now 23:30 hour, we have 25 Celsius 91% humidity, FB-DIMM 65 Celsius
picture s031-s036 shows the hardware reconfiguration of the V8-Xeon.
pictures say more than words :-)

Spectacular stuff.

One note:

Is not the left CPU fed with warm air from the right CPU?

Do you see much a temperature difference between the left and right CPUs?

Happy crunchin',
Martin

Hi Martin,
How you can see the left cpu get the warm airstream from from the right CPU, then to the backside fan, so the air runs in one direction backside out of the case. I believe that's the best alternative to have a constant stream. Today night I let run the two big case-fans additional with min. rpm to see the difference. With this combimnation I get now following core temps:
5:40 hour in the morning
Roomtemp=25 Celsius, humidity=95%, FB-DIMM sensor = 66 Celsius.

CPU 0
Core 0 = 63
Core 1 = 49
Core 2 = 60
Core 3 = 60

CPU 1
Core 0 = 57
Core 1 = 52
Core 2 = 58
Core 3 = 58

CPU 0 is the left one, we see not too big difference between them. Core 0 has a difference of ~6 degrees.
So we can now constate: without using the 2 big case-fans
(yesterday)
CPu 0 Core 0 was the hottest with 74 Celsius,
Today
CPU 0 Core 0 is the hottest with 63 Celsius, ~10 degrees cooler with this combination.
OK today it is 25 Celsius Roomtemp.
It is possible to run both CPU's without using the case-fans, so you have a very silent system.
The 2 big Case-fans are a bit noisy.
Conclusion: Both air-cooling solutions are possible, but I think if I want OC I must switch on the case-fans. But we will see what next days show.

Regards heinz

edit:
switched now the 2 big Case-fans off to have a compare with same roomtemp
after 25 min we see following values
CPU 0
Core 0 = 71
Core 1 = 61
Core 2 = 68
Core 3 = 68

CPU 1
Core 0 = 63
Core 1 = 58
Core 2 = 63
Core 3 = 63

so we can say: difference of core 0 is 7-8 degrees...

heinz

Now I let run the left big case fan with lowest possible speed
and we get following temperatures:
FB-DIMM sensor 70 Celsius raised up ca 6 degrees
warm air runs now onto the OCZ-RAM cooler !
CPU temps:
CPU 0
Core 0 = 66
Core 1 = 55
Core 2 = 64
Core 3 = 62

CPU 1
Core 0 = 63
Core 1 = 59
Core 2 = 63
Core 3 = 63

we have still 3 degrees difference between Cor 0 CPU0/CPU1
very interesting alternative, but the FB-DIMM temp increased..

heinz

shut down now, next thunderstorm is comming
heinz

10:00 hour
we are back again. Roomtemp = 25 Celsius, humidity = 90%
let the machine now run without the two big case-fans(quite modus)
FB-DIMM sensor shows 65 Celsius
CPU 0
Core 0 = 70
Core 1 = 60
Core 2 = 68
Core 3 = 67

CPU 1
Core 0 = 63
Core 1 = 58
Core 2 = 63
Core 3 = 63

heinz

How you can see the left cpu get the warm airstream from from the right CPU, then to the backside fan, so the air runs in one direction backside out of the case. I believe that's the best alternative to have a constant stream. ...

I agree on getting a constant steady straight-through airstream. The small 3 deg C difference that you see shows that the good airflow is working well.

For that layout, you have no good alternative for the airflow in any case...

How accurate are the temperature sensors?

Good luck on your experiments.

Happy crunchin',
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

How you can see the left cpu get the warm airstream from from the right CPU, then to the backside fan, so the air runs in one direction backside out of the case. I believe that's the best alternative to have a constant stream. ...

I agree on getting a constant steady straight-through airstream. The small 3 deg C difference that you see shows that the good airflow is working well.

For that layout, you have no good alternative for the airflow in any case...

How accurate are the temperature sensors?

Good luck on your experiments.

