Re: [gentoo-user] new machine : CPU : 22 nm vs 32 nm

[Michael Mol]

On Wed, Jul 25, 2012 at 4:32 PM, Florian Philipp <lists@×××××××××××.net> wrote:
> Am 25.07.2012 22:14, schrieb Volker Armin Hemmann:
>> Am Mittwoch, 25. Juli 2012, 16:05:29 schrieb Philip Webb:
>>> I've listed what's available at the local store,
>>> which I trust to stock reliable items, tho' I wouldn't ask their advice.
>>>
>>> All the AMD's are  32 nm , while the Intel recommended by one commenter
>>> -- Core i5-3570 4-Core Socket LGA1155, 3.4 Ghz, 6MB L3 Cache, 22 nm --
>>> is  22 nm : it costs  CAD 230  & they have  3  in stock,
>>> which suggests demand, but not the most popular ( 9  in stock).
>>>
>>> Isn't  22 nm  going to be faster than  32 nm  ?
>>
>> no
>>
>
> Lower transistor size gives you two advantages: Lower current (->
> potentially lower power consumption and heat) and more transistors to do
> something. The practical effects depend on what the chip maker does with
> this.

I second this; the feature size limit of the process isn't really
something a consumer should care about at _all_. Its only real impact
is on what architectural options are open to the manufacturer, which
in turn drives how much they can get out of a performance and feature
balance.

What you really care about is what the manufacturer builds, not the
tools and materials they had available to them.

>
>>>
>>> In the same price range, AMD offers  Bulldozer X8 FX-8150 (125W)
>>>  8-Core Socket AM3+, 3.6 GHz, 8Mb Cache, 32 nm  ( CAD 220 ,  2  in stock).
>>>
>>> How do you compare cores vs nm ?
>>
>> who cares?
>>
>
> You cannot really compare this. If you can use more cores, e.g. because
> you have an embarrassingly parallel application, by all means, get it.
> Otherwise you should probably care more about single core performance.

I'll note that emerge -e @world with parallel emerge and parallel make
qualifies.

So does running a browser like Chromium which gives each tab its own process.

>
>>> How far is cache size important ( 6 vs 8 MB )?
>>
>> depends on the architecture.
>>
>
> In short, for all three questions: Look at benchmarks and look at the
> TDP ratings if that is important to you.

Good points.

>
> nm numbers don't tell you anything that can be directly translated into
> performance or other qualities. They only allow educated guesses. If you
> really want to delve so deep into chip design, you could as well look at
> pipeline depths, cache associativity and such alike (not that you should).

Not that that isn't fun. ^^



-- 
:wq