I remember finishing the game on a Geforce 2 MX 400 64mb SDR 128-bit, 384mb SDR PC-133 ram and Pentium III 733mhz. Loading times took around 5mins. I literally have to lay down in my bed chill out for a bit before I could go on to the next stage. I was playing it at 640x480 at the lowest settings. Up to this day I still don't have a single clue how the heck did I manage to play it without my eyes burning out. Good times 😄
@@RAA12586Sure it was a pretty demanding game, but you were using terrible hardware from the offset! Radeon VE and then even a (minimum requirement) GF3!? You were behind the times! XD Respect.
@@blastfromthepast3073Dude cmon, nobody played at 20fps by 2003. I had a 22" Diamond Mitsubishi SB by then capable of 1920x1440@75Hz, and 160Hz@1280x1024. I was lucky that I live in the UK, and had access to decent hardware even working a terrible "minimum wage" job, but I'd left 20fps behind back in teh Doom 2 era of 1994/5. Quake was the game (specifically for multiplayer and Quakeworld) that cemented my need for 100fps at all times, making me upgrade my K5-PR133 to a Pentium 200Mhz (1996) and later a P2-266 (1998). Never looked back. FPS is king. Specially if you're talking iD Software multiplayer FPS games.
Not sure this is actually fair. The Voodoo card seems to not be able to use alot of the light technology that the Geforce does, and so has a much easier job. If the Geforce did the same as zvoodoo I think it would win easy.
3dfx was a driver for new tech, and nvidia allaways was marketing about 32 bit colour (it's correct naming 24 bits) and for now, we have 20 years without new tech's only lights and aliasing but finaly picture is same as was 20 years ago, so when glide die, we lost any chances to get something new and amazing, we lost magic...
@@liamiangaming7931 No it is not that simple. The game takes advantage of more features in the Geforce than the Voodoo so the voodoo gets of easy. Doom 3 was not suppose to be played on either card. If you run something which is supported by both cards or something like Warcraft 3, 3dmark2001 then you will see a completely different result.
@@andresjacobsen4551 Don't forget about Memorybandwith. Yes the GTS is a beast of Fillrate! I love the GPU for that. But you need so much faster Memory to enhanche (i dont know the spelling!) the full potential. So i checked a GF2Ti @ near DDR600... its a Monster. But a stock GTS can 't use the theoretical Fillrate. Its limited by "slow" Memory. The V5 and yes the Fury Maxx weren't bottlenecked so hard. It did not mean they are better of course not! The GTS has a Monster GPU!
@@user-yt3tj6df8m Not sure, the 2000s was pretty exciting. Ati (now owned by AMD) Radeon 9700 Pro released July 2002 helped personify the 2002-2006 generation. Nvidia Geforce GTX 8800 helped personify 2006-2011/2012 ish. Even though dx11 wasa out by 2009, we did not see adoption until 2011/2012. HD Radeon 7970 in late 2011 started a weird decade. It was a very powerful card that can still run well today with reduced settings, but it was an era of slowdown where people were starting to expect only 30% performance gains per generation instead of the traditional 60% or more. Of course the late 1990s saw massive gains per year in GPU and CPU.
I got the alpha leak from an FTP back in the day, only hit around 8-11fps on my Geforce 4 Ti 4800SE. When the full game came out, that number had only improved to around 16-20fps even on lowest settings in 640x480. The next week I went out and bought a 6800XT AGP and enjoyed the game in high settings, 1024x768 and pretty much a solid 60fps except maybe in Delta Labs.
thats kinda weird, i have a Geforce 4 4600 TI in my power mac g4 and it runs the game at around 30ish fps on medium settings i think, (its been a while dont hold me to that but i do remember the game being very playable) are you sure you weren't CPU bottlenecked or something?
@@flameboi4209 Absolutely sure. I was running a Dell Optiplex machine, lovely build quality a sort of workstation rig, with a P4 2.4Ghz CPU. As I said, installing the 6800XT massively improved Doom 3's performance. My FPS with the Ti4800SE was in line with what other people were getting with a similar set up.
Weird. On my GF3Ti200 oveclovked to Ti500 i had about 30fps 800x600 on high setting and it was playble. Maybe some other locations were more demanding and needed to set power detail.
