'&E' markings: SL enhanced CPU (Processor includes power management features).

The last i486 with core frequency equal to bus frequency, introduced in June 1991. It was followed 1992 by the DX2 with clock doubled core frequency.

To run it at its 50MHz, this Intel chip demands a heatsink. To over-clock this chip, it demands a heatsink-fan (purpose designed for the later DX2 -66 {and hotter} chips). I overclocked the only one I ever bought. It ran at 60MHz all day long. (with the heatsink-fan of course). I failed to overclock it at 66MHz. I never knew whether it was the m/board chipset chips, or the -50 CPU that failed at 66MHz. Could have been either, since the chipset chips were also stamped -50.
 
For a long time, Intel only supplied samples of the i486DX -50 CPU to OEMs. This went on for nearly one-year. The 50MHz M/Board was a real "trouble-maker" at the time, and no 50MHz supporting chipsets had ever been in use before this time. Until then, less reputable OEMs had made 4-layer M/Boards alongside the more-expensive 6-layer M/Boards. This had the double-expense to these low-grade OEMs, of needing to make 6-layer boards only, for 50MHz operation throuhgout, and to source expensive 50MHz supporting chipsets! With Intel's long holdup of commercial volumes of -50 CPUs, these nasty OEMs had time to re-develop stable 50MHz M/Boards. However, one-year later when consumer 50MHz 486 systems flowed freely in the marketplace, many "clones" were still very unstable, due to 4-layer designs persisting still, and due to non-use of a heatsink on the CPU (as extra cost-cutting)!!! The -50 was the first x86 CPU to demand a heatsink. N.B. the later-on i486DX2-50 did NOT require a heatsink, but usually came fitted with one anyhow. (I guess the system builders had gotten used to fitting heatsinks by now).

This chip was the great saviour for OEMs and system builders. Unlike the DX-50, the DX2-50 didn't demand a heatsink for use. Unlike the DX-50, the DX2-50 didn't demand OEMs to build quality advanced 50MHz-throughout M/Boards. Many early VESA VL-BUS add-on cards would mis-behave when the M/Board was running at 50MHz-throughout. By building a system based on a DX2-50, system builders could sell stable reliable 50MHz-CPU-systems that were only 3% to 15% slower in "real-world" apps. than when using same-for-same BIOS and logic M/Boards, built using DX-50 CPUs. Any DX2-50 end-consumer system (with VL-BUS cards) always ran all programs faster than any 33/40MHz end-consumer system using only 16-bit ISA cards. Thus the DX2 -50 was the saviour.
N.B. once the DX2-66 was common, slower 50MHz systems were "always" built with the DX2-50 or (rarely) the SX2-50. The DX-50 systems fell into non-supply except for the vary rare EISA BUS M/boards (they were mostly 50MHz, except for the EISA circuitry, which was still at 33MHz even then).
With Intel supplying the DX2-50 cheaper than it supplied the DX-50, it was one really really great chip.

An i486 DX2 manufactured by IBM

'&E' markings: SL enhanced CPU (Processor includes power management features).

When this chip came out, I had a "huge" form-factor (full-AT-size) 486 M/Board. It had sixteen SIMM sockets, but had no clock-crystal socket. This was before end-user jumpers for clock generators were around, thus I was locked into 33MHz, due to the crystal being soldered. I tested a DX2-66 CPU in this giant M/Board, but the (very very) early BIOS didn't let it work. If I wanted to use a 66MHz chip, then I needed a new M/Board that supported such things. By then, the newer M/Boards also had settings for DX4's as well.
The DX4-100 soon followed into the market. As a result, I bypassed the i486DX2-66 altogether, and bought the latest and greatest i486DX4-100. So, I never experienced to joy of owning a -66.

'&EW' markings: SL enhanced CPU (Processor includes power management features) with write-back cache.

This CPU-module is the processor inside my IBM Thinkpad 755C from 1995. To take a picture of it about 40 screws and lots of circuit boards, cables, plugs and shieldings had to be removed inside the Thinkpad - hard work.

When this chip was the latest and greatest new kid on the block, I upgraded from my then 486 33MHz system. I had to replace my then M/Board, as it didn't support "ANY" clock-multiplier CPUs. see comment for A80486DX2-66.
Within months of having my new speed, the AMD 5x86 and Cyrix 5x86 came on sale. The AMD was a clock-quadrupler. The Cyrix was still a clock-tripler, but advanced 5th generation style. Maybe the AMD should be more rightly called an AM486DX5-133, since inside it is still identical to all other 486's. My brand new M/Board's BIOS didn't know these two new chips, so I bought the AMD -133 chip and got a BIOS update at the same time. The AMD behaved unreliably in my system, so I exchanged it with the shop for the Cyrix. (The BIOS update included both). So to end my story, I only got about two-months use from this Intel A80486DX4-100 CPU. N.B. It is nearly a decade later, and I'm still using Cyix 5x86 CPU's.

