Graphics card upgrade

After I've tried everything to improve my current settings, I'm considering upgrading my graphics card. SL runs well on my computer right now , except for the occasional crashes at overpopulated sims, but my atmospheric shaders are grayed out, and I'd *really* want those without having to get a whole new computer. These are my current specs:

CPU: Intel Pentium 4 (3391 MHz)
Memory: 2047 MB
OS Version: Microsoft Windows XP Service Pack 2 (Build 2600)
Graphics Card Vendor: ATI Technologies Inc.
Graphics Card: RADEON 9250/9200 Series DDR x86/SSE2
OpenGL Version: 1.3.1072 WinXP Release

My graphics card slot is an old AGP one, so my options are quite limited. These are the ones I found at the shop, I have no idea at all what to choose:

ATI:
- SAPPHIRE 96000 (128MB)
- GECUBE HD2600 (512MB)
- SAPPHIRE2600PRO DDR2 (512MB)*

NVIDIA:
- POINTOFVIEW 7600GT (256MB)
- POINTOFVIEW 7300GT (512MB)
- BFG 7300GT DDR2 (512MB)*

(* - I'm clueless whether these ones would be compatible at all)

Which one would you recommend? What is your experience with these cards?
Thank you!

what about the nvidia 7600?

Go nVidia whatever you choose. SL is not ATI friendly from what I hear.

well, thing is, with my current ATI it hasn't been *that* bad, so I'd like to know if I'll actually notice an improvement with those

@ Morgaine: do you know for sure whether the 7600 is any good?

7600 is decent. Although you may need a better power supply. I'd recommend at least a 400W for your system with that card.

I also suggest the 7600.

But I also agree with Beezle, ensure that you've got at least a 300 watt power supply.

If you don't know off hand, open up your computer and look for a sticker on the power supply, it'll give you information about it. Peak wattage doesn't help you, look for sustained wattage.

Yes the wattage is important. I am still not an expert but when I installed the 7600 it was fabulous on the desktop.
I had to open the computer and had I think 350w?

thank you all for your help, I'll give that 7600 a go... after all, the price is quite reasonable (especially compared to ATIs), so I'll be able to change the power supply for almost the same price!

look at tiger direct, they have some 7600GT's ~ ati hd 3800's depending on you price range

personally id go with a nvidia, ati still has issues with drivers and certain games, but the hd 3800 is a pretty monster card if you are willing to plop down 200 bucks, where as the 7600GT is like as good leap above what you have now, but much cheaper

and dont look into anything less than 256mb nowdays, sl will drop a 128mb card pretty quick (i know i still have one)

And go with a bigger power supply.............take a look at my thread "What the heck does a PS have to do with it".

I saw an increase in performance due to having a under powered computer. If nVidia asys you 300 watts.........go to 400.

You'll understand if you read my last post in that thread. Good luck.

I just put in an 8600 GT (XFX). Upgrade from a 7600 256k.

I found that the larger memory (1 meg) really speeds things up and it looks gorgeous!
I always look for what's close to the$100 mark and gt the best one I can find. Also, I've had really good luck with XFX brand cards over the years.

I'm with the rest, Nvidia is the way to go.

Good luck!

Since the "PSU upgrades made my computer faster" thing came up, I should mention this.

There's no technical reason why a PSU upgrade would boost performance.

So if you're running a PSU that you know is underpowered (trust me, you'd know), then you should upgrade. But don't upgrade expecting it to speed up your computer.

I really do see it as a placebo effect.

There have been times when I've changed something in the BIOS that I know shouldn't make a noticable difference, but I could swear the computer is faster. You want it so much that it becomes true.

I don't want to go round and round with you again Robot. I understand exactly what you are saying about PSU's and performance increases. However, I did see a performance increase when I upgraded to a larger PS. The final post in my thread about it I tried to explain what I believe happened to cause that increase. Technically you are probably correct in your assesment that a PS has nothing to do with it........actually that was my thought when I posted the thread last weekend.

