GF3Ti500 vs. GF4Ti4200 in games

Just in case anyone wants to know what to expect from a vid card upgrade on a medium power system. Specs. - 1400Athlon, 512mb Simply did tests with the Ti500 then installed the Ti4200 (with proper drivers) and ran same tests again. Doom3 Alpha- 12x10 4200=map 1= 9 fps 12x10 500 =map 1= 9 fps 12x10 4200=map 2= 14.5 fps 12x10 500 =map 2= 14.5 fps Comanche4- 12x10 4200 w. AA on =25 fps 12x10 4200 w. AA off=38 fps 12x10 500 w. AA off=15.5 fps Mecwarrior3- 10x7 4200 w. AA on =25 fps 10x7 4200 w. AA off=49 fps 10x7 500 w. AA off=49 fps RalliSport Challenge- 6x4 4200 w. AA off=38 6x4 500 w. AA off=26 6x4 4200 w. AA on=38 6x4 500 w. AA on=26 12x10 4200 w. AA off=25 12x10 500 w. AA off=19 12x10 4200 w. AA on=19 12x10 500 w. AA on=19 16x12 4200 w. AA off=25 16x12 500 w. AA off=NA 16x12 4200 w. AA on=19 16x12 500 w. AA on=NA

Oh, I used 4S(the highest) AA with the 4200 and Quin AA with the 500. Destroy wrote:

Quoted text (51 lines)

> Just in case anyone wants to know what to expect from a vid card upgrade > on a medium power system. > > Specs. - 1400Athlon, 512mb > Simply did tests with the Ti500 then installed the Ti4200 (with proper > drivers) and ran same tests again. > > > Doom3 Alpha- > 12x10 4200=map 1= 9 fps > 12x10 500 =map 1= 9 fps > 12x10 4200=map 2= 14.5 fps > 12x10 500 =map 2= 14.5 fps > > > Comanche4- > 12x10 4200 w. AA on =25 fps > 12x10 4200 w. AA off=38 fps > 12x10 500 w. AA off=15.5 fps > > > > Mecwarrior3- > 10x7 4200 w. AA on =25 fps > 10x7 4200 w. AA off=49 fps > 10x7 500 w. AA off=49 fps > > > > RalliSport Challenge- > 6x4 4200 w. AA off=38 > 6x4 500 w. AA off=26 > > 6x4 4200 w. AA on=38 > 6x4 500 w. AA on=26 > > 12x10 4200 w. AA off=25 > 12x10 500 w. AA off=19 > > 12x10 4200 w. AA on=19 > 12x10 500 w. AA on=19 > > 16x12 4200 w. AA off=25 > 16x12 500 w. AA off=NA > > 16x12 4200 w. AA on=19 > 16x12 500 w. AA on=NA > > > >

Destroy wrote:

Quoted text (41 lines)

> Just in case anyone wants to know what to expect from a vid card upgrade > on a medium power system. > > Specs. - 1400Athlon, 512mb > Simply did tests with the Ti500 then installed the Ti4200 (with proper > drivers) and ran same tests again. > > Doom3 Alpha- > 12x10 4200=map 1= 9 fps > 12x10 500 =map 1= 9 fps > 12x10 4200=map 2= 14.5 fps > 12x10 500 =map 2= 14.5 fps > > Comanche4- > 12x10 4200 w. AA on =25 fps > 12x10 4200 w. AA off=38 fps > 12x10 500 w. AA off=15.5 fps > > Mecwarrior3- > 10x7 4200 w. AA on =25 fps > 10x7 4200 w. AA off=49 fps > 10x7 500 w. AA off=49 fps > > RalliSport Challenge- > 6x4 4200 w. AA off=38 > 6x4 500 w. AA off=26 > > 6x4 4200 w. AA on=38 > 6x4 500 w. AA on=26 > > 12x10 4200 w. AA off=25 > 12x10 500 w. AA off=19 > > 12x10 4200 w. AA on=19 > 12x10 500 w. AA on=19 > > 16x12 4200 w. AA off=25 > 16x12 500 w. AA off=NA > > 16x12 4200 w. AA on=19 > 16x12 500 w. AA on=NA

Ouch ! Could you post Doom3 at 10x7 with the 4200. Looks like i might have to upgrade my CPU. Great post. Thanks.

This link might be helpful http://firingsquad.gamers.com/hardware/gf4athlonscaling/default.asp

")-()-(" <[email hidden]> wrote in message news:[email hidden]...

