Some experiences make you head hurt, and I just had a doozy of one! Dealing with the unfamiliar hardware can sometimes result in a feeling similar to a "Brain Freeze"

I love high tech hardware of any kind, even if I don't completely understand it, I can always appreciate the end results.

Normally my world is focused on cranking that last 100 points from a benchmark so that I can get my games to run a bit faster, but this week I stepped over the line into something that has caught my eye before, but I had no way to explore it, till now!

I recently became friends with a PR member (Alpha) that is only a few hour drive from me, so he stopped by on Saturday with his Dual 2GHZ Opteron/ 3-D LABS Realizm 200 512MB rig for some informal Desktop Card Vs. Cad Card testing



Comparing disimilar in purpose systems is like swapping bathing suits with a hot chick in a Thong Bikini though...Some stuff that sticks out is realy nice to see, but other things are best left hidden!

That being said, I'll throw any benchmark I can get in my rig and have at it, and don't have a problem smackin around other peoples hardware!

First, I gave a benchmark he runs a try with my 6800 SLI/4000+ rig, although I think it only used one of my videocards, It may be possible to enable the other card according to something I recently read about the latest NVIDIA drivers, I'll have to look into that.

I'm sure some of you have heard of SPEC VIEW before..well, I dont know if this is good or bad for a desktop PC, but here it is!





Next, I loaded up a 3-D modeling program that is very commonly used in the Automotive industry called "Catia". He then loaded up a Dodge Neon 2.0 engine block a fellow student had reverse engineered in and we had a look.

He had his rig on a table next to me loaded with the same engine, so I went back and forth and looked at the subtle and not so subltle differences. I'll post some pics of it later, I had to get a copy of the engine from him, I didn't save it at the time.

First impression was suprise in that that my single core rig rig did not chug while using this program, I had been expecting it to look pretty horrifying, but it wasn't too awful!
To a gamers eyes, it looked like the difference between a DX-8 and a Direct X 9 game in visual quality.

Ok , now time to turn the tables!

We loaded up his rig with Far Cry and 3DMark 01, previous experience with 05 has shown his card was not capable of running it, 3DMARK 03 wil run, but we did not have time to try it.
He mentioned Doom 3 had also been tried with limited success, the lighting did not work properly though.

3DMARK 01 ran without fanfare on his Realizm, scoring a fairly low 9500 points with default settings, and no tweaking. It renders EVERTYHING with no optimisations, so that explains why it does not tear it up.



Far Cry ran with framerates were quite playable, but it looked realy weird! CAD cards will draw everything as its supposed to be truly rendered, so the "Algae" looked more like liver spots!



It did not move, looking like a semi solid cube with a texture stretched out over it, but if you got to the right spot underwater just below the surface, you could see the shader trying animate it!

He also was kind enough to give me a preview of his "Heat sink" that has caused a bit of a stir at PR in other threads, he explained the concept in complete detail, and showed me his current rendering of it...I think he's on to something!

But he is a bit over the top though, the design specs call for some pretty expensive manufacturing processes, and with material (Silver) that makes it cost prohibitave to making this a "Production" peice.

I think with a few revisions, something that is a bit more "Robust" (Read: nOOb Boo-Boo proof!) that could be brought to the market for under 75$, it would be a hit!

He is the middle of building a virtual "Working" PC case to model the airflow, it's pretty raw right now, there are no components like a motherboard or power supply (coming soon) however, I was impressed by the results from what he has so far, this can be a very usfull tool when it completed

He "Flipped" the side panel fan from intake to exhaust, and I was able to see in "Real time" just how this would effect the air flow patterns in the case!




And then exhaust




Oh, yeah..I'm hooked!

I have gotten quite serious about finding a way to make another dream come true, and this part of the equation to make that can make it happen, a way to test tweak my vision without having to build it!

I realy think the "Ludicrous" fan idea has a future in the Modding/Overclocking world, the first go around was just to see if it was worth the effort cooling wise, now I have refined it a bit..