I'll briefly say it again here since people are not likely to go read my thread to hear all the discussion. I have a relatively new computer (Mid December, 2007) that is a fairly powerful off the shelf Lenovo Desktop. The computer this one was replaced had an nVidia 8600GT in it. I swapped the card from the old one to this on before I even turned this computer on.......so I have no comparsion to measure a before and after. The computer came with a 320 watt PS. My nVidia card installation instructions said that the minimum PS was 300 watt. I figured I was good to go. My computer ran beautifully.....plenty of speed and I was very satisfied. Well, I'm looking at gettting a new video card (ATM, I'm looking at nVidia 9800GT or GTS...forget exactly which) and I knew my little 320 watt PS would not work with it so I bought a 650 watt Antec last Saturday in preparation for my eventual graphics upgrade. When I installed it in my computer and turned it on there was nothing noticable at all......it just worked so I knew I did everything correct. Okay, now I go to launch SL.......the time it took normally from clicking my shortcut to my log in screen was somthing over 45 secs........as much as a full minute. But when I launched it the first time with the new PS that time was cut in half. Okay.....still, just a pleasant surprise. No big deal. Then I break open GIMP and start playing with textures I've made. I thought I'd give my system a workout with a filter (map to cylinder) that requires much numbers crunching and graphics rendering. The texture I used to test was one that consistantly took nearly 10 mins to run to completion........but this time it took only about 3 minutes. I did it about 5 times and each time it was almost identical times.

It's then that I started wondering what the heck made the improvement? Only thing that changed was a bigger PSU. So I posted a question in this forum asking why. After all the discussion and me going to a website to calculate what size PSU in needed for my computer configured as it is that I think I found out why it happened. Everything in this computer right now needs 400 watts......it needed that when I had the 320 watt PSU in it. So, I figured that with the smaller PSU I was not getting full performance from either my CPU or my GPU (probably both). The increased power brought whatever it was up to speed and there was my increased performance.

So, I might understand what Robot is saying.......and I agree mostly. I also know what happened with my system.........so, if you are working with an underpowered system you may not be getting everything your computer is capable of. That is my only point of even mentioning it.

And, NO, you won't necessarily know if you are running under powered. I didn't know and I pay attention to performance. I'm sure eventually I would have known when the power supply failed due to over loading it........but, then it's sort of late.

"My RAM usage was 42% compared to the normal 50 - 55%. "
That's from your other thread, and makes everything suspect. Having more power available affecting your free RAM is absurd. In theory having more power could make components work faster (it doesn't, but we'll talk theoretically), but having more power available won't compress data in RAM, or make your computer load less data into RAM.

In fact, what it implies, is that you made other drastic changes to your configuration.

I won't go into why more power != more speed again, but it doesn't.

And yes, you do know when your computer is underpowered. Severe instability results from an underpowered system. Random stop errors (blue screens of death) with seemingly random stop codes (IRQL_Not_Less_Or_Equal, for example).

is your entire life based around that 1 thread and trying to prolong that argument?

ok so you dont understand basic logic gate thresholds or modern power management tech, i gave up cause its like talking to a brick wall

but here you are again trying to stir up shit for no other reason than to burn time

its kinda sad

Actually, it's because I would feel a personal guilt if anyone sees the posts about power supplies boosting performance, buys a nice power supply, and gets nothing out of it. I have a bizzare sense of nobility when it comes to wasting money on un-necessary component upgrades.

Whereas you're in the other thread talking about HDD's spinning slowly and CPU's having enough logic to stay at low speeds when in an insufficiant power situation.

You've above average with many things as I've seen from your forum posts, but you appear to be lacking a bit when it comes to a computers inner workings (specifically, power management and component power requirements, and the effect of low power conditions on components).

yea well i may be lacking with that masters in electronics engineering and computer sciences

besides that post was asking what could be the reason for the performance boost, not everyone go out and do it now


also im not the one that cant wrap their head around the fact that if a computer can dynamicly change speed to change power consumption that it can do it on post and just be stuck there

really come on, if a computer can operate perfectly fine at a reduced state just to save power, why cant it do it with out being dynamicly driven

(durh)

As I stated in the other thread, ACPI, APM, and other power management was designed to reduce power consumption actively, not passively.

Reducing CPU clock speeds, turning off HDD's (or in the case of the Western Digital Caviar Green series of drives, slowing the platters down) to save power is an active move, turning off other components, it is an active move.

Slowing things down, turning things off in response to an under powered PSU would be a passive move. It is something that could be put into a system, and some devices can function like this (some GPU's can in some cases deal with a low power condition, but not always), but the issue is with maintaining stability when it is determined that there is a low power condition. A low power condition can only be determined once it has occured, and once it has occured, you can no longer trust, for example, data in memory. If your memory (system or video) has encountered voltage droop, or the processor has (and did not fail), there is a very real possibility of corrupt data within RAM, which can have a number of effects (from odd behavior to a corrupt application corrupting data on disk to the corruption being exploited by an attacker (which is more common than you'd think, although the cause of the corruption is generally deliberate)).

Indeed, the only way to safely put the computer into a situation to survive a low power condition is to detect it before it happens...which isn't really possible.

