I created a Monster called "Peak of bottleneck" :

Valid : LINK
It's kinda funny how much a old CPU can b'neck a modern GPU 😀

Q : How Celeron 2,0GHz (Northwood), stacks against Pentium III class CPUs ?

agent_x007 wrote:

Q : How Celeron 2,0GHz (Northwood), stacks against Pentium III class CPUs ?

You have truly created a monster! That Celeron is probably slower than a Pentium 4 1.5GHz, so you do the math. Perhaps it's a little bit faster than a Pentium III 1000?

You have truly removed the GPU as a variable in testing your CPU's performance. I'm sure many tech-YouTubers would be proud. 🤣

That is stupidly stupid. So stupid it's funny. 😀

looking4awayout wrote:

A stupid thing I did many years ago was killing a working AT super socket 7 motherboard by putting both SDRAM and EDO modules with the hope to go beyond 256MB of RAM. After doing that the motherboard would no longer go past the Verifying DMI Pool Data screen. I weep about that moment still today... 😒

My shitty PC-Chips AT Socket7 board has two 72 pin and two SDRAM modules for a total of 256MB RAM. The thing is, this configration is official (i.e. described in the mobo manual), so maybe it is not a stupid thing to do, and your problem with the motherboard is probably due to something else.

PC Chips TXPro M560
ALI Aladdin IV+ ALI 1531 (TX Pro)
64MB x 2 PC100 SDRAM
64MB x 2 72 pin SIMM

nekurahoka wrote:

So, I was working on updating microcode to an Abit Is7-V2, pulled the flash chip to flash it using my USB flasher and broke the retaining socket on the motherboard. A whole corner came free and now when the chip is inserted it pushes the leads out far enough that there is no contact. Gonna have to rig something together or just wait until get the next mobo I've been eyeing, which already has the latest microcode.

I just wonder why people do that. 🤣 I always think that EPROM programmers are a last resort thing, and I should try to use first the mobo facilities and flash software to do anything with flashing BIOSes if they are available. Am I somehow wrong?

tayyare wrote:

I just wonder why people do that. 🤣 I always think that EPROM programmers are a last resort thing, and I should try to use first the mobo facilities and flash software to do anything with flashing BIOSes if they are available. Am I somehow wrong?

Nope, you're not wrong in most cases. In this instance I was in the middle of a lot of trial and error and had done several flashes. It was just easier to pull the eeprom and flash. I don't use floppy drives in any of my other systems, so we'd be talking about burning a boot CD each time. In adding the microcode the way I was at the time, I was also trying to bypass the boot blocker. The microcode was successfully added and working, but the BIOS's boot blocker was stepping in because of the hack.

This is a fun topic and I've been building/modding PCs since my family's first Packard Bell 486SX. So I've seen and done a lot of things...

In my 233MMX I replaced the primary slave with a CF->IDE adapter and installed a couple of OSs on CF cards. DOS/Win3.11 works great and works with the 18GB FAT32 slave disk. The OS/2 Warp4, though, works but didn't recognize the FAT32 disk despite having the driver installed. I poked around in the volume manager and created a volume on the 18GB disk. Still no joy. Back in DOS...Aargh! The 18GB disk was wiped! Lucky I had a copy of the contents on my main rig so I could pull the disk, hook it to the main via USB enclosure, and restore the contents.

Another time, I moved my Win98 disk into a different machine, from a 430VX mobo to a SiS768. It booted fine but finding drivers for the various mobo components was tough, and I never found working drivers for the built-in audio. On top of that, the supposed driver for the IDE controller crashed Win98 so it was stuck with a generic driver in compatibility mode. Copying the 400MB WINDOWS folder from C: to a backup on D: (secondary slave disk) in compatibility mode takes 20-30 minutes! Oh, the hours and days wasted trying to get DMA working... One day, while sifting through my closet I came across Biostar M7VIG mobo. I checked the driver CD and found full Win98 drivers. In two hours I had the rig completely rebuilt with the Biostar and all of the drivers installed, and full DMA. Why didn't I do that sooner...

One of my stupidest, though, has to be when I rebuilt a Sempron rig about 10 years ago. I put everything together, closed the case, pressed the power button and was assaulted by a scream from the speaker. I switched off the PSU as fast as I could but it was too late. The CPU was fried because I forgot the thermal paste.

F2bnp wrote:

agent_x007 wrote:

Q : How Celeron 2,0GHz (Northwood), stacks against Pentium III class CPUs ?

You have truly created a monster! That Celeron is probably slower than a Pentium 4 1.5GHz, so you do the math. Perhaps it's a little bit faster than a Pentium III 1000?

I doubt it'd be any slower than a 1.5GHz Pentium 4, I once benchmarked both a 1.6 GHz (Willamette) Pentium 4 and a 2 GHz (Northwood) Celeron in my Dell Dimension 4300S and the Celeron outperformed the Pentium 4. Then again, I only benchmarked 16-bit CPU and x87 FPU performance, so it may depend on how cache-dependent the benchmark utilities are since the Celeron has half the cache of a Willamette Pentium 4.

I'll have to check my NSSI benchmarks photos to see which Pentium III is close to my 2 GHz Celeron.

Still, that GTX 1080 benchmark on a 2 GHz (Northwood) Celeron made my day.

Those Commanche 4 benchmarks still haunt me.

Celeron 2.0 is certainly slower than a Willamette P4 1.6 in games 😀.

