
Intel 80387SX
Floating-point coprocessor for the Intel 80386SX
The Intel 80387SX is a floating-point unit (FPU) built to sit alongside the Intel 80386SX microprocessor and handle decimal arithmetic in hardware. Introduced in 1987, it was made in 16 MHz to 33 MHz versions and gave the 386SX far stronger maths performance than the older 80286/80287 pair.
- Floating-point coprocessor made only for the Intel 80386SX
- Introduced in 1987 by Intel
- Clock speeds from 16 MHz to 33 MHz
- Fits a 68-pin PLCC socket
- Three to five times the floating-point speed of 80286 with 80287
What it is
The Intel 80387SX is a floating-point unit, the companion chip that adds hardware maths to a computer built around the Intel 80386SX. It arrived in 1987 and does decimal (floating-point) calculations on its own, so the main processor does not have to fake them with slower software routines. It was built only for the 386SX, not for the wider 386DX. Intel sold it at speeds from 16 MHz up to 33 MHz, matching the clock range of its own 386SX chips. Paired with a 386SX, it delivered roughly three to five times the floating-point speed of the older 80286 and 80287 combination.
How it came about
Intel brought out the 387SX in 1987 to fill a gap left by its existing 80387. That earlier coprocessor could not talk to the 386SX, because the SX part had been given a 16-bit data bus in place of the 32-bit bus of the 386DX. A matching coprocessor was therefore needed. Intel also shipped a low-power 387SX to go with its low-power 386SX. Prices fell over time: the first 16 MHz part listed at US$506, a 25 MHz model appeared in 1991 at US$189, and a 33 MHz version followed in 1993 at US$109.
Inside the chip
The chip is a coprocessor, meaning it works beside the CPU rather than replacing it. It handles floating-point arithmetic, the kind of maths used for decimals, in dedicated hardware. It was built to match the 386SX's 16-bit data bus, which is why the wider 386DX and its 80387 partner were not compatible with it. The 387SX fits a 68-pin PLCC socket, the same package style used by some 80286 variants and the rarer 80186. Its clock range of 16 MHz to 33 MHz lines up with Intel's own 386SX speeds.
How it was made
Intel manufactured the 387SX itself. Speed grades ran from 16 MHz through 33 MHz, and third-party parts such as the IIT 3C87SX reached 40 MHz to match the fastest 386SX chips. The package is a 68-pin PLCC. No process node or transistor count is given here.
Where it went
The 387SX went into systems built around the Intel 80386SX, where it sped up floating-point work. With a 386SX in front of it, the pair offered three to five times the floating-point performance of the 80286 and 80287 combination it replaced. A low-power version was offered for machines using the low-power 386SX.
Versions and family
Intel sold the 387SX at several clock speeds, starting at 16 MHz and rising to 33 MHz, plus a low-power edition. Other makers produced compatible parts, including the IIT 3C87SX, which ran at up to 40 MHz. The chip's core design also appeared in the 80C187, a coprocessor for the CMOS Intel 80C186 that came out the same year and used an 8087-style interface.
Why it matters
The 387SX gave 386SX owners a hardware route to fast decimal maths, closing the gap with the larger 386DX platform. Its core design was reused in the 80C187, showing how one floating-point engine could serve more than one processor family. Falling prices over its life, from US$506 down to US$109, put hardware floating-point within reach of cheaper systems.
Questions people ask about Intel 80387SX
What does the Intel 80387SX actually do?
It is a floating-point unit, or FPU. It carries out decimal arithmetic in hardware alongside the main processor. That makes maths-heavy work much faster than doing the same sums in software on the CPU alone.
Why was a new coprocessor needed for the 386SX?
The existing 80387 could not talk to the 386SX. Intel had changed the data bus on the SX part from 32 bits down to 16 bits, so a matching coprocessor with the right bus had to be built.
How fast was it?
Intel sold versions from 16 MHz up to 33 MHz, matching its own 386SX clock speeds. Other companies made compatible parts too, such as the IIT 3C87SX, which reached 40 MHz.
How is it fitted to a board?
It uses a 68-pin PLCC socket. That is the same package style used by some 80286 variants and the less common 80186 processor.
Is it related to any other Intel coprocessor?
Yes. The 80C187, made to pair with the CMOS Intel 80C186, came out in the same year and uses the 387SX core, though it connects through an interface like that of the older 8087.
Good reading and sources
The facts on this page come from the Wikipedia article "Intel 80387SX", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryChip rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.
Like Intel 80387SX
All CPUs and microprocessors
🇺🇸 United StatesI486
32-bit microprocessor with on-chip cache and FPUCPU · x86 · 1989
🇺🇸 United StatesI386
First 32-bit x86 microprocessor by IntelCPU · x86 · 1985
More from Intel
All of them
WorldIntel 80387
Floating-point coprocessor for the Intel 80386 CPUCPU · x86 · 1987
🇺🇸 United StatesIntel i960
RISC microprocessor for embedded systems and military useCPU · RISC · 1988
🇺🇸 United StatesIntel 8284
Clock generator chip for early Intel processorsAnalog · 1986