Wanna a new display card

[Signal2Noise]

Do a search in this forum on video cards and ye shall be rewarded with a plethora of good info.

What app(s) are you using as far as CG goes?

If you ask me, except for your current vid card, there is nothing wrong with your system. Instead of "doubling" your CPU speed perhaps you can look at maxing out the RAM on your mobo. RAM has a significant impact on rendering speed. RAM is still fairly inexpensive these days would cause less headache getting more of that than swapping out CPUs.

Good luck.

[kiri]

speaking of ram, how much ram do you think would be needed to run maya pretty smoothly?
512 enough, or should i try and go all out for 1gig?
thanks

[Signal2Noise]

Kiri, my rule of thumb is: As much RAM as you can afford and/or the motherboard will hold. There can never be too much RAM.

[zcmaya3d]

thank you for answering
actully in China , or just my friends , there workstations has 1G or 2G RAM , RIAD , and professional display card , i am sure a better computer can give you better feeling , i do not know if it is needed to have a large memory , okay , maybe 512 is enugh ,
so how about my display card ?? any advise ?

[zcmaya3d]

yeah , i am a little agree with Signal to Noise,
in my mind :
the more RAM the better
:p
of course more money needed :hmm:

[GregHess]

I would save money up to purchase a new system. Your computer is about four times obsolete when it comes to intel cpu's and chipsets. Purchasing a new video card won't help performance too much, because your computer is cpu limited, not video limited.

You could go to a geforce2 ti, but even with that I doubt you'd see much of a performance increase...the cpu power just isn't there.

(Your cpu is roughly equivilant to a 1.0 gigahertz Thunderbird Athlon.)

So I suggest scrapping the whole computer. You won't be able to use the old ram in a new system, and everything else (sans maybe the HD, but there are much better versions avail), is pretty much obsolete or ebay material.

Putting a Wildcat on that system...first of all it wouldn't work at all, and second, if you did get it to work, it would be like putting a supercharger on your lawnmower.

So whats your budget for an entirely new workstation?

[Signal2Noise]

Greg, after reading several of your postings in regards to hardware it has become apparent you like to spend money. Or other people's money.Don't take this the wrong way. Your hardware advise seems to be always sound and solid and many, myself included, enjoy the knowledge gleaned from your suggestions and feedback.

However, I do have to question the necessity of always having to have the latest & greatest computers to run today's applications. I mean, if you look at the system requirements for certain CG apps most computers people have that you deem "obsolete" will more than adequately run them. Hell, I still have a PIII 600Mhz w/ 512MB RAM that runs Max 5 on it! And this desktop is over twice the recommended system requirements. Sure rendering times are slow compared to today's speed machines but a year or two ago this was fast because there was nothing to compare it to! For the CG hobbyist or part-time freelancer this is sufficient. I can see if the CG artist is running a money-making production house and time is of the essence to complete projects but then we'd also be moving into the world of render farms in addition to high-end work stations.

You may not agree but I always feel if the user has an already adequate system, which in this case it is (although we don't know which apps he/she is running), then the simple upgrades in memory and video cards will enhance the performance output a great deal and be easy on the pocket book.

But that's just my thoughts. Thanks for reading and I look forward to the continued dialogues to be shared.

[GregHess]

Signal,

I base my comments on a price/performance ratio. Its almost always centered around the individuals budget, which is why I routinely request "Whats your budget?".

Here's the way I see it.

1) User A complains of Render Time/Viewport Performance.

There are a variety of solutions to this problem. Of course the easiest is User A just learns how to use the Freeze/Hide/Dummy features of his/her program, and learns to optimize his scenes/models. The average 3d user though, would rather invest time and money into a faster machine, instead of just optimizing what they already have.

Advantages are...

1) Better program understanding.
2) Optimized geometry and scenes increases modeling skill.
3) Reduced rendering time, increased viewport speed.
4) No cost.
5) Addresses the problem at the source.

Disadvantages are...

