VOGONS


Reply 20 of 35, by Davros

User metadata
Rank l33t
Rank
l33t

looking at the 1024 2xaa you get 0.2fps for 12% increase in cpu speed
so a 3.4ghz cpu (if it scaled the same) would give you an extra 1.6fps

Guardian of the Sacred Five Terabyte's of Gaming Goodness

Reply 21 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t

I was going to add the following comments as an "edit" to my post above, but I wanted to attach another image, and so an "edit" would have made the post above look confusing.

(Edit) I managed to successfully increase the FSB overclock value to 155, and so I got a real CPU clock speed of 2326 mhz. I then reran two "no AA" tests again, which can be seen in the yellow highlighted right hand column labelled "2326 mhz". It looks like there's still a couple of frames per second in the 800x600 test to be gained from this small FSB adjustment, but the gain is less apparent in the higher 1024x768 resolution test.

Attachments

  • unreal2.png
    Filename
    unreal2.png
    File size
    2.64 KiB
    Views
    1803 views
    File license
    Fair use/fair dealing exception

Reply 22 of 35, by sliderider

User metadata
Rank l33t++
Rank
l33t++

Going back to your Voodoo3 card, how much memory does it have? If it's only 8mb then it might be a V3 1000. Since it's a Gateway model, though, it could also be an STB Velocity 100, which was released as an OEM only board at the same time that the V3 3500 was released. The STB Velocity is clocked the same as a V3 2000, but has one TMU disabled and has half the memory on board which might account for the slower than expected performance. You would think they'd have learned their lesson about crippling cards from Rush and Banshee, but I guess that's one of the reasons they're not around anymore.

Reply 23 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t

I'm fairly confident it's a 16MB Voodoo3 3000 model. Here's a photo of the actual item. The sticker on its BIOS says "G-WAY 2.15.07". Regarding the slower than expected performance, perhaps I need to reinstall the card and driver, and rerun those tests using the up to date and fully patched Unreal games DVD?

Attachments

  • v3.jpg
    Filename
    v3.jpg
    File size
    33.72 KiB
    Views
    1786 views
    File license
    Fair use/fair dealing exception

Reply 24 of 35, by swaaye

User metadata
Rank l33t++
Rank
l33t++

That heatsink usually indicates a 3000. The 2000 models have a smaller heatsink and the 3500 has TV.

The Voodoo3 1000 / Velocity looks quite different and typically has no heatsink (insanity).

Reply 25 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t

Problem solved. A few days ago, I was experiencing two problems with testing the V5. 1) Toggling between no AA and 4x AA did not show any visual improvement, and 2) FRAPS would not display its FPS score on the screen. I've just discovered why this is. Inside the V5 control panel, I had the V-Sync option set to Disabled. I did this for some benchmarking purposes. However, it needs to be set to Enabled, if the FSAA and FRAPS are to work correctly.

Reply 26 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t

I've managed to make my V3 card go nuts! I decided to uninstall the V5 card, and reinstall the V3. I wasn't sure what driver to use, so I went to the Italian 3dfxzone.it website, and downloaded a file called euro-1.04.02.exe. I think it was labelled as the last V3 reference driver, but as it was installing, the messages displayed on screen were "this is the 3500 TV driver". However, it installed and works OK. Anyway, I ran utbench.dem, using my mobile barton at its lowest setting of 133x6. No problems. I then increased the CPU's multi to its maximum (15x), and the utbench.dem locks up. I tried running the demo again, and I get a really weird "blue screen of death" complaining about not being able to write to drive C! I lower the CPU's multi to 12.5x, and rerun the demo. No problems. I'm sure I've read somewhere on the falconfly.de website that some old V1 and V2 cards don't like being run at very high CPU speeds. I reckon that I may have discovered the limit to the AGP V3 card: don't try them at 2 Ghz.

BTW, for 16-bit Glide @ 1280x1024, I get 26.9 FPS using utbench.dem in Unreal Tournament (Anthology/GOTY edition). Interestingly, that's roughly the same score I get with the CPU running much lower at about 800 Mhz. This means that running the V3 with a CPU speed of ~ 1.5 - 2 Ghz is totally unnecessary.

