The big issue for the K6-3+ at 600 is memory bandwidth.
Clocked at 5*120 or 4.5*133/135 with fast CL2 memory the K6-3+ is beating even a P2 450 in most tasks, even games.

Testing these things are not easy as just switching memory module with the same settings can change the performance alot.
Super Socket 7 platforms are very sensitive to hardware changes and to get the best performance alot of testing is needed.

candle_86 wrote:

actually the K6-III 450 smoked the Pentium II 450, a K6-III 400 as seen nearly caught up and beat a PII 450 lol […]
Show full quote

actually the K6-III 450 smoked the Pentium II 450, a K6-III 400 as seen nearly caught up and beat a PII 450 🤣

http://www.xbitlabs.com/articles/cpu/display/k6-iii.html

As long as the drivers and game supported 3dnow you where golden

go find something else thats 3dfx-free. 😠

In my tests, the K6-III-450 scored 20.4 in Quake II (software mode, not 3DNow! enhanced), while the PII-400 scored 23.4 fps (software mode). The resolution was 640x480. In my tests, 3DNow! demonstrated a 20% enhancement at 800x600 (GF2), and depended on the graphics card. The enhancement decreased to 10% for 1152x864. Some graphics cards, such as the ATI Rage 128VR, showed only a 4% increase at 800x600 when using 3DNow! ehanced Quake II. At best, the K6-III-450 equated to a PII-400 in games, but on average, was probably more like a PII-350.

Skyscraper wrote:

The big issue for the K6-3+ at 600 is memory bandwidth. Clocked at 5*120 or 4.5*133/135 with fast CL2 memory the K6-3+ is beati […]
Show full quote

The big issue for the K6-3+ at 600 is memory bandwidth.
Clocked at 5*120 or 4.5*133/135 with fast CL2 memory the K6-3+ is beating even a P2 450 in most tasks, even games.

Testing these things are not easy as just switching memory module with the same settings can change the performance alot.
Super Socket 7 platforms are very sensitive to hardware changes and to get the best performance alot of testing is needed.

I'd love to see some benchmarks! I had very few gains with my own K6-3+ at 5*120. 🙁

F2bnp wrote:

Skyscraper wrote:

The big issue for the K6-3+ at 600 is memory bandwidth. Clocked at 5*120 or 4.5*133/135 with fast CL2 memory the K6-3+ is beati […]
Show full quote

The big issue for the K6-3+ at 600 is memory bandwidth.
Clocked at 5*120 or 4.5*133/135 with fast CL2 memory the K6-3+ is beating even a P2 450 in most tasks, even games.

Testing these things are not easy as just switching memory module with the same settings can change the performance alot.
Super Socket 7 platforms are very sensitive to hardware changes and to get the best performance alot of testing is needed.

I'd love to see some benchmarks! I had very few gains with my own K6-3+ at 5*120. 🙁

I have posted some benchmarks before and I have reposted them like every time we have been on the subject.
I have not done any serious testing like feipoa and Im not going to, at least not in the near future.

I have got the best results with the K6-3+ using the Gigabyte GA-5AX and a single double sided 256MB module, with more memory or more than a single module the scores drop like stones even though I run with the L3 cache disabled. I think its the "write combining" that gets messed up.

If you enter "Gigabyte ga5-ax k6-3+ fsb site:vogons.org" in your Google search window you will find more information

I've seen these before, but they're synthetic benchmarks, why not bench some games with a 3dfx card? I think a lot of people here would find that helpful.

F2bnp wrote:

I've seen these before, but they're synthetic benchmarks, why not bench some games with a 3dfx card? I think a lot of people here would find that helpful.

The number of things to test for the benefit of glorious forums of Vogons are infinite while the amount of space and time availible to do so isnt.
It may seem like Im always here but thats just because Im always near a computer, not because I do not have a job or other things to do 😁

The next time Im playing with a K6-3+ in the Gigabyte board I will take more screenshots and write down more numbers whenever that might be...

Sure mate, whenever you feel like it. I didn't mean to give the impression that I don't really care about the benchmarks you've done already, it's just that I feel games are of much more significance, as it is largely why we build these systems and people have wondered for years how the K6 compares to the Pentiums. 😀

65 pages, 875 benchmark results of Voodoo. Lots of AMD chips as well.

However, I didn't tweak anything for AMD 3DNow!

http://www.philscomputerlab.com/voodoo-2-and- … ng-project.html

As far as I'm aware the only really beneficial 3DNow effort was Quake 2 and that was orchestrated by AMD to help sell K6-2. I'm not sure if other Quake 2-based games include the changes. id Software didn't officially support them.

Expendable also had good 3dnow support.