Sign in to follow this  
miamicanes

keep the X15 Cheetah, or dump it for an Western Digital 'JB?

Recommended Posts

Argh. I screwed up.

This evening, I had a moment of stupidity and ended up buying a first-generation Seagate ST318451LC Cheetah X15 on eBay (it was fairly cheap, but I didn't realize just *how bad* its performance was relative to the 15K.3, or even the Western Digital JB IDE drives for that matter).

Anyway, it's too late to do anything about it now and I'm stuck with the drive, so I basically have three choices:

* Sell it immediately to someone else on eBay and never use it. Estimated loss: $20-40 for shipping, paypal, and eBay fees.

* Keep it, but buy a $15-25 SCA2 80-to-68 pin adapter, the cheapest $15-20 ultra-wide SCSI-3 cable I can find, and use it with the old Adaptec 2940uw host adapter I threw in the closet in disgust a couple of years ago after my most recent ugly run-in with SCSI. Estimated cost: $30-50.

* blow another hundred dollars on an Ultra2 LVD SCSI host adapter, terminator, cable, and SCA2 80-to-68 pin adapter.

Unless someone can come up with a *really* compelling argument in favor of #3, I'm going to write it off as not even *remotely* worth the cost and classify blowing an additional $50 on a LVD Ultra2 controller as a *total* waste of money by virtue of the fact that even a lowly $110 Western Digital 800JB apparently blows the X15 away.

A quick survey of search results on google seems to suggest that the maximum transfer rate of the Adaptec 2940uw is 40 megs/second. On the other hand, the drive's manual ( http://www.seagate.com/support/disc/manual...i/83329484e.pdf ) seems to imply that 40 megs/second can't be achieved with a non-LVD interface... suggesting to me that SCSI might have multiple, incompatible 40 meg/second transfer modes, which once again would destroy #2 as a viable option. a 40 meg/second limit for a drive that can only do 42 megs/second under ideal conditions anyway is no big loss... but 20 megs/second is another matter entirely.

Assuming for a moment that LVD *is* an upwardly-compatible superset of Ultra Wide SCSI-3, and a nominally 160 meg/second LVD Ultra-2 drive can be coaxed into making the best of a 40 meg/second ultra-wide SCSI-3 interface if that's all that's available as long as I've got the proper adapter to kludge it to 68-pin ultra-wide SCSI-3 and a suitable cable to connect it (via the adapter) to the host interface, #2 might be worthwhile after all... particularly if the Cheetah's faster seek times make it faster than an 800JB after all in real use.

Ultimately, options 1 and 2 will probably end up costing me about the same amount of money -- #1 as an up-front loss, #2 in additional acquisition costs. The question is, which is likely to leave me happier -- dumping the X15 for an IDE Western Digital 800JB, or spending the minimum amount necessary to make it work with my old Adaptec 2940uw? It's really hard to tell from the benchmark database here, because the only three benchmarks the two drives have in common are the Business and High-end Disk WinMark 99 scores, and the STR graph... all three of which seem to suggest that the 7200RPM 800JB totally smokes a 15kRPM drive that was state of the art just 2 years ago and effectively consign it to the junk heap.

Share this post


Link to post
Share on other sites

Well, it looks like the 2940uw will at least do 40 megs/second with the Cheetah. I was initially scared by page 64 of the Cheetah's manual (which stated that the maximum speed was "20 megatransfers/second" with a non-LVD interface), but apparently that figure really means 20 megatransfers per second per 8-bits. Since the 2940uw is both ultra and wide, it's 16 bits, so the real speed is 20 x (16 / 2) = 40.

The jury's still out on whether I should still dump the drive or use it with the 2940uw, but I think I can safely eliminate buying an LVD controller as a serious option. For the hundred+ bucks it would cost to get the card, cable, and terminator, I could just buy a FASTER WD800JB and condemn the X15 to eternity in my web server, connected to the SCSI-2 Symbios card, cable, and passive terminator I already have in there for the DAT drive. A terrible waste of a once-proud drive, but probably a better alternative than wasting the cash on an expensive controller for a slow drive. Hmmm. I guess selling it on eBay *would* be merciful by comparison...

