Date: Sat, 20 Dec 97 08:35:22 EST From: Dwight McKay (The Moderator) Reply-To: Suns-at-Home@net-kitchen.com Subject: Suns-at-Home Digest V10 #44 To: Suns-at-Home-List Suns-at-Home Digest Sat, 20 Dec 97 Volume 10 : Issue 44 Today's Topics: IPC simms, SS10 vs SS5 monitor, simms, and mbus modules Ross HyperSPARC modules and SX graphics for SS10 - where to buy? Running Solaris R2.6 on older equipment (IPC/SS2/IPX/LX)? screen color question (2 msgs) Solaris 2.3 printed doc. sets for sale Suns-at-Home Digest V10 #44 (2 msgs) X Windows doc. set for sale +--------------------------------------------------------------------+ | Submissions: suns-at-home \ | | Requests: suns-at-home-request > @net-kitchen.com | | Archives: suns-at-home-archives / | | WWW Archive access: http://www.net-kitchen.com/~sah | +--------------------------------------------------------------------+ ---------------------------------------------------------------------- Date: Sun, 14 Dec 1997 23:29:49 -0800 (PST) From: James Lockwood Subject: IPC simms, SS10 vs SS5 To: Dwight McKay > >Believe it or not, the SM51 and SM61 CPUs seem to be quite a > >bit faster than the MicroSPARC chips found in the SS4 and SS5. > > Depends on the workload, but yes, for many things, the SM51 and SM61 CPUs > are faster than even the 110MHz MicroSPARC II in the SS4 and SS5. > > >We have an SS4 at work with the 110Mhz MicroSPARC CPU, and my home SS20 > >with the SuperSPARC SM61 feels almost twice as fast. > > Hmm, "twice as fast" is a bit of an exaggeration, I'd expect, but I suppose > there might be some jobs in which it'd be close to that. Especially since the Viking CPU's have such a sophisticated prefetch and caching scheme. They were amazingly complex and advanced for their time. FYI, here are some SPEC numbers in case anyone is still interested: System CPU ClkMHz Cache SPECint SPECfp Info Name (NUMx)Type ext/in Ext+I/D 92 92 Date ================= ========== ======= ========== ======= ======= ===== Sun SS20/51 SuprSP 40/50 1M+20/16 81.8 89.0 May95 Sun SS20/61 SuprSP 50/60 1M+20/16 98.2 107.2 May95 Sun SS5/70 MicroSP2 70 16/8 57.0 47.3 Mar94 Sun SS5/85 MicroSP2 85 16/8 65.3 53.1 May95 Sun SS5/110 MicroSP2 110 16/8 78.6 65.3 May95 The SS10/51 and /61 numbers are a little lower, but still quite good. My SS10/61 feels faster than the SS5/110 I used to have at work. > >SS10 also suffers more heat problems than the SS20, due to the > >internal layout of the board and fan units (SS20 should generally supply 3 > >fan units). > > You've got the two machines reversed. The SS20 suffers from more heat > problems than the SS10, because Sun only put two fans in the SS20 (there were > three in the SS10), to save money. Sun had a very difficult time making the > SS20-514 work properly, due to heat problems; eventually, they had to release > a special set of modules with better heat characteristics. The SS10-514 > worked fine from the outset. Frankly, if I were Sun, I'd have scrapped the > SS20-514 and just sold the SS10-514 -- it was just as quick, and I hardly > think a lot of people would be using a 514 as an SX workstation. This is 100% correct. When Sun first introduced the Turbo-ZX they found that it worked just fine in a SS10, but they had to replace the side-panels of the SS20 to keep it from melting down. IIRC, the TZX when specced for the SS10 would even run with only one auxiliary cooling fan. Agreed also on the SS20-514, since the mbus speed is 40MHz with the 51 series modules there was no benefit to the SS20 over the SS10SX. > >The MicroSPARC CPU in the SS5 I do not believe is easily upgradeable > >except later versions (85mhz +) I think are socketed, allowing > >up to 170Mhz. > > No, the 170MHz upgrade uses a motherboard swap, not a processor swap. True, but Fujitsu sells a 160MHz TurboSPARC upgrade that plugs directly into the CPU socket (the only other mod needed is a new boot PROM). A few months ago (before the 170MHz motherboard tradeins from Sun were quite as cheap) I upgraded 4 boxes that need to remain on SunOS 4 for at least another year, they have performed flawlessly. Be warned though, some (most) of the 85MHz and 110MHz motherboards do not have socketable processors. > >Personally, I feel the SS10 is better, but the SS5 is much newer technology. > >It was this factor alone (and the 24bit SX option) > >that made me go for the SS20 rather than the SS10 > >even though it was (not quite) double the price. > > Newer technology is important, but it's more important to consider the full > ramifications of the technology used in those