Instead what happens is we have voltage sensors on all of the PSU rails, generally within the PSU. The BIOS monitors the voltage on each of these rails, and there is a set tolerance that cannot be exceeded safely. If you go much beyond 12v+/- on an HDD, for example, you can physically damage the drive. Same goes for a CPU (modern Intel CPU's run at about 1.32v maximum, and if it's too high, damage can occur). What happens is similar for thermal monitoring. The BIOS will do a hard shutdown if it detects too much voltage droop, or too high a voltage spike. Or it may sound an alarm.

I really didn't bring it up in the other thread, but hardware damage can occur if your PSU gets into a situation where it maxed out (the situation where a component isn't getting enough power means the PSU is maxed out). When a PSU is maxed out, or exceeded, assuming the PSU doesn't shut itself down (which they will do if their limit is exceeded), you get unstable power from it. Unstable power, even within limits, shortens the life of components. It doesn't slow them down, it destroys them slowly.

So as you can see, it's better to bring the system down than damage it. There's no real need for a components ability to go into a low power state because of an emergency solution in a desktop, because the only solution is to replace the unit, otherwise you risk damage.

Me...I've only taken a few electrical engineering classes. Most of what I know I've learned in the field. I always do say that you learn more by going out and doing, than sitting down and reading.

I didn't have any of that. As I said, my computer ran beautifully at 80 watts (calculated) under what my system required. I checked earlier this evening at a event I was entered with about 30 avatars in close proximity. Lag was somewhat high but not so bad as to be annoying.......RAM usage was 45% the whole 2 hours I was there. CPU stayed at 16 to 18%. There was no changes to my system except the power supply.

Now here you go again......because you don't understand the why you doubt it happened. It did happen. And with all the talk you have done you still have not told me why. I guessed at the reason.....using a little logic. You have done nothing but doubt me. And tell me and others that my contention is "suspect". And, as Osgeld has said at least a couple times, I have not said nor implied that someone should go out and spend money on a power supply to boost performance.........I have said (and implied) that an underpowered computer will not perform to it maximum.

I would think with all the absolutes you use to explain the ins and outs of a computer you might have some idea why........but, as long as you simply choose to disbelieve what I've said you cannot even speculate. It's okay.........I feel better about my computer now. I feel I can safely upgrade my graphics card to what I want now. The PSU purchase was for that alone.........not to sup up my computer.

I haven't told you the "why" because the "what" makes no sense from a technical standpoint. Realistically speaking, what you are explaining is impossible. Higher GPU performance (which is the only really semi-realistic thing), higher CPU speeds, more RAM (or compressed RAM)...

I can speculate on certain things. An overclocked GPU, overclocked video RAM, and overclocked CPU can explain the improved performance there. I cannot speculate on lower RAM usage. There is nothing, absolutely nothing, that can explain that short of you having 8GB, and booting into a 64 bit OS with one PSU and a 32 bit OS with the other PSU.

I'm explaining the way these things work in the computers we currently have. Our current computers do not have mechanisms that slow everything down when the PSU isn't powerful enough.

I'm not making these things up.

http://www.extremetech.com/article2/0,1697,1815446,00.asp
http://www.realworldtech.com/page.cfm?ArticleID=RWT110100000000
http://www.politicalcrossfire.com/forum/viewtopic.php?t=81086
http://forums.anandtech.com/messageview.aspx?catid=31&threadid=2186907
http://www.accessmylibrary.com/coms2/summary_0286-6607594_ITM
http://www.motherboardpoint.com/t168157-symptoms-of-an-underpowered-x1950pro-512mb-agp.html

I'll stop there because I don't think any more proof is required, but I could keep going if need be.

I'm still thinking that it's a placebo effect (as I've said before, similar things have happened to me and others. I change something, and I can swear it's faster...even if it's not, or not faster to the point where it can be noticed).

What would make a serious difference is if the smaller power supply didn't have enough plugs to put a power plug in the video card. Without this being plugged in, the card will automatically down-clock itself so as not to fry the video card slot. Plugging this in with an adequate power supply will then cause a perceived "boost" which is just the video card running at the speed it's supposed to.

All-in-all, it's more like a "it even ran?" sort of thing.

Under-powered supplies usually involve lots of system instability (BSOD, random reboots, etc) before the p/s eventually burns itself out.

