Out of personal experience, Celeron 700 with 66MHz FSB performs about as well as, if not worse than, a Katmai PIII 500. It's pretty slow.

The Serpent Rider wrote on 2022-02-16, 17:29:

Are there any BX boards with UDMA66 and 3 ISA (or more) ISA slots?

No.

I have a ITOX CB60-BX-C which has 3 ISA slots and i think supports UDMA66

I have a ITOX CB60-BX-C which has 3 ISA slots and i think supports UDMA66

Doesn't look like 440BX and it's probably i815. Unless people from DFI were so lazy that this motherboard have universal AGP slot on unsupported chipset. In that case - still UDMA33.

bloodem wrote on 2022-02-18, 06:04:

Well, it's definitely something worth exploring (and I do have a 600 MHz Celeron SL3W8, which I've never tested - I hope it works 😁 ). My early estimate is that it should be slower than the Ezra-T, because of the 66 MHz FSB (which becomes a severe bottleneck).
The Celeron 600 would definitely be slower than the Pentium 3 "Katmai" 600 (which I will try and test today, before I leave for a short one week vacation). Even if the Celeron has 128 KB of on-die L2 cache, its 66 MHz FSB is most likely enough to keep it well below the Katmai.

A couple of years ago I benchmarked the Pentium 3 600 family:
ASUS P3V4X // PCMark 2002
oooooooooooooooooooo
Celeron 600Mhz 128Kbcache fsb66 on slotket:
CPU score: 1470
oooooooooooooooooooo
PIII 600 fsb100 Katmai:
CPU score: 1527
oooooooooooooooooooo
PIII 600B fsb133 Katmai:
CPU score: 1546
oooooooooooooooooooo
PIII 600E fsb100 Coppermine:
CPU score: 1599
oooooooooooooooooooo
PIII 600EB fsb133 Coppermine:
CPU score: 1623
oooooooooooooooooooo

The Serpent Rider wrote on 2022-02-18, 10:43:

I have a ITOX CB60-BX-C which has 3 ISA slots and i think supports UDMA66

Doesn't look like 440BX and it's probably i815. Unless people from DFI were so lazy that this motherboard have universal AGP slot on unsupported chipset. In that case - still UDMA33.

According to the manual it is 440BX.
Also it doesn't have an universal slot on most of the pictures I found.
But seems to be UDMA33.

Also it doesn't have an universal slot on most of the pictures I found.

Re: Socket 370 with BX440 Motherboards?

The Serpent Rider wrote on 2022-02-18, 13:30:

Also it doesn't have an universal slot on most of the pictures I found.

Re: Socket 370 with BX440 Motherboards?

I said MOST. Not ALL.

Which is irrelevant in this specific case. Earlier revisions/models were indeed 100% 440BX and UDMA33.

NEW VIDEO (Intel Pentium 3 "Katmai" 600 MHz): https://www.youtube.com/watch?v=2UkFnM_rYqE

TEST SYSTEM REMINDER:
MB: Gigabyte GA-6BXC rev 1.9 (VRM mod)
CPU: Intel Pentium 3 "Katmai" 600 MHz
RAM: 3 x 128 MB SAMSUNG SDRAM PC133
VIDEO: Asus V7700Ti GeForce 2 Ti (OC @ GeForce 2 Ultra clocks)
SOUND: Creative Sound Blaster Live 5.1 SB0220
HDD: Seagate 40 GB IDE/PATA

The Pentium 3 "Katmai" CPUs can usually be overclocked to 672 MHz (FSB 112). Some will go even higher. The best Katmai 600 that I have works fine at 700 MHz/FSB117 without increasing the voltage (on motherboards that support this pretty unusual FSB frequency).

Having said that, I've decided to test this CPU at its default clock speed because:
1. It's not a particularly interesting CPU in terms of it's capabilities and overclocking headroom.
2. The main goal/idea is to use the P3 600 as a "speed reference", and comparing it with other CPUs that are more interesting (like the VIA C3, that I've tested in the previous videos).

CONCLUSION: when the VIA C3 Ezra-T is overclocked, it trades blows with the Pentium 3 Katmai 600. Sometimes it wins, sometimes it loses, and sometimes they're within 1 - 2 FPS of each other. All in all, I'd call it a draw.
The Pentium 3 is quite a bit faster in Quake 2, but that's where Pentium II/III CPUs always shined. I should mention that, with the 3DNow patch, the Ezra-T comes very close of the Pentium (it scores almost 150 FPS in Quake 2). Unfortunately, I forgot to show this in the Ezra-T benchmark video.
As for the VIA Nehemiah... well, that CPU is in a completely different league when it comes to performance (it's usually between 30 - 50% faster than the Pentium 3 Katmai / VIA Ezra-T).

I'll be on vacation for the following 10 days. When I return, I will switch to SS7 for a bit, and benchmark the AMD K6-2/3+ on an Asus P5A, just to finally answer the question if it's indeed faster than the VIA C3 (Spoiler alert: it's not 🤣 ).

