Video card makes no difference in GP2.

I have been keeping up with this newsgroup for sometime now, but this is my first post. I always thought that upgrading my video card from a trident 9440 (worst) to a Herc Dynamite 128 (best) would make a significant improvement in the performance of GP2. I trusted the SIM RACING NEWS graphical comparison of video cards. I mailordered my Herc Dynamite 128 on Monday. Meanwhile, I heard a few posts from people, especially Majin, claiming that the video card makes no difference and I thought they were crazy. But they were right! This was all done on a P100 with 24mB ram. On my trident 9440, the video score was 70; with 3 textures on and low detail in SVGA the estimated frame rate was 14.2. Stating races like this the processor occupancy was around 120 -130 %. With my Herc Dynamite 128, the video score was two hundred something, and the estimated frame rate with this detail was 17.0 fps. However in the game, both video cards performed equally. When I started racing the processor occupances was still 120-130% with the Herc D 128. And throughout the race the occupancies and noticable slowdowns were nearly identical. I even replayed a crash that consistently ended with a processor occupancy of 233% with the trident. I was urked to see it still at 233% with the Herc D 128! I heard that the gP2 vid score is only for VGA, but no increase in SVGA, this is crazy! John Wallace, I love your sim racing news, but where did you get the information for that chart?

Oh, I forgot to say that I used the same frame rate of 12.0 frames per second for both of the vidoe cards even though the estimated frame rates were 14.2 (trident) and 17.0 (herc D 128).

Getting a new video card is not going to reduce processor occupancy. Setup up your game so you only see 100%+ in extreme circumstances. What you SHOULD see is better performance of a real world frame rate when the occupancy is 100% or under. You didn't say what kinda of CPU you had either or if your motherboard has PB cahce. There is where your bottle neck may be and nothing you do will help. But I think tweaking it so it's usually under 100% occupancy will help...good luck -- Adam M

Quoted text (13 lines)

> ... Herc Dynamite 128 (best) would >make a significant improvement in the performance of GP2. >...few posts from people, especially Majin, claiming that the video >card makes no difference and I thought they were crazy. But they >were right! >This was all done on a P100 with 24mB ram. >On my trident 9440, the video score was 70; with 3 textures on and >low detail in SVGA the estimated frame rate was 14.2. Stating races >like this the processor occupancy was around 120 -130 %. >With my Herc Dynamite 128, the video score was two hundred >something, and the estimated frame rate with this detail was 17.0 >fps. >However in the game, both video cards performed equally.

When set to 12fps >When I started racing the processor occupances was still 120-130% ... I don't know if your Herc is really the best with only 200+ video speed. My Daimond 64 2001 2Mb scores ~330 and posts here rate Matrox 300 just wont cut it. Still, very discouraging post. BTW I'm running a P5100 HX, 32M EDO 256KPB above Vid card. detail level. It's much like GP1 with better driving model, sound and car detail. Also running FROM Win95 (double click on there icon), haven't had any problems at all. Austin, Texas

I wouldnt say no difference, but if you are expecting to gain more than a couple of frames youre going to be dissapointed.

Quoted text (9 lines)

> This was all done on a P100 with 24mB ram. > On my trident 9440, the video score was 70; with 3 textures on and > low detail in SVGA the estimated frame rate was 14.2. Stating races > like this the processor occupancy was around 120 -130 %. > With my Herc Dynamite 128, the video score was two hundred > something, and the estimated frame rate with this detail was 17.0 > fps.

Hi I am grateful for this information. I have been on the edge of upgrading my video card for some time now and getting GP2 seemd like just the right incentive! In the last week or so I had seen the odd posting that made me feel that it might not be worth it for GP2 and you seem to confirm that. For most of my other games I have enough CPU/horse power to overcome the old video card so I might as well wait for the GP2 patch that Microprose are shortly going to bring out (vbg!!!!). BTW I think the video score figures for John Wallaces graph came from using the CBench test program which he was kind enough to post on the binaries group. I think the figures probably reflect the kinds of improvements that would be gained in "normal" games? Al

My Pentium 100 with 32mb Ram Video score of 314 I think. Matrox Millinium 2MB WRam.

