Isnt't this just a GA7IXE with Socket A instead of Slot A?
I had/have a GA7IXE running for years without probs. Maybe it is some kind of compatibility problem in combination with other hardware.

The socket A board can accept a 1.4Ghz T-Bird @ 100 FSB. (But these higher speed CPUs are rare and expensive.) Perhaps the slot A board only goes up to 1Ghz?

I tried to get 2 of these boards working today. As mentioned, no POST. I then transfered all of the testing components (RAM, gfx, CPU) to a third working board, and it POSTed and worked OK.

RG100, I want to see photos of your old hardware collection. Have you had to rent out a storage garage yet? 😁

Since the multiplier is coded inside the Slot-A package, the board might go up higher, though the fastest available Slot-A Athlon is 1 GHz.

Space is a major problem. I'm getting closer and closer to "full capacity"! I've had to do the following -

* As can be seen in the "distressing" photos, I've squashed my cards side-by-side in to boxes. 🙁

* Put all discs in to CD wallets, and discard all the plastic cases. Any accompanying printed material goes in to a box. 🙁

* Squash every game box so that they become flat "2D" rather than a proper box shape. 🙁

* Whinge whinge... 🤣

oh dear. I see some extra space in those photos!!!! 😀

What's in the gray boxes? Antistatic bags 🤣

I swear I can't get much more in, because this room is the attic. The problem is not space, but weight! Those gray boxes at the back: I can barely slide them across the floor. The floor is just basic plywood edit: it's called chipboard, not plywood. Already, there's 2 areas I can't walk on because the floor kind of "snaps", and if I go through it, I'll end up in the room below! 🤣

Edit: You see that space in the top left area. I put down some thin wooden sheets (for want of the correct terminology), and put some of my boxes on them. But I noticed that after a couple of weeks, the atmosphere in the attic (and the boxes too) became "clammy". I realised that putting stuff there was blocking the air flow through the eaves. So, unfortunately that space must remain empty.

No ESD protection? 😮

At least it looks organized 😀

When I want to dig out a card "I know I have somewhere" I needs days to find it, hehe thamkfully it will clear up after I've moved (2 weeks left and counting)

Amigaz wrote:

No ESD protection? 😮

All of my anti-static bags are in one of those gray boxes. 😉

Amigaz wrote:

At least it looks organized 😀

I'm looking forward to doing "round 2" of testing, sometime next year. Round 2 involves making sure I've downloaded all the right drivers and manuals for this stuff, and just generally "sorting things out".

Amigaz wrote:

When I want to dig out a card "I know I have somewhere" I needs days to find it, hehe thamkfully it will clear up after I've moved (2 weeks left and counting)

Good luck! Ah, your very own "retro lab". Nice! Wish that was me! 😁

@retro games
In my experience, LOTS MANY mobos do not work well if the little coin battery is low on power. It should be around 3v.
If lower than 2v, you might not get even a POST.

I've just checked 2 of these "dead" boards that I have inside my "stuff to take to the recycling centre" box, and it looks like I have used fresh batteries.

Regarding this Gigabyte GA-7IXE4 mobo (from the O.P.), I can't get the AGP to run in 2x mode. The mobo spec says: "AGP slot supports 2X mode", which can be seen here -

"http://www.gigabyte.com.tw/Products/Motherboa … ctName=GA-7IXE4

I've tried an nVidia MX 440 AGP (universal 8x) card, and also an ATI Radeon 7500. Both run at AGP 1x on this board. I tried uninstalling the graphics drivers, then using the basic VGA mode only, reinstalling the AGP mini port driver from the AMD chipset driver package. Then, I reinstalled the graphics driver. For the ATI Radeon 7500, there's something in the graphics driver package called "SmartGart", which tests your AGP speed and stability. I set it to 2x (from its current setting of 1x), then rebooted. When the desktop appeared, it looked a tiny bit "unstable" with a few bits of odd flickering. I went to look at SmartGart's settings, and saw that the setting had been reset to 1x. I guess that SmartGart thought that 2x was unstable, and decided not to go with that speed. That's disappointing.

I'm currently using 1 stick of pc-133 SDRAM, 256mb. The BIOS is the latest F9 version, and its settings are "optimised".

AMD750 Chipset tends to have higher noise on AGP than other chipsets which tends to make some combinations with graphics cards unstable. In a typical scenario it will freeze at some point with increased AGP traffic.
So Ati and NVidia included a drop down to AGP 1x in their drivers.