@@grzegorzjurczynski6897 No. It's been documented, verbosely, for the last 20 years. I mean, there is a chance that you had some alien technology GF3 that was somehow 3x faster than ANY OTHER GF3 in the entire universe, and those thousands of detailed benchmarks from back in the day are wrong.... ....
@@TheVanillatech well check on yt, your score was arround geforce 2 performance. Maybe you had weak cpu. I found on YT comparision between many geforces generation in doom3 ( from 256 to 7800). I had sempron palermo overcloked to 2.2GHz and this CPU was beast for that price.
@@armorgeddon yes of course. 😊 Voodoos scaling very well with fast CPUs. Nvidia did a great job with T&L for CPUs back in the Days. But it becomes a bit obsolet with raw CPU Power.
@@alexandernorman5337 Yes that should be the reason, but it also shows that apart from the missing T&L feature the VSA-100 was very good. It would make for a very interesting written report or video at which CPU the crossover point is for the few games/engines of that time. I've never seen someone have a go at that.
I´m sorry for deleting the first version of this video. But i was not happy with that. Now i added a side by side comparision at the end. I really hope you like it.
You should definitely delete this one too. As others have pointed out, this comparison is simply flawed from the get-go. As much as I love the 3dfx cards, the GeForce 2 GTs is a faster, better card than the Voodoo 5 in... basically every way.
@@ruxandy It was not my intension to make somebody angry. I only benched different cards with the game the same render path and showed up that the voodoo (wich even doesn't have newer drivers) renders doom3 faster as a factory oced Geforce2. So thats all. I never said V5 is technical better than Gf2.
I'd prefer GF6x for 2000s games an ultimate performance solution. And TNT2 Ultra for pre-2001 game because it don't has ugly S3 texture compression like GeForce
Doom3 against that cards is like trying to pull a trailer with a Smart. ^^ There would have been better bechmarks then a game released four years after the cards had been released. Max Payne (2001), Quake 3, 3D Mark 2000 or 2001, Unreal Tournament or Unreal for example.
The gts was very bandwidth limited as the memory was way too slow for the chip. The ultra was a far more balanced card and a vast improvement, even excluding the faster clocked core. The mx has lowest bandwidth and fillrate of all these cards so couldn't hope to compete.
I remember Geforce2 Ultra was sold for 1200DM here in Germany. Should be 600USD if i remember right, crazy! In future i will do a Geforce2 Video, maybe from MX200 to Ultra. It was a great GPU!
@@blastfromthepast3073 yes the Gef2Ultra went for insane prices. It was a great card, mainly I think because the gpu wasn't bottlenecked by the memory, or not much anyway. It was the same story with the original Geforce, the sdram seriously bottlenecked the gpu. Really, Nvidia just trying to make xmas $$. It was only when the DDR version was released early in 2000, that the real-life bandwidth met the theoretical bandwidth. Those who rushed to buy the sdr version must have been kicking themselves.
@@patrickct9386 Yep! Geforce SDR had the same Bandwith like V3 3000. For V3 it was okay with 333MTexel Fillrate. Geforce 256 had 480MTexel with same Bandwith and sucked! Geforce2 GTS had 1600MTexel Fillrate but near same Bandwith like Voodoo5. So GTS never could use the Full Fillrate, in real life maybe 700MTexel. Ok using a tiny Resolution and 16Bit increase Performance but who did that with a expensive Geforce2?
as 3dfx voodoo5 owner at release date, and UT99 pro player, i can confirm that v5 was fastest card in UT99, GF2 Hercules prophet was faster in general especially games supported T&L(one of fastest gf2 64mb ddr), other stats; v5 was 30% cheaper (250$ vs 350$) in my country v5 worked great with cpu-s available at time (had 900mhz athlon tb, while 1ghz was top of line p3/athlon) gf2 was bit more bottle necked with those cpu-s Doom3 is from 2004 and those cards were from 2000 Some of demanding games from their time: UT99,Quake3,NFS5 Porsche,Counter Strike,NOLF, Sacrifice(T&L),Midtown Madness, Giants(T&L),Deus ex, Driver, etc 99/00 3dfx fan. first two love, Banshee 16mb sg, v5 5500..than gf4ti
This game also depends on game version (there is some 1.0.xx , 1.3.1 exists) and used operating systems.. and when is benchmark run repeated there are different results.. Its not best benchmarking material.
