Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!agate!apple!ames!ncar!asuvax!mcdphx!mcdchg!ddsw1!ddsw1!karl From: karl@ddsw1.MCS.COM (Karl Denninger) Newsgroups: comp.sys.ibm.pc Subject: Comparison of ACB2372 vs WD1006V [was Re: Disk controllers for AT] Keywords: RLL, ESDI, SCSI Message-ID: <[4172.5]comp.ibmpc;1@ddsw1.MCS.COM> Date: 18 Jul 89 20:00:08 GMT References: <579@holin.ATT.COM> <1190@chansw.UUCP> Lines: 126 >----- >Response 4 of 4 (4172) by chan at chansw.UUCP on Tue 18 Jul 89 10:42 >[Jerry H. Chan] > >>In article <[4172.3]comp.ibmpc;1@ddsw1.MCS.COM>, karl@ddsw1.MCS.COM (Karl >>Denninger) writes: >> Try using that board with SCO Xenix and watch the performance disappear. > >Performance is pretty good on Interactive Systems 386/ix 2.0.x w/HPDD (High >Performance Disk Driver -- Fast File System optimizations). Maybe you need >a change in OS's, eh? ;-). Yeah, I want to change to an OS which doesn't know how to spare sectors right for anything except 3:1 interleave MFM drives! (inputting BFI values produce either garbage mapping, or valid values are rejected! This bug is _confirmed_ by Interactive, and is in 1.0.6 > 2.0.1! "mkpart -A" is also broken, and confirmed). Anyone who is selling 386/ix Unix for use with RLL drives is asking for some MAJOR headaches when customers can't get defects mapped right! >> Try using the WD1006-V/SR2 and watch all that performance come back and >> more, and in addition watch the controller work flawlessly with more drives >> for a longer time. > >Implied in your statement is that you've had trouble w/the WD1006 as well >as the ACB2372, only that the trouble was deferred for a longer period of >time. More detail about failures you've seen would be appreciated. Actually, we've had about a 20% failure rate on the ACB boards within the first year of ownership. We've had ONE board fail out of about 50 with the WD1006-V/SR2s. The ACB board does not have a track cache -- it achieves 1:1 interleave only if your driver can access the sectors fast enough. Under dos this _usually_ works. Under Unix it works once in a great while ;-) The WD is the clear winner here. The WD also has a superior ECC implementation, and a built-in surface test that really, truly works. > * Analog floppy data separator. This gives much higher data integrity than > digital data separators. Thus floppy data reliability and data > retrievability are improved over previous AT floppy controllers." Which is why I have an ACB board right here with unreliable floppy performance.... riiiight Adaptec! Ever hear of a thing called "drift"? I'll bet anything you wish to lose that the pot Adaptec uses isn't even CLOSE to the stability of a dedicated-value component in that location. So why doesn't Adaptec use a (insert value) 1% precision resistor in that place? Perhaps because they need to tweak each board before it goes out -- or it might not work at all on floppy drives! >> [comments about poor phone support deleted] > >I'm surprised that SCO Xenix is tossed for the same reasons :-) Really? Sco has ALWAYS been responsive to our questions and bug reports. Interactive (in California) pretends you don't exist, won't call you back, and in general is completely unresponsive; their east coast office tries to help, but lacks the manpower and knowledge to do so. (The east coast office has also told us TWICE that a fix for the bug was in a new release -- first with 2.0, then with 2.0.1 -- neither time was it really fixed). >I will take a close look at the WD1006V series (they are just becoming >available to my hw distributors in this area), and will run some benchmarks >on both ISC 386/ix 2.0.x as well as DOS 3.3 on various classes of machines. >Give me a couple of weeks, eh? :-) I would, except that we've been selling the WD boards for over 7 months! C'mon, they're not THAT new! I've had them in stock since December; they're a real hot item. Finally, the WD board will run in a high-speed bus (16 Mhz claimed). We haven't checked THAT one completely -- the fastest bus I have here is 12 Mhz. The ACB2372 board chokes completely with the speed cranked -- the WD1006 just returns clean data at ~700KB/sec! >Climbing out on a limb (i.e., speaking from personal memory / observations), >the ACB2372 is older technology and has been available for a while >(1+ years?), whereas the WD "V" series is *brand-spankin'-new*. I would Again, we've had 'em for seven months... >performance is *not* the *only* criteria on which one bases his purchases >(price, customer support, availability, reliability come to mind immediately). In that order? We base our choices on which product(s) to offer for sale to customers on several criteria as well, but we have different priorities. They are: reliability performance customer support (since we provide that in-house) price (yeah, it's important, but not in front of the others!) The WD V-series has consistantly come out on top of the "Standard" controller heap. If you want the _BEST_, you buy a DPT board with on-board cache (512K - 16MB!). But be prepared to spend through the nose for it. 95% of the users get along just fine with a WD1006V. If you choose your products based on price above all else, you _will_ get hosed. Thus, you need to build in some extra cash for insurance (and/or replacements) against that eventuality. If you DO build in that extra cash, you may as well have bought something better -- the first time. >the trash (was there an implicit smiley face?), being that the ACB2372 is >a fine, usable product -- can you fault a piece of equipment which does >what it was spec'ed to do? While the ACB2372 may not be leading the But it doesn't. I have had a lot of problems with Adaptec, dating back over two years. In addition their customer support STINKS. Getting busy signals for 4 or 5 hours straight, and then finally being told "too bad, don't use our board" in response to a question on the technical support line doesn't cut it. When we call a manufacturer's customer support line, it's because we truly need help. It is unacceptable, to us as a reseller, to be told to go stuff. We use, spec, and support the WD boards. They aren't perfect -- but they're a darn sight better Adaptec. And WD will talk to you if there ARE problems (they also tend to have a solution to the problem :-). -- Karl Denninger (karl@ddsw1.MCS.COM, !ddsw1!karl) Public Access Data Line: [+1 312 566-8911], Voice: [+1 312 566-8910] Macro Computer Solutions, Inc. "Quality Solutions at a Fair Price"