1) Requires a forward thinking and pro-active user.
2) Time. Time. Time.
3) Though it may reduce the problem, other factors may still require an upgrade.
4) Suggesting this to a user is seen as a direct insult. (usually)

The second solution is the individual invests "time" and "money" into upgrades for there machine. This has a variety of advantages and drawbacks.

Advantages are,,,

a) short term performance gain, and
b) reduced expenditure compared to a new system.
c) significantly easier to accomplish then building a new computer.

Disadvantages are....

a) Incompatibilities with previous hardware configurations.
b) Minor increases in performance compared to other solutions.
c) Time spent troubleshooting possible non functioning components.
d) Not actually addressing the problem.
e) Upgrades may be incompatible with future purchases.

Upgrades are usually most successful if the system is within a year timespan. Once you pass the 2-3 year mark you encounter a major problem. The cost/benefit ratio of upgrading/replacing starts equalizing, as well as possible upgrades to your machine being useless if you were to jump to a different platform or new chipset.

The third option is majority system replacement, or a completely new machine. I usually consider a system replacement when you have to replace the ram, motherboard, and cpu...cause that usually means you'll also need a new powersupply, and possibly a new case.

The Advantages of a system replacement/new computer are...

1) If its a new machine, you gain a second work box. Netrendering, Distributed Rendering, increased workflow.'
2) Massive performance increases compared to other solutions.
3) Increased window for Option #2 (Upgrades)

The Disadvantages are...

1) A large investment of cash is required.
2) If your building it yourself, a host of possible problems could occur.
3) Could miss certain key timeframes, and buy a machine right as a swapover is occuring (buying a i845PE now, when i865PE is only a month away for example), screwing yourself out of the upgrade window.

So with these methods understood, lets look at zcmaya3d's personal machine.

p4 1.5 GHZ
512 RAM
IBM 60 GB
ELSA Gefroce 2 mx 100/200 with 32 MB
ViewSonic G73f

Lets look at the three options for his computer. Option 1, is of course self explanatory.

Option 2: Upgrades.

This particular system is a mainstay of Pentium IV's build around 3 years ago. The chip in question is a first generation Pentium IV, the Williamante, its a Socket 423 Chip, with 256k of l2 cache. It runs at a 400 FSB, and either comes with Rambus (superior) or SDRAM (inferior) memory.

The maximum obtainable speed for Williamante is 2.0 Gigahertz. (If the motherboard supports that). Its also well known that the williamante is one of the slowest processors Intel has ever manufacturered, with some of the lower speed pentium III chips, actually outperforming them. (You PIII 600 is only around 133-200 megahertz slower then the 1.5 P4 is).

The Video card is a Geforce2 MX. Not exactly the best card ever made, but more then sufficent for 1024x768x16 work. If the individual wants to bump his display res to 1280x1024x32, he'll definitely need a faster accelerator.

So right here we've got...

1) Obsolete Processor.
2) Obsolete Ram.

The earlier Pentium IV motherboards required a specific mounting platform for ATX cases. In some cases these mounting systems are so complex, that you can no longer fix standard ATX boards in them...so replacing the entire case may be necessary...and I'm sure the power supply is underpowered for new gpu's/cpu's.

So then...Upgrades are immediately limited by available case and power supply. Possible GPU upgrades could not function.

So lets say he decides to upgrade.

He spends money on a 2.0 Ghz Williamante, and another gigabyte of RAMBUS or SDRAM.

What has just happened? He's screwed himself. Neither the williamante, nor the ram will work in a new system. The money he just invested in upgrading will be useless if he goes to another platform. Not to mention the performance difference between a 1.5 - 2.0 Williamante isn't that much either.

On the other hand, he could invest into a platform where he's actually see a performance benefit, as well as opening a much larger upgrade route.

An i865PE with a 1.8A Northwood would upgrade past 3.6 Gigahertz. Plus the DDR would work in any number of current motherboards, either Intel or AMD.

Remember Signal, though I tend to not elaborate on my posts (don't have the time to explain why always), I am making them with about 6 years of experience in the industry...