I benched the scene twice for a stable result but you're right. There little differences but it depends on your System too. Using a healthy clean System with enough and fast Memory will help a lot.
Geforce 2 cards were manufactured by many producers, with new cards still comming in spring 2001, while Voodoo 5 cards were made only by 3dfx/STB (There are some VSA-100 powered PowerColor cards, but they are Voodoo 4 and are quite rare) and all production was done by Christmas 2000. No suprised that there's a LOT more functional Geforce 2 cards. Buying STB not only killed 3dfx, it also made retrocomputing with late 90's/early 2000 machines much more expensive.
If runs on GTS so could run on a PS2 back in the day many people said only the xbox could run doom 3. Remember GTS has a high memory bandwidth that is why can still "run" doom 3 at least not at slow motion. So the PS2 memory badwitch is even much much higher than a GTS. But drawing shadows and light would have to be chaged completely using Vu0 like sillent Hil 2 did
Doom 3 would crush the PS2. It would crush the Gamecube too. The game's engine - idTech4 - was made to utilize 8bit stencil buffers for all the real time lighting and shadowing. The PS2 and Gamecube didn't have those. You could fake a hardware stencil buffer to try to run the engine on them, but there is a lot of processing overhead involved. And that would cripple those systems. I doubt you would get more than 7-8fps on either of those consoles. And you would have pathetic texture quality on the PS2 as well.
@@alexandernorman5337 Not if you rewrite the shadow renderer like Silent Hill 2 and Swat: Global Strike did using PS2 vector units. Look the shadows on those games
@@hugo-garcia You only have those two vector multiplication units in the PS2, and they are not very fast. The base geometry processing of the game would tax them fully. You really don't have a free resource for the stencil volumes - very expensive calculations because you have to project off of other surfaces - and so you cannot obtain a playable framerate. You also have a problem with memory. If your graphics hardware has an eight-bit stencil buffer, then you can just properly set that and associate that with the affected vertices. Those eight bits tell the renderer how much shadowing it needs to include for the shadow with the associated vertices. If you don't have that, then you need to create new vertices - with a proper offset to prevent Z fighting - and that means duplicating all the associated properties such as texture coordinates. That will cost you a lot more memory. And the PS2 doesn't have as much as Xbox. There is simply no playable way to run idTech4 on the PS2 hardware. Silent Hill 2 utilized a lot of blobs and other prebaked mapping operations to limit how often the game engine needed to do those projections. But every frame needs them - lots of them - with idTech4.
I remember finishing the game on a Geforce 2 MX 400 64mb SDR 128-bit, 384mb SDR PC-133 ram and Pentium III 733mhz. Loading times took around 5mins. I literally have to lay down in my bed chill out for a bit before I could go on to the next stage. I was playing it at 640x480 at the lowest settings. Up to this day I still don't have a single clue how the heck did I manage to play it without my eyes burning out. Good times 😄
Wow this is heavy! But back in the days we were able to play Games at 20fps and below. Good old times. ☺
I used a radeon 7500 then a geforce 3 on this games, was very system intensive in those times.
@@RAA12586Sure it was a pretty demanding game, but you were using terrible hardware from the offset! Radeon VE and then even a (minimum requirement) GF3!?
You were behind the times! XD
Respect.
@@blastfromthepast3073Dude cmon, nobody played at 20fps by 2003. I had a 22" Diamond Mitsubishi SB by then capable of 1920x1440@75Hz, and 160Hz@1280x1024.
I was lucky that I live in the UK, and had access to decent hardware even working a terrible "minimum wage" job, but I'd left 20fps behind back in teh Doom 2 era of 1994/5.
Quake was the game (specifically for multiplayer and Quakeworld) that cemented my need for 100fps at all times, making me upgrade my K5-PR133 to a Pentium 200Mhz (1996) and later a P2-266 (1998).