Share this post


Link to post
Share on other sites

Um, are you using XP? If so, the majority opinion is that SCSI is mabey not so hot for XP. :( The night I placed an order with my local vendor for my 15.3k I run across threads describing poor SCSI perfomance under XP. I had a 15.2k and suspected that something was awry with XP. Anywho, I puchased a WD400jb at googlegear for $92-free fedx. I think the 800jb is $111+50 cents fedx. My 400jb seems to be pretty smokin'! I dont do much benchmarking but I have the same software setup as I did with the 15.2k. The WD seems to work better with my current OS (XP Pro). And if you can't afford 100 bucks or so to try a drive that might perform well, you might have the wrong hobby.

Share this post


Link to post
Share on other sites

My experience is that the WD800JB and the X15 have similar performance. It varies depending on what you do with your computer. If I remember correctly, Tannin found the X15 to still be faster than the JB drives, for instance. I doubt there are many real-world situations where the X15 really gets blown away. I think the realistic worst-case scenario is that the X15 is a little bit slower, and the realistic best-case scenario is that the X16 is a little bit faster.

Assuming the acoustics of the X15 aren't a problem (I'd be much more concerned with that than with the performance), #2 isn't a bad option. Unless you need the absolute best in speed, the X15 is still a really good drive.

Share this post


Link to post
Share on other sites

Dude, you have serious problems with scsi. What are you doing that is so earth shatteringly important that you are driven to manic obsession with infintessimal differences in drive speeds? Are you a video editor billing clients at $600 per hour and need the fastest drive on earth? If its a money thing then why even consider scsi? You buy a fabulous scsi drive and then cheap out on a controller. If you want to play, you got to pay. You are probably the type of person who drives a piece of stinker $500 car with $2000 rims on it. Your rediculous postings about conspiracy theories of drive manufacturers colluding to keep scsi prices high just to rip you off is beyond childish. Sell all your scsi items and be done with it.

Share this post


Link to post
Share on other sites

Damn, that was pretty harsh. I have started to see a lot of those $500 cars with $4000 rims and $8000 cd player "I want to be deaf by 25" systems lately in the north part of town....

Share this post


Link to post
Share on other sites

Jhaislet, it rather sounds as though the thing that will make you happy will be selling the X15 - not for any technical reason, simply because you seem to dislike SCSI stuff. :wink:

I currently own a pair of first-gen X15s. One is in the machine I'm using now, connected via a the very same 2940UW controller I bought to go with my 4.5GB first-gen 10,000 RPM Cheetah 1. The other is at home and running off a Tekram Ultra/160 card. I can't tell them apart. I'm sure that I could measure a difference, but with the things I do with my machines (which are mostly not transfer rate intensive anyway) it just doesn't show up. And if the difference is so small that I can't see it, then I think it's too small to worry about. (The two machines are otherwise almost identical - Athlon 1900s, 512MB, Matrox G450s.)

A while ago I bought a 1000JB to slip into my home machine as a replacement for my old data drive (A 7200 RPM Atlas). I didn't need anything like that capacity, but I was seduced into spending the extra by the ridiculous "JB = 15K equivalent" hype that was going around at the time, and because that was the smallest JB available back then. I kept it for about two or three weeks and although I was initially very impressed - it was the fastest IDE drive I'd ever used - I soon became dissatisfied with its sluggishness compared to the X15. It wasn't that bad, but not what I expected from such an expensive drive. (Remember, it was vastly bigger than I needed - I was looking for speed but paying for capacity. JBs were expensive back then.) So I pulled it out and decided that, seeing as I wasn't going to get X15 performance anyway, I might as well at least satisfy one of my two desires, and have a cheap drive. So I replaced it with a 30GB 5400 RPM Samsung. And that was too slow, as much slower than the JB (or the old Atlas) as the JB was than the X15. A week later I wound up with a 40GB 7200 RPM Samsung instead, which is still there nearly a year later. That was close enough to the JB to be reasonable so far as performance goes, cheap, very quiet (as if that matters when the Martians come calling!) and (best of all) gives me that same feeling of confidence in the reliability of my hardware that Seagate SCSI drives do.