[loop29]

Yes, it´s more thrilling to put the most cheapest computer together and get solid performance for your needs. I tend to wait and see what the PC Industry is doing. I´m still on Athlon XP 1800+ and I won´t upgrade until the things that I want will show up on the market and didn´t cost me an arm and a leg. I would be really excited if they start focusing on PCI Bus and speed up PCI Express development or other bus systems, SATA will need some time to drop into market, I think it will take another year to get 300 Mb/sec and controller for next version.
You might upgrade your CPU and video card and get 10% to 20% increase with P4 2 - 2.26 GHz, depends what MB you have. You should check if it supports 533 MHz FSB for the faster Intel CPU´s. The Geforce 2 MX 100/200 is really no match for a newer video card, I would think of Geforce 2 Ti, GF 3 Ti 200/500 or Geforce 4 Ti. Geforce 2 Ti/Geforce 3 Ti 200/500 can be converted to Quadro completely with Softquadro. So you can get some increase let´s say for around 300 $´s or less.
Just my thoughts.

regards

[Signal2Noise]

Whew! Thanks for the good read over lunch, Greg.:)

You do make good points but with this philosophy people would be buying complete new systems every 6 or 12 months even if they aren't getting new software to go with it. Computers tend to get out dated faster than the applications.

I still stick with the fact that if the current system can be upgraded before having to shell out big money for a new system then that's the way to go. I agree about the processor issue but RAM and video cards are still a viable option. After all, that computer, although now 'obsolete' by today's standards, was purchased way back when with upgrading in mind. Just like if the user were to just buy a new system now with future upgrading down the road in mind. But then it comes to that time to actually upgrade but there's yet a faster machine on the market. It's a vicious circle. A catch 22. If the parts are available (and price is right) then why not take advantage of it? Albeit it will be for the short term but only if there are in fact plans to upgrade the application being used.

That's the defining factor for me: If the software is not being upgraded, is there a real reason in buying a completely new system?

I've been involved in the engineering design visualization industry for coming on 11 years now. We use extremely high end CAD stations, VR presentations, and animation software for construction study purposes. In those 11 years the majority of the software has only been upgrade about 3 times on average. Some not at all. The ones that did get upgraded did require either a system replacement or, 25% of the time, simply upgrades in RAM and such. The software that hasn't changed over time (and looks it, admittedly) are still on older machines that has seen some upgrades but are dependant on certain o/s versions to run and therefore the hardware has remained the same on the whole. No point in buying expensive hardware if the software hasn't changed. that's where I'm coming from.

Anyway, that's all I have to say. Thanks for taking the time to relay the reasoning behind the suggestions. It's still great advice, especially if one has the money to do it.



:beer:

[GregHess]

Hey Signal,

You do make good points but with this philosophy people would be buying complete new systems every 6 or 12 months even if they aren't getting new software to go with it. Computers tend to get out dated faster than the applications.

Not quite. The philosophy is as follows. If someone chooses to suddenly expunge their bank account towards a computer upgrade, its sometimes more cost effective to buy a new system (or at least, most of a new system) then upgrading an old one.

If

http://www.3dluvr.com/crossbow/incoming/math.jpg


Then buy the new system. I'm really bad at math, so that might not make any sense at all.

Remember that the main reason I am suggesting a new machine vs an upgrade for this individual is because his computer's age is almost 3 years old, or 36 months. If as stated in your suggestion, the age of the machine was just 6, or 12 months old, that would be a completely different story, and I'd probably just upgrade specific components.

I agree about the processor issue but RAM and video cards are still a viable option.

As you know, adding additional ram only helps if your running out of it. It doesn't increase performance.

As to video cards...most of the newer accelerators are CPU limited, meaning that the faster the cpu, the faster the video card. Buying something like a Quadro 900XL and attaching it to a P4 1.3 would be a waste of resources.

Now as mentioned in another post by loop (i believe), he could pick up something like a geforce2 ultra, or geforce2 ti, and softquadro it, for a decent boost in performance, especially if the res is higher then 1024x768.

