Here is one example:
http://www.ebay.com/itm/5PCS-W24257AK-10-Manu … =item1e8cf98c2d

Look for: W24257AK, AS7C256. The number (-10, -15, -20, -25) after the marking is the speed. You need at least 20ns, but 15 is usually best.

Also check in the manual what size should the TAG chip be. Usually can be the same as the others but sometimes it needs to be smaller capacity.

Thanks for the reply. Unfortunately I'm working with limited info since it's a Dell motherboard and no manual exists, or at least nothing I've been able to locate. Thanks for the information 😀

You may find markings printed on the motherboard itself that are searchable product codes on the net.
Also go to:
http://stason.org/TULARC/pc/motherboards/
and see if you can match the motherboard layout with a known product - they have Dell oem boards btw. If you find a match you'll get the vital cache jumper settings.

The cache seems to be of the 28-pin variety. The TAG chip probably goes in the 9th socket [upper left hand corner in your pic], and populate the rest with 8 more chips - probably of the same type as others have noted. The chips have a 'direction' denoted by a symbol at one end of the IC which needs to be lined in the slot in the correct orientation.

If you don't have a chip extractor tool already it's a good idea to get one. That way if you find you need to add/remove chips as you experiment on getting the cache in correctly you don't mangle the delicate pins - they look like this:
http://www.ebay.com/itm/New-Circuit-Board-IC- … =item566a9f3670

I did what you're doing for the first time last year - getting the jumper setting & IC spec was the trickiest part. I sourced the chips on eBay and found 9x 32kb 15n for around 12-15 USD shipped. 256k cache on a dx2/66 made things butter smooth, and even 128 would probably be just fine.

Good hunting...

Appreciate all the info, thanks

This layout looks like it might be the one?
http://stason.org/TULARC/pc/motherboards/D/DE … XPS-XX-MDT.html

Cache is overrated esp for a dx33

JaNoZ wrote:

Cache is overrated esp for a dx33

Disagree. Without cache is will drag just about anything short of 286 down to a fraction of its nominal speed

I have the feeling it doesnt, it will never have cache misses and always read instantly from main ram which gives a better overal throughput for slower cpu's. No additional latency due to not having to look in cache.
I have used a 486sx25 for many years without cache simple isa mainboard and it was always faster than my uncle,s 386dx33 which had also no cache. And that one should have at least outperformed any high speed 286.

I dont say it is bad thing to do, but the amount to spend on the sram to get for a dx33 seems a bit of a waste for me.
If we were talking dx2/66's now that is a different story, but still would not major improve the overal speed.
Cache was very expensive back in the day, and most systems without it performed just very good.

A clock increase has a much bigger impact on system performance.

My 486/33 benchs between half and one third normal when the cache is turned off. I would say that is a pretty big deal

In-game experience definitely can benefit from cache especially with flight & driving games. For the 'butter smooth" response it's well worth it - essential even.

Cache makes a huge difference. 😀

Yes yes yes, cache is good.

I did some extensive testing with a sis471 based VLB board lately and found out that particular board behaved much faster without cache chips installed than with cache chips installed and just disabled L2 in bios.
I don't know why, or it might be a fluke.
cachechk and speedsys are also reporting more ram speed without cache enabled or it is just me, but quake and doom ran only a few fps lower without cache on a dx2/66 so i just must assume the performance increase would be neglible.
I always thought the cache would only do a 5-10% increase, and around 5% for WT mode.
I have some desktops 486 from dell and other brand without being able to install a cache and there is 100mhz dx4 inside. why would they have skipped the cache option anyway? puzzling to me.

I woule like to know what the outcome of testing between cache removed and just disabled and enabled would be compared to a dx2/66. would be a nice read, i have some on paper but have to look for them might someone be interested.

Or i am just getting mad, maybe that is an option too.