heirich@beowulf.ucsd.edu (Alan Heirich) (07/31/90)
About a month ago someone posted information about bargain priced 80287. If I recall correctly, they were available for $100 each from a company in Chicago (?) Has anyone had experience using one of these chips in a 12 MHz AT? If so, can you you tell us if you had any problems, and repost the address/phone number of the supplier? thank you, Alan Heirich heirich@aurel.cns.caltech.edu ============================================ Program in Computation and Neural Systems California Institute of Technology Pasadena, CA 91125
dlg6627@uxa.cso.uiuc.edu (07/31/90)
>About a month ago someone posted information about bargain priced >80287. If I recall correctly, they were available for $100 each >from a company in Chicago (?) Look at PC Week. It is advertised in this magazine. >Has anyone had experience using one of these chips in a 12 MHz AT? >If so, can you you tell us if you had any problems, and repost the >address/phone number of the supplier? I believe my friend is in the process to order one. I forgot the address/phone number. /* End of text from uxa.cso.uiuc.edu:comp.binaries.ibm.pc.d */
jgarland@kean.ucs.mun.ca (07/31/90)
In article <1238400005@uxa.cso.uiuc.edu>, dlg6627@uxa.cso.uiuc.edu writes: >>About a month ago someone posted information about bargain priced >>80287. If I recall correctly, they were available for $100 each >>from a company in Chicago (?) > > Look at PC Week. It is advertised in this magazine. > > >>Has anyone had experience using one of these chips in a 12 MHz AT? >>If so, can you you tell us if you had any problems, and repost the >>address/phone number of the supplier? > > I got one here in Canada. As I reside in a rather remote location, I asked them to send it by mail. It arrived within 12 days (and Customs didn't even sock me for our 12% Federal Sales Tax, nor any duties if any are still around these days). It works fine on my Dell 200 12.5 Mz machine (AT clone). This machine can be set to run the coprocessor at 2/3 CPU speed *or* at CPU speed. AMD says that there aren't any heat risks (due to the fact it is a CMOS chip) from running the chip fast, but that there is a risk of errors at speeds greater than 10.75 Mz as that is the cutoff point for flunking their test suite. Experimenting on FRACTINT v.13 and Quattro,I have found no errors but also that the difference between running the 287 at 8.3 vs. 12.5 Mz is only about 10%. As such, I've decided to go with the slower speed as the big difference is just plain *having* the coprocessor in the machine in the first place. You are reading one satisfied customer's reply. John Garland Bitnet: jgarland@mun Internet: jgarland@kean.ucs.mun.ca
sonny@charybdis.harris-atd.com (Bob Davis) (07/31/90)
In article <12085@sdcc6.ucsd.edu> heirich@beowulf.ucsd.edu (Alan Heirich) writes: >About a month ago someone posted information about bargain priced >80287. If I recall correctly, they were available for $100 each [DELETIONS] >Has anyone had experience using one of these chips in a 12 MHz AT? >If so, can you you tell us if you had any problems, and repost the >address/phone number of the supplier? > HERE IS INFO I SENT A WHILE BACK TO OUR LOCAL PC USERS: Date: Mon, 2 Jul 90 11:22:52 EDT Subject: CHEAP ($99 half-price) 80C287 Math Coprocessors Call AMD at 1-800-888-5590 about their $99 deal on 10 MHz 80C287 Coprocessors. Saw this in #30005 comp.sys.ibm.pc. Accept VISA, MC. AMD = Advanced Micro Devices, Inc. Austin, TX These are CMOS chips requiring less power than vanilla 80287. The going rate in Computer Shopper for the last several months has been $215 or so. These are 10 mHz chips, suitable for use with 80286 running 12 mHz or less. Compatible with Intel 80287 (but NOT with 80287A for some reason). 