That's the defining factor for me: If the software is not being upgraded, is there a real reason in buying a completely new system?

Thats a very debatable statement. Were in the Cg industry. The faster our computers, the faster we can render, and the more clients we can pack in. There is a substaintal difference to the amount of work that can be accomplished on a Gigahertz thunderbird, as opposed to a Dual 3.06 Xeon. The question that has to be asked is...is the power REALLY needed?

The creed of the hardware industry is...

Buy when you NEED NOT when you WANT.

To further extrapolate this example.

Lets assume zcmaya3d has the following configuration.

An abit TH7 (nice board)
512 megs of PC800 Rambus (2x256 Must be in pairs)
1.5 Gigahertz Pentium IV Williamante.
250 Watt PSU
Geforce2 MX

Lets say we upgrade it to a 2.0 P4, with a gig of PC800, and a Geforce2 Ti.

2.0 P4 (297 USD...yes its more expensive then northwood 2.0's)
512 megs of PC800 (2x256) 180 USD
Geforce2 Ti 70 USD (64 meg)

Thats 547 USD to upgrade the machine.

For about the same amount of money you could get....

1 Asus A7N8X Nforce2 Motherboard (117 USD)
1 2400+ XP (130 USD)
2 512 meg Crucial PC2700 DDR (140 USD)
1 Antec 330 Watt Truepower PSU (57)
1 PNY Geforce4 Ti 4200 (120 USD)

564 USD.

Now do you understand my point? Not only is the newer system faster in all aspects (rendering/video), but it has higher quality components, supports new standards like AGP 8X, USB2.0, and is upgradable to well past a 3000+ XP.

[zcmaya3d]

oh, thank you all :applause:
from your words , i can see the P4 1.5 is not good , nearly the worst willamette P4 , and the card is bad too, i think i can feel that very much , i use MAYA4 and PIXAR RENDERMAN ARTIST TOOLS 5 , in maya when i wanna view the screen , or when i press the space with the view window change , it is so slow , and worse when there are more objoects . cetanily it is very very bad when i use the particles .
actully i do not like the AMD 's CPU , i do not know why , Intel is my favorate ,
i have read the list you wrote , and i wrote another one here , i am not professional , so please point out my mistakes :
P4 Dual 2.8 xeon
1GB RAM
120 GB HB
ELSA Qudro 4 900GXL

actually i still haven't dicied to update or buy a new one ~~~~~~
i think all of them have their advantages :(
at the same time , i take note if that all the hardwares are much more cheaper than in China , they are all very expensive here .
okay , thank you all for ansewring my questions :D:D:D

[loop29]

You have no other choice if you want to use 2.8 Xeons than buying a new motherboard !! It seems that you don´t have a tight budget, or you may not know how much 2 Xeons and a proper board will cost you.

regards

[zcmaya3d]

yes, i am a new one to them , so i wanna all of you give me advise

[GregHess]

A dual 2.8B Xeon will cost around 2,500-3,000 USD.

Possibly 3500-4000 USD depending on the ram and video cards involved.

[Sieb]

Greg, and anyone else, we need a Sticky for this forum about Vid cards and Setups, some guidlines. I think Greg already has something like this on Luvr, I don't remember. But we usually get one a week about either topic. What you guys think?

[GregHess]

I don't know how to sticky. Don't you need to be a moderator for that?

[zcmaya3d]

oh, yeah , that is really to evpensive , :cry:

[Sieb]

yea, only admins can sticky, but I was just saying we should probably start compiling something since its so common and we could always post updates to it with new hardware. It would just be easier instead of always haveing to post the same info over and over. I know new people dont like to search. :)

[zcmaya3d]

hi :
sieb:
i am sorry , okay , i see , i will do a seach later , you know my english is too bad , and my mother language is chinese, so i am a little difficult on that , so i use this easier way :p
but thank you for your advise , i will seach that :)

[PlanetMongo]

OO! Greg is spot on on this. Here, let me tell you what I ran into recently that's turned out to be a royal PITA.