Happy crunchin',
Martin

Hi Martin,
I use CoreTemp 0.99 to read the cpu temps, I have no idea how correct this is.
Speedfan can not found the fans on the motherboard, this program is not usable for D5400XS.
Today I made the BIOS update 1140 as Francois recommended. The description of the process was not helpful, first I thought I have killed the board. After the machine automatic shutdown I waited for a auto-restart, but it not happened. After one hour I switched reset, but nothing, then I touched power and after this I saw on the monitor how the bios files are updated. Then starts Vista and after login a windows opened and shows "your BIOS update was sucessful". uuhh this was a hard thing.
But now is all fine and I will look in the BIOS for some ocing.

regards heinz

How you can see the left cpu get the warm airstream from from the right CPU, then to the backside fan, so the air runs in one direction backside out of the case. I believe that's the best alternative to have a constant stream. ...

I agree on getting a constant steady straight-through airstream. The small 3 deg C difference that you see shows that the good airflow is working well.

For that layout, you have no good alternative for the airflow in any case...

How accurate are the temperature sensors?

Good luck on your experiments.

Happy crunchin',
Martin

Hi Martin,
I use CoreTemp 0.99 to read the cpu temps, I have no idea how correct this is.
Speedfan can not found the fans on the motherboard, this program is not usable for D5400XS.
Today I made the BIOS update 1140 as Francois recommended. The description of the process was not helpful, first I thought I have killed the board. After the machine automatic shutdown I waited for a auto-restart, but it not happened. After one hour I switched reset, but nothing, then I touched power and after this I saw on the monitor how the bios files are updated. Then starts Vista and after login a windows opened and shows "your BIOS update was sucessful". uuhh this was a hard thing.
But now is all fine and I will look in the BIOS for some ocing.

regards heinz

RealTemp is a more accurate utility for temps on 45nm chips. Your CoreTemp is probably reading 10C high.

F.

---

RealTemp is a more accurate utility for temps on 45nm chips. Your CoreTemp is probably reading 10C high.

thanks Fred,

RealTemp is not able to show 8 cores, it can still show 4 cores...
Therefore I used better CoreTemp to see all 8 cores.
I tried Everest Ultimate, but as freeware it did not show all values ;-(
So I must see how I come around...
Always surprizes with DX5400XS.
At the moment I have no cooling problem, so it is all ok I think.
But it would be nice to have better possibilities to control the machine.
Ok, if all is fine and stable and no change is done, nothing must controled always.
It's enough to have a look when and then onto the values.

heinz

18:30 hour
roomtemp = 28 Celsius, humidity = 85%, FB-DIMM sensor = 65 Celsius
Today I opened the case, a first try to oc did not have success, must reset the BIOS over jumper.
Exclusive photo of the running V8-Xeon after reconfiguration. (case open)

have fun

heinz

19:20 must shutdown, next thunderstorm is already here.

19:20 must shutdown, next thunderstorm is already here.

20:50 hour, it's over... uuuuh this was a big storm with masses of rain..I hear the firemans horn. In the neighbour street the water is 20 cm high.
But the electricity was always up. It's still raining, but slowly and the thunder is to hear.
Till to the next time.
We are up with the machines and all is ok.
Temp is a little bit down now, its 25 Celsius and 91% humidity.
Hoping for a silent night now.

heinz

Perhaps you should get a good quality UPS system for the V8-Xeon. Then you could survive the thunderstorms and have the system do a controlled shutdown if needed.

Perhaps you should get a good quality UPS system for the V8-Xeon. Then you could survive the thunderstorms and have the system do a controled shutdown if needed.

I should really do so.
This year it is extrem with the thunderstorms here in the Rhin valley.
Hoping end of the week it will be better.

we are down last night, thunderstorms are comming back again.
7:45 hour, we are back again.
Temp is 25 Celsius, 86% humidity, FB-DIMM 64 Celsius constant, thanks OCZ.
Now the machine runs (low noise) modus ;-)
The two "ThermalRight eXtreme" cooler are doing her job very well.
You can see it there --> Message 762117

greetings to all who are reading here.

