CompuServe Thread

#3DS Benchmark

21 messages in this thread
#193341From: douglas drossmanSep 30, 1995 7:38 PM
Hello there. I read the thread and thought you might be interested in my results. System info is as follows. SuperMicro CMS with 512K Pipe-line Busrt cache, 32 megs Ram(non-EDO), and a P-133. I used both Mapping-on and Mapping-off. Pixel sixe was 1.1 and resolution set at 640×480. I had display ON so maybe its a little faster but I doubt it since Im running a Diamond Stealth64 VRAM. Anyway here's what I got. MAPPING ON 1:05 ———–>Both with MAPPING OFF :49 ———–>Display on. Oh yea, Metal shading selected. Let me know what you think. Im using Qemm 7.0 but with EMM386 I feel I'd get the same results. Later. BTW, I should be getting the P-150 as soon as its available so the new results should be interesting. Douglas Drossman
#193404From: Stephen C. LevyOct 1, 1995 1:56 PM
1:05!?!?! THE NEW RECORD!!! That SuperMicro mobo must be THE ONE! I'm also waiting on the P150… Where did you get your 133? BTW, if you're interested in seeing how fast you can go (and who isn't?<G>) I've had luck on several machines with no memory manager at all — set up a separate boot menuitem that just invokes minimal TSRs (SCSI and video for me) to let you run 3DS. Everything loads low (No himem.sys or EMM386). If this setup leaves you with 250K low DOS available, it should shave a few more seconds off per-frame time. I only use it for big renders since you've got to reboot for any Winapps, but it's saved me 2-1/2 hours on a 1000-frame render! Thanx for the info. SCL
#193419From: douglas drossmanOct 1, 1995 3:48 PM
Hey there. I got the P-133 at a local hardware place in Queens, New York. My system is custom assembled so the parts are all premium. I feel the biggest speed boost besides the pentium chip itself, is the 512K Pipeline burst cache. (8ns.) It really speeds things up. Also, the P6 should be coming around DEC. to JAN. I might spend the money and go for a Quad P6 Processor Motherboard and start off with just one P6@150 Mhz. to start. I'll speak to you later. Bye. Douglas Drossman 74032,3552
#193431From: Stephen C. LevyOct 1, 1995 5:27 PM
Well, if you're in N.Y., I don't think I'll be patronizing your assembler. (I'm in L.A.) Are you on Intel's press release e-mail list yet? Their ads all have their web page address, and from there it's quick'n'easy to be listed. Re: P6s, IMHO, it'd be a good idea to wait until the second-generation (faster) chips come out. They've said that the first 150MHz chips will be built at their .5 micron fab, whereas later ones will go to .35 micron. (Kinda like the P60/66 to P90/100) Thanx, SCL
#193517From: Rick MillerOct 2, 1995 11:02 AM
Douglas, I just knew I should have bought the P133<BG>, I guess you don't need too many of systems when they run that fast. How much do you want for your P133 when you upgrade to the P150? I went from a 386-20 to my Pentium 100, when I rendered another file for comparison, the Pentium was 28 times faster. I don't know how I lasted so long on the 386-20, can't go back now. Regards………Rick – Technical Animations
#193535From: Stephen C. LevyOct 2, 1995 1:09 PM
Rick, 28 times?!?! Geez, and I've been snivelling 'cuz the P133 times are _almost_ 5 times faster than my poor, old DX2/66! <g> SCL
#193725From: Rick MillerOct 3, 1995 9:15 AM
Stephen, I didn't have to wait till I was rendering to take a break, just do a redraw on all four screens would do it.<BG> Regards…………..Rick – Technical Animations
#193861From: douglas drossmanOct 3, 1995 8:13 PM
I trade in my old cpu's for the new ones. A deal I have with my hardware place. I just pay the difference between the two chips. I expect the 150 to be out by Nov. So I expect a major speed increase. Looking foward to it! Later. Douglas Drossman 74032,3552
#193930From: Stephen C. LevyOct 4, 1995 1:44 AM
Douglas, Yeah, I'm hoping for Nov. too. Just out of curiosity, why do you expect the 150 to be a major increase? I'd expect a good 150 to (barely) break the 1-min. barrier on Chevy — Do you know something I don't? Thanx, SCL
#194248From: douglas drossmanOct 5, 1995 11:42 PM
Well I compared the speed increase from to 120 to the 133, a difference of only 13 Mhz. and yet a significant increase in performance. Therfore, the jump from 133 to 150, ie: 17Mhz. should yield a more significant boost. Also I read somewhere that the 150 will be a newer design rather than the older pentium design. Anyway, only time will tell. Speak to you later. Douglas Drossman 74032,3552
#194264From: Stephen C. LevyOct 6, 1995 2:38 AM
Douglas, Yeah, I think this rumor of a new chip design qualifies as something I didn't know! You could well be right — this could be more than just an incremental improvement — added impetus for me to wait as long as I can! Thanx again, SCL