Of course, in a way I cheated: by slipping that 5400 in first, I readjusted my expectations. I could have put just about any 7200 in and felt good about it!

Perhaps I'll slip in a gen-3 X15 one day soon, perhaps I won't. If I do it will mostly be to cut the noise down - the performance difference would be enough to make me want to spend the price of an IDE drive, not enough to buy yet another SCSI credit-card destroyer. I will almost certainly replace the Spinpoint with a bigger one when the next Samsung 7200 comes out, but only if I need the space (probable) or it's vastly faster (improbable), and only because it costs me virtually nothing to do that.

So there are really three questions here: is the X15 noticably faster than a JB? Absolutely yes. They are a long, long way apart for the things I do. I wouldn't give up either of my X15s for a JB in a million years. If you are into data-rate-intensive things then you may not feel the same way. (Note Barry's comments here: Barry knows his drives as well as anyone, and he felt that his two were broadly similar. Different folks, different strokes.)

The second question is: would you feel that the X15 was significantly faster, given your usage pattern and expectations? The answer to this one is "probably, but not for certain". Most X15 owners love them, a few feel underwhelmed.

And the final question is: if your answer to question two is "yes", would you feel that the extra cost of the X15 (as compared to a JB) was justified? Sure, it's faster, but is it faster enough? That's not a technical question, it's a psychological question. But for what it's worth, my guess is that, in your case, the answer is "no".

Go grab yourself a cable (you can always sell it when you sell the X15 and get half your money back on it) and try it out. I am guessing that you will try the X15 and shrug your shoulders and say "nice, but not nice enough", just as I did with my JB. But you owe it to yourself to find out.

Share this post


Link to post
Share on other sites
Dude, you have serious problems with scsi.  What are you doing that is so earth shatteringly important that you are driven to manic obsession with infintessimal differences in drive speeds?  Are you a video editor billing clients at $600 per hour and need the fastest drive on earth?  If its a money thing then why even consider scsi?  You buy a fabulous scsi drive and then cheap out on a controller. If you want to play, you got to pay.  You are probably the type of person who drives a piece of stinker $500 car with $2000 rims on it.  Your rediculous postings about conspiracy theories of drive manufacturers colluding to keep scsi prices high just to rip you off is beyond childish.  Sell all your scsi items and be done with it.

My serious problem with SCSI is based upon the fact that, historically, every time I've touched it in the past I've ended up spending twice as much money and ended up with slower performance than I would have gotten from IDE anyway. I have no objection to forking out cash (within reason), but I damn well expect to see some tangible, quantifiable, real benefit from it.

The X15's purchase was, in fact, a stupid mistake. I decided to check out Cheetah prices on eBay one last time, and tripped across an auction for one that was $80 and ending in 14 minutes. Based upon the limited information I was able to get in 12 minutes from Seagate.com, it looked like its performance was comparable to the current crop and mainly was just bigger and louder than the 15K.3 drives, so I jumped at it, and ended up getting the drive for a little more than a hundred bucks.

It wasn't until afterward, when I really had time to dig through the storagereview database and find the old review for the X15 itself in the old/legacy section that I discovered the horrible truth -- it's not particularly comparable to even the second generation drives, let alone the third. What really ruined my evening, though, was comparing its benchmark scores to those of an 800JB and discovering that the JB did, in fact, appear to beat the X15 by a fairly big margin in all of them.

I suspect its real-world performance for things I do might be a lot better than the STR and WinMark stats suggest, because my main activity is programming... which generally involves reading lots and lots of little files in a short interval of time before writing a small number of slightly-larger files. Unfortunately, the stats that would actually reveal the truth or falsehood of that hypothesis aren't present for the X15, so I'm mainly buying into this hypothesis as a matter of faith at the moment.