heinz

Preferences of my machines
If anyone is interested to have proven preferences for a "Working Horse" machine, this runs very well, I used the same on my P4 Northwood and the machines are crunching parallel with all what you want todo. I compiled programs, looked TV, cut films, short made all what I want todo and crunched parallel. ;-)
-----------------------------------------------------------------------------
03.06.2008 07:19:33||Starting BOINC client version 5.10.45 for windows_x86_64
03.06.2008 07:19:33||log flags: task, file_xfer, sched_ops
03.06.2008 07:19:33||Libraries: libcurl/7.18.0 OpenSSL/0.9.8e zlib/1.2.3
03.06.2008 07:19:33||Executing as a daemon
03.06.2008 07:19:33||Data directory: C:\\Program Files\\BOINC
03.06.2008 07:19:33||BOINC is running as a service and as a non-system user.
03.06.2008 07:19:33||No application graphics will be available.
03.06.2008 07:19:33|SETI@home|Found app_info.xml; using anonymous platform
03.06.2008 07:19:34||Processor: 8 GenuineIntel Intel(R) Xeon(R) CPU E5405 @ 2.00GHz [Intel64 Family 6 Model 23 Stepping 6]
03.06.2008 07:19:34||Processor features: fpu tsc pae nx sse sse2 pni
03.06.2008 07:19:34||OS: Microsoft Windows Vista: , Service Pack 1, (06.00.6001.00)
03.06.2008 07:19:34||Memory: 8.00 GB physical, 16.04 GB virtual
03.06.2008 07:19:34||Disk: 50.00 GB total, 22.61 GB free
03.06.2008 07:19:34||Local time is UTC +2 hours
03.06.2008 07:19:34|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: 4387433; location: home; project prefs: default
03.06.2008 07:19:34||General prefs: from SETI@home (last modified 20-May-2008 00:38:57)
03.06.2008 07:19:34||Host location: home
03.06.2008 07:19:34||General prefs: no separate prefs for home; using your defaults
03.06.2008 07:19:34||Preferences limit memory usage when active to 4094.35MB
03.06.2008 07:19:34||Preferences limit memory usage when idle to 7369.83MB
03.06.2008 07:19:34||Preferences limit disk usage to 22.61GB
03.06.2008 07:19:34|SETI@home|Restarting task 07mr08ac.15130.18477.14.8.23_0 using setiathome_enhanced version 528
---------------------------------------------------------------------
and you should set twice of your memory-value as "Virtual Storage" in your OS.
On my P4 this works with 1GB real-memory very well too.
I use pagedefrag from Windows Sysinternals to hold the paging-file not fragmented. !

have fun
heinz

---

D5400XS V8-Xeon

and you should set twice of your memory-value as "Virtual Storage" in your OS.

I have always disagreed with this rule of thumb.

The whole concept of virtual memory is to use your hard drive as RAM when you run out. This concept came about due to high memory costs whereas hard drive space has always been comparatively lower and more abundant.

Under that rule of thumb, a 32MB system would only need 64MB of virtual memory, but what if you have applications that require more RAM? Adding more memory of course would be the best option, but failing that, there's virtual memory.

This same rule of thumb says that an 8GB machine needs 16GB of virtual memory. Why? Unless you are running a huge database, most applications will run fine with 8GB of RAM while barely utilizing the paging file.

My professional opinion is to set the paging file to no large than you actually need (and if you need lots, consider a RAM upgrade). The best way to find out how much virtual memory you require, simply launch every application you plan on running (or even a few extra just to be safe) at the same time. Measure your RAM usage and take note of swap space being used. If you're not using that much, there's no reason to have a huge 16GB virtual memory file when you can probably get away with 2GB or less. Systems with less memory may need more virtual memory space. Take for example the aforementioned 32MB system. If you run quite a few applications and fine that you need 128MB of virtual memory, set it for that (and seriously consider a RAM upgrade).

Setting virtual memory for more then you need is just wasteful (with today's hard drive capacities, that is probably not an issue for most people), while setting it too low can be detrimental to your daily computing.

Just my opinion, of course.

and you should set twice of your memory-value as "Virtual Storage" in your OS.

I have always disagreed with this rule of thumb.

The whole concept of virtual memory is to use your hard drive as RAM when you run out. This concept came about due to high memory costs whereas hard drive space has always been comparatively lower and more abundant.

