Come in sit down and make yourself at home. Welcome to the humble abode that is MyOnlineTeam. anyone?

Good to see you Zirith, PioPicone, Dan Brinsdon, laughing66607 and blueKnight

I've (finally) created your member pages for you - they are now online - if you would like any extra info on there please drop me an email and I'll update it for you. My email address can be found on the teams website. Sorry it's taken so long (Holiday + New Job - excuses, excuses) but if you take a look you will see your MOT points graphs are well on their way


Thanks for choosing to crunch with the MOT Crew

They are the greatest bunch of crunchers on the World Community Grid
(I promise if you can suggest a better way to describe a gaggle of totally dedicated number crunchers I will use it )

Roll on tonight's stats update (We've been sabotaged by IBM as Tuesday is usually our best day)

Dave

...............................................................................................
..............................
Date....................Runtime...................Daily Points.......Daily Results.........Place
09/20/2005..... 0:296:13:34:28.......... 179,052................ 785................. #5
09/19/2005..... 1:063:19:40:58.......... 254,384............. 1,266................. #5
09/18/2005..... 1:036:12:54:53.......... 241,518............. 1,295................. #4
...............................................................................................
..............................
Stats-Image Generator by WCG Member Börni

.................A BIG ”My Online Team” WELCOME.................
.......................to our newest team member! ......................
................................ Alison Prior ............................

.__ Alison Prior .__

________ for Joining our Top 3 Team
__ All Our New Members - Please - Get Your Team Banners
___For all the Members who would like a Signature Banner in the Messages
_ Click Here and please follow the Instructions: My Online Signature Banners
__ Click Here for instructions to get the: Stats Generator in your Signature
_____Please check your Member Profile at: My Online Team Website
-Click to Email David Autumns our Captain with any Info you would like added

If you don't like long posts just jump down to the three statements at
the end of this post...

Hi David,

There is only one core but in HTT CPUs many CPU ressources are
doubled so given the fact that a CPU is only PROCESSING 30% of the
time it's doing other things 70% of the time. Doubling the ressources
needed for these 70% will increase overall throughput of the CPU if
there are parallel tasks. I showed a trick to get 1.8* the performance
in a pure crunching environment from a single CPU.

It's not only trickery, it's real physical redundance built onto the chip.

That's not true from my point of view. You can run a "tool" (VMware)
which is a WINDOWS program. It emulates one or more PCs on one
physical PC which you can (almost) use like real ones. You can set up
two virtual machines, install the client and get two new devices for WCG.
You don't need two user accounts, these are two new PCs! They do not
share memory, hard disk space or anything else the client needs to
crunch -- only CPU ressources and network card are shared. And as
with the network card (used at different times) the CPU ressources are
also used (often) at different times.

OF COURSE the virtual machines do more work (research) than the
initial PC as the CPU is crunching only 30% of it's time and shifting
around data all the other 70% (see Intel and others on the HTT concept).
The real processing core is IDLE quite often and HTT makes the two
virtual machines to better use the idle time of the processing unit. This
is clearly proved by the CPU score which is (correct me if I'm wrong)
not calculated but measured by repeated benchmarks!! If I get a CPU
score of 155 for the REAL WINDOWS SETTING and 134-135 for each
virtual machine (and these are benchmark results) then this is clear
proof of the concept! We are lucky enough to run an application (WCG
client) that seems to be PERFECTLY suited for being run in a HTT
environment.

Not really -- I just didn't want to wait for an unnecessary limitation
to fall. The limitation that doesnt't allow you to run more than one
client instance although many PCs have so much more idle ressources
on their machines. A Windows tool allowed me to do it.

I showed my system, the concept is NOT limited to this system:

1) OS change is not necessary, VMware for Windows is available just
like for Linux. Both work the same -- it's just that I run Linux as my
main OS. This is no limitation of my proof of concept.

2) I'm using 4 virtual machines (to get 3.55 x the output of the Windows
setting) because I have TWO CPUs!!! Quite a few systems now come
with Dual CPUs but that's not the point. I proved that you get 1.8
times the WCG processing power by using HTT in EACH CPU, it also
works for 1 CPU and only 2 virtual machines!! At least you have
enough RAM to drive both machines to the max -- I didn't so I "only"
got 1.42 times the processing power.

3) The XEON thing is only for 1 purpose: you need special Pentium
CPUs for DUAL boards, you can't use the regular ones. No limitation
of my concept.

4) My XEONs have only 1MB cache. Not more than the regular high-end
ones. When HTT is enabled the cache is shared (divided) among the
two pseudo-cores.

So if you have one or more HTT CPUs on your board or even one or
more of the new dual-core CPUs: try VMware and see that the second
client gives you about 80% more crunching power (less if you have to
lower memore assignment). Run two machines per physical CPU.

If you try it on a non-HTT CPU (or HTT is disabled, look at the BIOS settings) you can NOT win anything (nothing can run simultaneously
on those CPUs) -- you will only lose due to the switching overhead.

I'M SORRY IF MY POSTS ARE THAT LONG.

What I wanted to say is:

- it can run on YOUR MACHINE if you have at least one HTT CPU

- you don't need Linux, you only need VMware (free test for 30 days)

- it's not a point trick, it's really working faster

I wonder if CompuDude accepts the CPU score as proof of concept for
crunching advantage by using HTT -- I don't see any other reason for
my observations.

Best, Stefan.