Could someone with muchos graphics knowledge help me prove a point?

[krooton]

Some fool on another forum is positive that graphics cards make a noticable difference in the 2D rendering of standard apps.

I agree that there is, but for non graphic-intensive applications the difference would be unnoticable.

We used MS Word as our reference app.

I've tested on my work machine with it's intergrated intel graphics, and the load time of the app (through the netowrk no less as our profiles are web based) was 1.5 seconds.

He is certain that when he underclocked his 7900GT by around 100mhz, there was a multiple second increase in the load time.

Utter tripe.

The amount of gpu power needed to render the 2D layout of a MS Word document is almost nothing, as proven by my own work machine.

As such, if you had an identicle system in every aspect, but one had say an FX 5200, and the other that uber Nvidia card, the most powerful AGP one on the market, there would be no noticable difference in the draw time of loading of MS Word, correct?

 

[krooton]

That's just having no drivers though. I think nearly any display adaptor nowadays with the correct drivers would perform general gui functions fully without chugging.

Exactly the point I am trying to make.

He is adamant that upgrading some guys card from a FX 5200 to a 9600 Pro would have a noticable difference in the 2D rendering of standard desktop stuff.

 

[krooton]

I think desktop speed does very between ATI/Nvidia. I get to use many different cards/equipment and although I have never investigated Nvidia is more snapper in 2D than ATI. Just an observation. Its not seconds faster though.

The focus of the debate is that some dude that help spec an upgrade is still using his FX 5200.

He isn't a gamer, so it doesnt really matter, the card is more than up to any 2D rendering XP will throw at it, correct?

This other guy says that if the card was upgraded, there would be a noticable improvement to the 2D rendering.

I upgraded the guy to a A64 3200+ with 1G ram (he was using a Duron with 384MB ram).

 

[krooton]

Your right what your saying and your advice is sound. My response is to your initial statement.

"Some fool on another forum is positive that graphics cards make a noticable difference in the 2D rendering of standard apps."

There is a difference between Nvidia & ATI. I had a X1900GT at work on my development machine and I switched to an 7900GTO and immeadiately everything felt a lot snappier in 2D. Like I said just an Observation.

And as he already has an Nvidia card, that becomes moot point?

Everyone agrees then that there is no need for him to pay money to upgrade his graphics card if he isn't a gamer?

 

[krooton]

No need to upgrade, however it will limit his upgrade potential in future if he sticks with AGP here. If I remember from the other thread he doesn't upgrade much anyway so it doesn't matter hugely.

It was a case of him already having an AGP card and not being a gamer, so speccing him a PCI-E mobo would up the overall cost of the spec which had a tight budget.

He would also get next to nothing if he tried to sell the FX 5200 to fund anything else.

Plus, should he need to, 9xxx cards are going pretty cheap second hand for the time being, or he could the X1300 AGP route.

 

[krooton]

Apologies if I missed it but what exactly would the benefit be even if this guy was right? It isn't like Word taking a second longer to load up means anything to average user

He's the argumentative sort that never admits defeat

I would link to the thread, but I'm not sure if you are allowed to in OCUK, hehe

 

[krooton]

this is true, on my old cirrus logic 512KB ISA graphics card i remember dragging windows to be very slow and jerky/tearing like when you put a new graphics card in your pc and it doesn;t have its drivers intalled.
then switching to an ati rage 128 card mate everything ultra fans and thats all just in 2d apps.

so the guy on the other forums is making sense.

however the 2d speed of all gfx card seemed to have saturated at the point where the old riva tnt/matrox millennium/rage128 etc were out. now only the 3d speed increases since there is not much more than can be done with 2d.

if the guy is saying that going from 6600gt to a 7900gt will increase the rate at which you can scroll text in windows, them im sorry to say that he popped out of his mammas wrong hole

Haha, he quoted the first half of your post as if to say it proved me wrong, even when you clearly said that 2D rendering peaked in cards made 9 years ago in the second half.

He maintains that he is correct.

I tested the 'startup' time of MS Word on my home machine, give or take a second, but then the whole machine is a big step up from my work one, so if the performance gain is a measly 0.5 seconds, surely the fact I have an A64 at 3Ghz and 1GB ram would account for more of that than my X1800XT does.

What a noob!

 

[krooton]

In PCMark05, on the test where you have lots of small windows flying about everywhere I noticed a substantial difference in performance when switching from a 9800 pro to an X1900 XT. I doubt you'll ever do anything as intensive as that during normal desktop usage but the point is, a faster card can make a difference in 2D.