Oh, yeah, and a few pics with the Pentium 3 Katmai (this particular CPU has a big passive heatsink that I'm cooling with a very quiet 120mm Arctic fan):

Le PC pr0n! Merci monsieur! 🤣
Enjoy your break and rest well!!

Seems like the right moment to show some figures by a dude who knows what he is doing with a P5A to make sure OP doesn't sandbag. Note however that these tests were only intended to be self consistent, for the purpose of comparison between CPUs, cache sizes and bus speeds, so the only result that is 100% transferable is the 3DMark99 CPU score which is given in the 3DM99b column. While there are some numbers for a GF3 under the main table, we are not sure if this was done on the true default setting, or the method used in this thread which exploits the fault of 3DM99 giving higher scores if only the game tests are run or tests run individually.

https://drive.google.com/file/d/12_JgRPXbcj8j … 8-i7KYLC3w/view

So I have 2 c3 cpus that should be both be Nehemiah, but i'm not convinced since they seem the same physically but they have different cpu id:
https://imgur.com/a/RxtyAnC

I tested them today with the QDI Advance 10T, both at 1200 mhz (apparently the 800 mhz didn't complain), I used setmul to enable every kind of cache, brench prediction etc:

I also tested them with fastdoom and with quake, both had the very same performance.

I think the last figure of the CPUID is only going to be the stepping difference, not entire model/process difference. Nehemiahs are on the whole less awful than Ezras though.

BitWrangler wrote on 2022-02-18, 17:20:

Seems like the right moment to show some figures by a dude who knows what he is doing with a P5A to make sure OP doesn't sandbag.

Not sure why I would try to sandbag anyone, it's not like I'm getting paid by VIA to promote their 2002/2003 CPUs and badmouth the K6-2/3+. 🤣
As for the 3DMark 99 CPU scores, from my experience, they are completely off.
I mean, the P3 Katmai @ 600 MHz scores 9100 points (as seen in the video that I linked above), and, according to those charts (and even my own previous benchmarks), the overclocked K6-3+ can score as high as 9400 points. Just laughable to think that any K6-2 could ever achieve (or exceed!) the gaming performance of a 600 MHz Katmai.
I would say that the 3DMark2000 CPU score is more representative for the actual FPU and real world gaming performance.

Nah, that's the one that everyone complained about because PIIIs and Athlons getting everywhere from half to double on the same CPUs between motherboards/builds, resulting in virtually nobody listing results from it and complete withdrawal of the score in 3DMark2001.

But yeah, lets just move the goalposts closer together now you've taken your penalties.

BitWrangler wrote on 2022-02-18, 18:18:

Nah, that's the one that everyone complained about because PIIIs and Athlons getting everywhere from half to double on the same CPUs between motherboards/builds, resulting in virtually nobody listing results from it and complete withdrawal of the score in 3DMark2001.

But yeah, lets just move the goalposts closer together now you've taken your penalties.

🥴 Anyway... moving on.

It's a good idea to have some baseline results. In the persuit of accurate results in benchmark based on fairness I would expect to see both cpus overclocked. As VIA and AMD might have been a good value. A poor mans pentium like a overclocked celeron might have been a good value too when compared to a overclocked amd k6 or via. They say what's good for the goose is good for the gander

I don't see any value in using default resolutions in 3dmark when you are using a overpowered video card. Anything less than 1024x768 doesn't even stress the graphics cards. Id be in favor of running all tests at something like 1600x1200 to really stress the whole system then compare scores. It's only under stress sometimes that cpus and graphics cards really exhibit their flaws.

Warlord wrote on 2022-02-18, 20:30:

It's a good idea to have some baseline results. In the persuit of accurate results in benchmark based on fairness I would expect to see both cpus overclocked. As VIA and AMD might have been a good value. A poor mans pentium like a overclocked celeron might have been a good value too when compared to a overclocked amd k6 or via. They say what's good for the goose is good for the gander

For sure! I will start with the K6-2+ @ 616 MHz, a frequency which is usually achievable by many of these chips on decent motherboards.
I also want to compare the performance on VIA vs ALI based boards.
The K6-2 was my second CPU, on the first PC I ever built myself (back when I was 16), so it will always have a very special place in my heart. 😀

bloodem wrote on 2022-02-18, 18:13:

Not sure why I would try to sandbag anyone, it's not like I'm getting paid by VIA to promote their 2002/2003 CPUs and badmouth the K6-2/3+. 🤣

You're getting paid by the Chinese to sell their e-waste.
Admit it! 🤣

RaiderOfLostVoodoo wrote on 2022-02-18, 21:31:

bloodem wrote on 2022-02-18, 18:13:

Not sure why I would try to sandbag anyone, it's not like I'm getting paid by VIA to promote their 2002/2003 CPUs and badmouth the K6-2/3+. 🤣

You're getting paid by the Chinese to sell their e-waste.
Admit it! 🤣

That's a bit harsh .. Via chips are not E-waste .. yet, they have to spend some time as paper weights and key chains first.