickson@alee.UUCP (Lee C. Erickson) (03/12/91)
In article <1991Mar11.163108.12361@hollie.rdg.dec.com>, krauss@mansw1.enet.dec.com (Michael Krauss) writes: >In article <Mar.8.20.25.02.1991.5782@u2.rutgers.edu>, andrewsr@u2.rutgers.edu (Rich Andrews) writes: >> >>What is the difference between SCSI and SCSI-2? (My guess: SCSI is >>16bit and SCSI is 32bit). > >SCSI(-1) is 8 bit, SCSI-2 is 16 bit. > SCSI-2 supports 8-bit, and optionally 16- and 32-bit configurations. Also a "fast" logic mode. >> >>How much better is the performance of a SCSI-2 over a SCSI? >> >>If there are no good (or too expensive) SCSI-2 drives around, is it >>worth waiting for them to come out (or drop in price)? >> The speed of the bus generally isn't the limiting factor in such drives/small computer systems. ----- Lee Erickson uucp: erickson@alee.UUCP ..!psuvax1!burdvax!gvlv2!alee!erickson internet: erickson%alee@gvl.unisys.com usmail: 720 Raynham Rd., Collegeville PA 19426
pk@wet.UUCP (Philip King) (03/14/91)
In article <19745@cbmvax.commodore.com> daveh@cbmvax.commodore.com (Dave Haynie) writes: [...] >There are several SCSI-1 transfer modes. Every SCSI-1 controller implements >the standard asynchronous mode, which goes around 1.5 MB/s. Some have an >improved asynchronous mode which goes 2.5 MB/s; I don't know much about it, >other than the SCSI chip used in A3000 and A2091 supports this. Then there's >the synchronous mode, which, depending on the clock used, runs up to about >5 MB/s (I think the A3000 runs it at around 4 MB/s). SCSI-2 adds a high speed >synchronous mode, which kicks up to a maximum of around 10 MB/s. If you could >find a 32 bit SCSI-2 drive, you could expect a theoretical maximum transfer >rate of 40 MB/s. > >Now, none of these transfer rates really attack the question of drive speed. >Good SCSI drives manage bursts of 1.25-1.50 MB/s, straight from the disk. >Until any seeking is necessary. The higher transfer rates help when you have >multiple devices on the SCSI bus, and also to a degree with many of the modern >drives which, like the Quantum series, have some on-drive caching and won't >always need to burst data straight from the disk. But you don't expect in a >single drive system to double your effective disk performance going from >asynchronous to synchronous or from synchronous to high speed synchronous >transfers. > Dave, does this mean that Amiga hard disk controllers support synchronous transfers? I don't think I've ever seen any of the host adaptor manufacturers claim that. I happen to have a drive that supports this, and I wouldn't at all mind if I could speed it up! It's the Seagate/Imprimis ST-1239N/94350-230S model, 200MB 15ms. Apparently it's got an onboard cache memory like the Quantums, but it doesn't default to 'on' mode. Someone told me that there might be a way to activate it, if you could talk directly to it with some kind of program that issues SCSI commands. Any insight on how I might achieve that? Thanks in advance... Philip King pk@wet.uucp {ucsfcca,hoptoad,well}!wet!pk (My usenet site forces me to use more text in the followup than the text from the original article...and I already cut the heck out of it! Sheesh, it's kinda ridiculous to have to put a bunch of useless drivel at the end of my post, but otherwise it dumps me out of postnews and I have to start from scratch again! Ugh!