Here is a sneek peak at Version 2.0 during the "Mock up" stage...

Attached Thumbnails

     

     

 

Quote: Originally Posted by $SOLID$ Necro *INSERT PIC and SPEC VIEW BENCH RESULT HERE...will update SOON!*

Its probably still rendering the image...You and your slow computer

I would be quite interested in that airflow modeling program. It would save alot of time I've spent in real testing ,being able to make changes on the fly would be a massive timesaver.nice work alpha

OMG!!!! the Ludicrous Fan is getting Bigger!!!!
You need a Warning label on that Fan!!! Xtreme Speed Vary

Actually, this version is a bit smaller, possibly lower CFM, but it should be much quieter!!!

EDIT: OOPS, looks can be deceiving! I just put the two of them together, and the new fan is just a bit larger in diameter, but it's much lower in profile. just need to hook up the 110 and find a Grill!

Before this thing blows up, could you give us fair warning...er, I mean, an esimate on how big the blast radius is going to be? I just want to make sure that I'm on the outside of it when it does happen.

Man....you are nuts, you know that?

It's guys like you, though, that make PR what it is.

Rob

Quote: Originally Posted by Rob Before this thing blows up, could you give us fair warning...er, I mean, an esimate on how big the blast radius is going to be? I just want to make sure that I'm on the outside of it when it does happen. Man....you are nuts, you know that? It's guys like you, though, that make PR what it is. Rob

Ditto, Solid...you are nuts! But you have helped me 10 fold.
Thanks.

Quote: Originally Posted by $SOLID$ Necro Here is a sneek peak at Version 2.0 during the "Mock up" stage...

Being used to see 60 - 120mm fans, that photo makes the case look like you can hold it in your palm.

$Solid$
You certainly can not be accused of being afraid to try new things. Pushing the boundries to see where it leads is the kind of person that serves to inspire the rest of us. And being a bit nuts doesn't hurt either.

Quote: Originally Posted by RBIEZE I would be quite interested in that airflow modeling program. It would save alot of time I've spent in real testing ,being able to make changes on the fly would be a massive timesaver.nice work alpha

as most people know of my somewhat apparently AWOL, it's been a result due to a combination of school, life, work, other work, events happening in and around school, meetings, and such; so haven't really had time to post.

I have 4 programs that can do flow analysis now and each of them function slightly differently than each other. One of them, I haven't been able to get running yet becuase I don't have the proper pre-processor for it. Another one; I have gotten it to work, but I'm still learning the interface because the pre-processor is separate from the solver and post-processor. From what I've been told, that the latest updated release directly interfaces (appends) the solver and post-processor with the pre-processor; and also allows for multiphysics and coupled flow and structural analysis. (To those of you where all of the stuff that I just wrote is jibberish, ignore it. To those that DO know, it's Ansys 9.0 Workbench and Ansys CFX 5.7.1's integration into workbench for coupled CFD-FEA in additional to thermal, and electric fields/MEMS analysis. Another program that I know that has the same or similar abilities is Comsol's FEMLAB 3.1. I've yet to request the trial version to see how well it works, but I'm not in a big rush to do so.)

Two my other programs have been given me a hard time trying to solve for the heatsink that $solid$ references; and I am still working on trying to figure out why, but I'm near my wits end on that. (Unfortunately.)

I did load up the computer case (grossly simplified) model to illustrate that this is something that I do pretty much on a daily basis when I'm not in classes/school, and there's a lot more to it that I just haven't gotten around to yet. (i.e. again - for those who know - I am referring to rothalpy, point source thermal (energy) dissipation, trying another relaxation like RNG k-e turbulence model, to see if I can get some of the other runs that I've done already (about 32 in total so far since April) to converge. I DON'T thnk that it would fall under LES, but there's a lot more to it. The program comes with a training manual that's in a 3" binder, and 3 books that probably total nearly 1000 pages. In particular with STAR, they have a prostar-amm (formerly pro-am) for advanced automated meshing, so I am also going to try to figure that up to see if I can get it to clean up some of the runs that I tried doing before where it wouldn't work because it imported the mesh improperly. Currently using STAR-CD 3.22, STAR-CCM+ 1.04, star-cat5 (CATIA V5 plugin), provis).

