Hacker News Reader

Comment by jhallenworld | Hacker News Reader

Pentium Pro was really unbelievable at the time. Full Linux kernel recompile went from 15-20 minutes to less than 5 minutes. You had to use 'gcc -pipe' because the drives were then the limiting factor.

But I think it was not great on 16-bit code or something. The boost was not as impressive for DOS and Windows (still had lots of 16-bit code at the time).

I bought my Pentium Pro + motherboard and RAM at a computer show for ~$1000. This was a fairly good purchase. Around the same time I also bought an HP CD burner for around $1000- big mistake since it was slow and the prices soon plummeted.

View on HN

> Full Linux kernel recompile went from 15-20 minutes to less than 5 minutes.

We had a similar jump going from the 486 to the Pentium. I distinctly recall a Linux kernel compile being a 45-60 minute adventure on a 486DX2-66, but only 5 minutes on a Pentium 100. (I felt like a lot of elements of PC architecture improved together around that time as well.)

>45-60 minute adventure on a 486DX2-66, but only 5 minutes on a Pentium 100

Pentium was not that much faster than 486 - 486DX4-133 was faster than Pentium 90. I guess it had more memory

People don't realize how good Intel was (and is!). Take a good look at what the latest gen Xeon or i9 can do (and its SIMD instructions which are like a separate RISC computer inside) and you'll see why they're the leader. It wasn't just strongarming vendors, etc.

StrongARM is dead anyway. Take a look at what Threadripper and EPYC can do, especially for the cost - now that's what a real leader looks like.

One minor quibble. Intel pushed Itaniums hard at on point. The machines were huge and I had to have additional power routed to my office to run it. Thankfully AMD introduced the x64 architecture and the PC industry was saved. Between the power consumption of the Itanium and how hard the IA64 assembly was to debug. I atleast welcomed the X64 architecture.

IA64 assembly was "hard" because it wasn't expected to be something that humans would write or debug routinely - it was basically meant for compilers.

This also describes VLIW architectures more generally, and even more so wrt. current speculative developments like Mill, which AIUI doesn't even have a single, standardized "machine-level" assembly; what the machine runs at its lowest level is only defined by your actual chip configuration. (It is understood that GPU compute works much the same way, with the GPU driver acting as a "compiler" of sorts.)

Reply on news.ycombinator.com