CompuServe Thread

#THE Perfect System

29 messages in this thread
#190258From: MICHAEL CORTESESep 11, 1995 9:50 AM
We are looking for The Perfect System for 3D Studio 4 & 3D MAX. WE are looking for input on complete systems and/or individual components (Motherboards/Chipsets/Video Cards/BIOS Preference/) all to work well with 3DS4, 3D MAX, DPS products, Windows NT, Windows 3.1/etc. Any input or suggestions, whatsoever, is greatly appreciated. We are looking for info on any platform, but leaning towards PC(Intel). Thanks, Mike C.
#190394From: Stephen C. LevySep 12, 1995 1:34 AM
I've been researching the "perfect" system, with the hope of buying one that could take me through at least the first year of MAX. It appears that this "perfect" system does not yet exist because the only chipset that supports the new, faster EDO memory is the Triton chipset, and it DOESN'T support Symmetrical Multi-Processing. That said, the fastest Pentium with the most RAM (32 megs, minimum) and at least a 2-gig hard drive, with at least a 17" monitor, is about as good as it's gonna get until there's a successor to the Triton chipset, or the P6, or both.
#190957From: Michael LudwigSep 15, 1995 6:53 AM
PMJI, currently I'm working with P133 Triton,96MB……I tried EDO-Ram, but I haven't noticed any visible change, save the horrible price <g> Have you experienced any speedups, and what did you do for it ? Michael
#191016From: Stephen C. LevySep 15, 1995 2:58 PM
I haven't actually tried EDO yet; I'm trusting the veracity of PC Magazine (and others) test reports, which claim something like a 10% improvement — maybe 10 seconds off a two-minute frame render, if they're accurate, but to my mind, every little bit helps! What you're saying gives me pause, however. Have you tried rendering the same scene on a non-EDO 133 and seen no difference?
#191037From: Mark D. PippinSep 15, 1995 5:59 PM
Quad DEC Alpha…I feel the need for speeeeeeed……
#191096From: Stephen C. LevySep 16, 1995 3:51 AM
Aaaaaah, yes, but one of the dirty little secrets we are about to face is the fact that your apps under NT must be optimized for different CPUs. While it is likely, I hear, that a MAX for Alpha will be released, such has not been announced. Not being seriously knowledgeable about the issues at hand, I'm not even sure that a MAX for Intel NT would run at all on NT for Alpha. Perhaps some wiser head could enlighten me… SCL
#191214From: Mark D. PippinSep 17, 1995 1:56 AM
Talk to Martin Foster of Metrolight. I'm pretty good when it comes to hardware but he puts me to shame! Right now everything I want to use with NT is already out there. MS Word, Excel, and a few others. I loaded NT on my P100 at the office and everything (so far) has converted over. I ran four flics at once. Wild. Only thing that didn't was the Turtle Beach card and they are working on the NT drivers. I knew that beforehand. I noticed an increase in speed under NT for most things too. I didn't go all the way and reformat the hard disk. I wanted to keep DOS/WFWG as my main thing. The NT install was to start working with it. So now I can dual boot to DOS or NT. Frankly, the only point in doing the Alpha thing is for 3DS, otherwise I would do P6's. Drop me a note. Bye!
#191302From: Stephen C. LevySep 17, 1995 5:57 PM
Yes, I've played with machines running the same dual-boot configuration you're talking about. Perhaps I was insufficiently clear previously, but what I was asking about was whether, running NT on Alpha(s), apps needed a total recompile to run at all, or was it just an optimization issue to get the best performance. But thanx for your input!
#191337From: Mark D. PippinSep 17, 1995 11:09 PM
Supposedly, not all apps need to be recompiled. I don't know. Just the ones I've run are ok. Thanks.
