• 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.

Someone explain nm to me plxpxlxplx

Soldato
Joined
5 Jul 2003
Posts
4,250
Location
Larndarn
This is driving me mad now, i keep reading rave comments about how 90nm is gonna be reduced to 80nm processes and how this is really *** and i have nfi what it means.

why is 80nm better than 90nm. what does the reduction signify? I hae an inkling it has somthing to do with capacitor (?) and a manufacturing process (?) how does all this transform into improved performance?

thankyou, i feel better after my rant.

thank you all for any responses enlightening a grumpy reader :)
 
Associate
Joined
2 Jul 2006
Posts
771
Location
altrincham, manchester
Snax said:
so out of hte box there is no performance increase?
if 2 processors where exactley the same bar the nm process then theoretically the one with the smaller process should be ever so slightly faster, but thats not always the case, the jump from the northwood cpus to the presscott cores wasnt a great one 130nm>90nm, with most northwood processors outperforming there presscott counterparts in terms of clocking capability...
 
Caporegime
Joined
8 Jan 2004
Posts
31,993
Location
Rutland
LanceCrossfire said:
the jump from the northwood cpus to the presscott cores wasnt a great one 130nm>90nm, with most northwood processors outperforming there presscott counterparts in terms of clocking capability...

A 130nm and a 90nm will perform exactly the same at the same speed so long as their architecture is indentical.

The smaller the process used, the less heat output the chip should have and the greater headroom it should have. Smaller processes require less voltage, but don't tolerate voltage aswell either.

The Northwood/Prescott example isn't great, the process shrink allowed higher clocks, but the architecture change (longer pipelines) reduced the clock for clock efficiency and was horribly inefficient power wise. So not only was the CPU putting out more heat per clock, they had to ramp up the clock speeds to get better performance out of it, meaning yet more heat on top.
 
Soldato
Joined
24 Jul 2006
Posts
8,876
Location
Hoddesdon, London, UK
Smaller process means more die per wafer as well, so more cpus can be made. Similar in some ways to the Prescott saga was the AMD move from Palomino to Tbred A, which was'nt really much cooler than the Palomino and did'nt clock any better but was eventually sorted with the B revision, shows that sometimes process changes can be hard at first, later Prescotts, though still not as efficient as the Northwood clock for clock, ran cooler than the first batch.
 
Last edited:
Associate
Joined
17 Apr 2006
Posts
2,240
Location
Hampshire
I think the nm rating refers to the smallest bit of wire they put in the chip. A chip is essentially some clever bits, some memory (cache) and a lot of transistors. If the wires are smaller you can build a lot more transistors in the same space as the transistors are now smaller, making the 'silicon real estate' - space on the chip a lot 'cheaper' - more freely available. This can lead to cheaper chips on smaller 'wafers' - the silicon the chip is built on, more transistors, and more room for cache.

I don't believe overclocking potential of a chip is related to the nm rating, as with the improvements the stock speed goes up. Also, the price of the chip will initially be very high as of course it is always the enthusiasts who need the best kit NOW that pay for a large percentage of the R&D costs on any new product before the prices come down for everyone else.
 
Associate
Joined
6 Sep 2005
Posts
283
Example, the AMD Athlon 64 3500+ processor fabricated at 130nm and running at 2.2GHz, employs a nominal voltage of 1.5V – its maximum thermal design power is 89W, and processor current is specified at 57.4A. The same processor fabricated at 90nm only needs 1.4V – its thermal design power drops to 67W, and its current maxes out at 45.8A. This principle applies down to 35nm manufacturing.
 
Back
Top Bottom