Quoted text (13 lines)

> Destroy wrote: > > Doom3 Alpha- > > 12x10 4200=map 1= 9 fps > > 12x10 500 =map 1= 9 fps > > 12x10 4200=map 2= 14.5 fps > > 12x10 500 =map 2= 14.5 fps > > Ouch ! > Could you post Doom3 at 10x7 with the 4200. > Looks like i might have to upgrade my CPU. > Great post. > Thanks. >

Just thought I'd mention the fact that the Doom 3 Alpha is in no way representative of the performance of the Doom 3 retail version. It is optimised for the NV30, and there isn't much in the way of optimisation for anything else (it is an Alpha, and not a release - it was designed to run on the computers at E3 only). So just hang on a little before you buy that new computer...

"Destroy" <[email hidden]> wrote in message news:[email hidden]... > Oh, I used 4S(the highest) AA with the 4200 and Quin AA with the 500. Which of course, makes the comparisons rather useless. 4S is extremely bandwidth and fillrate intensive, it's by far the most resource-hogging AA mode on the GF4 line. If you're comparing performance, then keep the graphical settings the same.

>>Oh, I used 4S(the highest) AA with the 4200 and Quin AA with the 500. > > > Which of course, makes the comparisons rather useless. 4S is extremely > bandwidth and fillrate intensive, Well really doesn't seem to matter anyway. After playing a while, any AA with either of these cards on my system is not worth the performance hit IMO. AA is for cards that can handle - ATI 9700 and hopefully the upcoming GF FX.

On Sat, 30 Nov 2002 00:58:08 GMT, Destroy <[email hidden]> wrote:

Quoted text (9 lines)

>Oh, I used 4S(the highest) AA with the 4200 and Quin AA with the 500. > >Destroy wrote: >> Just in case anyone wants to know what to expect from a vid card upgrade >> on a medium power system. >> >> Specs. - 1400Athlon, 512mb >> Simply did tests with the Ti500 then installed the Ti4200 (with proper >> drivers) and ran same tests again.

Really a better upgrade for that system would be cpu. Notice on a couple test the results didn't change between the cards, this is because your cpu couldn't feed even the Ti500 data fast enough to take full use of it.

Y'kno... I just upgraded to a P4/2.66Ghz/GF4 Ti 4600/512Mb Corsair XMS 512Mb PC3500/Antec 430W system... built especially for racing sims... and I'm not so terribly impressed.... Yes, it is fast... I can run 1600x1200 (no-FSAA/4xAF) on a 19" CRT... (F1 2001/2002/GPL/MBTR/Heat/Thunder2003/SCGTat10x7) But somehow... I expected more... (4sFSAA/8xAF in SCGT with Mod cars is still too much...) I get just shy of 13,000 on 3DMark2001.. Compared to upper 8's on my old system.. (AMD 1600+/GF3 Ti200/512Mb Crucial PC2100/Enermax 350W) But, I see little real-world difference... On the 19", 1280x960 is fine... (AA&AF a bonus) Luckily, it's no loss... I'm considering dedicating the P4 to audio recording (w/ the GF3) and using the AMD (w/ GF4) for games... Real-time audio-processing will probably make more benefit of the CPU... But, I'd recommend anyone about to lay out hard-earned cash, to shoot lower in CPU speed... In practice, it didn't help me much... Regards... "Mark H." <[email hidden]> wrote in message news:[email hidden]...

Quoted text (8 lines)

> On Sat, 30 Nov 2002 00:58:08 GMT, Destroy <[email hidden]> wrote: > > Really a better upgrade for that system would be cpu. > Notice on a couple test the results didn't change between the cards, > this is because your cpu couldn't feed even the Ti500 data fast enough > to take full use of it. > >

"Glaspak" <[email hidden]> wrote in message news:asf6l4$1n$[email hidden]... > Y'kno... I just upgraded to a P4/2.66Ghz/GF4 Ti 4600/512Mb Corsair XMS512Mb

Quoted text (11 lines)

> PC3500/Antec 430W system... > built especially for racing sims... > and I'm not so terribly impressed.... > > Yes, it is fast... > I can run 1600x1200 (no-FSAA/4xAF) on a 19" CRT... > (F1 2001/2002/GPL/MBTR/Heat/Thunder2003/SCGTat10x7) > > But somehow... I expected more... > (4sFSAA/8xAF in SCGT with Mod cars is still too much...)