Never looked back. FPS is king. Specially if you're talking iD Software multiplayer FPS games.
Damn. I bought a 6800 just to play Doom 3....... Then I never finished it lol
Not sure this is actually fair. The Voodoo card seems to not be able to use alot of the light technology that the Geforce does, and so has a much easier job. If the Geforce did the same as zvoodoo I think it would win easy.
3dfx was a driver for new tech, and nvidia allaways was marketing about 32 bit colour (it's correct naming 24 bits) and for now, we have 20 years without new tech's only lights and aliasing but finaly picture is same as was 20 years ago, so when glide die, we lost any chances to get something new and amazing, we lost magic...
Well it is fair, they just needed to make the GeForce better.
@@liamiangaming7931 No it is not that simple. The game takes advantage of more features in the Geforce than the Voodoo so the voodoo gets of easy. Doom 3 was not suppose to be played on either card. If you run something which is supported by both cards or something like Warcraft 3, 3dmark2001 then you will see a completely different result.
@@andresjacobsen4551 Don't forget about Memorybandwith. Yes the GTS is a beast of Fillrate! I love the GPU for that. But you need so much faster Memory to enhanche (i dont know the spelling!) the full potential. So i checked a GF2Ti @ near DDR600... its a Monster. But a stock GTS can 't use the theoretical Fillrate. Its limited by "slow" Memory. The V5 and yes the Fury Maxx weren't bottlenecked so hard. It did not mean they are better of course not! The GTS has a Monster GPU!
@@user-yt3tj6df8m Not sure, the 2000s was pretty exciting. Ati (now owned by AMD) Radeon 9700 Pro released July 2002 helped personify the 2002-2006 generation. Nvidia Geforce GTX 8800 helped personify 2006-2011/2012 ish. Even though dx11 wasa out by 2009, we did not see adoption until 2011/2012. HD Radeon 7970 in late 2011 started a weird decade. It was a very powerful card that can still run well today with reduced settings, but it was an era of slowdown where people were starting to expect only 30% performance gains per generation instead of the traditional 60% or more.
Of course the late 1990s saw massive gains per year in GPU and CPU.
Is it true that 3dfx has a better score in higher resolution ?
WTF??? What are the actual settings???
I got the alpha leak from an FTP back in the day, only hit around 8-11fps on my Geforce 4 Ti 4800SE. When the full game came out, that number had only improved to around 16-20fps even on lowest settings in 640x480.
The next week I went out and bought a 6800XT AGP and enjoyed the game in high settings, 1024x768 and pretty much a solid 60fps except maybe in Delta Labs.
thats kinda weird, i have a Geforce 4 4600 TI in my power mac g4 and it runs the game at around 30ish fps on medium settings i think, (its been a while dont hold me to that but i do remember the game being very playable) are you sure you weren't CPU bottlenecked or something?
@@flameboi4209 Absolutely sure. I was running a Dell Optiplex machine, lovely build quality a sort of workstation rig, with a P4 2.4Ghz CPU. As I said, installing the 6800XT massively improved Doom 3's performance. My FPS with the Ti4800SE was in line with what other people were getting with a similar set up.
Weird. On my GF3Ti200 oveclovked to Ti500 i had about 30fps 800x600 on high setting and it was playble. Maybe some other locations were more demanding and needed to set power detail.
@@grzegorzjurczynski6897 No. It's been documented, verbosely, for the last 20 years.
I mean, there is a chance that you had some alien technology GF3 that was somehow 3x faster than ANY OTHER GF3 in the entire universe, and those thousands of detailed benchmarks from back in the day are wrong....
....
@@TheVanillatech well check on yt, your score was arround geforce 2 performance. Maybe you had weak cpu. I found on YT comparision between many geforces generation in doom3 ( from 256 to 7800). I had sempron palermo overcloked to 2.2GHz and this CPU was beast for that price.
Does that renderer use the hardware T&L in the GeForce cards?
Yes i think so 🙂
@@blastfromthepast3073 If that's the case the Voodoo5 is very impressive in this game.
@@armorgeddon yes of course. 😊 Voodoos scaling very well with fast CPUs. Nvidia did a great job with T&L for CPUs back in the Days. But it becomes a bit obsolet with raw CPU Power.