Under that rule of thumb, a 32MB system would only need 64MB of virtual memory, but what if you have applications that require more RAM? Adding more memory of course would be the best option, but failing that, there's virtual memory.

This same rule of thumb says that an 8GB machine needs 16GB of virtual memory. Why? Unless you are running a huge database, most applications will run fine with 8GB of RAM while barely utilizing the paging file.

My professional opinion is to set the paging file to no large than you actually need (and if you need lots, consider a RAM upgrade). The best way to find out how much virtual memory you require, simply launch every application you plan on running (or even a few extra just to be safe) at the same time. Measure your RAM usage and take note of swap space being used. If you're not using that much, there's no reason to have a huge 16GB virtual memory file when you can probably get away with 2GB or less. Systems with less memory may need more virtual memory space. Take for example the aforementioned 32MB system. If you run quite a few applications and fine that you need 128MB of virtual memory, set it for that (and seriously consider a RAM upgrade).

Setting virtual memory for more then you need is just wasteful (with today's hard drive capacities, that is probably not an issue for most people), while setting it too low can be detrimental to your daily computing.

Just my opinion, of course.

I agree, if you have enough RAM you can todo so, but in "My case" the machine is a "working-horse" and not a dedicated cruncher, so I have a reserve, if some app want more ram, and remember, I divided the ram half for crunching and half for the apps and on this machine I run more app's parallel, crunching, cut films, looking TV, compile some stuff, etc., and all this is running parallel.
Sure the best way is always to get the machine more hardware-ram as any virtual solution.
I have already thought about this, 16GB is the max for D5400XS.
So I have the possibility to install a second 8GB-Kit.
Additional I can install some 16 or 32GB CF-disks, accestime=0ms...if this should necessary, the machine is prepaired todo so.
This machine is equipt with 3 Terrabyte diskspace.
We will see what happen when I install Server2008(with some virtual machines) and Fedora9.

Attention: pagedefrag from sysinternals is not running under Vista.

heinz

The earliest mention of the 1.5 to 2 times system ram virtual memory size that I recall, stems from readme files in early windows versions of Adobe photoshop, at a time when typical photoshop image processing far exceeded the typical RAM configuration of a machine. Tasks such as those are trivial nowadays by comparison, however, we now tend to process equivalent ratio audio/video information in a similar fashion.

With a Virtual memory configuration larger than 2x system ram, the system ram becomes occupied by more page tables, so diminishing ram expansion / capability / performance ratio is encountered at that point making 2x system ram still a reasonable 'suggested' upper limit.

A smaller virtual memory setting than 1.5x however, will depend, as Ozzfan describes, on what your processing needs are. As hard drive space is cheap, and I tend not to edit large video, 1.5 x for me is fairly 'standard' and I would consider 1.0x a minimum to avoid excessive kernel time & hard drive page file restructuring under burst periods of heavy load. This multiplication factor ( of 1.0x - 2.0x) being a function of usage intensity by application demand, by content size.

Note that historically, the performance under win95, IMO from where these rules of thumb originate, was unstable enough to warrant minimising disk access, and OS issues addressed by programs such as photoshop suggested using a 'fixed size' to minimise kernel page file restructuring time during critical operations. This is largely reduced to almost non-existant, however 'page faults' are still the most expensive kind of cache miss you can generate in a system, so in that context a larger 'fixed' (& defragmented for that matter) page file on the fastest drive still makes sense if you are editing large content or many programs at once...

... but probably not for typical Windows users (like my mum) who have just freecell, outlook express and a browser running ;P) leaving it at 'Managed by system' is just fine these days, as improved speed and reliability at all hardware and software levels mitigate the risk of paging & restructuring.

Not discounting, of course, that I still think typical 'off the shelf' systems are sold with far too little ram for the product to function 'as advertised' ...

Jason

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

and you should set twice of your memory-value as "Virtual Storage" in your OS.

I have always disagreed with this rule of thumb.

The whole concept of virtual memory is to use your hard drive as RAM when you run out. ...

