3 ms·

While I share your feeling about x86, I don't feel PCs have stagnated since their standardization, at least not up until a few years ago. You are certainly corr

by DeepDuh 14y ago

While I share your feeling about x86, I don't feel PCs have stagnated since their standardization, at least not up until a few years ago. You are certainly correct that IBM PC at first was a horrid standard and until the Win95 generation other products have been ahead at least in certain aspects. However, I can't imagine the subsequent technology race between Intel, AMD, later 3Dfx, NVIDIA and ATI without that standardization. You just can't build chipfabs for billions of dollars if the market is splintered. Today we effectively got hundreds of GFLOPS accessible at our fingertips on commodity hardware, over a TFLOP if you count peak performance. Granted, bus architecture is still a bit stuck in the 80ies, which has now become a bottleneck with our SSDs - hence why Apple chose the proprietary route to get more out of their Macbook Airs.

rbanffy 14y ago
> I can't imagine the subsequent technology race between Intel, AMD, later 3Dfx, NVIDIA and ATI without that standardization Before the x86 dominance, you'd have Motorola in this group, making 68K chips (or, maybe, 88K). Processors today would probably be as fast as they are, maybe faster because they wouldn't need to be compatible with binaries written 30 years ago.
- DeepDuh 14y ago
  I didn't mean that the progress we've seen is because of x86, but because of the PC standardization. Imagine if we had a couple of competing systems like IBM PC, Atari, Commodore and Mac, each with 20-40% market share, each with their own chipsets. Do you think we would have 30 GFLOP single threaded / 700 GFLOP multithreaded performance built in commodity systems now? I don't think so. In mobile space it's kinda similar: Only because ARM was able to get such a dominant position in chip design we have the current speedup rate. The thing is however, that the main chipset is not the whole story, it seems to me that the way those dozens of subsystems work together nowadays just isn't designed well - as you said it just barely works. Having a standard would help in that every subsystem manufacturer could just provide drivers for his part and the OS would then know what to do with it, it's not necessary anymore for the assembler to do all that work.

by DeepDuh 14y ago

rbanffy 14y ago

> I can't imagine the subsequent technology race between Intel, AMD, later 3Dfx, NVIDIA and ATI without that standardization Before the x86 dominance, you'd have Motorola in this group, making 68K chips (or, maybe, 88K). Processors today would probably be as fast as they are, maybe faster because they wouldn't need to be compatible with binaries written 30 years ago.

DeepDuh 14y ago
I didn't mean that the progress we've seen is because of x86, but because of the PC standardization. Imagine if we had a couple of competing systems like IBM PC, Atari, Commodore and Mac, each with 20-40% market share, each with their own chipsets. Do you think we would have 30 GFLOP single threaded / 700 GFLOP multithreaded performance built in commodity systems now? I don't think so. In mobile space it's kinda similar: Only because ARM was able to get such a dominant position in chip design we have the current speedup rate. The thing is however, that the main chipset is not the whole story, it seems to me that the way those dozens of subsystems work together nowadays just isn't designed well - as you said it just barely works. Having a standard would help in that every subsystem manufacturer could just provide drivers for his part and the OS would then know what to do with it, it's not necessary anymore for the assembler to do all that work.