Just to up the ante for anyone who is shopping around for a new proc/mobo combo, here is an article reporting on the new core-duo performance from the Washingtom Post. The story refers to a recent article in PCWorld.


Intel's New Core 2 Duo Processors


*can't get BBCode to stick :(

Intel's New Core 2 Duo Processors

Maybe this will do.

And Yes My next cpu is an X6800 and the motherboard is an Asus P5WD2-E Premium. It should do nicely.

If you're interested, X-bit labs has a new review of the new Conroes on an Asus P5W DH Deluxe motherboard. They mention the Core 2's lower latency resulting from the new shared cache.

---

Just to up the ante for anyone who is shopping around for a new proc/mobo combo, here is an article reporting on the new core-duo performance from the Washingtom Post. The story refers to a recent article in PCWorld.


Intel's New Core 2 Duo Processors


*can't get BBCode to stick :(

Intel's New Core 2 Duo Processors

Maybe this will do.

And Yes My next cpu is an X6800 and the motherboard is an Asus P5WD2-E Premium. It should do nicely.

---

My Amazon Wishlist
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

Just to up the ante for anyone who is shopping around for a new proc/mobo combo, here is an article reporting on the new core-duo performance from the Washingtom Post. The story refers to a recent article in PCWorld.


Intel's New Core 2 Duo Processors


*can't get BBCode to stick :(

Well just to terminate my original question that started this thread, I wound up buying instead of building. I got a Dell Precision M65 laptop with a dual core Intel 1.67GHz & 2G RAM and it is running circles around the 3.4GHz single CPU Intel at my office. They both get about the same amount of SETI computer time. The laptop seems to be outperforming the desktop computer at an unrealistic rate but there it is, I cant say why. (excuse the lack of apostrophes, my home computer is being weird when I type one). I also had to hide my computer stats (sorry) because the computer names started showing up and had my companys name in them which I dont feel comfortable showing -- too bad, I dont know why SETI@home decided to do that.

FYI, your computer names and IP addresses only appear when you're logged into your account. Other users can only see the computer ID # when we look.

Alinator

That M65 looks neat. Good choice.

I'm trying to decide on a computer also and I've narrowed the choice between the HP dv5000t and the Dell Inspiron 6400. I was wondering If I should go Intel or AMD (it has to do with the 64 bit future issue). The HP is availible in a AMD Turion version if I choose to go with 64 bit AMD.

---

Well just to terminate my original question that started this thread, I wound up buying instead of building. I got a Dell Precision M65 laptop with a dual core Intel 1.67GHz & 2G RAM and it is running circles around the 3.4GHz single CPU Intel at my office. They both get about the same amount of SETI computer time. The laptop seems to be outperforming the desktop computer at an unrealistic rate but there it is, I cant say why. (excuse the lack of apostrophes, my home computer is being weird when I type one). I also had to hide my computer stats (sorry) because the computer names started showing up and had my companys name in them which I dont feel comfortable showing -- too bad, I dont know why SETI@home decided to do that.

---

I wonder how you optimize e@h. Do you have an "reoptimizer" program to optimize as you wrote? What does it do to a binary executable and to what degree does it optimize? It seems to optimize Windows programs.

I did it by hand, for fun.
Used programs: Borland's Turbo Debugger and Flat Assembler (both are free).
..snip...

Hi Akos,
It sounds very interesting to know hand optimization is effective on E@h! I once tried hand assembly with sse2 instructions in a function of seti@home (non-enhanced), but it's slower than than the compiler's optimization. It has turn out to be because inline assembly disables "inline function" Intel compiler automatically applies, and the effect of inline function was bigger (maybe because P4 has a long pipeline)!

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

Hi Tetsuji, ML1 and the others...

I wonder how you optimize e@h. Do you have an "reoptimizer" program to optimize as you wrote? What does it do to a binary executable and to what degree does it optimize? It seems to optimize Windows programs.

I did it by hand, for fun.
Used programs: Borland's Turbo Debugger and Flat Assembler (both are free).

I am asking this because usually source level optimizing with optimizing compiler has much advantages, because compilers can use efficient vectoralization, and IPO to eliminate inefficient call/jmp.

The sourcecode optimization should give the best performace improvement, of course.

