BillA/Overclockers reviews the OCPC Atlantis

thermal-management-testing.com

e-mail is badams@ stuck in front

BillA said:

re the 'old' and the 'new' data
yes there are differences in the equip, not really so huge, but differences
these were described in Hoot's wb construction thread

Click to expand...

Think you mean Joe's thread :
http://forum.oc-forums.com/vb/showthread.php?s=&threadid=141676&perpage=30&pagenumber=6
Bottom of page.

thanks Less
(its all becoming a blur ??)

be cool

AHHH! AT LONG LAST, I FOUND IT! WOOOOHAHAHAHA! A THREAD BILLA POSTED IN WITHOUT ENDING WITH THE "BE COOL!" ladies and gentlemen, this is a site to behold. j/k

edit:spelling.

ohmygod, another bubble burst
I'm imperfect !!!

be cool
be cool

hint: try procooling

Well I for one like how the roundup articles are done, only thing I'd like to see are graphs done with metric and imperial units together like HERE. If anybody knows the software to make graphs like that please speak up.

I'll admit there is a bit of a learning curve in understanding the articles, but putting "soft numbers" would result in "soft conclusions". It would be nice if some enterprising member would submit an article on "How-to: Read BillA's Test Results" or "BillA's Test Results 4 Dummies" HaHa, I'm having a bit of fun here but really somebody could do a nice service for the community and possibly walk away with $100 for the effort, see HERE.

peace.
unloaded

if i ever get to have a special name i'm gonna be a senior bubble burster.

I hate to join the minority here but I'm with Axle. I read through the review a bit and at the end I'm like, "Uhhh, so it's a good block?" I'm not too picky with my stuff.

--Illah

I think the majority of people aren't able to read the graphs. But I don't expect bill to put up something simple like idle/load temps (nor do I expect him to start). I don't get the graphs that much either but I'm trying to learn how to read them.

I'm sure if we do some reading on Bill's site, it'll become more clear to us

I looked over the C/W vs. Flowrate charts for the Hoot's block and the Atlantis.
Hoots block- (about) .205 C/W @ 7.5 LPM
Atlantis block- (about).245 C/W @ 7.5 LPM

So Hoot's block is capable of better cooling.

Am I reading these correctly?

JML said:

I think the majority of people aren't able to read the graphs. But I don't expect bill to put up something simple like idle/load temps (nor do I expect him to start). I don't get the graphs that much either but I'm trying to learn how to read them.

I'm sure if we do some reading on Bill's site, it'll become more clear to us

Click to expand...

The graph to really compare TEMPS of the WB's in BillA's testing is the C/W vs. Flowrate graph. The lower the C/W the better. You can pinpoint various C/W's for various flow rates. I would guess for most hobby pumps, ie., Eheim 1250, via aqua etc that the corresponding flow rate in lpm would be between 4 - 6 or so. Maybe BillA will chime in if he knows...

Check out the post before your's nikhsub1, I think you posted your response like 5 seconds after I posted my second reply

IF "the majority of people aren't able to read the graphs"; then such is a sickening indictment of our educational system

I see these comments and am dismayed, other threads have been consumed in flames regarding this
I will try to moderate my commentary to a dull (a crappy pun ?) roar
N.B. I am not slamming those who cannot read graphs, its their teachers and admins that have me going

I am (obviously) not the best guy to write a graph reading intro - but Dave Smith (of amdmb fame)doesn't hang here much - so I'll attempt to keep it simple

a graph is a picture composed with numbers - full stop
the numbers are grouped according to their 'type'
the numbers are arraigned in 'some' order
two sets of numbers may be arraigned to illustrate a 'relationship' between them

not so difficult, eh ?
some examples:



here there are 10 'numbers of interest', the dark blue points
their 'value' (or magnitude) is given by the vertical scale on the left - and this is the basis for each dot’s vertical positioning

these 10 readings were taken one after the other and the horizontal numbers indicate the reading's order in the sequence
- the vertical 'Error Bars' are an indication of the uncertainty in the measurement based on a defined type of statistical analysis
- the black trend line is the representation of another statistical analysis (linear regression) showing how the average of the readings is changing as the sequence progresses
(in this case the increase suggests something was changing - or I started to get a bit sloppy towards the end, the most likely; consistent mounting is very difficult)

it is worth understanding that the above graph is NOT of a measurement, shown is the product of 4 physical measurements which were then used in a series of calculations (2 temps, voltage, and current) AND that each of the actual measurements had a whole host of 'corrections' made to the raw value before it was used in the calculation

so why the graph ?
'cause it cuts through all that number business so that the 'relationship' can be seen at a glance

let's try one more complex


whoops, where'd this come from ? (time travel)

"C/W" on the vertical axis, pressure drop on the horizontal
(what you're looking at, should also be indicated in the title of the graph)

here you can see that:
- for all the wbs, their "C/W" decreases as the pressure drop (across them) increases
- and since each of the 'dots' represents a measurement (or calculation !) at a specific flow rate, one can see that each of these wbs has a different pressure drop at the same flow rate
(by comparing the location on the horizontal axis of the corresponding 'dots' - look at the last ones for all 4 wbs, taken @ 2gpm)

some of these graphs will 'display' a number of relationships simultaneously
-> spend some time, work through it
(this last one is very interesting, let's hope JoeC gets back to start putting some of this stuff up)

be cool

I think the issue here is one of scale.

C/W doesn't mean much by itself when someone is asking "so how many degrees C will block X gain over block Y".

The answer is, of course, in BillA's discussion, where he states that standard conditions are a 70W heat source which correlates to roughly around 100W according to Radiate or any number of other CPU thermal calculators.

In other words, for a modern CPU that is overclocked/over-volted to a moderate extent (not extreme) that's under full load, there will be a 0.7C difference in temperatures for every 0.01C/W difference on BillA's testbed (note - lower C/W's are better).

Now the C/W difference applies only to BillA's testbed and not directly to real CPU's, but given that BillA's heat-die size is 100mm^2 which treads the middle ground between AthlonXP/P4/T'Bred/Duron/0.09u P4 successor, the logical correlation is close enough to give a very good indication of what a particular waterblock will do for you.

The challange now comes with matching up pressure-drop vs predicted flow rate with a particular pump to figure out whereabouts on a flow/CW line a system will sit. I've just woken up after 24 hours of travel (with numerous link-up delays) and will attempt to tackle an explanation to this later when my brain clears.

Does anyone else get the feeling that some threads are actually a rehearsal for an AFLAC commercial?

Hoot

What's AFLAC for us international forum members?

sorry hoot, i didnt understand. My english is improving....

well, my English is fair to middlin' and I missed it also ???

AFLAK

be cool

isn't aflack the insurance company?

Yes AFLAC is the insurance company, they are secondary insurance... The commercial is with that stupid duck that screams "aflac" "Aflac" "AFLAC" and no one ever seems to hear him...