[N&B]

cpu voltage: 1.5375

To me it is a high value, at least for air cooling.
I have read across several places on the web that 1.45 volts is a maximum for safe operation with aircooling.
Watercooled pc can go higher without problem.

What are your temp readings when you render the speedtest scene? (coretemp is a good app to get these)

EDIT: At 3ghz I'm stable with the default cpu voltage (1.325 v), I think you could be stable with a lower setting and save some heat. Try something like 1.35v or 1.38v

[deadalvs]

i'll move my answers to this thread here:

http://www.maxwellrender.com/forum/view ... p?p=206010

else i have to reply in two threads and here, it's OT... see ya there...

[GM5]

[quote="ivox3"]thanks Greg, .. ...another 2GB stick is on it's way. ...I'll do it again then.


More RAM info... This shows RAM configurations that will achieve ideal performance when:

Riser A contains:________Riser B contains:
2 x 512MB---------------------2 x 512MB
2 x 1GB-------------------------2 x 1GB
2 x 2GB-------------------------2 x 2GB
4 x 512MB---------------------4 x 512MB
4 x 1GB-------------------------4 x 1GB
4 x 2GB-------------------------4 x 2GB

DIMMs must be installed in pairs of equal size. For instance, you must not have one or three DIMMs on either riser at any time.

[deadalvs]

okay... i don't understand this world anymore...

can anyone help me interpret this picture, please... ?

* * *

the only thing i did was setting the task priority to "above normal" in the task manager, after maxwell started... is that allowed ???

(i did not do that with my last official benchmark that is on the list !!)

it seems strange to me that the benchmark of 213 is only little higher than 210, mine on the list...

*HELP* *THIS IS STRANGE*

[deadalvs]

ok.

i ran the test again. i didn't alter priority this time, just as normal.

i just did some rearrangements with the overclocking specs.

intel core 2 quadro 2.66 GHz @ 3.33 GHz
(4 cores)
win xp 64
4 GB Ram
Time: 30 min 54 sec
Benchmark: 214.01

cinebench: xcpu: 1810

[ivox3]

deal ...

I've done the same and more alterations, ..yet didn't receive such varying results, ... what up ??

[deadalvs]

bo ... ? nooo idea...

* * *

but is setting up the priority legal in this test ?

24 min 05 sec can not be, can it ?

[Thomas An.]

Processor: Twin Clovertown E5345's [ 2.33Ghz /8MB cache / 1333Mhz FSB] 8 Cores

RAM: 2GB / FB 667Mhz Dimm (single stick)

OS: XP64

Time: 32m 04s

BMark: 202.65

Shameful
With the octacore you should have broken the 20min barrier ! ... now go back to your room and reflect on what you did wrong

[ivox3]

... I know. ...still pondering.


________

Dealdvs, ... If it actually makes a difference, ...there's certainly not much been documented about it, ..so, with no factual side of things outlined it remains just some kind of strange peripheral experiment. I believe all the other speedtests are without that type of adjustment. I still think it's worth investigating though ... like I said, ..I tried it as well, ...but not with the same degree of sucess that your reporting.

As far as it's submission legality, ...couldn't tell ya, ... but I know that 'not' reporting it would be a little misleading.

*** 24m 05s time, ... ??? hmmm, ... I don't know. ...right now, ..it just seems like an abberation until other tests can validate similar results.
Any chance you could put an actual stop watch to it? ..I've been doing that just to check 'Maxwell time' to real time, ... so far, ..it's essentialy been aligned.

[Thomas An.]

Wow, if I were to upgrade now to a quadcore it would be equivalent to an automatic 6.35x speed boost for Maxwell.
(This is one reason to wait as long as possible before getting a new system ... it makes the new much much more appreciated )

CPU Type: AMD Athlon 64, 2200 MHz (11 x 200) 3500+ 1core
System Memory: 1024 MB (PC3700 DDR SDRAM)
OS: Windows 2000
Time: 3h.16m.50s (196min50sec)
Benchmark: 33.57

[ivox3]

believe me, .. I waited.

[JDHill]

@deadalvs:

...all that running on one priority or another is going to do is to make it more or less possible for other processes get time on the processor(s)...it shouldn't make much difference...unless you feel like you absolutely have to check your email while running the benchmark.

If I was testing my machine to see how changes affected its speed, I'd:

> open Task Manager and kill all process which aren't absolutely necessary
> Task Manager > View > Select Columns > check CPU Time
> run mxcl from the command-line without -d
> close Task Manager
> leave the machine alone until the render is finished

now, if you're running multiple/multicore processors, you'll need to divide the CPU Time by the number of processors, because the number represents total processor-time, not clock time. When mxcl is finished, the command window will say:

>> Render Successful.

and the machine won't put another click on the CPU time. Now, you've got something a bit more consistent to compare between different hardware/software configurations. I suppose to play fair in the competition, you should be soaking up a few cycles running the mxcl UI, though, as it seems that's how people are doing it. I'm not 100% on how well the mxcl timer correlates to benchmark...if I want empirical results, I'll just do as above and read the actual CPU time taken for the given task.

[Hervé]

Thomas, I am waitng too... hheeh.. right now, it seams I would go 10 x faster... wowowo... too fast.. almost.. when I'll reach 15 x diference with the first one.. I'll upgrade.. maybe... (addictive situation..)

Chris, can you launch missiles with your rocket PC...

[ivox3]

Hi Hervé, ....no rockets, .. but I can bake a hellified number of croissants in a given hour, ...more than a dozen hungry and drunken frenchman could possibly consume in a week.

[Hervé]

LOL