I use the Voodoo tweaking utility called V_Ctrl, and increase the V3's clock from 166 to 180. I rerun the UT utbench.dem benchmark, and get a score of 29.3 FPS. That's about 2.5 frames faster.

Reply 27 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t

I installed the game Unreal (Return to Na Pali) from the Unreal Anthology DVD, and ran the "timedemo 1" castle fly over benchmark. This is using the V3 card. I waited until the FPS values section displayed "Cycle #2". The FPS results are -

800x600 = 75.6
1024x768 = 54.1

I OC'd the V3 card from 166 to 190, and reran the 1024x768 test. I get 62.0, which isn't too shabby! I then pushed my luck, and OC'd the V3 to 200 mhz. Wow, it doesn't go mad. I rerun the 1024x768 test and get 65.6. I then OC the V3 to 210 mhz, and the whole system goes 😖

Reply 28 of 35, by sliderider

User metadata
Rank l33t++
Rank
l33t++
retro games 100 wrote:

I've managed to make my V3 card go nuts! I decided to uninstall the V5 card, and reinstall the V3. I wasn't sure what driver to use, so I went to the Italian 3dfxzone.it website, and downloaded a file called euro-1.04.02.exe. I think it was labelled as the last V3 reference driver, but as it was installing, the messages displayed on screen were "this is the 3500 TV driver". However, it installed and works OK. Anyway, I ran utbench.dem, using my mobile barton at its lowest setting of 133x6. No problems. I then increased the CPU's multi to its maximum (15x), and the utbench.dem locks up. I tried running the demo again, and I get a really weird "blue screen of death" complaining about not being able to write to drive C! I lower the CPU's multi to 12.5x, and rerun the demo. No problems. I'm sure I've read somewhere on the falconfly.de website that some old V1 and V2 cards don't like being run at very high CPU speeds. I reckon that I may have discovered the limit to the AGP V3 card: don't try them at 2 Ghz.

BTW, for 16-bit Glide @ 1280x1024, I get 26.9 FPS using utbench.dem in Unreal Tournament (Anthology/GOTY edition). Interestingly, that's roughly the same score I get with the CPU running much lower at about 800 Mhz. This means that running the V3 with a CPU speed of ~ 1.5 - 2 Ghz is totally unnecessary.

I use the Voodoo tweaking utility called V_Ctrl, and increase the V3's clock from 166 to 180. I rerun the UT utbench.dem benchmark, and get a score of 29.3 FPS. That's about 2.5 frames faster.

What version of Windows are you running? I vaguely recall that Win 9x,ME freak out on CPU's faster than 2ghz, so it might be an OS issue and not a video card issue.

Reply 29 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t
sliderider wrote:

What version of Windows are you running? I vaguely recall that Win 9x,ME freak out on CPU's faster than 2ghz, so it might be an OS issue and not a video card issue.

I am using Win98SE, but that issue you recall is related to installing Win9x on a system faster than 2 Ghz. Apparently, sometimes you see a show stopping error pop up as you are installing the OS. But actually using win9x with a CPU faster than 2 Ghz isn't a problem.

Perhaps my system is a bit unstable at 2 Ghz? I uninstalled the V3 card, and installed a GF3 Ti500. Everything was fine with it, but when I increased the CPU's multi to 15x (giving a real clock speed of 2 Ghz), the system rebooted. It could be a power problem. The PSU is very good, but it is used and might be showing some signs of age.

Reply 30 of 35, by ux-3

User metadata
Rank Oldbie
Rank
Oldbie
retro games 100 wrote:

I decided to uninstall the V5 card, and reinstall the V3.

As a hint: I am using the latest original drivers for both cards concurrently. I can swap cards and win98se comes out with the right card. No more installations when changing V3 to V5 or vice versa. I like that.

