Other than game server hosting?

Well, it's useful for regular hosting... but, I don't think it's particularly needed in most cases.

I've heard P3's and athlons outperform P4's for regular hosting. What is everyones take on this?

Originally posted by AcuNett
I've heard P3's and athlons outperform P4's for regular hosting. What is everyones take on this?

Well, that seems to be what people heard, but where are the facts/benchmarks/etc (P3 vs P4)? Of course, for the hosting industry, we are not interested in the 3D benchmark nor Office 2000 benchmark and the likes. I just can't imagine how apache or mysql will perform slower under P4 than under P3. And I won't believe if people say P3 is a better alternative for a webhost than a P4. Give me some facts (not hearsay) and I will believe it.

And without any benchmarks nor facts, you think P4 is only better for game server? :rolleyes:

BTW, those hearsay are probably outdated already; it may be the case that P3 was performing better than P4 but perharps because there was no optimized OS (linux, FreeBSD, etc) for P4 yet. But I would be surprised if under current versions of OS, P3 will still perform better. I will be amazed ....

There was a thread on it recently. P3 does outperform some P4's, but it depends on many other factors. Otherwise, not much difference between Athlons and P4's as far as I see.

Summary: I think lately we are seeing far more processor power and cache than most people need. I also think people are too concerned about this. Some celerons and durons work great!

There's much more then just MHz in determining a faster processor. (Intel with the P4 doesn't want you to know that, almost all of this crowd does though)

P4s run hotter then P3s... this makes them less suitable for a hosting environment. Other then that, with the MHz lead they have over P3s, they're even better at the much higher MHz. (the 512kb cache versions)

Now for XPs... they run cooler then older Athlons but still hotter then P4s. So they're even less suitable for a hosting environment when not cooled properly. (which I'd imagine is very hard in a 1u case.) They can do more calculations in the same amount of time then a P4 though, so they are faster at the same MHz. I don't know of any "web hosting" benchmark programs, but overall other benchmarks prove this. So Just consider an XP at it's AMD rating when compared to a P4.

The P4 will help on servers which process lots of data, e.g. many MySQL processes.

Clearly, the intent behind the P4's design was to improve multimedia performance. It was also intended to improve multitasking. These enhancements can only boost overall performance.

I'm not certain that heat is a real issue with the P4. The .13 micron data path was put there to alleviate heat (P3's is .18).

The word on Intel chips is that you really can't judge true performance on the early stampings because they're not produced by the "real" production facilities. A couple months after initial release, the chips are made at the regular production lines that have better lithography, etc., and are more representative of the product. This has been true since the very first Pentiums.

*rememberin' the good old days with the P2s... .35 microns n lovin' it... lol

Originally posted by netdude
*rememberin' the good old days with the P2s... .35 microns n lovin' it... lol

at the time also i was unable to see tracks of processor :(

Actually, I think Cyrix with 1024 cache's are best. :)

I've got a P4 w/rackshack...there 600GB deals a while ago....it runs a ton better than their Durons. I have a p3 1.13mhz too, and the P4 seems to handle better. I continually pump out about 1.2mbs, peak at around 2.4mbs and the load stays < 1.0 usually

Hello

Everyone :) Please bother yourself and see benchmarks for the internet jobs on Intel's own site.

Their own benchmarks show higher performance for P4.

Mac

Originally posted by Get-Hosted.com
P4s run hotter then P3s... this makes them less suitable for a hosting environment.

Which P4 and which P3? I am not sure about P4, but I do think P4 has a newer 0.13 micron technology which will allow you to have higher clock and less power consumption. Which P3/P4 are you comparing?

dude... do you really think intel's benchmarks on internet performance would be really valid when considering the processor for hosting... unless they designed the processor for hosting, i really doubt the benchmarks would be really valid...

come on, they made the chip for the consumer level, the benchmarks would reflect that... (consumer applications)...

Doesn't P4 have that netburst or is it burst.net technology? :)

Is that like deuwwwwed, man? WTH is a dude anyway... a new kind of bologna sandwich? ..and while we're at it, tell me... Why do you always find beer whereever you find poor people?