Quoted text (24 lines)

>> This was all done on a P100 with 24mB ram. >> On my trident 9440, the video score was 70; with 3 textures on and >> low detail in SVGA the estimated frame rate was 14.2. Stating races >> like this the processor occupancy was around 120 -130 %. >> With my Herc Dynamite 128, the video score was two hundred >> something, and the estimated frame rate with this detail was 17.0 >> fps. >Hi >I am grateful for this information. I have been on the edge of >upgrading my video card for some time now and getting GP2 seemd like >just the right incentive! In the last week or so I had seen the odd >posting that made me feel that it might not be worth it for GP2 and >you seem to confirm that. For most of my other games I have enough >CPU/horse power to overcome the old video card so I might as well >wait for the GP2 patch that Microprose are shortly going to bring out >(vbg!!!!). >BTW I think the video score figures for John Wallaces graph came from >using the CBench test program which he was kind enough to post on the >binaries group. I think the figures probably reflect the kinds of >improvements that would be gained in "normal" games? >Al

<Lots of "benchmarks" deleted> The principal reason that the video card seems to little difference to performance is the way that GP2 accesses SVGA. It is rather erm, antiquated....It uses VESA 1.2 to setup video modes and access video memory. VESA 1.2 is a standard for accessing SVGA cards from the days when SVGA cards were an extension of VGA cards and only had a 64K byte window onto the 1,2,4,8 meg of video RAM. Nowadays with Local bus or PCI cards the whole range of video memory is available to applications through the new(ish) VESA 2.0. VESA 2.0 is used by games like Duke Nukem(yukk), Quake(mearly as good as GP2)...only old/conservative games use VESA 1.2..mainly because most gfx cards support it whereas VESA 2.0 is still a bit on the new side. hope this is of some help, Dave Fiddes

Quoted text (16 lines)

>>The principal reason that the video card seems to little difference to >>performance is the way that GP2 accesses SVGA. It is rather erm, >>antiquated....It uses VESA 1.2 to setup video modes and access video >>memory. VESA 1.2 is a standard for accessing SVGA cards from the days when >>SVGA cards were an extension of VGA cards and only had a 64K byte window >>onto the 1,2,4,8 meg of video RAM. Nowadays with Local bus or PCI cards the >>whole range of video memory is available to applications through the >>new(ish) VESA 2.0. >>VESA 2.0 is used by games like Duke Nukem(yukk), Quake(mearly as good as >>GP2)...only old/conservative games use VESA 1.2..mainly because most gfx >>cards support it whereas VESA 2.0 is still a bit on the new side. >However, the video card doesn't seem to affect Quake either (according >to the Quake frame rate page). ICR2 is reported to be faster on >Hercules and Matrox card, and I believe ICR2 is VBE1.2. >--KCI

I believe that Icr2 is VBE 2.0 and the support is linked directly to the executable. Run the Uvconfig that comes with the game, looks alot like the one that comes with Univbe 5.1.

Quoted text (14 lines)

>>>However, the video card doesn't seem to affect Quake either (according >>>to the Quake frame rate page). ICR2 is reported to be faster on >>>Hercules and Matrox card, and I believe ICR2 is VBE1.2. >>>--KCI >>I believe that Icr2 is VBE 2.0 and the support is linked directly to >>the executable. Run the Uvconfig that comes with the game, looks alot >>like the one that comes with Univbe 5.1. >I don't have ICR2 on my HD anymore. You maybe right but the shareware >version of UNIVBE only supports VBE1.2 after the trial period is >expired. The Need For Speed comes with UNIVBE and it's 1.2. >By the way, GP2 does get performance increase from loading UNIVBE. >Weird, but true. >--KCI

That may explain the moderate performance boost that I got after installing the Lightspeed. I wasnt using Univbe before.

VBE1.2 after the trial period is expired. << Where there's a will... PF