derSammler wrote:

They actually had WB cache before "normal" CPUs had it.

I wouldn't take Wikipedia's word for it. Not sure who added that little nugget about the WB cache in there, but I am pretty sure that is incorrect. The POD5V63 and 83 which plugged into Socket3 on the other hand DID support write back cache...but those are Pentiums not 486s. If you can find any Intel documentation showing a 486 overdrive chip supporting WB cache I'd love to see it.

Also, I noticed there are actually three different DX2 overdrives.

ODP for SX pinout
ODP for DX pinout
ODPR for DX pinout

I guess the reason ODPR for SX pinout doesn't exist is because many OEM SX systems had the CPU soldered to the board.
As far as I know all the DX4 overdrive pinouts are of the DX type.

appiah4 wrote:

mpe wrote:

There are different versions (steppings) of the DX2 and the overdrive. That’s what creates the difference.

The only obvious difference (apart from the integrated heatsink) between any DX2 and DX2ODPR is absence of the boundary scan feature on the overdrive (apart from certain sample chips).

This is interesting, I never knew this. Are there any real world implications of this ommission?

Boundary scan is used during systems testing. The 486 chip has a number of testability pins and these don't work on production overdrive chips

https://en.wikipedia.org/wiki/Boundary_scan

DX2-66 replacement overdrives (DX2ODPR) have been produced in at least 4 different versions. The regular DX2-66 possibly in even more. Some of these might or might not have SL enhanced mode, WB L1 cache, CPUID instruction, different CMPXCHG opcode, etc. It is hard to compare DX2ODPR overdrives with regular DX2 without saying which exact versions you compare.

derSammler wrote:

From: https://en.wikipedia.org/wiki/Intel_80486_OverDrive […]
Show full quote

Anonymous Coward wrote:

None of the Overdrive chips have write back cache, even as an option.

From: https://en.wikipedia.org/wiki/Intel_80486_OverDrive

Intel's i486 OverDrive processors are a category of various Intel 80486s that were produced with the designated purpose of being used to upgrade personal computers. The OverDrives typically possessed qualities different from 'standard' i486s with the same speed steppings. Those included built-in voltage regulators, different pin-outs, write-back cache instead of write-through cache, built-in heatsinks, and fanless operation — features that made them more able to work where an ordinary edition of a particular model would not.

They actually had WB cache before "normal" CPUs had it.

That article on Wikipedia is a bit misleading. Going to fix it 😀

For whoever is still interested in this topic...

I just got from Ebay an Intel 486 DX266 overdrive (DX2ODPR66) to replace my Intel DX266 (SX911) and the results are exactly the same, unless I'm missing something that CHKCPU.exe is not detecting... (I don't think so).
Landmark benchmark was the same too.
I had WB set in the BIOS but it still shows as WT as you can see in the image below.
I was tempted to remove the heat sink to see what it says but didn't pull the trigger...

Aside from the multiplier, is there actually any other difference between a DX2ODPR66 and a DX4ODPR100?

Has anyone tried overclocking their DX2ODPR66 by setting a 3x multiplier? (Or is the multiplier locked?)

Intel DX4 ODP/ODPR100 have 16KB L1 cache

I don't have the ODP/ODPR66's but the multiplier is labelled undefined in the manual for the 3x position. Low = 2x (default). High = 3x (DX4 CPUs). Not to say that it definitely won't work on all CPUs.

Does the DX2ODPR66 also have 16KB L1 cache or was that only reserved for the DX4 Overdrive?

The Overdrive name seems to be somewhat misleading becaue one would assume it is an enhanced performance version of a regular DX2 or DX4 processor. (Even the DX4ODP which wasn't a "replacement" CPU .....If I understood the earlier comments correctly)
Has it been confirmed that there is no performance difference at all between a regular Intel DX4 100 with WB cache compared to the DX4ODPR100? (I mean aside from the heatsink of course. Which I suppose is quite nice as well)

byte_76 wrote on 2025-03-28, 10:31:

Does the DX2ODPR66 also have 16KB L1 cache or was that only reserved for the DX4 Overdrive?

The Overdrive name seems to be somewhat misleading becaue one would assume it is an enhanced performance version of a regular DX2 or DX4 processor. (Even the DX4ODP which wasn't a "replacement" CPU .....If I understood the eqrlier comments correctly)
Has it been confirmed that there is no performance difference at all between a regular Intel DX4 100 with WB cache compared to the DX4ODPR100? (I mean aside from the heatsink of course. Which I suppose is quite nice as well)

The Overdrive CPU's, in many (most?) instances allowed an upgrade of performance considerably beyond what was otherwise possible, so in that context, I think the name was OK. A system with a soldered-on 486SX-25, which would otherwise only be upgradable to the 487SX-25 (math co-pro) could suddenly be upgraded to a DX2-50 or DX4-75 with the ODP. If it had a selectable FSB (which some did), the options opened up even further, all the way to the DX4-100. And then of course there are the Pentium Overdrive CPU's, which provided a considerable performance boost to compatible 486 systems.

The OPDR, which are the ones that go into the regular socket, yeah, the Overdrive moniker is perhaps a bit misleading.

byte_76 wrote on 2025-03-28, 10:31:

Does the DX2ODPR66 also have 16KB L1 cache or was that only reserved for the DX4 Overdrive?

The DX2ODPR66 only has 8KB L1 cache.

The marketing was aimed at 486SX owners which had no heatsink & fan, so I believe the heatsink was provided so the user didn't mess up putting a heatsink on and then asking for a refund.

https://www.os2museum.com/wp/486-overdrive/

As I understand it, the main point of the Overdrive (ODPR) was that it worked in 5V sockets because the voltage regulator built onto the CPU reduces the voltage to the 3.3V that is actually required by the processor. So, I suppose in that respect, the Overdrive naming is understood becaue it allows upgrading those single socket machines with only 5V support to processors that are much higher performing than the normal 5V processors the board was actually designed for.
However, I'm sure there were many consumers who had no idea what all of that meant and simply thought that the Overdrive was an enhanced CPU. I'll admit that it was my first thought when I saw the advertising before I did any research.

Speaking of voltage, has anyone tried a DX4ODPR100 with the voltage set to 3.3V on the board? Does the CPU work that way or would the board need to be set to 5V?

I’ve had ODPR (5v) chips run on 3.3 volts, most run fine and maybe a little cooler but technically that is underclocking because the regulator drops a little voltage even when it’s not needed.

Going the other way.
3.3v 486 chips don’t really like 5 volts.

Some folks obviously ran for years that way on an AMD 5x86 or Intel dx4 but it would gimp the chip and usually overclocking at 5 volts didn’t work out well.

That said I think some confusion is because of AMDs late entry advanced 486dx2’s that had 16kb of cache (WB OR WT) and required 3.3 volts.

The "Overdrive" name just means "upgrade chip", and what exactly the upgrade did changed from product to product. The original overdrives ran natively at 5V just like regular 486s, and the big upgrade was clock doubling.