SCGT is a poorly optimized sim, and when you mix in high polygon-count cars, even a Cray will bog down. I used to get better fps with a PIII 500 and voodoo5 than I do now with a xp 18+/gf3ti200

Quoted text (15 lines)

> > I get just shy of 13,000 on 3DMark2001.. > Compared to upper 8's on my old system.. > (AMD 1600+/GF3 Ti200/512Mb Crucial PC2100/Enermax 350W) > But, I see little real-world difference... > On the 19", 1280x960 is fine... (AA&AF a bonus) > > Luckily, it's no loss... > I'm considering dedicating the P4 to audio recording (w/ the GF3) > and using the AMD (w/ GF4) for games... > Real-time audio-processing will probably make more benefit of the CPU... > > But, I'd recommend anyone about to lay out hard-earned cash, > to shoot lower in CPU speed... > In practice, it didn't help me much...

your gf 4 is fast when used in lower resolutions. ONce you climb over 12x8 and begin using fsaa and anistrophic stuff, it overwhelms the card. You did right by going for a fast cpu. You just need to realize the other limits your sytem still has. Turn off the Fsaa and other effects and the card should scream (just not with SCGT){biggest frame-rate improvement for scgt is to use the 1/4 size mirror from the uspits. Try benchmarking with n2k2 or the 3dsetup program for F1 2k2. I'd guess without fsaa enabled, you should get over 40 million triangles in f1 2k2. dave henrie > > Regards... > >

"Dave Henrie" <[email hidden]> wrote in message news:_BFG9.200606$NH2.13870@sccrnsc01... > SCGT is a poorly optimized sim, and when you mix in high polygon-count > cars, even a Cray will bog down. I used to get better fps with a PIII 500 > and voodoo5 than I do now with a xp 18+/gf3ti200 I agree Dave... I have a Voodoo5 that runs the original SCGT in Glide at 800x600 beautifully (PIII/733/256Mb) > your gf 4 is fast when used in lower resolutions. ONce you climb > over 12x8 and begin using fsaa and anistrophic stuff, it overwhelms the > card. You did right by going for a fast cpu. You just need to realizethe > other limits your sytem still has. Turn off the Fsaa and other effectsand > the card should scream (just not with SCGT){biggest frame-rateimprovement > for scgt is to use the 1/4 size mirror from the uspits. > Try benchmarking with n2k2 or the 3dsetup program for F1 2k2. I'dguess > without fsaa enabled, you should get over 40 million triangles in f1 2k2. > dave henrie I do get 43+M-tri in F1 2K2 at 1280x960x32... But at 1024x768 w/4xFSAA&8xAF it's too slow. (complain...complain... I know) It is Fast... (very fast) I guess that I was just expecting a HUGE jump in performance... Which I didn't get... Like I said... no loss... I'll use this system for something else.. The AMD1600+ with a GF4 Ti4600 should run everything "well enough"... (1024x768x32) Btw... I do use the 1/4 size fuzzy-mod mirror patch in SCGT (60fps with no mirror)... I even reduced smoke t bump the FPS... (btw2... do you notice a scruntched effect with the dash-view in the newly-released Nordscleife?) Thanks for the input Dave...

"Glaspak" <[email hidden]> wrote in message news:asf6l4$1n$[email hidden]... > I get just shy of 13,000 on 3DMark2001.. > Compared to upper 8's on my old system.. > (AMD 1600+/GF3 Ti200/512Mb Crucial PC2100/Enermax 350W) > But, I see little real-world difference... > On the 19", 1280x960 is fine... (AA&AF a bonus) What resolution are you running you old AMD + GF3 at? I have a P4 2.7G + GF3 Ti 200 overclocked to almost Ti500 and I only get 8500 in 3D Mark 2001 under 1280X1024. Seems to me your old 3D Mark number is really high.

"Glaspak"

Quoted text (8 lines)

> Btw... I do use the 1/4 size fuzzy-mod mirror patch in SCGT (60fps with no > mirror)... > I even reduced smoke t bump the FPS... > (btw2... do you notice a scruntched effect with the dash-view in the > newly-released Nordscleife?) > > Thanks for the input Dave... >

Nope I missed that, I had not run SCGT for quite some time and so I dloaded an Oreca Viper just for the Nordscliefe.(had to put up with an ART Viper dash tho...tsk tsk) :) So I didn't perceive the scrunched view. Perhaps they messed with the FOV slightly. dave henrie

On Mon, 02 Dec 2002 18:05:28 GMT, "Dave Henrie" <[email hidden]> wrote:

Quoted text (16 lines)

> >"Glaspak" >> Btw... I do use the 1/4 size fuzzy-mod mirror patch in SCGT (60fps with no >> mirror)... >> I even reduced smoke t bump the FPS... >> (btw2... do you notice a scruntched effect with the dash-view in the >> newly-released Nordscleife?) >> >> Thanks for the input Dave... >> > Nope I missed that, I had not run SCGT for quite some time and so I >dloaded an Oreca Viper just for the Nordscliefe.(had to put up with an ART >Viper dash tho...tsk tsk) :) So I didn't perceive the scrunched view. >Perhaps they messed with the FOV slightly. >dave henrie >

There is a patch to fix this scrunched view. http://garage.racesimcentral.com/ -- karkrazy AT softhome DOT net http://karkrazy.m4driving.sm

I had an AMD 1800XP, GF3 Ti 200 (o/c'd to 210 / 500) - 512 Mb DDR PC2100 running FAST at Cas 2 etc. 3DMark 7750 (ish) (all this on a 133 FSB). 30.82 reference Detonators. I wouldnt worry if you score 8500 - that sounds about right for the Ti 500 at default clock speeds. Though your investment in a 2.7G P4 deserves a beefier GF4 Ti card - a 4200 o/c'd to Ti 4400 levels will bring you up to 13000 ish imho (with 40 Detonators). "Goldfinger" <[email hidden]> wrote in message news:KNLG9.209253$[email hidden]...