#193958From: Rick MillerOct 4, 1995 9:35 AM
Douglas, Sounds like a good deal for you, thanks for getting back to me. Regards……….Rick – Technical Animations
#194249From: douglas drossmanOct 5, 1995 11:46 PM
Umm, what exactly did I tell you? I hope I didn't blow you off. Let me know, ok? Later DMD
#194299From: Rick MillerOct 6, 1995 10:52 AM
Douglas, No problem on this end. I was only partly serious about about buying your P133? when you upgraded. Regards…………….Rick – Technical Animations
#194662From: Timothy P. SmithOct 9, 1995 11:37 PM
Here is a new benchmark, Super Micro P55CMS 100Mghz, w/256k 8ns PBC. Win95 clean boot except mouse driver. Phong Res. All Off All On Shadow Off Just Mapping ____________________________________________________________ 640 0:20 1:22 1:06 0:34 1024 0:40 2:48 2:07 1:08 I recently built this system myself, specs: SMP55CSM w/8ns pbc "Module" Standard Ram Western Digital 1.6g Enhanced IDE Diamond Stealth 64 w/4mg vram Panasonic 4x cd-rom ide US robotics sportster 28,800 SoundBlaster multimedia 17" Acer 76i I plan on upgrading to a true 180-200 mghz cpu (when intel sees fit to ship them <g>) and adding F/W scsi2 among other things. Since my ide performs at around 9 ms access with 16.6 mb/s burst transfer rate I feel I really need a hot scsi subsystem to match it. Any suggestions, comments, questions? Timothy NVision Inc. & Digibotics
#194695From: Rick MillerOct 10, 1995 9:17 AM
Timothy, I commend you for having the knowledge and time to put together a system yourself. It looks like the SuperMicro motherboard is definitely one of the best to have. The times seemed to be within a few seconds of the times I had, although its hard to know for sure if we are comparing apples to apples. If you like, try rendering with the following settings, then your times will be comparable to most everyone in the thread: ____________________RENDER DIALOG BOX___________________ Phong | View: Camera03 On | Video Color Check: Off On | Gama: On On | Fields: Off On* | On | Off | Off | Show | Tile | * With mapping turned off this setting would change to OFF. Here are my results for the rendering benchmarks: Pixel Size:1.1 Display:Null Save File:No Shading:Phong Mapping-Off Mapping-On 640x480x1.0 ( ) ( ) 1024x768x1.0 ( ) ( ) "( )" Put your times here Regards…………………Rick – Technical Animations
#194725From: Timothy P. SmithOct 10, 1995 2:01 PM
Hi Rick, >>I commend you for having the knowledge and time to put together a system… It's easier than it looks. However if you need to buy a complete system then it is much cheaper, and less time consuming to buy it ready to use. >> …its hard to know for sure if we are comparing apples to apples… Sorry about that. I checked my settings and the only thing that differed was that I had save last frame on. So I re-tested with those settings. Mapping-Off Mapping-On 640x480x1.0 0:20 1:22 1024x768x1.0 0:39 2:45 Timothy NVision Inc. & Digibotics
#194761From: Rick MillerOct 10, 1995 5:48 PM
Timothy, >>>> Mapping-Off Mapping-On 640x480x1.0 0:20 1:22 1024x768x1.0 0:39 2:45 <<<<< The times with and without mapping are sure different. My times with mapping on were typically about 30% slower, your times are about 75% slower. Maybe the difference is in the hard drive interface. Any ideas on your end? Regards………………Rick – Technical Animations
#194767From: Timothy P. SmithOct 10, 1995 7:03 PM
Hi Rick, >>The times with and without mapping are sure different… your times are about 75% slower. Maybe the difference is in the hard drive interface… The obvious differences that I can see are Dos 7.0 w/ Win95 and the (recently released) 8ns pipeline burst cache as opposed to the 16ns or slower pbs. I asume that if everything takes place in ram the hard drive never gets a single hit during a test like this. But for the sake of argument I am only using the built in ide controllers on the mb. It supports up to four drives and or cd-roms. I set up a ms-dos session from win95 that reboots the system clean (except for the mouse drv) into dos 7.0, then executes 3ds.exe directly. nothing fancy. Timothy NVision Inc. & Digibotics
#194746From: Stephen C. LevyOct 10, 1995 4:43 PM
Hi, Rick I think most of the times uploaded were with Metal shading. BTW, I've been talking to Mark DeMazza. Everything looks good! SCL
#194760From: Rick MillerOct 10, 1995 5:48 PM
Stephen, >>I think most of the times uploaded were with Metal shading.<< Oops, I thought the orginal call for times was asking for Phong shading. It did tend to shift to Metal as the thread got longer. Mark will give you a good system, let me know how it goes. Regards…………..Rick – Technical Animations