I'm not doubting that faster cards would make a difference when rendering multiple changing frames over a period of time, but our debate is whether the singular rendering of a standard app would be noticably changed on it's load, such as running MS Word.

There is no way a 9800 XT would render a MS Word document noticably faster than an FX 5200 would.

 

[krooton]

Here's how it started:

General Zod wrote on 12-10-2006 12:45 PM

Soliloquy wrote on 12-10-2006 11:52 AM

Stakker wrote on 12-10-2006 11:49 AM

i dont, but it its slow on other things, i am gonna change it.

other things like what? *** GFX card would only effect graphics rendering. it shouldnt have much (in any) effect on the net / MS office / music production or even photoshop

You what? Shouldn't have much affect on Photoshop? The GFX is a very important aspect of a PC, and regardless of the backend, if a poor performance card is installed the entire experience is affected. For example, my machine instantly loads Word however after reducing the clock speeds the load time increases by a few seconds. [Takes longer to display].

Then

Soliloquy wrote on 12-10-2006 12:50 PM

i didnt say it wouldnt affect system performance, i said it shouldnt have a noticable effect. besides, if all you are doing is loading word up then a GFX card from 5 years ago would be powerful enough.

as for photoshop - as far as i'm aware it doesnt call any GPU specific instructions during image rendering. maybe the application has intergrated itself with the GPU more closely since when i used to use the program (which admittedl was a few years back)

Then this (The guy is running a 7900GT)

General Zod wrote on 12-10-2006 12:52 PM

I reduced my clock by 100Mhz and memory by 200MHz though I was troubleshooting. Try running Windows in both Safe mode and normal mode and compare the difference in display times. The HD load will remain identical but the application will take longer to display on the screen.

Then me:

kr00t0n wrote on 12-10-2006 01:03 PM

If it occured in safe mode, that would be because it isn't using the card's drivers, and just the native vga settings in XP.

As Soliloquy says, old graphics cards render it near instantly, even intergrated onboard gfx like in all our work pcs do.

He replies nonsensically:

General Zod wrote on 12-10-2006 01:15 PM

Which is precisely the point. A machine installed with high performance HD, RAM, and CPU would experience a bottleneck when displaying 2D/3D images whether in safe mode or Windows mode.

Which was responded to with:

Soliloquy wrote on 12-10-2006 01:23 PM

actually that wasnt his point - you've misinterpreted his post.

his point is that in safe mode any GFX card will display slower as it's running in a native VGA mode with no GPU chipset specific functions (et al) being called. plus more than likely no graphics card memory being utilised and the GPU clock speed throttled back as well.

safe mode is called that for a reason - it uses minimum resources to help diagnose or correct problems safely

a geforce 10 with 10GB of memory and a 9THz GPU clock speed (if such a device existed) would more than likely render the desktop at the same speed as a Geforce 2 MX

I reply:

kr00t0n wrote on 12-10-2006 01:41 PM

Exactly, safe mode put any sort of processing, graphical or not, on the cpu and cpu alone.

With any sort of gpu, be it an onboard SIS chip in a mobo, a PCI S3 Virge, a Voodoo 3, a 9800 pro, or a pair of 7900GTX's in Sli, the difference in the speed rendering of non-graphic intensive applications would be would neglible and not something to be perceived by the end user.

And he replies with this, and I lol'd:

General Zod wrote on 12-10-2006 01:50 PM

If one compares two identical machines with alternate GFX' installed [7900GT and 7950GT] then the difference is 100% noticeable when comparing side by side. Granted, its minimal however a program is considered to be loaded when displayed on the screen and if a delay within the rendering/refresh/memory dump occurs then the load times are considered slower.

Noticable difference between those two cards in the same machine?

Yeah, right, whatever

And he continued for 3 pages.

 

[krooton]

Ah but the OP was making a point about the gfx cards affect on 2d gfx, which clearly it does.

The thing is, if you check the convo I quoted, we never denied that it does, just that in most cases it isn't noticable.

 

[krooton]

He is certain that when he underclocked his 7900GT by around 100mhz, there was a multiple second increase in the load time.

He moaning about this quote, as I forgot to mention his card seems to be having memory issues.

Regardless, of course a faulty card will have a noticable impact on anyhting graphical, but a standard, working card, of recent enough spec, even if it is quite poor in 3d stuff, woudl still not offer any noticable loss in the rendering of a 2D windows desktop application compared to a higher end one of the same socket.

 

[krooton]

Thanks guys