need CPUs for Dell 4500

[johnk320]

I need a 1.0 and 3.0GHz CPU for my Dell 4500 (which runs a 2.0GHz CPU) to do a science experiment for my son. We want to run a recursive algorithm (tower of hanoi) on a consistent platform, changing only the CPU, and time the results.

I called Dell and they told me I can only use a 1.7 and 2.8GHz. They also quoted me ridiculous prices.

My question is this: Can anyone direct me to a source for chips of this speed that will work with this board?

Your help is appreciated.
_________________
Thanks,
JohnK

[Google]

[fussnfeathers]

Unfortunately, Dell is correct. Your motherboard will only support 400/533mhz P4 procs on the Intel 845E chipset, and only at speeds from 1.6ghz to 2.8ghz (1.7ghz only for a Celeron).

The only 1ghz procs Intel made were P3's and Celeron B's (I believe), which will not fit your motherboard (you have a Socket 478, the 1ghz proc you want is a Socket 370).

Your board also doesn't support HyperThreading, so anything over a 2.8ghz is out of the question, as is anything in the "C" range of procs.

Processors are expensive.......and depending on the one you asked about, Dell might not be that far off. You're looking at between $175-$200 for a 2.8 proc that will run in your machine, if you can find one.

If you were to want to run an even test on your system, I'd buy a 1.6, a 2.2 (or close to it) and a 2.8, for an even spread. Expect to spend upwards of $500 to do that, though.
_________________
Big enough to scare you

[johnk320]

fussn,

Thanks for the info. It doesn't sound like there would be a single m'board that could accomodate my requirements?
_________________
Thanks,
JohnK

[fussnfeathers]

Nope. What you're asking for is three completely different motherboards, chipsets, and sockets.

You could come close, though.......and honestly, if I were trying to show how technology has gotten faster while simultaneously smaller, I'd use three machines. On P3 1ghz at its best, one Williamette based P4 (say 1.5ghz), one P4 Northwood, and one P4 Prescott. That's Intel only, of course.

Tell me exactly what you're trying to do, if I don't have a workable idea, then somebody else here probably will. Or we'll put our collective brains together to come up with somethng...........TN'ers love a challenge.
_________________
Big enough to scare you

[johnk320]

fussn,

I truly appreciate your ( and anyone else who might jump in on this) help with this.

What we were hoping to do is this: Use one computer. This eliminates variables due to m'board mfgs, different brands memory, bus speed, etc. three CPUs (1, 2 and 3 GHz), and one algorithm (in this case the tower of hanoi). We start by using the computer equipped with 256MB RAM and the 1.0GHz CPU. We start the algorithm using 50 disks and time it from start to finish 3 times and find an average(theoretically, it shouldn't be any differernt between runs). This sets our baseline for the rest of the experiment.

Next, we switch out the CPU for the 2.0 GHz. We run the exact same test and compare the results. Theoretically, the time should be half of the base line results.

Finally, we swap out the 2.0 for the 3.0 GHz CPU and run it all again. Again, theoretically speaking, we should find that these results are one-third the speed of our baseline results.

The reason I want to use CPUs at each different level is that there is a clear delineation that should be evident in the times. I could probaly do the experiment using a 1.7, 2.0 and 2.8 CPU as Dell suggested and adjust the results mathematically but It would be tougher to see a difference especially between the 1.7 and 2.0 speeds. Again, the reason I'm so stuck on using a 2.0 CPU is that is what's currently in my system.

The application in real world usage would be for those users who need to process large amounts of data quickly and may be wondering if they can justify the cost of upgrading.

That's it in a nutshell, guys!

thanks in advance for any input...
_________________
Thanks,
JohnK

[fussnfeathers]

There's already a ton of websites that compare proc to proc variations. There's alot more variables than you think..........

Windows is not a static OS, so each time you reboot, the memory, hard drives, IRQ assignments and the like all react slighly differently. Figure on a midrange system, you should allow for a =/- 2% change per reboot.

One other thing to consider is the differences in procs. It's not always just the core clock speed that changes, it's alot of other variables as well. Number of stages in the pipeline, L2 (and even L3 cache on some procs) vary from 256k to 2mb, FSB speeds vary from 400mhz to 1066mhz (higher in the case of AMD FX procs).

Also bear in mind that what you want to do is one-sided......you're not comparing AMD procs as well, which run at lower core clock speeds, but can outperform an Intel at nearly 1ghz faster.

Couple that with the fact that you're using last-gen hardware, and you're not really accomplishing anything. Hardware like processors change in rather dramatic ways with each new generation, and not just in clock speed. For example, a 2.8ghz Northwood will spank the heck out of a 3.06 Williamette, because of the larger FSB, smaller die size, and increased on-die cache.

While I think it's a cool idea, and a great one for a science experiment, you're leaving out too many variables that are more important to upgrading than just the number on the box.

If you asked me, I would say this................is a purely core clock speed upgrade worthwhile? No. While you will see some performance gains, it won't be a whole lot........within the same family of processor (say, upgrading from a 1.6ghz Williamette to a 2.4).

On the other hand, is upgrading from a last-gen proc to a new, latest release proc worth the money? Yes, yes, yes.........you'll see a dramatic performance increase going from a 2.0ghz Williamette to a 2.8ghz Northwood, and even more going to a Prescott core with the 1066 cache. And that's just within Intel.

Also figure that the newest processors being pushed by system makers are dual core, which adds another level to the whole "should I or shouldn't I" scenario.

What I would do, if I were you, would be to find people with three similar systems, but based on three different generations of proc and RAM. Find an older 1.4ghz that runs on DDR133 with a 64mb vid card, with similar RAM capacity and a good HDD, a Northwood based 3.06 with a 128mb vid card, DDR 333 or 400, etc, and a new, midrange machine with, say, a Prescott 3.2ghz or so, DDR2 566, blah blah......basic midrange setups. Run your tests on those three machines, and compare the results. Remember that if you're doing this to prove to businesses or non-gearheads that an upgrade is worthwhile, a very large amount of those people won't be bothering to upgrade just a part, but will buy a whole new machine instead.

Doing it this way (sortof a group effort, and I'm sure some of us here would be willing to help out.....we loooove benchies!) you could also toss in AMD based machines, of similar vintages, to level the playing field.

Give it some thought, and let us know.
_________________
Big enough to scare you

[2old2care]

And John, (welcome by the way)
If you were to use "B" version northwoods in that dell, the results you seek will be nearly linear based on clock speed because the processors would be the same.
Similar to taking a 2200b and over and under clocking it from 1.6 to 2.6.
_________________
.
Liquid-Cooled Q9450 and an EeePC
.

[fussnfeathers]

It'd still be a bit skewed, if he were looking for a purely upgrade-based result. (Underclocking a proc allows it to run smoother due to decreased thermals).






Oh, yeah,







Welcome to TN. I got a little excited.
_________________
Big enough to scare you