There was an early 2000’s Intel roadmap that used 686/786/886/986 monikers to refer to future Intel chips projected up to 10ghz by its end .

Was an amusing place in history made moot by the stagnate and malaise era with parallelism and 64bit bloat taking the place of real progress .

rmay635703 wrote on 2022-08-07, 22:49:

There was an early 2000’s Intel roadmap that used 686/786/886/986 monikers to refer to future Intel chips projected up to 10ghz by its end .

Was an amusing place in history made moot by the stagnate and malaise era with parallelism and 64bit bloat taking the place of real progress .

64bit bloat as you call it was a required step and as the designers quickly discovered you cant keep driving up speeds without it costing you in other areas quickly so multi core CPUs got around the lack of speed increases by spreading the load.

Its real progress regardless of how you feel about it.

except Itanium 64 .. that was truly bloat.

I remember when they canned Tejas and Jayhawk. I'm sure that was for the best.

Chips & Cheese did a nice retrospective on Netburst.
https://chipsandcheese.com/2022/06/17/intels- … on-for-success/

swaaye wrote on 2022-08-07, 23:59:

I remember when they canned Tejas and Jayhawk. I'm sure that was for the best.

Chips & Cheese did a nice retrospective on Netburst.
https://chipsandcheese.com/2022/06/17/intels- … on-for-success/

There is a lot of hate for Netburst and a lot of it isn't misplaced but looking at what Intel was trying to do you can see some of the crazy advancements they had in the core, many of which were backported to the Pentium M core and combined to produce the Merom Core Solo line of mobile CPUs which in turn became Conroe and Woodcrest.

Without Netburst Core simply wouldn't exist as it does now.

Another problem with "Pentium 5" is that'll cause confusion with the first pentiums. and now that this is 'exposed', we'll see more of that now.

rmay635703 wrote on 2022-08-07, 22:49:

There was an early 2000’s Intel roadmap that used 686/786/886/986 monikers to refer to future Intel chips projected up to 10ghz by its end .

Was an amusing place in history made moot by the stagnate and malaise era with parallelism and 64bit bloat taking the place of real progress .

Haha. No.

rmay635703 wrote on 2022-08-07, 22:49:

There was an early 2000’s Intel roadmap that used 686/786/886/986 monikers to refer to future Intel chips projected up to 10ghz by its end .

Was an amusing place in history made moot by the stagnate and malaise era with parallelism and 64bit bloat taking the place of real progress .

Do you even know what multithreading and 64-bit is for?

cyclone3d wrote on 2022-08-08, 18:25:

rmay635703 wrote on 2022-08-07, 22:49:

There was an early 2000’s Intel roadmap that used 686/786/886/986 monikers to refer to future Intel chips projected up to 10ghz by its end .

Was an amusing place in history made moot by the stagnate and malaise era with parallelism and 64bit bloat taking the place of real progress .

Do you even know what multithreading and 64-bit is for?

To fulfill Parkinson’s Law, nothing more.

https://en.m.wikipedia.org/wiki/Late_binding

In my world…64 bit has primarily provided an expansion of the memory bus past 32 bits which ironically isn’t 64bit wide and was possible with 32bit cpus. A flat address and software segments over 4gb perhaps the other side imagined effect.

The instruction set and registers expanded making things like compilers more complex and less efficient per megabyte of data while allowing for a brute force of more performance accompanied by a disproportionately large increase in needed overhead, speed, space and complexity to provide a comparatively small gain in performance.
Day to day I gained nothing useful from the additional instruction set, the added on die security features, sse or most of what I am told are game changers.

Multi threading has allowed you to take quadruple the power to do twice (or less) the work running multiple tasks separately while sharing the environment and process resources which need to bloat up to support the parallelism.

Excluding 3D video cards
parallelism as implemented has been a very poor way of hacking performance into the Pc world especially when combined with all the performance detractors that come along for the ride removing most of the gains.

So It works but you need 100x the transistors to do 10x the work assuming your code is compiled correctly.

The universal complaint I have seen since the dawn of time is why is my multi threaded code slower?, looks like about a million copies of the same question every year and not without reason.

Reminds me of the 3ghz P4 w/8gb of ram getting a hacked up modern version of Windows and loosing its ability to run OEM solitaire in a playable fashion . Progress

SMH

I seem to recall some magazines predicting that Prescott, with its even longer pipeline, double stuf L1 and L2 caches, SSE3, PCI Express chipset, AMD64 support, and this weird new socket with pins on the motherboard, would be called Pentium 5.

But then it came out on Socket 478, failed to outperform Northwood, and totally ignored AMD64. Yep, just another P4. 😜

Edit: I haven't watched the video yet, but looking at the TOC, it probably covers Prescott.