System suggestions for rendering

[cellbiogeek]

My thesis advisor is putting together a confocal fluorescence microscope workstation and he wants me to put together the computer and software end of it (since I built my last computer he thinks I have experience or something). I'm trying to get a computer with just enough performance so I can save money for the software. Basically all that will be required is image processing but occassionally we will need it for rendering 3D images. Confocal microscopes make images of several slices through a 3D specimen and the computer is required to put these slices back together for a 3D image. I'm interested in suggestions for what kind of system I should build so this doesn't take forever.

 

[FUGGER]

It seems your budget oriented, and you need rock solid stability in your next decision.

I would recomend P4 1.8Ghz northwood and a Asus P4T-E motherboard. since your not playing games a $150 Geforce3 Ti200 should fit in your budget. 512MB RDR, whatever drives you want to buy, we dont have enough info to base any size on.

Ask your advisor what they think about AMD, most professionals avoid them and will kick your ass if you try to slip them one.

You can pretty much throws those parts in a box and shake it twice and it will fall together, onthe other hand if you buy AMD your gonna be taking a crap shoot on if its gonna work when you put it together. Take a few minutes and read a few pages back on this forum (any catagory as they are all full of AMD problems)and notice that 99% of the problems are AMD related. most them are non booting, crashing, random lockups, random reboots, incompatabilities, overheating and so on. can you afford these problems with your advisors computer? I dont think so.

Avoid AMD If you dont have time to screw with it and fix problems. There is no money to be saved buying AMD. RDR and DDR ram prices are almost the same. you will spend more from AMD approved power supply and possibly heat sink upgrade from crapola stock heat sink.

If your a first time builder, the choice is clear. get the Intel based system. your not playing games like most of the people here who will try to sell you an AMD because they own one too.

When I install a machine, I never want to see it ever again. Our company will never sell AMD due to the above said problems, we do not have the money or recources to support AMD. and many vendors have dropped AMD due to the costs involved with eating bad product and high number support calls.

Time for the AMD lemmings to chime in and insult me, call me a liar, and jump up and down franticly.

I need to find transcripts of the hypertransport consortium brawls over platforms. You think were bad on this forum, they are in each others face. SGI is a member of the hypertansport consortium.

 

[Kelledin]

Hmmm...what kind of software is this? What system will it run on? Can you compile it yourself?

If you can compile it yourself, or it comes SSE2-optimized, a P4-based system might be worth considering. You could possibly get yourself a copy of Intel's C/C++ compiler for Linux or Windows. I don't know about the Windows licensing, but the Linux licensing specifies that as long as you're using the Intel compiler for something non-commercial, you don't have to pay for it.

If you can't get P4 optimizations into this rendering software, don't even bother with a P4. The Athlon is a far better choice and will probably crush anything P4-wise in this arena, even Northwood. Plus, you can go dual with Athlons without paying an arm and a leg.

If you feel you must go the Intel route, but can't get P4 optimizations, then you're probably better off with a P3 Tualatin-based system. It won't be nearly as powerful as a similarly-priced Athlon system, but it will go easier on your electric bill.

FUGGER:
"Most professionals" also frequent web forums such as this one. Judging by just about every web forum around, the Athlon has a much greater acceptance rating among "most professionals."

cellbiogeek:
Of course, whatever decision you make, run it by your advisor first. Better to get his opinion before you commit.

<i>If a server crashes in a server farm and no one pings it, does it still cost four figures to fix?

 

[cellbiogeek]

The software package is called Metamorph (though I don't know any details about the bits and pieces that make it up) plus we will also be using NIH Image and photoshop. Metamorph runs on Win98 or 2000. I want to go with AMD because of price (and I'm a lemming as Fugger would say) but I was wondering how big I need to go for occasional rendering. unfortunately this university's #@$!#% IT department will probably only service PCs from an OEM like Dell so this is a moot point unless we buy from a company with good customer service.

 

[AMD_Man]

Fugger, why do you have to revert to your trollish nature? I liked you as a un-biased, level-headed individual you were for that short period of time. Please try to return to that state of higher-leveled thinking as you'll gain more respect that way. I'm no AMD-lemming, but I can tell you that any sensible person would realize that you are significantly distorting the truth. In any case, a P4 will probably be faster in 3D rendering if the app is SSE2 optimized, which many are, nowadays.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[eden]

Rendering IMO is definitly the highest point of AthlonXPs.
These things at low MHZ can ramp up all rendering in no time compared to Northwoods. Some tests showed an AXP 2000 (1.66GHZ)competing a 2.2GHZ Northwood and guess what? The AXP rendered a 3d Studio Max scene in less than 3 minutes of difference! This is a 533MHZ less PC and it beats it. Notice I am saying PC, because if someone will come up to tell me IPC has nothing to do here and that we might as well compare Mac G4s, I'd tell them that PCs are PCs and Macs are Macs, so we compare PCs together is fully logical.