30 day warranty. They will refund money for defective chip within first 30 days. LIFETIME GUARANTEE: After 30 days, AMD will replace defective chip. I could not resist. +6% tax + $6 regular mail shipping. I RECEIVED THE CHIP VIA REGULAR MAIL IN 10 DAYS AND THEN POSTED THIS LOCALLY: I installed the AMD 10MHz 80C287 Math Coprocessor in my 12MHz 80286 machine last night. This chip, which AMD is currently selling for $99 (compare INTEL 80287 @ $210), passes INTEL's CHKCOP v. 2.00 Coprocessor diagnostic. It also passes MicroWay's 87TEST 8087 diagnostic. AMD's chip is touted as being completely compatible with the INTEL 80287. I ran my benchmark of 128 1024-point FFT's after installing the 80C287 and got the following results: No Coprocessor : 945 seconds With Coprocessor : 202 seconds This represents a speedup by a factor of 4.7. The AMD 80C287 CMOS coprocessor felt barely warm to the touch. So far, I'm happy. ______________________________________________________________________________ Bob Davis \\ INTERNET : sonny@trantor.harris-atd.com | _ _ | Harris Corporation, ESS \\ UUCP : ...!uunet!x102a!trantor!sonny |_| |_| | | Advanced Technology Dept.\\ AETHER : K4VNO |==============|_/\/\/\|_| PO Box 37, MS 3A/1912 \\ VOICE : (407) 727-5886 | I SPEAK ONLY | |_| |_| | Melbourne, FL 32902 \\ FAX : (407) 729-2537 | FOR MYSELF. |_________|
lhughes@b11.ingr.com (Lawrence Hughes) (07/31/90)
I just received my $99 AMD 80c287 chip - although they "lost" my first order, it showed up two days after I checked back in to see what happened.... The chip installed in my 10MHz AST Premium 286 with no problem - it operates at 2/3 the 286 clock speed, so actually this part should work ok in systems with up to 15(?)MHz clocks (you might get by with 16MHz.. haven't tried it). I found factor of 2 to 6 speed-ups over 286 alone, depending on how FP intensive each APP was... well worth the $99 price... AND NOW FOR SOMETHING COMPLETELY DIFFERENT (or at least I haven't heard anyone mention this yet): I also have a JDR 386/25 MB that has both 387 and 287 FPP sockets - the manual says you need a 25MHz FPP (!), but if you check with JDR, they confirm the info in the Intel books that indicate you CAN do this if you run the 287 socket with a lower speed clock (AIN'T no such thing as a 25MHz 287...). I tried my 10 MHz AMD 80c287 in this socket, and by golly, not only did it work, but all timing tests showed about a 1.5 factor improvement over the 286/287 combo... basically on FP intensive stuff, typical timings were: System Time (sec) 10 MHz 286 alone 10.0 25 MHz 386 alone 4.5 286 / 287 3.2 (287 actually running at 6.67 MHz?) 386 / 287 2.4 The improvement in the 286 was greater, but the ABSOLUTE speed in the 386 was even greater - gee-- which one should I leave it in? Long term no question, AMD is rumored to be offering a 387 deal soon... Also, Windows 3.0 (and everything else I tried) found the "80x87" and used it correctly in both cases (with 286 and with 386). BTW - anyone know how the AMD part compares with Intel, Weitek, IIT or other 287s? Some of them claim higher throughput at same clock compared to Intel... For now, however, call 800-888-5590 for the 80c287 offer...
brad@looking.on.ca (Brad Templeton) (08/02/90)
But what about the difference in speed between the AMD 287 and the Intel 287? Everybody's posting speed diffs between software math and AMD's 287. That's boring, because if you knew how bad most of the software math packs are, you would not be impressed by a 5 time speedup. -- Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473
davidsen@sixhub.UUCP (Wm E. Davidsen Jr) (08/02/90)
Enough
Let's move this to the comp.sys.ibm.pc.hadware or somewhere. It has
nothing to do with any binaries posted here.