This same rule of thumb says that an 8GB machine needs 16GB of virtual memory. Why? Unless you are running a huge database, most applications will run fine with 8GB of RAM while barely utilizing the paging file.

My professional opinion is to set the paging file to no large than you actually need (and if you need lots, consider a RAM upgrade). ...

Setting virtual memory for more then you need is just wasteful (with today's hard drive capacities, that is probably not an issue for most people), while setting it too low can be detrimental to your daily computing. ...

Sorry, but that sounds rather confused. I'm confused from reading it. Do you have a definite recommendation?

There's the historical view of requiring a page file (swap space for virtual memory) that had to be at least the same size as physical RAM. That was to have enough space to do a core dump (now there's an old term!).

There's the other aspect that for most systems for general use, allocating swap space that is more than twice your physical RAM (if fully used) would give such poor performance as to be not usable anyway.

And the killer is that if you run out of memory space, then your processes get killed.

Hence the reasoning behind the very easy and very general rule of thumb of setting virtual memory (swap space) to be x1 to x2 of physical RAM...


As you mention, disk space is silly cheap, so why not take advantage of it? Even if it might never or only rarely get used. That rare occasion might otherwise kill your work.

As for any general rule, there's always good reason to ignore it, if you have good reason.

And does it matter? Most home users would likely not notice any difference for such as surfing the web in any case! (Provided that is, they do always have enough physical RAM for all occasions.)


My summary is: Setting x1 to x2 virtual RAM is useful silly cheap 'insurance' even if never used. For performance, ensure you have enough physical RAM. (But then also, unused and costly physical RAM is definitely wasted money. HDD space is much cheaper than physical RAM.)

Happy crunchin',
Martin

(Just my slightly contrary opinion ;-) )

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Thank you all for your opinions.
I would not say that everybody had todo so. Everybody can do what is the best for him. I talked from a real "Working-horse-machine". Here some sample tasks:
My films from TV are always recorded in DVD quality(720x576 7000, audio 224kbps) as mpg2-files. A normal film (30)min is ca. 1GB, My largest(4hours) are 8GB in one file. Cut the films and render a DVD in high quality und crunching parallel to this with 100%.
----------------------------------------------------------------------------
Here are my Preferences for seti

Processor usage
Suspend work while computer is on battery power?
(matters only for portable computers) yes
Suspend work while computer is in use? no
'In use' means mouse/keyboard activity in last 1 minutes
Suspend work if no mouse/keyboard activity in last
(Needed to enter low-power mode on some computers)
Enforced by version 5.10.14+ --- minutes
Do work only between the hours of (no restriction)
Leave applications in memory while suspended?
(suspended applications will consume swap space if 'yes') no
Switch between applications every
(recommended: 60 minutes) 60 minutes
On multiprocessors, use at most
Enforced by version 5.10 and earlier 8 processors
On multiprocessors, use at most
Enforced by version 6.1+ 100 % of the processors
Use at most
(Can be used to reduce CPU heat)
Enforced by version 5.6+ 100 percent of CPU time
Disk and memory usage
Use at most 100 GB disk space
Leave at least
(Values smaller than 0.001 are ignored) 0.001 GB disk space free
Use at most 50% of total disk space
Write to disk at most every 180 seconds
Use at most 75% of page file (swap space)
Use at most
Enforced by version 5.8+ 50% of memory when computer is in use
Use at most
Enforced by version 5.8+ 90% of memory when computer is not in use
Network usage
Computer is connected to the Internet about every
(Leave blank or 0 if always connected.
BOINC will try to maintain at least this much work.) 3 days
Maintain enough work for an additional
Enforced by version 5.10+ 0.25 days
Confirm before connecting to Internet?
(matters only if you have a modem, ISDN or VPN connection) no
Disconnect when done?
(matters only if you have a modem, ISDN or VPN connection) no
Maximum download rate: no limit
Maximum upload rate: no limit
Use network only between the hours of
Enforced by version 4.46+ (no restriction)
Skip image file verification?
Check this ONLY if your Internet provider modifies image files (UMTS does this, for example).
Skipping verification reduces the security of BOINC. no
--------------------------------------------------------
you can, but you must not todo so

heinz