on my slot 1 board I also have to change a jumper when using FSB66 for the AGP speed, if I leave it at default (for fsb100) not sure if it's going to fail to run or just run very slowly!?

Gahhhrrrlic wrote:

I didn't realize the ram would run at 66. I thought there'd be some 2x multiplier or something at play.

693A can run memory asynch, but only FSB+33, so you can run it at 100MHz. Won't do you much good, as it makes bad latencies worse and the bottleneck remains the 66MHz FSB,

The 4 in 1 drivers appear to be for 98 and below only. I am doing my tests on XP (because it is more stable for my sound drivers) so can I even install the 4 in 1 on XP?

Depends on the version. During the XP era, Via renamed 4-in-1 to Hyperion, so you need either. There's huge discussions on which is best, but anything that supports your OS is better than nothing. Are you now saying you're running XP with no chipset drivers? That will nerf performance for sure.

SPBHM wrote:

on my slot 1 board I also have to change a jumper when using FSB66 for the AGP speed, if I leave it at default (for fsb100) not sure if it's going to fail to run or just run very slowly!?

That's the AGP divider selector. i440BX can run at 1:1 or 2:3. If you have 66MHz FSB, you want 1:1, if you have 100MHz FSB, you need 2:3 so AGP stays at 66MHz. There's no 1:2 divider, so if you overclock FSB to 133MHz (doable on most BX boards), AGP runs at 88MHz. Not great, but not disastrous as usually AGP cards can handle high bus clocks. Note that the actual card/GPU clock is independent of the bus clock, so you don't automatically OC that.

BX also has a PCI divider, which can be 1:2, 1:3 or 1:4, although not all boards implement the last one. That makes a big difference - most AGP cards accept 88MHz AGP, but very few PCI cards work well over 40MHz PCI, and 133/3 = 44MHz 😮

Any recommendations for a good cheap P3 at 133 FSB that's as close as possible to 700 MHz? I see there's 1 at 733MHz that should work on this board. If I had to buy a new one I might consider something like this.

Via 693A is a bad performer regardless, but that matters least with as high an FSB as possible with a relatively low multiplier. That P3-733EB would do fine. It would perform only marginally worse than a P3-700E would on a BX board (such as that P2B).

66MHz fsb holding back unreal tournament ? That's nonsense ! I've been playing ut99 on a pentium 2 266MHz at 1024*768 with a voodoo 3 3000 with acceptable framerate. I mean sure it wasn't 60fps, but it was also far from 15 (except when 5 person shoot at you and everything is exploding around you for sure)

I got UT when it was on sale on GOG not too long ago. I only played it on my main system with 2010-ish specs, but after reading through this thread I decided to give it a try on an older setup to see if it gives any insight into your performance woes. Here is my old P3 system:

440BX, Pentium 3 600e
192MB PC100
Windows 2000 SP3 or SP4 (I forget which)
GeForce FX 5200 64-bit

Bit of a swap fest in the beginning with only 192MB, but after the game got going it was fine at 800x600, 32bit, high, high. I'm guessing around 40fps on average, dropping somewhat when there was a lot of action.

It's true that Celerons are crippled by slow FSB but if you're only getting 5 fps I'd say you have worse problems than just that. How does it do with other games? Maybe you could post some benchmark results to shed some light on this. Maybe 3Dmark '01? And in particular, what do you get from the fill rate and high polygon count tests? How about SuperPi?

Actually it's kind of crap with most games. Anything Unreal related blows. HL1 is ok but does have some noticeable stutter around corners and such, Lithtech games like blood 2 are slow, haven't tried NOLF yet. Quake 2 is fine but... it's quake 2. Quake 3 was surprisingly not totally fluid either. RTCW was even slower. Dark engine games like Thief, Thief 2 and System Shock 2 are also choppy. I'd say everything developed after 1997 is not smooth.

I'd say if I tinkered with each game to reduce CPU load to an absolute minimum, most of these games are "playable" but not fun to play.