The main criteria for decisively eliminating option #3 was simple: A LVD Ultra2 controller, cable, and SCA2 adapter will realistically cost around $100, give or take ten bucks. For $11 more, I could buy an 800JB from NewEgg.com. Think about it for a moment -- $11 more, apparently faster performance than the X15, an additional 60 megs, and no SCSI grief. I could agonize a while and still rationalize buying the faster card if doing so would give some real, tangible benefit over cutting my losses, selling the X15, and buying a JB -- but all the evidence I've seen suggests that, at best, I'd get comparable performance and wind up spending twice as much as I'd have spent to get the JB in the first place if I were to actually go all the way.

On the other hand, option #2 emerged as a possible deal-saver by virtue of the fact that I do, in fact, have a 2940uw sitting in an antistatic bag in a box in my closet, and its capabilities appear to be only a tiny bit below the drive's best-case transfer rates anyway. If the X15's faster seek times were to actually translate into superior real performance compared to a JB for reading lots of little files in a short period of time (which I suspect it might, but can't prove at the moment), it just might be worth spending the $40-50 to get the adapter and cable to use the drive with it.

And, for your information, I happen to drive a fully-loaded (except for antilock brakes, which were impossible to get without waiting a month) 2001 Hyundai Tiburon, chosen mainly because its performance and general fun factor substantially exceed that of most cars costing 50% more. That's not to say it's their absolutely equal peer in every way, nor that it beats them across the board... but I have no complaints with it. It can effortlessly hit 120 if I feel like risking a major ticket, has no problem accelerating from a full stop at traffic lights as fast as I think I can get away with without arousing the ire of local law enforcement officials, and generally is quite fun to drive... and leaves me enough cash to support my hardware habits and still take a few decent vacations every year. For the most part, I do a pretty good job of identifying areas where I can make a tiny performance sacrifice and save a lot of money relative to what I'm giving up. Occasionally, I screw up like I did last night and need to dig my way out of it.

Insofar as my alleged conspiracy theories go, my biggest objection to SCSI is, and has always been, the fact that the unavailability of non-SCSI fast drives throws a needless barrier to entry into the equation by forcing the expenditure of more than a hundred dollars of additional hardware, none of which provides any tangible benefit when used to interface a single drive, and does, in fact, impose real, quantifiable consequences that (more often than not) seem to negate most of the benefits that motivated the purchase in the first place.

Finally, the decision to go ahead with the purchase of a SCSI drive in light its problems under XP was mainly the seduction of blazingly-fast reads. I've got a gig of ram and run with virtual memory disabled, so write performance isn't quite as important to me as read performance. Starting from a clean slate based on what I know now, the purchase of an XP15 to use with a 2940uw would have been eliminated instantly as a serious option. The fact that the drive is on its way and might be able to redeem itself by virtue of faster reads of small files muddies up the equation a lot and leaves me without a clear-cut superior option to pursue.

Share this post


Link to post
Share on other sites

A year ago the original X15 was blazingly fast, it's no slower today than it was then. Although I have a second gen x15 as well (no 3rd yet, drool!), I still boot and run my os off of the original 18gb x15 and it's PLENTY fast.

quit obsessing over this - in real world use it's plenty fast for you. The real issue is whether you want to invest in a newer controller. By the time you add up costs, you'll be cheaper just to buy the IDE drive. I buy SCSI and love it, but you have to invest in the card.

You might check out the hypermicro site - they did have a free LVD controller card deal with a drive purchase - you might pick up a second drive and a newer controller for a good deal there.

Share this post


Link to post
Share on other sites

i will have to agree with geckert's assesment of the situation. i think this dude needs to switch to whatever works for him which obviously ide, end of story.

Share this post


Link to post
Share on other sites

paying over $100 for a first gen 15k Cheetah?

You should have bought a Atlas 10k III for less and beat the stinker out of your WD.

François

Share this post


Link to post
Share on other sites

I have both of these drives, a Cheetah X15 and a WD 1800JB. My Cheetah has WinXP, and I use the WD for my programs. However, I bought a Adaptec 19160 (Ultra160) controller so I could get the max performance.

