'&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.