-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Conroe's true 64-bit processors ?
- Thread starter curvey
- Start date
Associate
- Joined
- 4 May 2006
- Posts
- 499
I may be wrong, but wasn't sse added to the pentium 3? I thought that was the only difference between the first model pentium 3s and latter model pentium 2s.
On the other hand who started the mhz war, i believe that was AMD with the athlon. Intel had to react as the mainstream market just assume 4ghz = twice as good as 2ghz (not unreasonable when you think about it).
On the other hand who started the mhz war, i believe that was AMD with the athlon. Intel had to react as the mainstream market just assume 4ghz = twice as good as 2ghz (not unreasonable when you think about it).
I think you're right (and NathanE, good contribution both of you) about us not agreeing, but there's a couple of things I'm going to have to address....
Netburst was NOT a fundamentally flawed architechture when the initial assumptions were taken into account. The fact that the assumptions proved false (and by extention the chip design showed flaws) doesn't make it quite as simple as you try to paint it. Intel did not sit there and say "I know, lets make a crap chip that just seems to go really fast." Contrary to what you may believe, they aren't that dumb.
There were two big assumptions that proved seriously false that were made by Intel designers originally.
1) Good code design and optimisation will reduce the impact of increased pipeline length.
This is actually perfectly true, and is frequently used on specific function, deep pipelined chips. A deep pipeline in itself is not a problem, apart from when the pipeline stalls, which can be minimised by optimised code. The problem intel had was that (a) it's optimised compiler produced code that ran great on a P4 and crap on everything else, and therefore wasn't really used, and (b) Trying to optimise code across a PC is nothing like optimsing code on a computing unit trying to do a specific function. A PC is a multipurpose, multifunction device that is frequently performing lots of different, unrelated functions at the same time. Even with each program correctly optimised, you can still end up with serious problems as you cannot optimise the programs to truely work with each other.
2) We should be able to scale this CPU to 10ghz or so.
Obviously, this never happened. Various things foiled this plan, including (1) above, heat problems, manufacturing issues etc etc.
Vertigo1 said:
Netburst was NOT a fundamentally flawed architechture when the initial assumptions were taken into account. The fact that the assumptions proved false (and by extention the chip design showed flaws) doesn't make it quite as simple as you try to paint it. Intel did not sit there and say "I know, lets make a crap chip that just seems to go really fast." Contrary to what you may believe, they aren't that dumb.
There were two big assumptions that proved seriously false that were made by Intel designers originally.
1) Good code design and optimisation will reduce the impact of increased pipeline length.
This is actually perfectly true, and is frequently used on specific function, deep pipelined chips. A deep pipeline in itself is not a problem, apart from when the pipeline stalls, which can be minimised by optimised code. The problem intel had was that (a) it's optimised compiler produced code that ran great on a P4 and crap on everything else, and therefore wasn't really used, and (b) Trying to optimise code across a PC is nothing like optimsing code on a computing unit trying to do a specific function. A PC is a multipurpose, multifunction device that is frequently performing lots of different, unrelated functions at the same time. Even with each program correctly optimised, you can still end up with serious problems as you cannot optimise the programs to truely work with each other.
2) We should be able to scale this CPU to 10ghz or so.
Obviously, this never happened. Various things foiled this plan, including (1) above, heat problems, manufacturing issues etc etc.