systems, and its geneology, not > just how new the particular machine is. The SS20 is just a minor tweak of > the SS10, which in turn was a tweaked 600 with the VME bus removed; there's > not much real difference between them. The 600 was a revolutionary machine > for its time, and, in many ways, the design has aged well. Similarly, the > SS4/SS5 is a scaleup of the Classic/LX, which was a very uninteresting > machine for its time; its only interesting feature was low manufacturing > cost. In other words, the SS5 is a newer model of a technically > uninteresting family of machines, while the SS10 is an older model of a > technically advanced family of machines. Frankly, I prefer the SS10. I agree here. I remember when the 600 came out and I thought "wouldn't it be neat if they built a 4/610?". This is basically what the SS10 is. The SS10 was a top of the line desktop machine when it was built, and the quality shows through well. The SS20 was also a top of the line machine, but by that time Sun had decided that the Ultra was the way to go and that mbus would be dead for new products. Keep in mind that a fully decked out SS10 (4 200MHz HyperSPARCs) can give any desktop Ultra a run for its money. No SS5 will ever come close. I'm willing to bet that in 6 months mbus modules will have dropped to half their current price in the used market. As far as newer technology being better goes, I just retired a 4/330 at home yesterday. This system originally retailed at around $30k in 1989, and the fact that it was still doing useful work 8 years later is a testament to quality engineering. -James [56 SIMM slots go a long way] ============================================================================= James D. Lockwood The Getty Information Institute System Administrator 1200 Getty Center Drive, Suite 300 james@gii.getty.edu Los Angeles, CA 90049-1680 - ------------------------------ Date: Mon, 15 Dec 1997 16:19:29 -0500 From: adh@an.bradford.ma.us (Sandwich Maker) Subject: monitor, simms, and mbus modules To: Suns-at-Home@tigger.net-kitchen.com - ------------------------------ Date: Thu, 11 Dec 1997 01:37:11 -0800 From: Jeff Fredrickson Subject: monitor & 3/60 boot error questions I'm (still) looking for a monitor for my SPARCclassic which unfortunately has been gathering dust for months since it lacks one... Does anyone have any comments on the 17" GDM-1662B (part # 365-1330)? What about the 15" Nokia monitor (part # 365-1286)? Are there any monitors I should avoid? you can use a sony svga monitor, if you can find one of those 13w3-hd15 connector adapters. i've got such a setup and love it. And finally... if I can't fix the 3/60, can I use its simms in a SPARCstation 1, 1+ or 2? afaik yes. - ------------------------------ Date: Tue, 9 Dec 1997 10:54:23 -0500 From: "Helen C. O'Boyle" Subject: Ross HyperSPARC modules and SX graphics for SS10 - where to buy? Be that as it may, I will soon have an SS10-51. My immediate thought was, "Gotta upgrade this." A bit more thought netted some questions, since I've never been around an SS10 that anyone's bothered to upgrade: 2. When one buys a HyperSPARC module, does one _add_ it to what's already there, or does one take out the original Sun processor module? (I've never been around an SS10 with more than one processor, either.) all cpu modules must be identical. are you running 4.1.4? if so, you can install a second supersparc 51 module. we had such a machine at my previous job - the module was floating around after an old upgrade of another machine to dual hypersparcs, so we stuck it in when we upgraded to 4.1.4. worked fine; so did a ss20-612 with similar raison d'etre. both had uptimes of over three months at the time i left, and they were running heavy chip cad/sim software. of course it applies to solaris2 also. also remember that clock for clock, ti supersparc modules are much faster than ross hypersparcs. [eg you have to go to ross 125s to beat ti 85s] - ------------------------------ ___________________________________________________________________________ Andrew Hay Dear Diary: Late to work today -- internet rambler Watch stopped, thought time did too. adh@an.bradford.ma.us Was I ever embarrassed! - ------------------------------ Date: Sun, 14 Dec 1997 23:10:01 -0800 (PST) From: James Lockwood Subject: Ross HyperSPARC modules and SX graphics for SS10 - where to buy? To: Dwight McKay > Hi, > > I seem to have acquired an SS10 for the same couple-of-hubcaps price a > previous poster reported for his IPX. :-) :-) Of course, with my luck, I'm > living in the backwoods and the fact that Ameritech offers flat-rate ISDN > will be irrelevant to me until 1999, unfortunately... Geez, I paid about $800 for a bare SS10 (sans module) less than 6 months ago and considered it quite a good deal. :( > Be that as it may, I will soon have an SS10-51. My immediate thought was, > "Gotta upgrade this." A bit more thought netted some questions, since I've > never been around an SS10 that anyone's bothered to upgrade: First of all, try it out with enough memory. A SS10/51 is just a trifle slower than a SS5/110 for most things, and is IMHO quite usable. > 1. In a previous digest post, someone said Ross HyperSPARC modules were > falling in price. Where? The prices on the Ross page are in almost all > cases more than I paid for the entire 10-51 with monitor and a couple gig of > disk, and it makes an upgrade sound cost-prohibitive. That was I. For a while MemoryX (http://www.memoryx.com) had 90MHz HyperSPARCs for $180/ea which was quite a steal. They've run out, but with the introduction of some new Sun models in January (I'm not under NDA, so I can chat about the Ultra-5 if I feel like it :) I expect mbus module prices to go down considerably. > 2. When one buys a HyperSPARC module, does one _add_ it to what's already > there, or does one take out the original Sun processor module? (I've never > been around an SS10 with more than one processor, either.) Ok, crash course in 2-slot mbus suns (Galaxy, SS10, SS20) follows: You've got 2 mbus slots, each of which can hold an mbus module. Each module can hold 1 or 2 CPU's. Since you've got a SS10/51, you've got a SS10 with one SM51 mbus module, which is a 50MHz Viking SPARC CPU with 1mb ecache. For upgrades, you could either slap another SM51 module in there (all SM51's are MP-capable, some SM40's and below are not) and get a 2-cpu SS10 system, or ditch the SM51 module and get one or more new modules. Another SM51 would be cheap (<$200) but would only increase your total processing power for tasks that can be spread across CPU's (good for multipurpose servers, not as good for desktop workstations). A 100MHz HyperSPARC would be a bit more expensive (closer to $300) but would be probably 40% faster. To quote a bracket-racing friend of mine, "speed costs money, how fast do you want to go?". IMHO if you've got 64mb or less the first thing you should spend money on is memory. > 3. Where is a good place to find the SX graphics option (used is fine)? > The machine has a TGX, but alas, that's 8-bit.... Don't knock the TGX, it's quite a decent 8-bit framebuffer. If you want 24-bit graphics, the best way is IMHO to go with the MGX+, which can now be found used in the $500-600 range. The sbus SX seems to be a rare bird, I haven't seen many in the used market. > Even sans upgrades, an SS10 is going to be quite a step up. My current > 4-years-old-to-me Sun machines are 1+'s. :-) Heh. My SS10/61 was quite an upgrade from my SS2 and 4/330's. :) -James [but even my SS1+'s make darn good PPP firewalls] - ------------------------------ Date: Mon, 15 Dec 1997 14:23:47 -0500 From: Ken Hansen Subject: Running Solaris R2.6 on older equipment (IPC/SS2/IPX/LX)? To: Suns-at-Home@tigger.net-kitchen.com Hello, I am curious if anyone has run Solaris R2.6 on older Sun equipment (IPC/SS2/IPX/LX) - if so, has performance improved vs. Solaris 2.6 and/or SunOS 4.1.X? I am considering going to Solaris R2.6 for the new goodies, but curious if it will make these machines *even* slower. Thanks in advance, Ken khansen@njcc.com - ------------------------------ Date: Sun, 14 Dec 1997 12:48:29 -0500 From: "Joseph E. Vornehm Jr." Subject: screen color question To: Suns-at-Home@tigger.net-kitchen.com In response to someone's question about Netscape and color maps, this is a problem on lots of *nix platforms, esp. with 8-bit color. There are a few command-line options to alleviate the problems; try "netscape -help" to see them. Here are the ones I find useful: -install to install a private colormap. -no-install to use the default colormap. -ncols when not using -install, set the maximum number of colors to allocate for images. I use -install, but some people don't like the effect it has on the rest of the screen when Netscape has the focus. (BTW, I think -no-install and -ncols are new to NS 3.x, and weren't present in v2.x.) Hope this helps! --Joe Vornehm jvornehm@ece.neu.edu - ------------------------------ Date: Tue, 16 Dec 1997 05:31:25 -0800 (PST) From: David Wolfskill Subject: screen color question To: gary@sabot.com >Date: Fri, 5 Dec 97 