Quoted text (15 lines)

> > "Glaspak" <[email hidden]> wrote in message > news:asf6l4$1n$[email hidden]... > > > I get just shy of 13,000 on 3DMark2001.. > > Compared to upper 8's on my old system.. > > (AMD 1600+/GF3 Ti200/512Mb Crucial PC2100/Enermax 350W) > > But, I see little real-world difference... > > On the 19", 1280x960 is fine... (AA&AF a bonus) > > What resolution are you running you old AMD + GF3 at? I have a P4 2.7G + > GF3 Ti 200 overclocked to almost Ti500 and I only get 8500 in 3D Mark 2001 > under 1280X1024. Seems to me your old 3D Mark number is really high. > >

I think you might have your wires crossed... why is it optimised for the nv30 when it was played on the r9700? Mike "Nick" <[email hidden]> wrote in message news:asa7e2$ps4i8$[email hidden]...

Quoted text (19 lines)

> ")-()-(" <[email hidden]> wrote in message > news:[email hidden]... > > Destroy wrote: > > > Doom3 Alpha- > > > 12x10 4200=map 1= 9 fps > > > 12x10 500 =map 1= 9 fps > > > 12x10 4200=map 2= 14.5 fps > > > 12x10 500 =map 2= 14.5 fps > > > > Ouch ! > > Could you post Doom3 at 10x7 with the 4200. > > Looks like i might have to upgrade my CPU. > > Great post. > > Thanks. > > > > Just thought I'd mention the fact that the Doom 3 Alpha is in no way > representative of the performance of the Doom 3 retail version. It is > optimised for the NV30, and there isn't much in the way of optimisation

for > anything else (it is an Alpha, and not a release - it was designed to runon > the computers at E3 only). So just hang on a little before you buy thatnew > computer... > >

Because John Carmack is the guy all video card manufacturers, especially ATI and nVidia, go to during development. If you read his .plan updates, you'll get a feel for this. The fact that E3 is about upcoming software, not hardware, means that no NV30 technology was given to the D3 booths at E3, so they had to use the best available hardware to get a decent framerate, maybe spend a week writing temp optimisation routines for the 9700 so the journos don't get upset. Remember glQuake? id Software supported 3DFX, and so all the other developers did, and so all the customers bought one. PowerVR missed out, despite being better and cheaper. Now JC is the guy you need to impress if you want to be a successful video card company. Surely you knew all this? "Mike P" <[email hidden]> wrote in message news:wHbL9.6304$[email hidden]...

Quoted text (28 lines)

> I think you might have your wires crossed... why is it optimised for the > nv30 when it was played on the r9700? > > Mike > > "Nick" <[email hidden]> wrote in message > news:asa7e2$ps4i8$[email hidden]... > > ")-()-(" <[email hidden]> wrote in message > > news:[email hidden]... > > > Destroy wrote: > > > > Doom3 Alpha- > > > > 12x10 4200=map 1= 9 fps > > > > 12x10 500 =map 1= 9 fps > > > > 12x10 4200=map 2= 14.5 fps > > > > 12x10 500 =map 2= 14.5 fps > > > > > > Ouch ! > > > Could you post Doom3 at 10x7 with the 4200. > > > Looks like i might have to upgrade my CPU. > > > Great post. > > > Thanks. > > > > > > > Just thought I'd mention the fact that the Doom 3 Alpha is in no way > > representative of the performance of the Doom 3 retail version. It is > > optimised for the NV30, and there isn't much in the way of optimisation > for > > anything else (it is an Alpha, and not a release - it was designed to

run

Quoted text (8 lines)

> on > > the computers at E3 only). So just hang on a little before you buy that > new > > computer... > > > > > >

"Billy-Bob" <[email hidden]> wrote in message news:[email hidden]...

Quoted text (8 lines)

> On Tue, 17 Dec 2002 18:52:50 -0000, "Nick" <[email hidden]> > wrote: > > >PowerVR missed out, > >despite being better and cheaper. > > And many people hated PowerVR because they used underhanded marketing > techniques.

Everybody uses underhanded marketing techniques if they know they are second best in peoples' minds. Remember the UT test CDs which still had the demonstator documentation on? Phrases like 'show them the big pipe in the middle of the level and pass it off as curved surface rendering'. Priceless.