I believe so. That's why the Voodoo is faster. The fast CPU is actually faster than the GeForce 2's T&L engine.
@@alexandernorman5337 Yes that should be the reason, but it also shows that apart from the missing T&L feature the VSA-100 was very good. It would make for a very interesting written report or video at which CPU the crossover point is for the few games/engines of that time. I've never seen someone have a go at that.
No over clocking or straping a fan to the Voodoo card to make it run at its full potential ?
Voodoo 5 cards go for over 400 euros and overclocking a 20+ years old GPU is a fast way to kill it.
I´m sorry for deleting the first version of this video. But i was not happy with that.
Now i added a side by side comparision at the end. I really hope you like it.
You should definitely delete this one too. As others have pointed out, this comparison is simply flawed from the get-go. As much as I love the 3dfx cards, the GeForce 2 GTs is a faster, better card than the Voodoo 5 in... basically every way.
@@ruxandy It was not my intension to make somebody angry. I only benched different cards with the game the same render path and showed up that the voodoo (wich even doesn't have newer drivers) renders doom3 faster as a factory oced Geforce2.
So thats all. I never said V5 is technical better than Gf2.
Hi Citizen, your test is based with which cpu and ram ?
Hi! This was tested with Pentium4 3.46 and 512MB DDR333 if i remember correctly.
3dfx is only good for 90s games, after year 2000 I had geforce 2 gts ddr 64mb and that was smart choice
I'd prefer GF6x for 2000s games an ultimate performance solution. And TNT2 Ultra for pre-2001 game because it don't has ugly S3 texture compression like GeForce
what is better geforce 6100 IGP or geforce 2TI
This is a good question. Has the 6100 dedicated Memory?
@@blastfromthepast3073 system memory usulay ddr400
@@antyKNP hm i would say Geforce2 Ti should be faster. But 6100 has better visuals.
@@blastfromthepast3073 geforce 6100 has 4ROPS 4TMU clocked at 400MHz
@@antyKNP without Memorybandwidth there is no Fillrate :)
Doom3 against that cards is like trying to pull a trailer with a Smart. ^^
There would have been better bechmarks then a game released four years after the cards had been released. Max Payne (2001), Quake 3, 3D Mark 2000 or 2001, Unreal Tournament or Unreal for example.
The gts was very bandwidth limited as the memory was way too slow for the chip. The ultra was a far more balanced card and a vast improvement, even excluding the faster clocked core. The mx has lowest bandwidth and fillrate of all these cards so couldn't hope to compete.
I remember Geforce2 Ultra was sold for 1200DM here in Germany. Should be 600USD if i remember right, crazy! In future i will do a Geforce2 Video, maybe from MX200 to Ultra. It was a great GPU!
@@blastfromthepast3073 yes the Gef2Ultra went for insane prices. It was a great card, mainly I think because the gpu wasn't bottlenecked by the memory, or not much anyway. It was the same story with the original Geforce, the sdram seriously bottlenecked the gpu. Really, Nvidia just trying to make xmas $$. It was only when the DDR version was released early in 2000, that the real-life bandwidth met the theoretical bandwidth. Those who rushed to buy the sdr version must have been kicking themselves.
@@patrickct9386 Yep! Geforce SDR had the same Bandwith like V3 3000. For V3 it was okay with 333MTexel Fillrate. Geforce 256 had 480MTexel with same Bandwith and sucked! Geforce2 GTS had 1600MTexel Fillrate but near same Bandwith like Voodoo5. So GTS never could use the Full Fillrate, in real life maybe 700MTexel. Ok using a tiny Resolution and 16Bit increase Performance but who did that with a expensive Geforce2?
as 3dfx voodoo5 owner at release date, and UT99 pro player, i can confirm that v5 was fastest card in UT99, GF2 Hercules prophet was faster in general especially games supported T&L(one of fastest gf2 64mb ddr),
other stats; v5 was 30% cheaper (250$ vs 350$) in my country
v5 worked great with cpu-s available at time (had 900mhz athlon tb, while 1ghz was top of line p3/athlon)
gf2 was bit more bottle necked with those cpu-s
Doom3 is from 2004 and those cards were from 2000
Some of demanding games from their time: UT99,Quake3,NFS5 Porsche,Counter Strike,NOLF, Sacrifice(T&L),Midtown Madness, Giants(T&L),Deus ex, Driver, etc 99/00
3dfx fan. first two love, Banshee 16mb sg, v5 5500..than gf4ti
i guess any game that supported Glide API was faster on Voodoo
This game also depends on game version (there is some 1.0.xx , 1.3.1 exists) and used operating systems.. and when is benchmark run repeated there are different results.. Its not best benchmarking material.