I was surprised by Celeron's performance in Sky Diver's graphics tests :
They were NOT a "slide show" 😁

Valid : LINK

Here's how a Pentium 4 HT looks like in this test :

Valid : LINK

3,41GHz (OC from 3,06GHz), so 70% gain is from clock speed, rest is from cache...
So 2.43x speed up (overall) in that 2x performance in Physics test, and 2.75x increase in GPU performance 😁

I think GPU Usage meter gets confused by 2D/3D clocks and/or Turbo Boost technology...

Thank you all for info about Celeron vs. Pentium III(-S).

Another one today - I was trying out a new board, fitted it into the case, turned it on and the dreaded puff of magic smoke came along with a horrendous smell and the computer shut off...

I had repaired the CPU fan earlier in the day when the cables detached but didn’t bother to use a light... turns out I somehow managed to bridge the positive and negative pads on the fan and burnt out the fan control transistor on the main board... so none of the fans worked!

Cleaned all the burnt scuzz off the board and just bridged the collector and emitter pads for now but the fan doesn’t shut off in sleep mode so I will be replacing it at some point!

Hopefully, this topic is not dead yet. 😀

Today, I accidently found out that my PSU in my P3 system prevents the motherboard from POSTing, and yet another PSU does allow it to POST, so now I'm waiting for a new one to show up.

Any who, that's not the stupid thing, really. While looking at the innards of my P3 PC, I noticed the jumpers on the motherboard, and then I thought, "Wait, I just upgraded to a P3-800E 800MHz CPU from a P3-600E 600MHz one a few weeks ago. Isn't there something with the jumpers that needs to be changed?". Turns out, I've had the jumpers set to the old multiplier of 6x instead of 8x (which is what the CPU needs to be ran at).

Now, combine that fact with the fact that I ran benchmarks on it to update the topic dedicated to that P3 build, and I think you realize why it's stupid. Now I have to change the jumpers and re-run the benchmarks. Yipee. 😀

tayyare wrote on 2018-05-09, 08:35:

I just wonder why people do that. 🤣 I always think that EPROM programmers are a last resort thing, and I should try to use first the mobo facilities and flash software to do anything with flashing BIOSes if they are available. Am I somehow wrong?

It depends on the BIOS ...
A BIOS with protective tape over a window: these must be removed, erased with UV light, and programmed. This is an EPROM (Erase-able Programmable Read Only Memory).
A BIOS with no window might be a ROM, a PROM, or an EEPROM:
If it's a ROM, you have to program a new one and replace the existing one.
I recently encountered a PROM - these are write-once, so you have to remove it to replace it.

What you are talking about is an EEPROM - electronically erase-able. You can probably reprogram these with software, but the board has to support it. I couldn't say, put a pin compatible EEPROM in the place of a PROM in a 386DX board and suddenly be able to program it.

Last year, I had to put new glue on the front panel of one of my computers because the glue that held the crack on it dried up. To make things safer, I removed the front panel from the computer before applying glue to it. I made note of how the front panel connected to the motherboard before removing it. Everything went smoothly, except for when I was testing the computer to make sure it still worked, I used one of my LCD monitors without realizing that the computer was set to a refresh rate that only my CRT monitors would cooperate with (85Hz), which resulted in the "Cannot display this video mode" message. 🤦

Because of that, I now keep my retro computers at the 75Hz refresh rate, which has less flicker than the standard 60Hz rate while still maintaining compatibility with most LCD monitors, just in case I have to connect an LCD to one of my computers for testing purposes.

I bought a couple Original XBOX controller breakout cables with the intention of making a USB adapter cable out of one of them by following the wiring diagram in this link. I already had the XBCD drivers installed on the machine I was testing it with (my Athlon 64 rig) and I wired the USB plug accordingly to the colors in the diagram, so I thought it'd be good to go. I plug it in and Windows XP detected the controller as an "Unknown Device", I was confused as to why it didn't detect it until I started to smell smoke, so I immediately unplugged the computer and inspected its USB headers.

Thankfully, there was no damage done on the computer. I've tested every USB port on it and they all worked. I tested the cable with a multimeter in continuity mode and found out that the colors on the wires in my cable didn't match the diagram at all. To add insult to injury, one of the wires for the USB protocol on the cable turned out to be the yellow wire, which is normally used for lightgun sync. I cut the yellow wire off on the USB connector end when I was doing the cable because I was confident that the wires would match, so now I'm going to have to strip more insulation off the cable and rewire it accordingly.

A PSA to everyone making cables: Check your pinouts with a multimeter before testing, because there's always a chance that the wire colors DO NOT match the ones in your pinout diagram.

EDIT: I opened up the controller I used to test the cable and found that a small surface mount capacitor has blown, I think it was because of how my adapter cable was wired. Thankfully I have other XBOX controllers.

This is why I usually draw schematics before I get to the nitty gritty. Glad your PC didn't lose its USB portion.

I did two stupid things with some old PC hardware. Both actually worked, but I don't recommend trying them.

1. I had a Packard Bell Legend 286-12 PC. The I/O controller on the motherboard failed. It handled floppy and IDE hard drives (and maybe serial and parallel. I can't recall). I tried putting an expansion card in, but it conflicted with the chip on the motherboard. My solution was to physically rip the chip off of the motherboard. This actually worked and I was able to use an expansion card to replace the functionality of the chip.

2. I had a PC chips 486 motherboard that needed a BIOS upgrade. Unfortunately, the BIOS chip on the motherboard was not flashable. I had a compatible chip from another motherboard that was, but no way to flash it. I decided to try to hot swap the BIOS chip after booting the machine. This worked and I was able to flash the BIOS to the new chip.