:eek:

I think p4's are pretty good for servers, but that's just me!

i have one p4 1.7ghz from rackshack
its prettty good

i also have another server that is 2ghz p4 -- works great too

I've said it before, I'll say it again... P4's (in none Xeon version) cannot compete with P3s.

We upgraded a P3 to a P4, very few more sites have been added to it, it pushes around the same amount of traffic and the CPU reports loads of at least double what they were on the P3. I know this isn't concrete evidence due to the lack of figures - but from personal experience, this is the case.

Originally posted by voxtreme-matt
I've said it before, I'll say it again... P4's (in none Xeon version) cannot compete with P3s.


If you're using SDRAM with a P4 and not DDR with proper chipset, you've basically got the throughput of a P3. (133mhz RAM vs. 533mhz RAM).

I don't dismiss what you're saying, but I'd like to see your specs and performance numbers. (But I'm sure that's confidential info - no problem.) Just based on the design specs, the P4 has to be at least *somewhat* better than a P3 in a hosting environment.

You can't give much credit to manufacturer reviews/benchmarks. Everyone knows that the manufacturer always bumps things up a notch or two. You have to read the real world tests on some of the other boards to get the truth.

The P4 uses less power than the P3. I've built a lot of both and the P4 runs hotter than the P3, just as they both run hotter than the celeron, which in my opinion is a much more stable cpu and has a longer lifespan in the workstation enviroment than the P3 or P4. Maybe I'm wrong, but I think in a normal server enviroment, the celeron would last longer than the others. Mainly due to the cooler temps.

When running RDRAM with the P4 in a pc, there is no way for the P3 to compete and at a minimum, you need to be running DDR memory or you might as well have a celeron. Using PC-133 memory with a P4 defeats the purpose of having the P4. I have found that a good, inexpensive solution is a 1.6 P4/ASUS P4S533 or SHUTTLE MS51N/512MB Samsung PC-2700. The shuttle board has Audio/Video & LAN.

The other thing to remember is... Unless you get Xeons, I don't believe P4s are SMP capable. Which means that if you ever decide to upgrade to a dual-processor box, you'll have to get all new processors, as opposed to popping in a second processor. I got a dual Athlon system when P4s were starting to come out, as my choices were AthlonMP, P3, or P4... P3s didn't do DDR, and P4s couldn't go dual... I'm quite happy with dual Athlons. (They, of course, generate a lot of heat... Throw some huge heatsinks on 'em, and fill 'er up with the biggest, most powerful fans you can. :) )

From my experience, my P4 server (with DDR RAM) performs faster than my P3 server. So if I'm given the choice between P4 and P3 servers with quite similar pricing and specs, I'd definitely choose P4 over P3 for hosting.

Originally posted by voxtreme-matt
I've said it before, I'll say it again... P4's (in none Xeon version) cannot compete with P3s.

We upgraded a P3 to a P4, very few more sites have been added to it, it pushes around the same amount of traffic and the CPU reports loads of at least double what they were on the P3. I know this isn't concrete evidence due to the lack of figures - but from personal experience, this is the case.

We run mostly all p3(1-1.4Ghz) servers and decided to try out the p4s a few months back. We had higher loads on the p4s(freebsd 4.4. loads were very jumpy) but we had heard about this from other people also. BUT data was still served correctly and without hesitation, what we did notice was when it came time to process logs and other large files. The P4 whooped the P3s ass. As for them both being plenty of power, yes they both work and we have no preference a few seconds on a logfile wont change the world.

let's not forget there are two types of caches available for the p4, 256KB L2 cache and 512KB L2 cache. you'd see a performance increase with the latter on the p4 :cool:

Voxtreme's "real world" experience seems very valid to me. I'd take a lot of stock in a host's experience over any benchmarks.

BTW, I'm told DDR is faster than RDRAM by CompUSA techs. I don't see how though, with the fastest DDR-PC2700 running at 333MHz, and RDRAM running at 800MHz.

The other thing to remember is... Unless you get Xeons, I don't believe P4s are SMP capable. Which means that if you ever decide to upgrade to a dual-processor box, you'll have to get all new processors, as opposed to popping in a second processor.
But if in the future you decide to upgrade from single to dual, by that time the $ value of the CPU you have is very little, maybe between $100-200 which is small compared to the cost of the motherboard, probably new RAM, etc.