#191232From: Larry MintonSep 17, 1995 8:48 AM
An Intel CPU 3DS Max will _NOT_ run on an Alpha CPU box. The CPU instruction sets are different. Instructions that exist on the Intel don't exist on the Alpha (and visa versa). The CPU and register sets are completely different. LAM
#191260From: David GouldSep 17, 1995 12:23 PM
<<Aaaaaah, yes, but one of the dirty little secrets we are about to face is the fact that your apps under NT must be optimized for different CPUs.>> That is no secret. It's really simple. You simply produce a different version of the program for each different CPU type. You simply recompile the program for a different CPU. So you end up with a different program for each different computer. You then put all the different versions on a CD and then the user installs the program that is suited to his/her particular computer. There really is no big deal in recompiling the program for different computers. Many of the compilers now produce code for different machines automatically, though often you have to spend a day or more fine tuning it for the different architectures. Gary, himself, said that they would just recompile the code on the Dec Alpha to see how well it going on the Alpha architecture. <<Not being seriously knowledgeable about the issues at hand, I'm not even sure that a MAX for Intel NT would run at all on NT for Alpha.>> A MAX version compiled for Intel processors would run on a Alpha, but not as native code, instead it would be run like the Alpha runs all Intel code, interpreted, so you wouldn't be taking full advantage of the Alpha. Instead if the program is recompiled specifically for the Alpha then it would run at it's maximum. DaviD "one in – multiple outs" GouID
#191287From: Mark D. PippinSep 17, 1995 2:34 PM
Thanks for the tip. I know Gary said they were going to support Intel only to start. Come on Alpha! Mark
#191474From: Stephen C. LevySep 18, 1995 5:17 PM
Yes, what you're saying is what I suspected. Thanx for the additional info, too.
#191387From: Michael LudwigSep 18, 1995 9:06 AM
>>same scene on a non-EDO 133 and seen no difference? exactly ! Concerning EDO – if you have no cache, EDO gives you the 10% burst, but if you want to get the best of your money, buy pipeline cache…. and forget edo, it gives around 1% if PCache is in. Michael
#191473From: Stephen C. LevySep 18, 1995 5:11 PM
Veeeerrrry interesting! You may've just saved me a bunch of $$$. Thanx!
#192581From: KYLE A. STRAWITZSep 26, 1995 3:45 AM
I'm intrested in buying a new system to run NT and 3DS MAX. Right now I am trying to make the best economic and system architecture decision I can. Dual Pentium capable systems are a bit expensive because they fall usually into the server class. With the P6 about to be released, and its true 32 bit data path I wonder if it makes any sense at all to run dual Pentiums. I am really undecided as to if I should buy a 133mhz single Pentium capable system or a 100mhz dual capable machine. Some of the vendors I have been looking at are Digital, HP, ALR for dual and Compaq for single processor systems. If you have any insight into this matter I would really appreciate it. Thanks, Kyle
#192715From: Stephen C. LevySep 26, 1995 7:14 PM
That's exactly the conundrum I'm finding myself in! Micron is advertising a dual-133 configuration that would come in right around five figures set up as a dedicated rendering slave (small monitor, plain vanilla video card, SCSI w/2gigs, 4x cd-rom but forget the audio stuff, and 128 megs o' RAM). That looks like the best dual system I've seen, not counting custom stuff. Re the P6: I'd wait 'til spring to get one, if you can wait that long… Remember the Pentium 60/66? Also, it sounds to me like MAX won't be out 'til Feb. or Mar. anyway. Of course this is just my opinion… For myself, I've decided to wait for a P150 as a stop-gap, then when MAX comes out I'll start to investigate P6 versus Alpha… Your thoughts?