dwjz@bcarh660.BNR.CA (Doug Zolmer) (03/14/91)
In article <1991Mar11.163108.12361@hollie.rdg.dec.com>, krauss@mansw1.enet.dec.com (Michael Krauss) writes: |> In article <Mar.8.20.25.02.1991.5782@u2.rutgers.edu>, andrewsr@u2.rutgers.edu (Rich Andrews) writes: |> |> >Where can I get SCSI-2 drives? (Looking for ~200MB, maybe Quantum). |> > |> >How much more expensive is SCSI-2 than SCSI? |> > |> >If there are no good (or too expensive) SCSI-2 drives around, is it |> >worth waiting for them to come out (or drop in price)? |> > Why do people feel the need to get SCSI-2 over SCSI? Properly implemented, SCSI will give you performance that is more than adequate for 99% of what anybody would ever want or need. Commodore's A2091 and MicroBotic's Hardframe 2000 can give transfer rates as high as the bus will allow. In everyday use, my Hardframe 2000 gives me read rates of 975 kilobytes/sec. Believe me, this is plenty fast - much faster than I've seen on any other consumer-grade personal computer. SCSI-2 was designed for very high throughput devices for things like real-time transaction processors (Tandem springs to mind), used by banks (for automated teller machine transaction handling) and telephone companies (for LIDB and 800 database transaction handling). It allows disk shadowing and is designed for fault tolerance - hardly things that most personal computer users need. For the added expense, SCSI-2 will not give any marked advantages over regular SCSI in the average personal computer. -- Doug Zolmer | My opinions are | Bell-Northern Research Ltd. uunet!bnrgate!dwjz%bcarh660 | mine only and do | Dept. 7N53, Mailstop 089 dwjz@bcarh660.BNR.CA | not reflect BNRs. | P.O. Box 3511, Station C Voice: +1 613 763 8217 FAX: +1 613 763 3292 Ottawa, ON, K1Y 4H7, CANADA
dltaylor@cns.SanDiego.NCR.COM (Dan Taylor) (03/16/91)
In <6034@bwdls58.bnr.ca> dwjz@bcarh660.BNR.CA (Doug Zolmer) writes: >For the added expense, SCSI-2 will not give any marked advantages over regular >SCSI in the average personal computer. Not strictly true. If you have multiple SCSI-attached devices, such as laser printers, tape drives, network interfaces, or multiple disks, then properly designed SCSI-2 (fast or wide, or both) devices with caching, or at least, large buffers, will show noticeable performance increases. I did see the word "average", though. Since the "average" (SCSI-equipped) personal computer is a plain, old, dumb, slow Mac, you're right. But this question really was asked about the A3000, hardly an "average" personal computer. SCSI-2 would be an asset, there. Even the newer Macs, if they had it, and EISA PCs, some of which do, will show a performance with SCSI-2. The real advantage of the higher transfer bandwidth is the ability to move large blocks between buffers/caches on the disk, etc. and similar buffers/caches on the host, without hanging up the requesting process for a lot longer than necessary. At 800Kb/s, a typical 512 byte block of data takes 640usec of data transfer, while at 4Mb/s, I can move 2048 bytes in 512usec. If the 2Kb is cached locally, then we never have to go through the request phase of the SCSI bus for the other 3 512 byte blocks, which is a real savings, since the first block didn't take any longer to arrive with SCSI-2, than -1. Remember, our Amigas are multitasking, we often have several SCSI transactions pending. Faster transfers allow closer interleaving of transactions, again a savings. The extra cost would be noticeable on an A500/A590 class machine, but in my 2500/30, or an A3000, particulary with multiple drives, the added cost would be negligible. Dan Taylor * My opinions, not the companies.