Gahhhrrrlic wrote:

I should have mentioned I installed the via_hyperionpro_v524a.zip drivers from phil's lab

it doesnt work
you still didnt say what graphic card you are using

anyway its not the celeron 700 at fault, its either no drivers/bad drivers/bad gpu, or some crap running in the background

Yeah because I remember playing half life on a i440lx motherboard with a Mendocino celeron running at 466MHz and it was more or less the same as playing on my pentium II 450

rasz_pl wrote:

you still didnt say what graphic card you are using

Radeon 9500 was mentioned earlier in the thread.

bakemono wrote:

rasz_pl wrote:

you still didnt say what graphic card you are using

Radeon 9500 was mentioned earlier in the thread.

sorry, missed that
so yeah, you need to fix crap via agp driver and/or gpu driver (install older etc)
most reliable fix is always switching to 440bx board

rasz_pl wrote:

sorry, missed that so yeah, you need to fix crap via agp driver and/or gpu driver (install older etc) most reliable fix is alway […]
Show full quote

bakemono wrote:

rasz_pl wrote:

you still didnt say what graphic card you are using

Radeon 9500 was mentioned earlier in the thread.

sorry, missed that
so yeah, you need to fix crap via agp driver and/or gpu driver (install older etc)
most reliable fix is always switching to 440bx board

I'm using the modified Omega drivers that turn the 9500 into a 9700. Surely those can't be bad.

And the hyperion drivers had an AGP driver in them, which did install, because now my video settings tabs all have some yellow via logo instead of the red ati logo (that and the fact that a message window said that agp drivers had been installed).

I agree that the CPU really should not be too slow for UT99 but I can't find any evidence to suggest that it's anything else. I just ran my P400 with V2 SLI and it's nice and fast with UT99. It has even less RAM (128MB) and of course the V2 only has 8MB available for textures, not 64 like the 9500 so unless Glide is 10x faster than DX, I dunno what else to blame here.

Gahhhrrrlic wrote:

I'm using the modified Omega drivers that turn the 9500 into a 9700. Surely those can't be bad.
And the hyperion drivers had an AGP driver in them, which did install

doesnt matter, 5 fps means they are bad/broken

Gahhhrrrlic wrote:

I agree that the CPU really should not be too slow for UT99 but I can't find any evidence to suggest that it's anything else. I just ran my P400 with V2 SLI and it's nice and fast with UT99. It has even less RAM (128MB) and of course the V2 only has 8MB available for textures, not 64 like the 9500 so unless Glide is 10x faster than DX, I dunno what else to blame here.

Any chance you could try the Celeron in the BX board? Or don't you have a slotket for Coppermine?

dionb wrote:

Via 693A is a bad performer regardless, but that matters least with as high an FSB as possible with a relatively low multiplier. That P3-733EB would do fine. It would perform only marginally worse than a P3-700E would on a BX board (such as that P2B).

Is the 693A really that much slower, that even with 33% higher FSB and 33 MHz extra on the cpu, the BX will still win? Wow

H3nrik V! wrote:

Gahhhrrrlic wrote:

I agree that the CPU really should not be too slow for UT99 but I can't find any evidence to suggest that it's anything else. I just ran my P400 with V2 SLI and it's nice and fast with UT99. It has even less RAM (128MB) and of course the V2 only has 8MB available for textures, not 64 like the 9500 so unless Glide is 10x faster than DX, I dunno what else to blame here.

Any chance you could try the Celeron in the BX board? Or don't you have a slotket for Coppermine?

I don't have the hardware for that conversion unfortunately.

Gahhhrrrlic wrote:

H3nrik V! wrote:

Gahhhrrrlic wrote:

I agree that the CPU really should not be too slow for UT99 but I can't find any evidence to suggest that it's anything else. I just ran my P400 with V2 SLI and it's nice and fast with UT99. It has even less RAM (128MB) and of course the V2 only has 8MB available for textures, not 64 like the 9500 so unless Glide is 10x faster than DX, I dunno what else to blame here.

Any chance you could try the Celeron in the BX board? Or don't you have a slotket for Coppermine?