I benched the scene twice for a stable result but you're right. There little differences but it depends on your System too. Using a healthy clean System with enough and fast Memory will help a lot.
We've been lied to
2021: Geforce2 GTS ~ 25 $. Voodoo 5 5500 ~ 400 $
Thats the sad truth.
Good, A better anti aliasing has his cost. lol
Geforce 2 cards were manufactured by many producers, with new cards still comming in spring 2001, while Voodoo 5 cards were made only by 3dfx/STB (There are some VSA-100 powered PowerColor cards, but they are Voodoo 4 and are quite rare) and all production was done by Christmas 2000. No suprised that there's a LOT more functional Geforce 2 cards.
Buying STB not only killed 3dfx, it also made retrocomputing with late 90's/early 2000 machines much more expensive.
Hercules 3D Prophet II MX - uses a weaker GeForce 2 MX 400 chip.
This is not the GeForce 2 GTS...
GTS comes at last
@@blastfromthepast3073 GeForce 2 GTS - Apr 2000,
Voodoo 5 5500 - Jun 2000,
GeForce 2 MX - Jun 2000,
GeForce 2 MX 400 - March 2001...
So Voodoo 5 can beat Geforce 2 GTS after all.
Too bad that it took over 2 decades to do so...
If runs on GTS so could run on a PS2 back in the day many people said only the xbox could run doom 3. Remember GTS has a high memory bandwidth that is why can still "run" doom 3 at least not at slow motion. So the PS2 memory badwitch is even much much higher than a GTS. But drawing shadows and light would have to be chaged completely using Vu0 like sillent Hil 2 did
Don´t forget that PS2 has much less System Memory! I remember shadow of the colossus wich gave PS2 a very hard time 😅 But it´s still impressive.
Doom 3 would crush the PS2. It would crush the Gamecube too. The game's engine - idTech4 - was made to utilize 8bit stencil buffers for all the real time lighting and shadowing. The PS2 and Gamecube didn't have those. You could fake a hardware stencil buffer to try to run the engine on them, but there is a lot of processing overhead involved. And that would cripple those systems. I doubt you would get more than 7-8fps on either of those consoles. And you would have pathetic texture quality on the PS2 as well.
@@alexandernorman5337 Not if you rewrite the shadow renderer like Silent Hill 2 and Swat: Global Strike did using PS2 vector units. Look the shadows on those games
@@hugo-garcia You only have those two vector multiplication units in the PS2, and they are not very fast. The base geometry processing of the game would tax them fully. You really don't have a free resource for the stencil volumes - very expensive calculations because you have to project off of other surfaces - and so you cannot obtain a playable framerate. You also have a problem with memory. If your graphics hardware has an eight-bit stencil buffer, then you can just properly set that and associate that with the affected vertices. Those eight bits tell the renderer how much shadowing it needs to include for the shadow with the associated vertices. If you don't have that, then you need to create new vertices - with a proper offset to prevent Z fighting - and that means duplicating all the associated properties such as texture coordinates. That will cost you a lot more memory. And the PS2 doesn't have as much as Xbox. There is simply no playable way to run idTech4 on the PS2 hardware.
Silent Hill 2 utilized a lot of blobs and other prebaked mapping operations to limit how often the game engine needed to do those projections. But every frame needs them - lots of them - with idTech4.
@@alexandernorman5337 ruclips.net/video/laWWmT5k4Cw/видео.html
Should have used Quake 3 . Not Doom 3 More suited to the hardware
We saw hundreds and hundreds of Q3 benchmark results. At least this is original and unusual.
Quake music on Doom video shameful.