erickson@alee.UUCP (Lee C. Erickson) (03/18/91)
In article <2216@wet.UUCP>, pk@wet.UUCP (Philip King) writes: >In article <19745@cbmvax.commodore.com> daveh@cbmvax.commodore.com (Dave Haynie) writes: >[...] >Dave, does this mean that Amiga hard disk controllers support synchronous >transfers? I don't think I've ever seen any of the host adaptor >manufacturers claim that. The newer GVP controllers support synchronous mode. > >I happen to have a drive that supports this, and I wouldn't at all mind >if I could speed it up! It's the Seagate/Imprimis ST-1239N/94350-230S >model, 200MB 15ms. Apparently it's got an onboard cache memory like the >Quantums, but it doesn't default to 'on' mode. Someone told me that there >might be a way to activate it, if you could talk directly to it with some >kind of program that issues SCSI commands. Any insight on how I might >achieve that? Thanks in advance... You would have to issue a drive-specific "mode select" command. ----- Lee Erickson uucp: erickson@alee.UUCP ..!psuvax1!burdvax!gvlv2!alee!erickson internet: erickson%alee@gvl.unisys.com usmail: 720 Raynham Rd., Collegeville PA 19426
daveh@cbmvax.commodore.com (Dave Haynie) (03/19/91)
In article <2216@wet.UUCP> pk@wet.UUCP (Philip King) writes: >Dave, does this mean that Amiga hard disk controllers support synchronous >transfers? I don't think I've ever seen any of the host adaptor >manufacturers claim that. Sure do. Apparently, though, the our host adaptors won't normally request synchronous transfers. Some drives lock up, or otherwise misbehave, if you ask for synchronous transfers. There's a bit somewhere in A3000 battery backed RAM, last I heard, that will tell the scsi.device to automatically request synchronous transfers. If you set that, and you have the proper drive out there, you should be OK, but some drives will find this to be a problem. Of course, that's How I Understand It. I'm not the SCSI.Expert around here, others may have more to add to this. >I happen to have a drive that supports this, and I wouldn't at all mind >if I could speed it up! Chances are, going to synchronous mode won't result in much of a speedup anyway in a single drive system. In most cases, the drive is the limiting factor, not the SCSI bus. It might make a visible difference on high end drives with large caches. Randell or Steve may have looked at this question in greater detail. -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "What works for me might work for you" -Jimmy Buffett
jesup@cbmvax.commodore.com (Randell Jesup) (03/19/91)
In article <19941@cbmvax.commodore.com> daveh@cbmvax.commodore.com (Dave Haynie) writes: >In article <2216@wet.UUCP> pk@wet.UUCP (Philip King) writes: > >>Dave, does this mean that Amiga hard disk controllers support synchronous >>transfers? I don't think I've ever seen any of the host adaptor >>manufacturers claim that. > >Sure do. Apparently, though, the our host adaptors won't normally request >synchronous transfers. Some drives lock up, or otherwise misbehave, if you >ask for synchronous transfers. Yup. We will do synchronous if the drive requests it (quantums can be told to request it: mode select page 55, byte 2, bits 4 and 5). There may be a battmem bit for it; ask steve. Some older scsi drive (guess who) will crash and take the bus down with them if asked to do synchronous. >Chances are, going to synchronous mode won't result in much of a speedup >anyway in a single drive system. In most cases, the drive is the limiting >factor, not the SCSI bus. It might make a visible difference on high end >drives with large caches. Randell or Steve may have looked at this question >in greater detail. Also correct. Without synchronous, my Quantum 210S gets 1.2M/s max. With it, I get 1.2M/s max. The win to synchronous is when you have multiple drives in use: a given drive stays on the bus less. It can also be a win when you get a cache/preread hit in the drive cache. These things do not usually show up in disk speed test, but might when running compiles on two drives at the same time, for example. The big limiting factor is usually the speed at which bits come off the platters. The new CDC/Seagate ST1480, for example, spins at 4400 rpm and can do 1.7M/s through the FS. -- Randell Jesup, Keeper of AmigaDos, Commodore Engineering. {uunet|rutgers}!cbmvax!jesup, jesup@cbmvax.commodore.com BIX: rjesup The compiler runs Like a swift-flowing river I wait in silence. (From "The Zen of Programming") ;-)
pk@wet.UUCP (Philip King) (03/22/91)
