wgreene@amelia.nas.nasa.gov (William H. Greene) (06/17/88)
I've seen very little posted about floating point hardware/software issues with Unix/Xenix on 386 machines. So, I'm performing a little survey to see what people's experiences are. Eratum 21 A few people have mentioned this well-known (?) bug in the current crop of 386 chips which occurs when a program is paging and doing floating point operations on the 387. The 386 math adapter board available from Bell Tech has also been mentioned. I'm using one of these boards in a CSS Labs 386/387 box with Microport Unix and it seems to solve the problem. I think I have also seen that there is another source for such an adapter but can't remember where. What 386 systems/motherboards experience this problem ? I'm thinking about a new machine and would like to buy one that has been designed around this problem. What about the Compaq 386/20 ? Anyone know when Intel will be selling chips with this bug fixed ? Weitek 1167 A lot of companies are talking about this coprocessor board. Is there anyone out there actually using it under Unix/Xenix ? Are there versions of Unix/Xenix that support the 1167 available now ? I know that Microport 2.2 does not but I've heard rumors about it being supported in the future. I saw an announcement from SCO about 1167 support in an upcoming version of Xenix. Microway says that they are running the 1167 with a pre-release version of ATT Unix. I've heard rumors of overheating with the 1167 and when I look at Compaq's mechanical design of their board with the massive heat sink at the top, I tend to wonder more. Anyone have any experiences with this ? Any information on these topics (or related ones) would be very much appreciated. Please email it to me directly and of course I will summarize for the net. Bill Greene
neighorn@qiclab.UUCP (Steve Neighorn) (06/20/88)
In article <682@amelia.nas.nasa.gov> wgreene@amelia.nas.nasa.gov (William H. Greene) writes: > > Weitek 1167 > > A lot of companies are talking about this coprocessor board. Is >there anyone out there actually using it under Unix/Xenix ? Are there >versions of Unix/Xenix that support the 1167 available now ? I know >that Microport 2.2 does not but I've heard rumors about it being >supported in the future. I saw an announcement from SCO about 1167 >support in an upcoming version of Xenix. Microway says that they are >running the 1167 with a pre-release version of ATT Unix. I've heard >rumors of overheating with the 1167 and when I look at Compaq's >mechanical design of their board with the massive heat sink at the >top, I tend to wonder more. Anyone have any experiences with this ? Regarding the overheating of the 1167 chip : It can be a problem, and those vendors who are not placing the 1167 chip set/80387 board in a heavily heatsinked vertical arrangement may encounter serious heat problems. A case in point : The Everex Step 386/20 machine uses one of the familiar 1167/387 sockets (the kind where the 387 sits one pin row in on every side) and places it directly below the left-hand hard drive bay. I know one installation at Tektronix that installed both the 387 and the weitek board, and the machine would work for a few minutes before crashing. The kludge solution was to cut a hole in the case and add a separate muffin fan blowing directly on the coprocessor board. This solved the heat problem, made a mess of the case, and proved that the 1167 is hot in more ways than one. Rumor has it that Everex is changing their motherboard arrangement on the Step 386/20, and going with an arrangement similar to that of Compaq. Regarding 1167 support and 386 Unix: The Green Hills C-386 C compiler has two flags referring to the 1167. -X143 specifies the generation of Weitek floating point code instead of 80387 code. -X213 specifies using the Weitek software emulation library, and requires -X143 to work. My GHC manual states that V/386 does not support the Weitek math coprocessor at this time, but that it will be supported in a future release. The version of C-386 I have is 1.8.3A. -- Steven C. Neighorn !tektronix!{psu-cs,reed,ogcvax}!qiclab!neighorn Portland Public Schools "Where we train young Star Fighters to defend the (503) 249-2000 ext 337 frontier against Xur and the Ko-dan Armada"
dave@micropen (David F. Carlson) (06/21/88)
In article <1397@qiclab.UUCP>, neighorn@qiclab.UUCP (Steve Neighorn) writes: > In article <682@amelia.nas.nasa.gov> wgreene@amelia.nas.nasa.gov (William H. Greene) writes: > > > > Weitek 1167 > > > > A lot of companies are talking about this coprocessor board. Is > >there anyone out there actually using it under Unix/Xenix ? Are there > >versions of Unix/Xenix that support the 1167 available now ? I know > >top, I tend to wonder more. Anyone have any experiences with this ? > > -X143 specifies the generation of Weitek floating point code instead > of 80387 code. > > -X213 specifies using the Weitek software emulation library, and > requires -X143 to work. > Steven C. Neighorn !tektronix!{psu-cs,reed,ogcvax}!qiclab!neighorn There is no kernel support in 2.2 SV/386 for Weitek. There is a module but it contains stubs and is only 888 bytes long! A beta kernel for an unreleased version (2.3!!! ~fall 88) contains a whole weitek driver (14k bytes). The GHC *will not* work without this kernel as floating point is protected in this 386 version of UNIX (for right or wrong). Actually, the "big" win of protected floating point is that weitek, 287, 387, etc. or emulation can be used without modifying executables (ie changing to -lm68881 as our Motorola cousins have to do.) Look for full weitek support real-soon-now. Seriously! One problem I haven't seen discussed much is that the weitek costs 2X a 387 and performs add/sub/mlt/div ~4X but floating point conversions, trig and other transcendentials much slower that the 387. What is the "real" world speed up over its Intel competitors? Not as much as one might think I posit. -- David F. Carlson, Micropen, Inc. ...!{ames|harvard|rutgers|topaz|...}!rochester!ur-valhalla!micropen!dave "The faster I go, the behinder I get." --Lewis Carroll
jimw@bucket.UUCP (Jim Willing) (06/23/88)
In article <682@amelia.nas.nasa.gov> wgreene@amelia.nas.nasa.gov (William H. Greene) writes: > Eratum 21 > > A few people have mentioned this well-known (?) bug in >the current crop of 386 chips which occurs when a program is paging >and doing floating point operations on the 387. . . >What about the Compaq 386/20 ? Anyone know when >Intel will be selling chips with this bug fixed ? Compaq has reciently released a bulletin referring to this problem, and has issued a plug in module to correct it. They claim that only the 16mhz system is affected, and the 20 & 25mhz systems do not show the problem. The update board is available at no charge from Compaq to people who have had the problem (or have the potential). The problem had never made itself real apparent on my system, but since I was running software that was prone to causing the problem (Xenix/386 + heavy co-proc use) they sent me the board. No recient info from Intel tho... BTW: The problem (when it shows up) will affect systems with 387 OR 287 co-processors... -- usually found at: ..tektronix!teksce!bucket!jimw alternatly found at CBBS/NW (503) 284-5260 never found where you would expect him to be...