When Centaur disclosed the technical details of WinChip 4(C4) at the end of 1998, it declared that they hoped to increase the processor clock to a level exceeding 500MHz (AMD only showed a 500MHz K7 sample at that time), so the depth of the instruction pipeline was increased to 1 1. But here's the problem. Centaur, which insists on "simplicity, rapidity and cheapness", is unwilling to redesign the new floating-point arithmetic unit of the deep pipeline to keep up with the target clock that has multiplied. What should we do?
Centaurs simply adopted the "big saw arrow method", and all costs were paramount. WinChip 4 uses WinChip 3' s floating-point arithmetic unit intact, but the clock only runs half of the main frequency, and everything is happy. Then a few years later, the floating-point efficiency of VIA x86 processor was repeatedly rubbed on the ground by Intel AMD, and it was not until later that VIA C3(C5XL) improved a little. It's really a great "raiders", so powerful that people can't laugh at Koko's smile.
Via acquired Cyrix and Centaur on 1999. The long-delayed core of Cyrix Cayenne was renamed twice to become the VIA Cyrix III processor with Socket 370 pin and code name "Joshua", and finally became a real product on February 22nd, 2000.
Strictly speaking, Cayenne is a complete micro-architecture of Cyrus M2. It has pipeline floating-point operation and MMX execution unit, and also supports MMX-FP instruction set created by Cyrix, which is comparable to Intel Pentium II in the same period. However, because it really dragged on for too long, when it appeared, it was three years later than the set Intel opponent. But on the bright side, at least, this is a decent ultra-pure non-sequential execution processor, not "high-tech version 80486".
Hard technology: Cyrix of x86 Kamikaze special attack team at the end of 20th century (I) Hard technology: Cyrix of x86 Kamikaze special attack team at the end of 20th century (II) Hard technology: Cyrix of x86 Kamikaze special attack team at the end of 20th century (II)
However, on June 6th, Cyrix III was occupied by pigeons, and was replaced by the internal codes C5A and VIA of Centaur, which were named after the biblical character Samuel. Based on the great concept of "simplicity, rapidity and cheapness", Cyrix III, the core of C5A, did not completely inherit all the features of WinChip 4, and cut corners from the long-awaited ultra-pure instruction execution and dynamic branch prediction specifications. It is still a superscalar x86 processor.
Although the brand "VIA Cyrix" still exists, it has existed in name only. By 200 1 year, which entered the 2 1 century, Cyrix even disappeared completely. The motivation behind VIA's decision is "simple, fast and cheap, with a strategy beyond transistors", and Centaur is more in line with VIA's high-level tastes. After the completion of Cyrix III, VIA couldn't keep anyone, and the existing Cyrix R&D team was dissolved, which is probably the real main reason. Therefore, VIA's x86 processor entered the 2 1 century with a half-speed floating-point operator, and it was not until late June 22nd, 2003 that VIA C3 (C5XL core) was "unsealed".
By the way, on February 20 19 1 day, AMD announced the termination of Geode product line originated from Cyrix MediaGX, which also meant that Cyrix's technical blood was completely cut off.
With "simple, fast, cheap" and "better than transistor strategy", we have achieved great success-even if it didn't happen in the market. This really makes people feel like science. It remains to be seen how far Centaurs can go under the banner of VIA in 20 years.
Finally, please recall the existence of VIA x86 processor, and then shout: Simple! Hurry up! Cheap! Strategy is better than transistor!
If it is Keke, it is necessary to track hard technology.