In 1998-99 I started playing with BeOS, a really nice SMP-aware OS. I decided to build a nice dual proc system to really make it go, and as the first Celeron hacks were around, I decided to just mod a couple Celery's. So, I picked up a Supermicro P6DBE and a couple slot1 Celeron's (because socket 370 celerons were not in existance yet). (Also note that my reasoning also was supported by the "Hey, I can always upgrade to "real" PIII's when I decide to upgrade later" mentality). I dorked around and never got around to modding the Celerons, but wouldn't you know it? The Socket 370 Celeron's came out (and to my chagrin, the Abit BP6 which would have been much better, IMHO) and.. slotkets! So, I found a friend who took the celerons off my hands (at great discount) and I picked up a couple Asus slotket's with Celeron 400's (jumpered for SMP), added half a gig of RAM, and lived happily ever after. Kinda. WEll, last year, I decided to upgrade my now increasingly "behind the times" computer. I ordered a couple PIII 750's (slot 1's) because the price seemed right. I got them in, installed them.. and nothing. WTF? Did some research and found I had an early revision of the mobo and it wouldn't take 750's. :: sigh :: So, I order a new revision P6DBE (for around $250, I think). Get all that installed. And yay. Now I have a 1.5ghz machine that is still a bit behind the curve, except I'm SMP. I just ordered another 512 megs of RAM from crucial, and come to find that it's not compat with my "old" ram (New ram is ecc registered, etc, what the manufacturer recommends), and it was $150 for the two 256meg DIMMS (because my mobo doesn't support the density-thingie that would allow me to install a couple half-gig DIMMS). Anyway, at this point, the outlay is something like:
$250 new mobo
$250 new CPU's
$150 RAM (half gig)
$150 video card (agp 8x, which my mobo doesn't support)
For that same amount of money, I could've grabbed a dual athlon board (or a nice nForce board w/top of the line Athlon), a full Gig of RAM, and maybe some 2nd or 3rd tier Athlons and had one *hell* of a better machine. So now I'm "stuck" until I can free up some more cash to do a "proper" upgrade, although in reality I'm kinda waiting to see if nVidia releases the rumoured SMP nforce chipset.

:: sigh :: Learn from my mistakes. Sometimes "upgrading", as Greg pointed out, is more expensive and more limiting in the long run.

[Sieb]

zcmaya3d: Dont worry, my posts weren't directed at you, but people like you, nothing bad. I just thought it would be easier for new people if us tech guys got together and made a nice "How To" or "FAQ" post for everyone and cut down on the repeating threads. Plus, it would be one thread we can all post updated questions in.

Organization and Optimization are my thing, hehe..

Mongo: Sucks dude. I remember the good ol Cele 300s. hehe. Good days. But yea, upgrading can suck if your behind a few years. Your better off with a whole new rig if you can afford it. And, this is the very reason I prefer to stick to AMD procs. The chipset doesn't change with every new processor revision like Intels.

I agree, Greg is on target with his post. You have a pretty good upgrade window if your system is 6-12 months old. Rule of the system is, "Your hardware is superceeded the minute it goes on the store shelf. In 6 months, your system is obsolete." Its the price we pay for pushing the barrier. More than a year old, your better off getting something new. Especially if its Intel.

AMD is a little more lenient since chipsets dont change much thus most processors are backwards compatible. I am still running a 1.4 Tbird on a KT3 Ultra ARU, well.. for one more week. Over the last two years, I have made a progressional upgrade from my 1.4 on an Abit KT7R, went from a GF2 to a GF4MX to a GF4Ti4600, from 30 gig drive to dual 30 gigs to dual 120 gigs.. Then from Abit to MSI board. Not to mention the case changes.

And now I finally move over to my Nforce Shuttle Cube with a 333 2600, dual 120gig drives and my Ti4600. Cost? $850
Cost of Progressional upgrade totaled over two years? ~$1200

Good example? hehe

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.