The 486SX processors were the same as their DX counterparts in design, but the built in math co processor was disabled. There was a 487 which was actually a 486 with the math co processor enabled. If the motherboard had this slot for the 487, the SX cpu would be disabled and the 487 cpu would take over all processing.

Was the i486SX "Cripple-chip" really a planned release, or was it actually as rumoured to be, a genuine i486DX -33, that failed the FPU production-line tests???? I was donated an i486SX-20, and I was also donated a "full-size-AT form-factor" 486-33 m/board. This crazy-large m/board had an 80486 socket, and 4167 socket too. It also had sixteen (count them, 16) SIMM sockets, cause, what was a 16MByte SIMM??? It only understood 256kByte and 1MByte SIMMs, and it only understood the 9-chip SIMMs, not the later 3-chip SIMMs. This proved to be a bane to me, as the 9-chip SIMMs cost more than the newer/faster 3-chip units.
I put two-and-two together, and that little PGA SX-20 chip ran at 33MHz just fine. As a test, I tried to smoke-it-out by covering it with cardboard etc. to fry the thing. It would not stop running normally! (Without the cardboard), I ran it at 33MHz for years, before I sold that sytem. Last I heard, it still was fine.

This is an SMD package CPU mounted on a PGA board to fit into a standard desktop socket.

Great little sub-unit this! Overclocked it to 40MHz no probs. Used it for years and years. No heatsink needed at 40 either! Not bad for a so-called -25.

I lusted for one of these! Time marches on, and I upgraded to Intel DX4 -100 after all. Never got to see one of these SX2's.

First CPU specifically designed for Notebook PCs.

This is an OverDrive processor for an additional OverDrive socket on the mainboard.
Upgrades 25 MHz Intel 486SX based systems to 50 MHz.

Note the extra pin. It is a kind of "tell the other guy to shut down" pin. Implemented so that the end-user can just "add" this new chip into the spare upgrade socket, without any need of interfering with the originally supplied SX CPU. Ie. what end-user can remove a PLCC or even PQFP chip.

Upgrades 25 MHz Intel 486DX and SX based systems to 50 MHz.

Upgrades 25 MHz Intel 486DX and SX based systems to 50 MHz.

Upgrades 33 MHz Intel 486DX and SX based systems to 66 MHz.

Upgrades 33 MHz Intel 486DX and SX based systems to 66 MHz.

Upgrades 33 MHz Intel 486DX and SX based systems to 66 MHz.

This is an OverDrive processor for an additional OverDrive socket on the mainboard.
 
Upgrades 25 MHz Intel 486DX and SX based systems to 50 MHz.
 
The Processor is still sealed in its original box.

This is an OverDrive processor for an additional OverDrive socket on the mainboard.
Upgrades 33 MHz Intel 486DX and SX based systems to 66 MHz.

Note the extra pin. It is a kind of "tell the other guy to shut down" pin. Implemented so that the end-user can just "add" this new chip into the spare upgrade socket, without any need of interfering with the originally supplied SX CPU. Ie. what end-user can remove a PLCC or even PQFP chip.

Upgrades 33 MHz Intel 486DX and SX based systems to 66 MHz.
 
There is not much difference between an DX2ODPR66 and a standard 486DX2-66. Intel was marketing their speed doubling technology in different products. DX2 chips were sold to Intel OEMs and OverDrive chips were sold directly to end-users. The DX2 chips typically have no built-in heatsink and have a distinctive DX2 label. The OverDrive chips have a distinctive OverDrive label and may have a built-in heatsink.
The 'V3.0' and 'V4.0' marks the third and fourth stepping of the processor since introduction.
 
References:  pcguide.com  Intel

Upgrades 33 MHz Intel 486DX and SX based systems to 66 MHz.

 
Upgrades 25 MHz Intel 486DX & 16/20/25 MHz 486SX based systems to 75 MHz.
The Processor is still sealed in its original box.

Upgrades 25 MHz Intel 486DX & 16/20/25 MHz 486SX based systems to 75 MHz.
 
DX4 OverDrives typically have qualities different from standard 486s of the same speed. These include built-in voltage regulators, different pin-outs, write-back cache instead of write-through cache, built-in heatsinks, and fanless operation.
 
References:  pcguide.com

Also will upgrade DX-66 systems to a stable 100 mhz.
This is probably an overclock of the overdrive itself.

Forgot to add, 486 fan was attached to heatsink for extra cooling and stability.

The Intel DX4ODP75 was designed for changing 25 MHz 486 systems to run at 75 MHz.
BUT. The v1.1 allowed 33 MHz systems to run at 100 MHz. The first version (1.0) could only bring the speed of a 33 MHz system up to ~83 MHz.

Upgrades 33 MHz Intel 486DX & SX based systems to 100 MHz.

Upgrades 33 MHz Intel 486DX & SX based systems to 100 MHz.