I don't have the hardware for that conversion unfortunately.

What a shame - it would've taken all unknowns out of the equation ..

H3nrik V! wrote:

dionb wrote:

Via 693A is a bad performer regardless, but that matters least with as high an FSB as possible with a relatively low multiplier. That P3-733EB would do fine. It would perform only marginally worse than a P3-700E would on a BX board (such as that P2B).

Is the 693A really that much slower, that even with 33% higher FSB and 33 MHz extra on the cpu, the BX will still win? Wow

only if you average fps over 24h and count all the bluescrens/freezes against it 😀

rasz_pl wrote:

H3nrik V! wrote:

dionb wrote:

Via 693A is a bad performer regardless, but that matters least with as high an FSB as possible with a relatively low multiplier. That P3-733EB would do fine. It would perform only marginally worse than a P3-700E would on a BX board (such as that P2B).

Is the 693A really that much slower, that even with 33% higher FSB and 33 MHz extra on the cpu, the BX will still win? Wow

only if you average fps over 24h and count all the bluescrens/freezes against it 😀

🤣 Actually, one of my greatest arguments of using Intel processors, was the ability to use an Intel chip set .. 😀

H3nrik V! wrote:

dionb wrote:

Via 693A is a bad performer regardless, but that matters least with as high an FSB as possible with a relatively low multiplier. That P3-733EB would do fine. It would perform only marginally worse than a P3-700E would on a BX board (such as that P2B).

Is the 693A really that much slower, that even with 33% higher FSB and 33 MHz extra on the cpu, the BX will still win? Wow

Dionb's assessment is actually on the conservative side. Here some outcomes (taken over time) from 3DMark99 cpu score:
3DMark99 cpu score // mobo // cpu // chipset

9514 // Chaintech 6ATA2 // 850/100 MHz FCPGA // VIA Apollo Pro 133 (693A)
9579 // Chaintech 6ATA2 // 850/100 MHz FCPGA // VIA Apollo Pro 133 (693A)
9640 // Chaintech 6ATA2 // 850/100 MHz FCPGA // VIA Apollo Pro 133 (693A)
9646 // Chaintech 6ATA2 // 850/100 MHz FCPGA // VIA Apollo Pro 133 (693A)
9696 // Chaintech 6ATA2 // 850/100 MHz FCPGA // VIA Apollo Pro 133 (693A)
9706 // Chaintech 6ATA2 // 850/100 MHz FCPGA // VIA Apollo Pro 133 (693A)

12471 // ASUS P2B-S // 850/100 MHz FCPGA // 440BX
12538 // ASUS P2B-S // 850/100 MHz FCPGA // 440BX
12667 // ASUS P2B-S // 850/100 MHz FCPGA // 440BX
12690 // ASUS P2B-S // 850/100 MHz FCPGA // 440BX
12598 // ASUS P3B-F // 850/100 MHz FCPGA // 440BX
12681 // ASUS P3B-F // 850/100 MHz FCPGA // 440BX
12736 // ASUS P3B-F // 850/100 MHz FCPGA // 440BX
12758 // ASUS P3B-F // 850/100 MHz FCPGA // 440BX

12330 // ASUS CUV4X // 850/100 MHz FCPGA // VIA Apollo Pro 133A (694)
12345 // ASUS CUV4X // 850/100 MHz FCPGA // VIA Apollo Pro 133A (694)
12381 // ASUS CUV4X // 850/100 MHz FCPGA // VIA Apollo Pro 133A (694)
12382 // ASUS CUV4X // 850/100 MHz FCPGA // VIA Apollo Pro 133A (694)
12383 // ASUS CUV4X // 850/100 MHz FCPGA // VIA Apollo Pro 133A (694)
12464 // ASUS CUV4X // 850/100 MHz FCPGA // VIA Apollo Pro 133A (694)
12524 // ASUS CUV4X // 850/100 MHz FCPGA // VIA Apollo Pro 133A (694)

Another 693A board that I have here (Jetway 993AN) shows a similar performance gap.