Try this (in this order):
Install Chipset driver
Install Graphics card driver (will default to AGP 1x)
Install a third party tool to set AGP to 2x again, typical PowerStrip or RivaTuner
with PowerStrip you can change this live.

Run some tests with 3D Mark f.e. if it works reliable or not.
I had a GF2 GTS from Elsa these days running and it froze once a week (with 6+ h of 3d gaming each day)
If you think it is ok, then configure powerstrip to set AGP 2x on startup.

On the other hand, benchmark between AGP 1x and AGP 2x setting, there might be alot less difference than you would expect.

OK! 😀 I downloaded PowerStrip version 3.87, and also 3DMark2001 (not the Second Edition version). Time for some benchies! I'm using a Palomino 1500 rated CPU, which gives a real clock speed of 1300, and runs at 100FSB (200 DDR). Its CPU code is AXL1500DLT3B. The gfx card is an ATI Radeon 7500; DirectX 7. RAM is 1 stick of 256mb PC-133.

AGP x2 mode score = 4031
AGP x1 mode score = 4019 - a decrease of (I think) a fifth of 1 percent

I also DL'd 3dMark2000 (version 1.1). Here are its scores -

AGP x2 mode score = 6833
AGP x1 mode score = 6826 - a very tiny decrease

Sandra 2002 Pro scores (with AGP mode set to x2)

dry = 3598
wet = 1794
multi-m (i) = 7151
multi-m (f) = 8416

I then OC'd the system, by setting some dip-switches on the mobo. I increased its speed from the default setting of 100mhz bus speed, to 110mhz bus speed. (I did try the mobo's maximum OC setting of 115mhz, but 3dMark2000 went mad, so I abandoned that setting.) I then reran the 3dMark2000 (version 1.1) speed tests:

AGP x1 mode score = 7070

Sandra 2002 Pro scores (with AGP mode set to x1)

dry = 3922
wet = 1957
multi-m (i) = 7797
multi-m (f) = 9164

So, the difference between AGP x1 and AGP x2 is incredibly small. Why is that?!

Edit: With the mobo's OC setting left on 110mhz, I shut windows 98 down to DOS mode, then ran Quake 1.06 shareware. I put the game in 640x480 mode, then ran timedemo demo1. I get 57.7 fps. Also, I DL'd SiSoft Sandra 2001 standard freeware version, then ran its memory speed test and got these scores -

Int ALU = 455
Float FPU = 487

Using SiSoft's Sandra 2002 Pro version, I ran a similar memory benchmark test, and it gave me the following larger scores. I don't know why these numbers are larger than the 2001 version.

RAM int = 805
RAM float = 796

Make absolutely sure to install the latest BIOS version and latest motherboard chipset drivers for AGP motherboards. Many of them have terrible performance and/or compatibility until you install motherboard-level drivers.

Well, the values fit with the performance I remember. I think the latest AMD driver pack wont change this, if it hasn't been used yet.
The Irongate has also no Sideband and Fast Write support, so there is nothing to gain there.

It is still one of the early AGP boards, the chipset was made for the Slot Athlons, like 500 Mhz.
You put a Palomino on it and a GFX-card that is far more powerful than a TNT2 or GF1.
This somehow reminds me of putting the adapter tower to put a Tualatin on a PPro board 😉.

Well at least it shows how this chipset scales from 500 Mhz to 1300 MHz.

retro games 100 wrote:

So, the difference between AGP x1 and AGP x2 is incredibly small. Why is that?!

From this result you can infer that the extra bandwidth afforded by the 2x spec isn't needed. Not surprising considering the card and processor used.

Some food for thought:

Even at 1x, AGP is still running at 66mhz, instead of 33hmz as with the PCI bus. This means the theoretical bandwidth is 266mbps, compared to 133mbps.

Another design characteristic is that AGP and PCI are separated, so again theoretically, saturating one bus shouldn't effect the other. The connection of both buses to the processor and ram is much faster than the two combined.

Even with that being the case, the volume of information generated and transferred back and forth isn't enough to max it out.

I was hoping that this AMD mobo would beat a Powerleap Intel based system, but it seems that the Powerleap 1.4Ghz upgrade really is a screamer -

http://www.roberthancock.com/dell/plip3t/powerleap-v2.htm (PL-iP3/T v2.0 & Celeron 1.4 GHz)