Reply 31 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t

Correction: I was wrong about thinking the V3 could not tolerate a system running at 2 Ghz. This is not true. However, I was right about something else - it was a PSU problem. I replaced the PSU with another one, and all is well. Using the first "bad" PSU, I noticed that inside the BIOS set up area, the 5V reading was 4.66. This is too low, IMO. Using the second "good" PSU, I noticed that the 5V reading was 4.93. That's better! I then installed a Voodoo3 2000 model card, just for fun. I set the mobo's FSB to 158 (from 133), and set the mobile barton's multi to 15. I now have a real clock speed of 2373 mhz. The V3 card works fine. I'm not joking, but I decide to "play with fire" and crank up the V3 2000 model's clock from 143 to 185 and run Unreal Tournament utbench.dem. It worked! I got 31.2 FPS using Glide mode @ 1280x1024. (Before OC'ing the V3, I got 24.2.) I then hastily turn off the system before it catches fire!

@ux-3, thanks for the hint. I decided to install the driver called "3dfx 1.07.00 WHQL" from the 3dfxzone.it website. I think this is the final original driver release.

Edit: I suspect that my Enermax "fat 5V amp rail" PSU may be failing, in the 5V rail department. Is there a way of fixing this potentially failing aspect of the unit?

Reply 32 of 35, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie
retro games 100 wrote:

Edit: I suspect that my Enermax "fat 5V amp rail" PSU may be failing, in the 5V rail department. Is there a way of fixing this potentially failing aspect of the unit?

That's not really a failure, most PSU's behave like that when connected to a board like the EP-8KTA3 Pro. These boards are capable of running 70+ watt CPUs, but they draw all the power for the CPU from the +5V rail, unlike later systems, which use an extra 12 V power connector. Plus, the voltage conversion isn't very efficient. That's why you need at least 30 amps on the +5V rail to get a stable system.

There actually was a popular mod for the Enermax PSUs to cope with the "sagging 5V rail" problem: It is possible to adjust the output voltage by replacing a resistor in the PSU with a potentiometer. I've done this with one of my Enermax PSUs, it really helps with these boards.

Reply 33 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t
5u3 wrote:

There actually was a popular mod for the Enermax PSUs to cope with the "sagging 5V rail" problem: It is possible to adjust the output voltage by replacing a resistor in the PSU with a potentiometer. I've done this with one of my Enermax PSUs, it really helps with these boards.

Excellent. Thanks. That's another job I need to add to my list of jobs to do. 😀 BTW, both PSUs mentioned in my post above have 40A's on their 5V rails. I guess the Enermax PSU suffers from the "sagging 5V rail" problem more than its replacement.

BTW, how's your Epox board? (Also, the same question goes to Vogons user Bestemor.) I was thinking of getting an xp-m 3000+, but they are expensive, 5V power hungry, and probably run rather hot. Also, I wonder how much they can be OC'd by? I am wondering if the barton core has a "speed ceiling" of around 2.4 Ghz? Just a guess. I OC'd my 2 Ghz 2600+ chip to 2.403 Ghz, but didn't leave the machine switched on for very long, fearing a fire might start!

Reply 34 of 35, by retro games 100

User metadata
Rank l33t
Rank
l33t

I got a PCI Voodoo5 from ebay. Please see photo. I've just tested it. It works, but I've got a ringing sound in my ears. Imagine the noisiest hair dryer imaginable. I tried overclocking the GPUs using the VSA100 OC'er. That's the utility that OC's both chips. I set the value from 166 to 183. It worked. I ran this V5 on a humble slot 1 600MHz system. I tried Unreal. Before OC'ing I get 50.4 FPS on the timedemo fly over scene; 1280x1024. After OC'ing, I get 54.something.

Attachments

  • v5.jpg
    Filename
    v5.jpg
    File size
    40.56 KiB
    Views
    1486 views
    File license
    Fair use/fair dealing exception

Reply 35 of 35, by WolverineDK

User metadata
Rank Oldbie
Rank
Oldbie

retro games 100: if the fans are the problem (the hair dryer sound/tinnitus you got), could it be solved by getting a new pair of fans, even though they are not the original fans ? since I had a problem with my former Gforce 8800 GT fan, which a friend of mine and his father removed and replaced with a fully working fan. So I do not have that noise in from my pc anymore, and the best part of the whole "surgery/operation" on the card is this. The fan is going even faster and does not make any kind of noise, in comparison to the old fan.