Page 11 of 11

Posted: Fri May 20, 2005 10:46 am
by foetz
R-ten-K wrote:
foetz wrote:
R-ten-K wrote:
foetz wrote:so if the altix can get somewhere near the gene with 1/3 cpu count it's the better machine.


Nope....


why?


Because Columbia will not be anywhere near what the BlueGene/L at LLNL offers flop wise, so that was a moot point.


no doubt, gene has the max peak but nobody denied that.
anyhow it does not mean that the gene is the 'best' machine.

Posted: Fri May 20, 2005 11:54 am
by R-ten-K
The thing about the Blue Gene is that it offers a density of computation that is very very very hard to beat. Both Columbia and Blue Gene are geared towards similar algorithms, so there is no clear design win from Columbia's more flexible communication/memory hierarchy in each of the members of the cluster when it comes to compare it against BlueGene.

In any case, Blue Gene L provides 100+ Tflops in less space than columbia :(

Posted: Fri May 20, 2005 2:59 pm
by quasi
R-ten-K wrote:The thing about the Blue Gene is that it offers a density of computation that is very very very hard to beat. Both Columbia and Blue Gene are geared towards similar algorithms, so there is no clear design win from Columbia's more flexible communication/memory hierarchy in each of the members of the cluster when it comes to compare it against BlueGene.

In any case, Blue Gene L provides 100+ Tflops in less space than columbia :(


Not all algorithms are the same. Not every problem can be divided into small bits for MPI to shuffle around. Or not every one has the man/hours to divide a problem into small bits.

Another adventage for the Altix is the overall memory. Columbia has 20TB of RAM compared to 8TB for the Blue Gene (it's not publisized, but you can add it up, 512MB/node, 2CPU/node 32768 CPUs). And Columbia is no where maxed out, by putting the maximum density RAM, you could go 24 times the RAM.

Everything is relative on what you want to do with you supercomputer.

Posted: Fri May 20, 2005 7:51 pm
by R-ten-K
The thing is that both Columbia and Blue Gene L are both used as large clusters, under algorithms that are fairly embarrasingly parallel so their algorithmic use is fairly similar. Memory wise, Blue Gene L has actually 16+TB of RAM, you also need to account for the I/O nodes.

Posted: Fri May 20, 2005 9:36 pm
by 87Porsche
I don't like the chocolate Pop-Tarts. The strawberry ones are better.

:)

Posted: Fri May 20, 2005 10:05 pm
by GeneratriX
87Porsche wrote:I don't like the chocolate Pop-Tarts. The strawberry ones are better.


Yup!; pretty much the same! Two kids fighting! :lol: :lol: :lol:

Posted: Sat May 21, 2005 9:14 am
by foetz
should we bet?
how about some quotes? :lol:

Re: :)

Posted: Thu May 26, 2005 3:47 pm
by R-ten-K
Diego wrote:
87Porsche wrote:I don't like the chocolate Pop-Tarts. The strawberry ones are better.


Yup!; pretty much the same! Two kids fighting! :lol: :lol: :lol:


No, it is two PEOPLE having a normal technical discussion... a pretty civilised and normal thing really. People having different points of view, whoah what a concept!

I don't really appreciate being called a "kid" OK?

Re: :)

Posted: Thu May 26, 2005 4:25 pm
by GeneratriX
R-ten-K wrote:I don't really appreciate being called a "kid" OK?


Sorry, was not my intention to offend any of both; really. But It's just that these angle of discussion SO MANY TIMEs was exposed here in Nekochan.Net, and the only thing that changes actually are the posters, but always are saying the same things...

"that", and only *that* is the thing that many times exposes the people posting those things as being kids fighting by different brands of theirs dady's car, or who can piss to longer distance. None other thing. I'm not implying that the discussion is with a low techinical level, or even a low intelectual level!; Nothing of this!

I'm not questioning the quality of the posts, OKAY?
Cheers! ;)

ATI

Posted: Thu May 26, 2005 9:19 pm
by GeneratriX
Well; if we can back to the topic, I'm interested to ask if anyone here in the forum currently has tested the new "Silicon Graphics Prism Deskside", as far as I know, some companies are buying the new workstation from SGI, to accomplish several kind of tasks.

So, the cover of the SGI main site shows a few of them, where I'm specially interested on the following:
SGI and VRSim Announce North American Reseller Agreement and Debut VRSim Mock-Up on Silicon Graphics Prism

Anyone here was trying at some point the FireGL Graphics Card X2PRO-256, or X3-256 ? :roll:

Please, I want just objective results, benchmarks, or any kind of relevant tests against other graphic subsystems from other actual vendors.Please, avoid any kind of off-topic flames about MIPS Vs. Intel, or ATI Vs. VPro. I already know what is IRIX, MIPS, and CRM or VPRO, so, just avoid the distractings about "Coneheads Vs. Roundheads", since we already have enough of this! Another thread could be created to discuss topics as "BlueGene L Vs. Columbia"! ;)

So, who truely knows here how good is This compared to previous generations of ATI Graphics Adapters?

Thanks in advance!
Cheers! ;)

Re: ATI

Posted: Thu May 26, 2005 10:01 pm
by unixmuseum
Diego wrote:Well; if we can back to the topic, I'm interested to ask if anyone here in the forum currently has tested the new "Silicon Graphics Prism Deskside", as far as I know, some companies are buying the new workstation from SGI, to accomplish several kind of tasks.

I have played with one a little, especially with the IRIX binary x-lator... Looked to work OK to me with a visualization application (although not extremely compute intensive, but it was still an IRIX binary as this particular application does not run on Linux). I'm going to get a loaner to further test with different IRIX-only applications... I am still questionning the reasoning behind an IA64 desktop as I don't think there are that many desktop type of applications that support IA64, so we might end up running Open Office for IRIX through the x-lator :lol:

I think a better solution for my needs is an Altix and other deskopts (Tezro maybe, or dual 600 Octane2)...

Re: ATI

Posted: Fri May 27, 2005 2:29 am
by GeneratriX
unixmuseum wrote:I have played with one a little, especially with the IRIX binary x-lator... Looked to work OK to me with a visualization application (although not extremely compute intensive, but it was still an IRIX binary as this particular application does not run on Linux). I'm going to get a loaner to further test with different IRIX-only applications...


Hi UM! ;)

okay, so, you're the guy. Let us know then what do you think about their global stability, and specially the graphic subsystem stability under stress conditions. I'm really interested. Even more: use the Quick-Transit translator as you have done! :D

...I know that these software does not converts a Prism physically on MIPS/IRIX box, but seems that really goes closer as possible.

unixmuseum wrote:I am still questionning the reasoning behind an IA64 desktop as I don't think there are that many desktop type of applications that support IA64, so we might end up running Open Office for IRIX through the x-lator :lol:


Well, I can't mention a single one reason of the top of my head now, but if they are really speedy, that could be one. But you are right about the applications, there is only a few certified applications currently. That's the point of my whole "back to the topic", I want to know how strong is the new platform to re-think about ports for my current developments.

unixmuseum wrote:I think a better solution for my needs is an Altix and other deskopts (Tezro maybe, or dual 600 Octane2)...


I'm really inloved of Tezro, and seems a really great tool, in fact a dream box for any desktop... I don't think I could change a Tezro by any other thing on this computers universe today.

In fact my O2/600MHz performs so well, that when I think in four 1GHz processors running inside a bigger blue toaster my face presents a smile gesture very difficult of define! :lol:

(By the way, you have P.M.! )
Cheers! ;)