--
bill davidsen - davidsen@sixhub.uucp (uunet!crdgw1!sixhub!davidsen)
sysop *IX BBS and Public Access UNIX
moderator of comp.binaries.ibm.pc and 80386 mailing list
"Stupidity, like virtue, is its own reward" -mesonny@charybdis.harris-atd.com (Bob Davis) (08/02/90)
In article <1990Aug01.175639.10728@looking.on.ca> brad@looking.on.ca (Brad Templeton) writes: [Some stuff deleted] > >Everybody's posting speed diffs between software math and AMD's 287. That's >boring, because if you knew how bad most of the software math packs are, >you would not be impressed by a 5 time speedup. >-- >Brad Templeton, ClariNet Communications Corp. -- Waterloo, Ontario 519/884-7473 Yes. Sometimes the SOFTWARE emulation of FULL coprocessor hardware-precision is VERY slow. Perhaps there is no way around this--I do not know. What I do know is that in Microsoft FORTRAN, doing full-blown coprocessor accuracy using software emulation routines, a certain floating point-intensive benchmark of mine takes: 945 seconds (full coprocessor accuracy FP software routines) Adding the AMD 80C287 coprocessor reduces the time by a factor of 4.7: 202 seconds (80C287 coprocessor, full floating point accuracy). HOWEVER, if one accepts the reduced accuracy available with the Microsoft FORTRAN Alternate Math Library (/FPa option), the benchmark time WITHOUT A COPROCESSOR drops to: 290 seconds (No coprocessor, reduced accuracy FP routines). I have found in several spectral analysis applications making heavy use of floating point-intensive FFT analysis routines, that the accuracy using /FPa is completely adequate. For such applications where one can use the reduced accuracy software floating point routines, one will be much less fascinated with the speedups associated with an expensive coprocessor (in my case, quoted above, I went from 290 seconds with no coprocessor, to 202 with it--both at acceptable floating point accuracies). The bottom line is: use reduced accuracy, hence faster, software routines if you can accept the results. You may find that a coprocessor is more of a luxury than you first thought. However, if money is no major obstacle, go for the coprocessor--there WILL be applications where the full floating point accuracy is required. ______________________________________________________________________________ Bob Davis \\ INTERNET : sonny@trantor.harris-atd.com | _ _ | Harris Corporation, ESS \\ UUCP : ...!uunet!x102a!trantor!sonny |_| |_| | | Advanced Technology Dept.\\ AETHER : K4VNO |==============|_/\/\/\|_| PO Box 37, MS 3A/1912 \\ VOICE : (407) 727-5886 | I SPEAK ONLY | |_| |_| | Melbourne, FL 32902 \\ FAX : (407) 729-2537 | FOR MYSELF. |_________|
doerschu@rex.cs.tulane.edu (Dave Doerschuk) (08/02/90)
>About a month ago someone posted information about bargain priced >80287. If I recall correctly, they were available for $100 each >from a company in Chicago (?) Here's the scoop on the AMD 80287-compatible coprocessor. I ordered one Monday, and am expecting it Friday. Company: Advanced Micro Devices (AMD). Chip: 80C-287 Math Coprocessor (low-power CMOS). Speed: 10Mhz (since 2/3 rule applies to 80286 machines, this chip will run in machines where the CPU runs up to 16 MHz. Yeah, I know, 2/3 * 16 = 10.66, but the AMD guy claimed that the chip was good to this tolerance (only 6.6% out of spec).) Compatible: Supposed to be 100% compatible; since someone on the net mentioned that AMD was using Intel code under license, I would imagine that they are as compatible as you can get. Shipping: In stock, shipping immediately. Warranty: Lifetime (maximum two replacements) Price: $99.00 Tax: They add state tax to the price. Phone: 1-800-888-5590 Credit: Visa/MC accepted at no extra charge, COD add $4.00. Other speeds may be available but I didn't ask. Hope this helps, good luck! Dave Doerschuk
cth_co@tekno.chalmers.se (Christer Olsson. MedNet, G|teborgs Universitet) (08/08/90)
In article <8405@b11.ingr.com>, lhughes@b11.ingr.com (Lawrence Hughes) writes: > AND NOW FOR SOMETHING COMPLETELY DIFFERENT (or at least I haven't heard anyone > mention this yet): I also have a JDR 386/25 MB that has both 387 and 287 FPP > sockets - the manual says you need a 25MHz FPP (!), but if you check with JDR, > they confirm the info in the Intel books that indicate you CAN do this if you > run the 287 socket with a lower speed clock (AIN'T no such thing as a 25MHz > 287...). I tried my 10 MHz AMD 80c287 in this socket, and by golly, not only > did it work, but all timing tests showed about a 1.5 factor improvement over > the 286/287 combo... basically on FP intensive stuff, typical timings were: Most 386 mothercards with both 387 and 287 sockets runs the 387 syncronous with the 386 (at same speed) and the 287 asyncronous (slower speed). 8Mhz are very common and the timings below seems to showing the 386 / 287 running with the 287 at 8Mhz. > > System Time (sec) > > 10 MHz 286 alone 10.0 > 25 MHz 386 alone 4.5 > 286 / 287 3.2 (287 actually running at 6.67 MHz?) > 386 / 287 2.4 If the 386-mothercard are made with CHIPS & TECH chipsset, it's maybe possible to change the 287 clock source beetwen 8Mhz bussclock, 1/2 cpu-clock or 1/3 cpu-clock. On my old 386, I can set FPP-speed to 1/2 cpu-clock (20Mhz) which gaves me a FPP-speed at 10Mhz. 1/3 (20/3 = 6.67) are also possible and even 8Mhz from the bussclock. If I run the CPU-clock at 16Mhz, it's possible to run the FPP at slowy 5.33 Mhz :-)