[again, if you're reading it and thinking "blah blah blah" ignore it. or just treat it as just blah, blah, blah.] From the looks of it, based on conversations with $solid$, it is possible that the second revision of his fan idea will be using those programs/tools to help in the design phase.

Poor $solid$ has to deal with me when I start asking him some pretty heavy duty questions regarding performance specifications, and product development and engineering process questions. lol....I guess that's what happens after being in product engineering for two years on and off.

I think that if we really worked on it, I would be able to crank out a rev. prototype A (PA or YA) in about 6 weeks. Nothing would be validated, but it would definitely gives us a starting point; which then we can go back and revisit to iron out some of the finer details. (the concept-to-prototype phases in the PDP).

P.S. sorry $solid$ I tried to keep it in plain regular English for ya. The good news is...if it's "blah blah blah" in your head, it's something that you can completely skip over. (The detail stuff was primary for/to the guy that asked.)

Quote: Originally Posted by $SOLID$ Necro And then exhaust

Dude What program is that! I must learn how to use it!

Damn Pat if that fan was any bigger the case would hover off the ground

Quote: Originally Posted by j-dogg Damn Pat if that fan was any bigger the case would hover off the ground

quadriplex fly-by-wire system powered by an single AMD Athlon 64 4000+? sweeeeeeeeetttttt....I LIKE that idea.

(i.e. the stabilization system on the Lockheed F-117A, because of the stealth characteristics, it was horrid for aerodynamic stability that it needed a quadriplex (which I picture as a 4-layer, cross-linked and interconnected) stabilization system that constantly made adjustments that helps/allows the plane to fly.)

Can be done.

Quote: Originally Posted by Nerdz Dude What program is that! I must learn how to use it!

Short answer. CFD program.

Long answer. what you're seeing is the post-processor program called Pro-vis. It is the means to see the results from the flow analysis. The program that actually did the solving is STAR-CD 3.22, STAR-CCM+ 1.04, STAR-CAT5 plug in (CATIA V5 CAA plug in), and Pro-vis by/from CD-adapco. (http://www.cd-adapco.com) They have a lot of computational fluid dynamics programs, as that is what they're known for and that is their forte.

[Translated into English - it's a computational fluid dynamics program that I run directly from the CATIA V5 (CAD) environment. Historically, you had to export the models, then re-import them and believe it's not always fun and games doing that. With the plug-in, it eliminates the need for the translation. There ARE some limitations as to what it can do depending on the different calls that it makes to the solver, but for what I'm using it for right now, it's plenty sufficient minus a few hiccups here and there.]

STAR-CD 3.22 is their latest solver that and also run as a standalone. Very nice program. Not quite as easy to use as the plugin, but infinitely more powerful. Often found to be running AS a standalone on Linux and UNIX systems.

STAR-CCM+ 1.04 is their newer generation of solver code. Written in C/C++ (where as STAR-CD was written in FORTRAN77 and FORTRAN90), CCM+ takes advantages of the more efficient C-based code and is also much more prevalent/dominant now. Eliminates the need for a separate FORTRAN compiler which is really expensive. Also allows for polyhedral meshing (i.e. hexagon elements) in addition to tetrahedral (triangular pyramid elements). The idea behind that is it should be able to reduce the number of elements need for the same given volume of space; which reduces computing time, and is more efficient. I've also been told that it's also substantially more accurate results (i.e. >95%) where as tetrahedral is between 80-85%.