In article <19967@cbmvax.commodore.com> jesup@cbmvax.commodore.com (Randell Jesup) writes: >In article <19941@cbmvax.commodore.com> daveh@cbmvax.commodore.com (Dave Haynie) writes: >>In article <2216@wet.UUCP> pk@wet.UUCP (Philip King) writes: >> >>>Dave, does this mean that Amiga hard disk controllers support synchronous >>>transfers? I don't think I've ever seen any of the host adaptor >>>manufacturers claim that. >> >>Sure do. Apparently, though, the our host adaptors won't normally request >>synchronous transfers. Some drives lock up, or otherwise misbehave, if you >>ask for synchronous transfers. > > Yup. We will do synchronous if the drive requests it (quantums can >be told to request it: mode select page 55, byte 2, bits 4 and 5). There may >be a battmem bit for it; ask steve. Some older scsi drive (guess who) will >crash and take the bus down with them if asked to do synchronous. Randall, perchance were you referring to ***gate when you alluded to some drives taking the bus down if you requested synchronous? I have an Imprimis/ Seagate ST-1239N, which is a 200M 3.5" drive that supports synchronous transfer (there's a jumper you add). I presume that if ***gate was the party in question, then my drive should be fine, since it's really originally a CDC/Imprimis piece? Does anyone else know if other host adaptors support Synchronous? I am using an ICD Advantage 2000, which is otherwise a very fast host adaptor. I also wondered if anyone else had any more specific information on how to activate my drives read-ahead cache, with the 'mode select' command Lee Erickson and Dave were referring to earlier. Is there a program that allows me issue these commands? Philip King pk@wet.uucp {ucsfcca,claris,hoptoad}!wet!pk
jesup@cbmvax.commodore.com (Randell Jesup) (04/02/91)
In article <2242@wet.UUCP> pk@wet.UUCP (Philip King) writes: >In article <19967@cbmvax.commodore.com> jesup@cbmvax.commodore.com (Randell Jesup) writes: >> Yup. We will do synchronous if the drive requests it (quantums can >>be told to request it: mode select page 55, byte 2, bits 4 and 5). There may >>be a battmem bit for it; ask steve. Some older scsi drive (guess who) will >>crash and take the bus down with them if asked to do synchronous. There apparently isn't such a bit in the driver, unless I missed it. >Randall, perchance were you referring to ***gate when you alluded to some >drives taking the bus down if you requested synchronous? I have an Imprimis/ >Seagate ST-1239N, which is a 200M 3.5" drive that supports synchronous transfer >(there's a jumper you add). I presume that if ***gate was the party in >question, then my drive should be fine, since it's really originally a >CDC/Imprimis piece? If I've read the source right, the controller won't request sync xfers. Some drives (like Quantum) allow you to specify that the drive should request them. Some drives probably request sync xfers all the time. It's not a big issue usually on amigados, depending on how busy your disks tend to be (though it may well help when backing up to scsi tape). Ah, the wonders of supporting such "standard" drives as the ST225N. >I also wondered if anyone else had any more specific information on how >to activate my drives read-ahead cache, with the 'mode select' command >Lee Erickson and Dave were referring to earlier. Is there a program that >allows me issue these commands? Keep your eyes open, I've been up late nights... -- Randell Jesup, Keeper of AmigaDos, Commodore Engineering. {uunet|rutgers}!cbmvax!jesup, jesup@cbmvax.commodore.com BIX: rjesup Disclaimer: Nothing I say is in anything other than my personal opinion. Thus spake the Master Ninjei: "To program a million-line operating system is easy, to change a man's temperament is more difficult." (From "The Zen of Programming") ;-)
pk@wet.UUCP (Philip King) (04/04/91)
>In article <2242@wet.UUCP> pk@wet.UUCP (Philip King) writes: >>In article <19967@cbmvax.commodore.com> jesup@cbmvax.commodore.com (Randell Jesup) writes: > > >>I also wondered if anyone else had any more specific information on how >>to activate my drives read-ahead cache, with the 'mode select' command >>Lee Erickson and Dave were referring to earlier. Is there a program that >>allows me issue these commands? > > Keep your eyes open, I've been up late nights... > >-- Okay! Eyes open, Scanners UP! Philip pk@wet.uucp {cca.ucsf.edu,claris,hoptoad}!wet!pk (Sorry, for some reason I have to edit the attribution and expand my contribution like crazy on my machine...otherwise it'll kill my post!)