--
The other day I heard an explosion from the other side of town.... It was a 486 booting up...

 

[lhgpoobaa]

Have a look at metamorph...
see if its got 3Dnow, SSE or SSE2 enhancments.

if you DO have to get a dell, just try to make sure its got a decent amount of ram in it, and NOT REPEAT NOT a pentium 4 with SDRAM.


The lack of thermal protection on Athlon's is cunning way to stop morons from using AMD.

 

[Kelledin]

Just going by Metamorph's web site, they don't tell you anything about P4 optimizations. The requirements (listed in the PDF brochure) just specify a "Pentium-class" processor, which isn't that enlightening.

Chances are it doesn't support SSE2. You won't know until you call them yourself though...

<i>If a server crashes in a server farm and no one pings it, does it still cost four figures to fix?

 

[Intel_inside]

I'd agree with fugger here, you want a p4 northwood system. Although athlons have won a few specific rendering benchmarks, it would not be worth the trouble to get one of those systems. At least not for serious work. For renderfarms they are alright, since you have many pc's and it's not as big a deal if one of them craps out. But for this specific requirement you would want the more solid intel hardware, packed with as much rdram as you can afford.

<i>Hi I am from Canada, I don't use amd cause they melt my igloo eh.</i>

 

[Kelledin]

For renderfarms they are alright, since you have many pc's and it's not as big a deal if one of them craps out.

*cough*

*heh*

Ok...

*snort* LoLoL!

RoTFLMAO!!!

Hahah...heh...

Ok, so I just couldn't keep a straight face there. And I'm not even going to bother to be nice, because this fool's really not worth it. cellbiogeek, if a thorough verbal reaming annoys you, I suggest you skip this post.

intel_inside, your statement basically demonstrates that you know precisely jack about renderfarms or clustering. Average stability, uptime, and reliability are far more important in clusters than in any other setup, simply because the average number of problems you'd have with just one system are multiplied by the number of nodes in the cluster. Basic math there.

If you have a system that crashes about once a day, and you have fifty or so systems like that one, that means you have to hire a round-the-clock "button monkey" to find and hit a "reset" button somewhere every half hour or so (or pay extra for watchdog hardware). Or actually pull the thing out of the rack and service it. If you're lucky, or the software's designed well, then the crashing system doesn't disrupt the entire cluster too badly, but that's a big if and another possible high cost.

This, and the fact that Athlons do so well in clusters, suggests that Athlons are a lot more reliable than you'd like everyone to believe. It would be well within your interest to know what you're talking about before you make such a fool of yourself again.

<i>If a server crashes in a server farm and no one pings it, does it still cost four figures to fix?

 

[AMD_Man]

Ok, I'll be 100% honest with all newbies here, but those who claim AMD instability and Intel solidity, are merely stating an opinion. Ok, realize this: as this is an un-moderated, opinion-based community, freedom of speech is all the applies here. Whether that person is telling an opinon or fact cannot be known. So therefore I urge you to take all advice on this community with a grain of salt. Please realize that you won't be getting the best advice in the world here. I came in here expecting professionals who like to contribute information in the spare time, but I was confronted by ameteurs and highly biased individuals, as well, as highly immature individuals who have no respect for others.

AMD technology + Intel technology = Intel/AMD Pentathlon IV; the <b>ULTIMATE</b> PC processor

 

[Intel_inside]

Lol, nice long pointless post.

If you thought about it for one second you would understand what I am talking about. If you have one system that you are using for a specific purpose and it fails then you lose time and money. If you have several you can simply reconfigure the setup and keep going.



<i>Hi I am from Canada, I don't use amd cause they melt my igloo eh.</i>

 

[Kelledin]

If you have a rendering cluster with a node so much as crashing once every half hour (no repair needed, just a reset), the cost of just being there to find and press the reset button usually outweighs the cost of a single non-clustered render station's downtime. The average cost of maintaining 50 nodes is 50x the average cost of maintaining a single node. This drastically outweighs the cost of a single render station's occasional downtime. <b>Do the math.</b>

<i>If a server crashes in a server farm and no one pings it, does it still cost four figures to fix?