STAR-CAT5 is the CATIA V5 plugin developed with CAA. Dassault Systemes have allowed companies like CD-adapco to get direct access to portions of the CATIA code with CAA allowing for tight integration with CATIA. Therefore; for product development (and even research), you can run your simulations all from within ONE environment and eliminating data translation layers. (Which I hate to do.) Down side is that in it's current state, it can't do everything that STAR can, but I would expect that future releases to be catching up to it. It also helps to dramatically reduce the amount of time it takes to prepare an analysis case, and does a lot of stuff for you automatically. There aren't as many buttons, or options, but it uses/assumes the most common settings that people use based on their experiences.

Pro-vis is the visualization post processor. From there, you can view the results of the analysis. I've yet to be able to find a computer than can animate 40,000 vectors in realtime. Most of it is more like a slide show, which makes it somewhat difficult to see. Also has some limitations in it's capabilities. Apparently, I should be able to also use STAR-CCM+ to run it as a standalone itself, but I've never been able to get it to automatically do a number of iterations by itself. Probably user error on my part, but I'm still learning.

There are others that are available. Most of them now are pretty similar so, in reality, any one of them would work/do. I've tried a mixture partially to verify the results, and partially because if one has troubles/errors, that maybe another one would be able to solve because it approaches it differently. Or alternatively, it might give me a more detailed description of the error.

I don't like to spend a heck of a lot of money if I can avoid it, and the analysis programs is to me a cost-save tool because I don't actually have to phsyically make or buy stuff in order to find out it doesn't work. I figure that if I can do 80-90% of the development work on the tube, that I would be willing to part with financials in and that last 10% would be more of a reality check to tell me whether I'm still crazy and completely off the ball, or whether I'm actually pretty good and in some cases, dead on, in my solution.

You've probably heard me talk about CFD stuff a lot. Yes, I do work with a fair amount in thermal sciences. But not nearly enough on a daily basis to be considered an expert in the field yet. Thermal sciences and thermal management is a pain, because there are a lot of factors involved in it. So, I guess that I must be a glutton for punishment because, with it, I can decipher between what to do and what NOT to do.

you know alpha, youve taken a lot of crap for that heat sink idea (which i have yet to judge as i havnt seen it) but if solid says you might be on to something then maybe we were all wrong to judge you on it. you definatley seem to know your stuff. if i made fun of you in one of those other threads (i dont think i did but i honestly dont recall) then im sorry, and i look forward to hopefully learning more about computer thermaldynamics from you.

Quote: Originally Posted by 4rm4g3dd0n you know alpha, youve taken a lot of crap for that heat sink idea (which i have yet to judge as i havnt seen it) but if solid says you might be on to something then maybe we were all wrong to judge you on it. you definatley seem to know your stuff. if i made fun of you in one of those other threads (i dont think i did but i honestly dont recall) then im sorry, and i look forward to hopefully learning more about computer thermaldynamics from you.

I'm working on learning others as I go along. Hopefully that eventually, that I would be proficient enough with the industry leaders that it really doesn't MATTER what a company would be using. Just covering my grounds; it's all.

If you want to learn more about it, you're more than welcome to pick my brain at however little it is that I know about it. I'm alright running into all sorts of issues with more complex cases and I haven't even hit the REALLY hard stuff yet. Like the rest of us here; I'm still learning too.

Here are a few more airflow studies I asked for using two 80MM (27CFM) Intake and exhaust fans, and a single 80 side fan.

Attached Thumbnails

     

     

 

When can we get our hands on that program??

Quote: Originally Posted by $SOLID$ Necro EDIT: Here ya go.. High Quality (4.6MB) http://68.61.120.116/necro/FM6000.avi Low Quality (982K) http://68.61.120.116/necro/Untitled.WMV

Well since you helped me out necro I hope this helps you out. I uploaded it to a faster host.
http://defcondesigns.com/cptplanet/FM6000.avi

Quote: Originally Posted by Sn0wMan When can we get our hands on that program??

Available now.