Ya'll can go back to beating the dead horse now.

ok your starting to really piss me off



this is simple, power on, post, cpu found but not working right, reduce speed untill it is go on with life, nvidia chipsets have been doing this since model 1



were not talking about active power management end of story, were not talking about green hard drives that really dont exist, were talking basic logic gate functions and improvements to internal chipsets

[

are you really that hard set that everything in a computer is 100% perfect 100% of the time?

my voltages go up and down several times a second, if what you were saying were true a computer could not operate at all, were talking about components that can operate at a wide range of voltages and at all kinds of wattages depending on load

you your self just told someone to notch down their cpu voltage, DO THAT IN SOFTWARE DURING POST problem solved

lower voltage * same amps = less wattage

and how far you can go is totally dependent on the chip design


your right there, thats why its done before the computer even starts

since you have no tangible knowledge of how a computer actually works outside of your hobby/breakfix experience ill fill you in on the fact that the computer does several test's even before you see the video bios pop up, thats how it knows if you set your cpu wrong, or your ram isnt installed ect...


were talking about going lower, and the power regulators take care of the rest to keep it stable



no sh* sherlock thats why the motherboards power regulator (you know those big lm314 power regulators on every mainboard) does that task, if the world was perfect, 7200 rpm didnt = 6892 or 7345, and 110v didnt = 92 ~ 138 we wouldnt need them

the only thing you can harm is your supply, nothing else, as much as you like to think so


you can not harm any thing else but the supply by not feeding the system enugh power, period

unless you operate the system in a reduced state, then the only risk is speed loss



thanks, good advice, ill just go tell my boss that i dont know what im doing cause im a book baby, and all those embedded systems ive made will simply not operate becuase the power cannot provide perfect output 100% of the time

call me when you biuld a rf router in the feild with 2 mcu's, a lm314, and a cigarette lighter type butane torch, to fix service to millions of cell phones while the real part is "on order"

i have already told you i recently poped a power supply, yea it died due to overload (after 2 years) and installed a new weak one i had laying around, i get a message with the old nvidia drivers that came on the cd that this exact thing is happening, my drives do not sound the same (if you really want me to ill go get my dads strobocon, tell you the pitch differences and calulate the rpm difference), my fans spin slower and my cpu and video card is running at about half speed just to keep a stable system

i didnt loose anything on my drive, and the system is perfectly stable

everything you say cannot happen did happen with no ill effects other than being slow

now quit hijacking threads to prolong a null argument becuase you dont know basic digital electronics, and refuse to accept that the world is not a perfect place

i can only imagine what the op thinks, sorry about all of this

When a CPU error is detected on POST, startup halts and you get a beep code. The specific beep code varies depending on manufacturer.

Never heard of nVidia chipsets allowing a malfunctioning CPU to pass post. Mind posting a source on that?

Oh, they exist. Good idea, too. They don't save much power, but a little in abundance really adds up.
http://www.wdc.com/en/products/products.asp?DriveID=336

We're talking about more than gates and logic. It's not even limited to chipsets, although they certain factor into it.



No, and I didn't imply that. Computer components are built to operate within certain tolerances.

Everybodies voltages fluctuate. How much they fluctuate depends on the quality of the PSU and the load it's under.

As long as the PSU stays within tolerances (and even poor quality PSU's stay within tolerances if you don't load them too heavily), then everything is golden. Hell, my 12v rail runs at 12.5v.

im so sick and tired of you

go get a degree, make more than 9 bucks an hour to hand hold grandma tru AOL and quit calling other ppls experiences bogus becuase you refuse to understand


im not making stuff up, fuck ill mail you the computer, its just a junk extra anyways just to shut you up...

its happened to 2 ppl in the last month, both must be wrong

your god please forgive me, how dare i question you

ill go back to desiging my cell phone network, you can go to hell with your ignorance

and your the one that entered this thread screaming bullshit eventho its not

I've got a degree.

I make $15 an hour. Not much, but it's something.

I've never done phone support professionally (only as favors).

I haven't called it bullshit because I don't think it's on purpose. I believe it's a placebo effect that she has taken to heart by now. Happens a lot. I don't blame the person, they just get defensive when something they believe to know is called into question.

Two people, huh. Only two people I've ever heard of this happening to then in my six years of doing this stuff, then. Not making that up, either. I've seen many people troubleshooting crashes only to find that they installed another component (generally a video card) only to have their system start crashing due to an underpowered PSU, or troubleshooting random system crashes only to find out the person never plugged in the video cards aux power connector. But I've never heard of anyone gaining performance.

Interestingly, you're getting extremely defensive. Possibly trying to put me on the defensive, or potentially trying to discredit me. Perhaps you're aware of inconsistencies or falsehoods within your claims, and are trying to point attention away from them and toward falsehoods you're making up about me.