1. This generation Cheetah is very loud, and has a constant click sound, which is normal.

2. SCSI feels a little more responsive, but not too impressive to me. Meaning, if my Cheetah is loading something which takes a 30 seconds, I can open up Outlook Express or Internet Explorer and not have to wait a long time for it to come up. But, it isn't amazingly responsive.

3. When copying a file to itself, the Cheetah is takes aprox. twice as long! The same goes for unzipping/unraring a large file. This is the WinXP bug I guess and miserable write speed. To me, this is real-world stuff (maybe not the copying), and the SCSI drive is jus dismal, and happens to be rather common task.

4. Tranfer Rates (winbecnh99) however on both drives shows the Cheetah starts at 60 MB/sec and can sustain that fairly well. The WD drive starts at 50 MB/sec and has a sharper drop as the test progresses. So I'm confused why I get a much better transfer rate through this benchmark for the SCSI drive.

I'm planning on installing either XP and/or Windows2000 and boot off of my WD drive and see how the system feels. For all I know, the SCSI drive is performing very well for OS functions and maybe I have a more responsive system than I realize because I'm just use to running off of a SCSI drive.

If you want to run XP, and you wanna buy now and not wait a couple months for MS to release some kind of "fix", I'd say get one of the new IDE drives with the 8 MB cache. They seem very good, and you gotta love that massive storage!

Share this post


Link to post
Share on other sites

Given logicprobe's 50+ meg/second performance, I'm considering spending an extra $10-15 now to hedge my bets and buy a cable and terminator that could be used with a future cheap Ultra2 LVD or Ultra160 card I might trip over. On the other hand, I want to make sure using an Ultra2 LVD or Ultra160 cable and terminator won't end up dropping the 2940UW down to 20 megs/second in the meantime.

Am I correct in assuming that 68-pin SCSI cables are perfectly downward compatible with one another? In other words, an Ultra160 cable can be used to connect an X15 to an Ultra2 LVD or an Ultra-Wide SCSI-3 card, and an Ultra2 LVD cable can be used to connect the X15 to an UltraWide SCSI-3 card?

How about terminators? Can an Ultra160 terminator be used to terminate an X15 that's connected to an Ultra2 LVD or UltraWide SCSI-3 card? Can an Ultra2 LVD terminator be used to terminate an X15 connected to an UltraWide SCSI-3 card?

At the other extreme, I'm wondering whether I can get away with using a SCA (NOT SCA2) 80-to-68 adapter with built-in active termination ($19.95) and 1-unit SCSI-3UW cable ($12.95). I know the X15 has a SCA2 connector, but I'm not sure whether it's SCA2 to support Ultra160 (and maybe Ultra2 LVD), but a plain SCA adapter would work fine for the purpose of connecting it to a SCSI-3 UW controller because SCA2 might be a superset of SCA, or whether a plain SCA adapter wouldn't work at all with the X15, period. Sigh. If only the terminators weren't so **** expensive...

Share this post


Link to post
Share on other sites

I would like to repeat exactly what logicprobe just wrote, almost word for word. Except my WDs are 1000JBs and 800JBs and I'm not even considering installing XP on them when I love its (system partition only) performance on the X15.

I have a spare 68->80 pin converter laying around. I'd be happy to buy that 1st gen. X15 off you to match my other one, miamicanes... though you're probably better off with ebay than with some wacko from a forum :wink:

Share this post


Link to post
Share on other sites

Go to www.hypermicro.com and tell them what you have and what you want. They will sell you a converter (20bux), a cable (15bux) , and a controller (79bux).

What are you worried about? You're a man it's a machine, it's a machine. go for it. Under heavy abuse I killed 6 IDE hard drives 2 of them were JBs. SCSI handeled them fine. (I was recommened to use a RAID-5 array for what I do is not healthy for drives).

