[RndTbl] new ECC computer options (jd)

jd jd at wcgwave.ca
Mon Jan 30 05:00:53 CST 2017


Not much on ECC from intel, is there?

Brace yourself for Xeon pricing;
e.g: octa-processor E7-8890-v3 (12-core) .. usd$ 7,174 !

For intel, a coupla notes:
- Skylake CPUs (Lga1151) all gots VT-d (virtualisation optim) in 100series
- Haswell is still around - may find bargains
- Broadwell is mainstream - mature?
- looks like it's back & forth (&back&forth) between
(say..) Broadwell CPUs and Chips, until you find the pairing you want,
THEN  go looking for mbrds based on that chipset,
AND which implement your features.

- - - -

With AMD .. waayyy easier.
Survey FX/Am3+ mbrds, and then .. pick a Bulldozer.
Dunno much about Opteronions.
ECC is clearly documented as supported by CPUs, so just make sure mbrd 
implements that.

..and.. at these upper classes, is DualChannel assumed?
Make sure mbrd offers it.
Your workstation's data-width ('swath') to/from RAM
"don't need no steenkin' zipper-merge"

Many but not all CPUs come with HeatSink/Fan kit.
TDP (Thermal Design Props) are all encouraging.
AMD is 32nm tech, but still needs serious cooling.
AMD supplies cooling kit on (only?) three of their mid-upper 'dozers.
intel's modern CPUs are 22nm (and 14nm!squint!), but again, gotta be 
cooled right.
Many models will push that pesky 3gHz 'barrier'

I would like to bet you a beer you will find worthy candidate mbrds in 
uATX,
and worthy boxen in that format, w/ adequate space, therein,
to warm up your sandwiches.

jd

Oh.Yes: Parity was an extra 9th-bit chip, in DIP days. On write, the 
9th-bit would be made to agree' with the byte's being odd or even. If a 
read cycle showed any dis-'agree'-ment, we'd be better off .. half the 
time.
IBM had some purdy schmaart dudes. If they want to save us half our 
Time, we should be grateful.
Parity survived into SIM (&SIP) "module'/'panel'days.
Later in that SIM aeon, around the time of [Nibble+Nibble+Parity] SMT 
DIPs on each SIM, it became ..uhm.. expedient to install a simple logic 
trick-chip (instead of a real bitwise chip), to calc the SIM's parity on 
any read, and present that to the eager, expectant mbrd parity guts.
On write, of course, it was still told what to remembory but .. what 
were we talking about?

Oh, Yes. I think them was days of "fast page" and EDO and paying a bit 
more for 70ns MT, man,
but not for 70ns EDO .. long story, man.

We were incensed when SDRAM/sdimm came along and it don't got no parity, 
man.

Mere parity was a laugh to VAXmen. man.

Macs no likey (any kinda) parity sims, I learnt. Yet them fruity 8bit 
sims were always more shekels.

Go figure.

_ _ _ _ _ _ _ _ _ _ _ _ _

2017-01-29 12:00 PM, roundtable-request at muug.ca wrote:
> Send Roundtable mailing list submissions to
> 	roundtable at muug.ca
>
> To subscribe or unsubscribe via the World Wide Web, visit
> 	https://muug.ca/mailman/listinfo/roundtable
> or, via email, send a message with subject or body 'help' to
> 	roundtable-request at muug.ca
>
> You can reach the person managing the list at
> 	roundtable-owner at muug.ca
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Roundtable digest..."
>
>
> Today's Topics:
>
>     1. Re: new ECC computer options (Trevor Cordes)
>     2. Re: new ECC computer options (Trevor Cordes)
>     3. Re: new ECC computer options (Kevin McGregor)
>     4. Re: new ECC computer options (Hartmut W Sager)
>
>
> _______________________________________________
> Roundtable mailing list
> Roundtable at muug.ca
> https://muug.ca/mailman/listinfo/roundtable

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://muug.ca/pipermail/roundtable/attachments/20170130/0f5fc847/attachment.html>


More information about the Roundtable mailing list