Pentium II MMX 233 is kind of a Slot CPU, not Socket 4-8!

btw, my i5-450M is ridiculously slow! Perhaps it's because of intel HD graphics? I wish I could turn HD5450 on in DOS.

RacoonRider wrote:

Pentium II MMX 233 is kind of a Slot CPU, not Socket 4-8!

Little mistakes can happen 😀

I will fix it.

Mau1wurf1977 wrote:

RacoonRider wrote:

Pentium II MMX 233 is kind of a Slot CPU, not Socket 4-8!

Little mistakes can happen 😀

I will fix it.

I did not mean to sound hostile, I understood that something was wrong only when I looked at the chip set.

Edit: Intel Pentium II 333 also got mislabeled:)

vetz wrote:

I have an Epox MVP3 board in addition to my Aopen AX59-PRO. I'll bench it today.

The preliminary results are in.

On MVP3 boards, do NOT run with AGP cards (except if you have the WTF board called DFI K6XV3+/66 B1 which breaks all logic compared to other MVP3, makes me question if the bench is correctly entered). The PCI counterparts give loads of extra performance.

Generally PCPBench and Quake is the only two benchmarks that doesn't have huge variances and which I trust the most. Look at the single 503 FPS score in 3DBench on a 400mhz stock setting. The average result is 310 for that speed. Same goes for Doom. Here we have the previously mentioned 120FPS score for the DFI board and 153 & 136 for two ALi V boards with the average around 90-105. Only the Quake and PCPbench scores does not have huge spikes in the results for SS7.

vetz wrote:

On MVP3 boards, do NOT run with AGP cards [...]

what made you conclude that? I am very curious because on my tests AGP or PCI riva 128/tnt/tnt2 cards score almost the same 😕 😕 😕
my HOT-591P with an AGP Riva128ZX scores pcpbench/doom/quake 49.2 / 90.42 / 39.2 with a p200mmx

keropi wrote:

vetz wrote:

On MVP3 boards, do NOT run with AGP cards [...]

what made you conclude that? I am very curious because on my tests AGP or PCI riva 128/tnt/tnt2 cards score almost the same 😕 😕 😕

My two MVP3 boards behave the same when it comes to the difference between AGP and PCI.

with what cpu? can you please write here your doom/quake fps? The bench page is a mess to read :\
At this point the results need some grouping options mainly by cpu and chipset IMHO

I have never run a PCI card on any of my SS7 boards. Why would you?

A proper database would be nice Keropi, but Google docs doesn't have one and it also requires a lot more preparation because you would have to enter all possible choices for users to select in order to have a proper database with integrity. Someone calling the CPU a Pentium 3 and someone else a Pentium III would be unacceptable for a database. Now this is something that can be tackled down the track, no doubt. All of this is way to early to worry about. Entering as much data is priority number 1. Anyone can download it and analyse it using Excel if you find Google docs to limiting. At the moment we do not have enough raw data to draw any real conclusions.

Something I haven't done is compare motherboards against each other using the same amount of RAM and graphics card. I do know that the DFI is slower with Caches turned off compared to others I have used in the past.

It can be seen that Doom can sometimes be "slow", while PCPlayer and Quake can be normal compared to the other systems. At least with AMD Socket A/754 that I tested today.

Duron 1200 on Abit KT7A has only 105FPS in DOOM, while PCPlayer has 287 and Quake 180.
Sempron 2600+ (@1600MHz, s754): DOOM=128, PCPlayer=341, Quake=235
For instance Intel P-MMX @300MHz has: DOOM=130, PCPlayer=80, Quake=65

😲

Also... In 486 chart, the comment about the system, the video card data is missing - it shows DOOM FPS instead.

kixs wrote:

Also... In 486 chart, the comment about the system, the video card data is missing - it shows DOOM FPS instead.

Fixed! Thank you for spotting this 😀

PS: Just saw that you used a FX-5200. That card is, for whatever reason, a little bit slower than a TNT2 or MX440. Still plenty of grunt though and, like you mentioned elsewhere, can do VGA and DVI at the same time, so for me it's a keeper.

Because of Vetz's statement regarding MVP3 pci/agp stuff, I did some more testing today:

Basically the only direct comparison is the R128 AGP and R128 PCI columns, both have the same chip obviously and the same rating 4MB vram.

I also benched some more cards (I did not bench more of the nvidia family because their vga core is pretty much the same , there was no point for faster GF cards. Sadly I couldn't get a GF2MX/PCI to boot 😢 and I did not bench any s3 cards prior to the VirgeDX of the Stealth2000pro, they would behave more or less the same or slower IMHO) and saw no surprises , things seem normal and nothing really tells me that agp support is flawed/slow 😕

Hey nice benchmarking 😀

I

Also did some benchmarking:

Yes, it's another video but it has funky retro MIDI music 🤣

Phil's ISA VGA Roundup - 7 cards - 1 winner

The ET4000 lacks the 640x480 pcplayer bench score.

elianda wrote:

The ET4000 lacks the 640x480 pcplayer bench score.

It has a zero, like others, because it couldn't run the benchmark.

Of course it can, just load the VESA TSR from the installation disks.
ftp://78.46.141.148/driver/TSENG/ET4000/dos/TLIVESA/
As a general rule of thumb: If the VGA card has more than 256 kB of memory there usually exists at least a TSR that enables VESA 1.x support, if the bios does not.
There is also one for the Trident cards, which is VESA.EXE.

I added the highest clocked stock K6-2 system and some links to images of the boards/cards as comment at the cells.

Of course you can load UNIVE and other, which by the way didn't work with most of these cards and I tried all three versions.

See the chart as "out of the box experience" and some cards support 640 x 480 straight out the box. If you build a system now you want to choose the best card and having to load software to do what other cards can do out of the box might be important. But it's a good point so I added a caption to the video explaining this.

Now I could have plastered a big "doesn't work" sticker on these cards, but a discrete 0 score seemed more appropriate 😀

Very impressed with the CL card. It's cheap, fast and very compatible.

Univbe requires VESA 1.x to be already present. That is why it didn't worked.

(Deleted. No longer relevant anyway)