-----------------------

The difference between common sense and paranoia is in common sense you think people are out to get you. Well, it's true, that's the way the world works. In paranoia you think they are all working together to get you. :D

Share this post


Link to post
Share on other sites

Well, based upon this afternoon's encouraging news (the faster seek times do mitigate the lower raw STR vs a JB), I went ahead and ordered a SCA-2 80-to-68 adapter and Ultra160 cable w/built-in active terminator at the end.

Sigh. 36 bucks down the hole just for a cable, adapter, and terminator. Hopefully, money well spent...

Share this post


Link to post
Share on other sites
4. Tranfer Rates (winbecnh99) however on both drives shows the Cheetah starts at 60 MB/sec and can sustain that fairly well.  The WD drive starts at 50 MB/sec and has a sharper drop as the test progresses.  So I'm confused why I get a much better transfer rate through this benchmark for the SCSI drive.
Interesting. This is not my experience at all. I've never seen my X15 exceed 40MB/sec in the real world, whereas my JBs nearly hit 50MB/sec more often than you'd expect.

Doing one STR-intensive task simultaneous with another task (STR-intensive or otherwise), the X15 blows away the WD800JB. The 800JB has higher STR in real life though. So if you just compute the MD5 checksum of a .iso, the WD800JB will absolutely blow away the X15. If you do that while you're burning another .iso onto CD at the same time, the X15 will blow away the WD800JB by a similar amount.

In any case, I should try comparing my X15 and my WD800JB with a recursive diff of two huge source code trees when I get a chance (perhaps later this week or early next week). That's the most time-consuming task I care about, speed-wise, and it would be highly interesting to see how the two drives compare.

Share this post


Link to post
Share on other sites

An interesting comparison indeed, Barry. I'll watch this thread for your results.

Share this post


Link to post
Share on other sites

Get both IDE and scsi drives and use them together so when you get mad at scsi you can smile at IDE--and when you get mad at IDE then you can smile at scsi. Problem solved, dude. :D

Share this post


Link to post
Share on other sites

I'll always have a mixed system. SCSI for main boot drive for speed and responsiveness. IDE for mass storage b/c its sickening how cheap it is per mb compared to SCSI. Besides, who needs a 15K SCSI drive to play back mp3's or movies?

Share this post


Link to post
Share on other sites

SCSI most definitely is faster in heavily loaded systems with lots of disk activity. Whether you will notice the difference on an office machine is rather questionable.

Share this post


Link to post
Share on other sites

Adaptec, Mylex, Tom's Hardware, Storage Review, etc., all have information on scsi and raid concepts. Spend some time becoming acquainted with the basics so you can make better decisions. RE: your question on scsi compatability: the nice thing about scsi is that any scsi device is compatible with any scsi controller (with correct adapter) and its performance is limited to the 'weakest link' in the chain. You get backwards and forward compatibility. :P

Share this post


Link to post
Share on other sites

"Third, we discovered a coefficient between SCSI and IDE drives: one Cheetah X15 36LP equals at least four Maxtor 6L020J1"

Taken from Xbit Labs IDE RAID vs SCSI article.

Nuff said. I still can tell HUGE differences between my 1st Gen X15's and the WD JB Drives.

Share this post


Link to post
Share on other sites
Guest Eugene
"Third, we discovered a coefficient between SCSI and IDE drives: one Cheetah X15 36LP equals at least four Maxtor 6L020J1"

Taken from Xbit Labs IDE RAID vs SCSI article.

Nuff said. I still can tell HUGE differences between my 1st Gen X15's and the WD JB Drives.

The flaw in this presentation is the assumption that performance scales upwards in most typical uses in RAID0. One Cheetah X15-36LP could very well equal at least four Cheetah X15-36LPs as well.

Share this post


Link to post
Share on other sites

dhman,

it's not an apples to apples comparison, i'm sure you can find an application in which x15 1st gen won't be any faster than a 800jb.

and such quotes are non more than trendy trash you see all over hardware review sites.

how other drive's benchmarks you can fit into another drives benchmark.. this reminds me of MST3K's Crow T Robot when he had a fad with turkey volume guessing. TVGM - TURKEY VOLUME GUESSING MAN!

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this