And I think it's difficult to analyze it in the binary level using "reoptimizer".

It's true, but it would be a funny program.

Could it be added as a post-processing step on the GNU gcc compiler?

I think it would be possible.

Best,
Akos

Toms Hardware have just put up another comparison article:

AM2: AMD Reinvents Itself - Conclusions

This compares the new AM2 socket CPUs against Intel's present offering.

And this reveiw looks at Intel's upcoming offering.
The next 6 months or so are going to be very interesting.

---

Grant
Darwin NT

Toms Hardware have just put up another comparison article:

AM2: AMD Reinvents Itself - Conclusions

This compares the new AM2 socket CPUs against Intel's present offering.

At the moment, looks like AMD offers the best performance/heat and keeps the lead on highest performance. Meanwhile, Intel's pricing undercuts AMD on the mid-range CPUs.

One thing to note on the (extensive) discussion on there about what DDR2 bandwidth is achieved, is that the AMD CPUs are nothing like as sensitive to FSB speed as is Intel's P4 series. Hence why AMD has had no great need to use DDR2 thus far. Look instead at the overall system performance for the task you need to run.

Happy crunchin',
Martin

---

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

I saw that the current code isn't too optimal for AMD, but i'm thinking on a 'reoptimizer' program (for all type instructions) that can eliminate some disadvantages of the compilers.

Hi akosf,

welcome to s@h!

I wonder how you optimize e@h. Do you have an "reoptimizer" program to optimize as you wrote? What does it do to a binary executable and to what degree does it optimize? It seems to optimize Windows programs.

I am asking this because usually source level optimizing with optimizing compiler has much advantages, because compilers can use efficient vectoralization, and IPO to eliminate inefficient call/jmp. And I think it's difficult to analyze it in the binary level using "reoptimizer"....

regards,

-Tetsuji

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

hey! double post!!!! please ignore! Sorry for your inconvenience.

---

Luckiest in the world. WMD = Weapon of Mass Distraction.
Click this table.

I vote for Intel.

So do I, but then they give me a paycheck twice a month. :-)

---

I vote for Intel.

Looks like the question comes down to heat & motherboard costs.

Also, whether you are building a dedicated s@h cruncher or whether you also require your machine to be general purpose.

My view is that AMD have a very good lead for general computing. The AMD systems are also very good when you need to "move data around" rather than concentrating on just numerical processing.

Happy crunchin',
Martin

---

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

Welcome to seti, btw :o)

Are you going to spread your magic here, too?

Your reoptimizer project sounds interesting. Will it work on already compiled binaries?

It's only a plan at moment.
It would work with PE header EXE binaries.

Could it be added as a post-processing step on the GNU gcc compiler?

(And congrats again on your Einstein trip!)

Regards,
Martin

---

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

I vote for Intel.

Welcome to seti, btw :o)

Are you going to spread your magic here, too?

Your reoptimizer project sounds interesting. Will it work on already compiled binaries?

It's only a plan at moment.
It would work with PE header EXE binaries.

After googling for a while I found a (not very detailed) image depicting this:

...

I think that picture is illusory.

This one is a bit closer to the reality.

Time will tell.

Welcome to seti, btw :o)

Are you going to spread your magic here, too?

Your reoptimizer project sounds interesting. Will it work on already compiled binaries?

Regards Hans

---

After googling for a while I found a (not very detailed) image depicting this:

...

I think that picture is illusory.

This one is a bit closer to the reality.

hmm..either way the unit of "clock" in Core (solo/duo/2) is different from P4 (core is much slower than P4), and overall performance is necessary. Anyway I am satisfied my current machines, two P4's and one celeron :)

IMHO, on the controversy on Intel vs. AMD, as soon as one buys a machine, one will lose to the late buyers. Such is a competition of high-tech machines.

I assume that 'Core' will be the best core for SETI for a longer period (more than a year), so i will buy a Conroe.

But as for optimized applications of s@h(_enhanced), Intel has advantage because of Intel compiler and IPP, I think. This is important on boinc.

I saw that the current code isn't too optimal for AMD, but i'm thinking on a 'reoptimizer' program (for all type instructions) that can eliminate some disadvantages of the compilers.