#193631From: S WhiteOct 2, 1995 8:46 PM
A few comments: A recent benchmark study in Computer Shopper put the Micron 133 at the top of the heap. And most of the trades are printing anti-P6 studies. Seems that the P6 is having trouble keeping up with the midrange Pentiums. Even in native code some of the apps tested barely were up to the 133 speeds. Just a few migrant thoughts….SW
#193682From: Stephen C. LevyOct 3, 1995 1:49 AM
Thanx; I've read the same article. And that just corraborates what I've been saying all along (for the last month, anyway) about the P6. It's gonna be the P5/60/66 all over again. The earmarks are all there — (relatively) large-trace fabrication process (equals bigger, more power-hungry and hotter) and clock speeds identical to existing pieces. The fact that they're pitching it as a "server chip" means nothing, as I'm sure you know — they did that with the 286, initially. Betcha we don't see "real" P6s (.35 micron or smaller fab, 200+MHz clock speed) before Mother's Day. They just recently closed out _applications_ for beta testers. I forget the exact terminology they used but that was what it amounted to… Sounds to me like they're running behind their announced schedule, particularly if they want to protect what's left of their reputation after the Pentium FPU mess. Sorry, didn't mean to get on my soapbox, but you started it!<bg> SCL
#193987From: Mark D. PippinOct 4, 1995 11:54 AM
I have just one thing to say… GO ALPHA! GO ALPHA! GO ALPHA! Iwanta8plex300MhzDECAlphaboxwithaOpenGLcardthatsHDIcapableandaGIGoframandaPVRca dwiththree9gigMicropolisA/VdrivesandIwantitnowSanta!Ican'twaitforChristmas! Steve, I'm going to the Autodesk MAX demo in Austin TX this week. I'll get the straight poop! Mark…
#191294From: Tim Krego, [pro.image]Sep 17, 1995 3:47 PM
>Triton chipset, and it DOESN'T support Symmetrical Multi-Processing.< Triton2 chipsets will support SMP. I don't know when they will be availabel, but since the SMP 120/133 chips just came out I would say real soon.
#191476From: Stephen C. LevySep 18, 1995 5:20 PM
Thanx! This is the first I've heard of the Triton2 chipset. Perhaps I'll get on the Web and see what I can turn up on Intel's home page. I'll let you know if I find anything interesting/pertinent! SCL
#191531From: Mark D. PippinSep 18, 1995 11:05 PM
To ALL, As of today, the Triton chip set is in a holding pattern since there are bugs still in the SMP routines. So no dual or quad P6's for the moment. That is from this week's Infoworld.
#191557From: Stephen C. LevySep 19, 1995 2:07 AM
Well, you can bet somebody's having their feet held to the fire until the problems get solved! Thanx for filling me in. SCL
#192028From: Tim Krego, [pro.image]Sep 21, 1995 8:13 PM
>As of today, the Triton chip set is in a holding pattern since there are bugs still in the SMP routines. So no dual or quad P6's for the moment. That is from this week's Infoworld.< The P6 uses the Orion chipset, the Triton2 is for Pentiums. That is what I saw in Infoworld.
#192051From: Mark D. PippinSep 21, 1995 11:17 PM
Your right, dummy here made a boo boo, sorry! By the way, did you read the latest BYTE? Says P6 ain't so hot, especially for 16-bit stuff.
#192147From: Jason LangeSep 22, 1995 3:45 PM
Yes they did say it choked on 16-bit apps, but was still easily twice as fast as the Pentium with native 32-bit coded stuff. Which is exactly what NT & MAX are so their shouln't really be a problem. JAson
#192288From: Mark D. PippinSep 24, 1995 2:20 AM
Yep, you're right about the native 32-bit stuff. If you have been reading any of the threads for the past few weeks you'll see there are a lot of people interested in the Alpha. I think its going to be awile before we see quad P6's because of the Orion chip set problems. No multiprocessor boards. However, I can build a 8-plex DEC Alpha 275Mhz right now. Now if Adesk will just release the native Alpha version… I think Intel will ask a premium for the P6 at least for the first half of '96 so if possible I'm going Alpha. We'll see… Drop me a note. Mark
#192294From: David GouldSep 24, 1995 3:12 AM
<<Yes they did say it choked on 16-bit apps, but was still easily twice as fast as the Pentium with native 32-bit coded stuff. Which is exactly what NT & MAX are so their shouln't really be a problem.>> These are exactly my sentiments as well. 16-bit applications will be slowly phases out so it is only natural that all the latest processors be optimized to run 32-bit at maximum performance. But in any case since the P6 will be faster than the P5 the lost of speed using 16-bit applications will be made up by the difference in performance on the P6, i.e. 16-bit applications will run faster on a P6 than a P5. DaviD "adding a few more bits to the puzzle" GouID