08:55:24 EST >From: gary@sabot.com (Gary Sabot) >Is there any way to deal with this (either recover when netscape >messes things up, or ask netscape not to mess things up)? What I do to try to mitigate this aspect of excessive resource consumption on the part of a Netscape product is to always invoke the Web browser with the argument "-install". That argument tells the beast to use a private colormap. As a result, I get the "color flashing" as I switch focus from the Web browser window to something else (& back), but at least it interferes (otherwise) minimally with other things I run. david -- David H. Wolfskill david@dhw.vip.best.com As a computing professional, I believe it would be unethical for me to advise, recommend, or support the use (save possibly for personal amusement) of any product that is or depends on any Microsoft product. - ------------------------------ Date: Fri, 19 Dec 1997 08:51:17 -0800 (PST) From: rodgers@cgl.ucsf.EDU Subject: Solaris 2.3 printed doc. sets for sale To: Suns-at-Home@ecn.purdue.edu Dear Fellow Netizens, We have two nearly complete sets of Sun Solaris 2.3 manuals for sale: These are in mint condition, many of them still in their original shrink-wrap bundles. Although we will consider selling them singly, we would prefer to sell them as sets. Send me email if you want a complete list of the individual documents in the sets. We are asking $150 per set. The books (weighing approx. 65 lbs. per set) can be shipped by any means specified by the buyer, who will be expected to cover shipping costs. Regards, Rick Rodgers (rodgers@maxwell.ucsf.edu) - ------------------------------ Date: Mon, 15 Dec 1997 09:07:28 -0500 (EST) From: Rick Leir EPS Subject: Suns-at-Home Digest V10 #44 To: Dwight McKay > But even after I exit netscape, regular applications that usually work > fine now have trouble allocating the 1 or two colors that they need > (i.e. emacs can't allocate the color to highlight headers in email). > > Is there any way to deal with this (either recover when netscape > messes things up, or ask netscape not to mess things up)? This surprizes me. Colors get deallocated even if the client crashes. Remember that Netscape takes a while to exit. If you still have a color problem, use xcolor(1) to debug it. You will see Netscape allocate its colors. Afterwards, put the pointer into the xcolor win to see what colors are free. cheers -- Rick Rick Leir rleir@igs.net http://www.igs.net/~rleir 613-828-8289 7951 Rocinante It's not the size of the dog in a fight that counts, but the size of the fight in the dog. -- Jim Barksdale - ------------------------------ Date: Mon, 15 Dec 1997 09:18:25 -0500 (EST) From: Rick Leir EPS Subject: Suns-at-Home Digest V10 #44 To: Dwight McKay > Be that as it may, I will soon have an SS10-51. My immediate thought > was, "Gotta upgrade this." A bit more thought netted some questions, Why? for most home use, you are flying. I use one at work all day. If you could get me one for a nod and a wink I would be grateful! Do you have an application for 24 bit colour? All the apps I use are fine with 8 bit colour, though I admit that surfing would be more fun with 24. cheers -- Rick Rick Leir rleir@igs.net http://www.igs.net/~rleir 613-828-8289 7951 Rocinante It's not the size of the dog in a fight that counts, but the size of the fight in the dog. -- Jim Barksdale - ------------------------------ Date: Fri, 19 Dec 97 14:18:27 PST From: rodgers@maxwell.ucsf.EDU (R. P. C. Rodgers, M.D.) Subject: X Windows doc. set for sale To: Suns-at-Home@ecn.purdue.edu Dear Fellow Netizens, We have the following O'Reilly X manuals for sale. They are in pristine condition. We are asking $6.00 per book or $45.00 for the set: Volume 0. X Protocol Reference Manual for X Version 11 Release 4 (1990) Volume 1. Xlib Programming Manual for Version 11 (1990) Volume 2. Xlib Reference Manual for Version 11 (1990) Volume 5. X Toolkit Intrinsics Reference Manual for X11 Release 4 (1990) Volume 6a. Motif Programming Manual for OSF/Motif Release 1.2 (1994) Volume 6b. Motif Reference Manual for OSF/Motif Release 1.2 (1993) Volume 7. XView Programming Manual for XView Version 3 (1991) Volume 7. XView Programming Manual An OPEN LOOK Toolkit for X11 (1990) Companion to Volume 7. XView Reference Manual for XView Version 3 (1990) The books can be shipped by any means specified by the buyer, who will be expected to cover shipping costs. Regards, Rick Rodgers (rodgers@maxwell.ucsf.edu) - ------------------------------ End of Suns-at-Home Digest ******************************