P4 FACTS

[Guest]

http://www.sharkeyextreme.com/hardware/guides/pentium4/2.shtml

for those of you who would like to get trueful factual info about the P4 and not rumor, or lame online reviews who do not compare the p4 code optimizations on the p4
using dd3d 8a etc


the PEntium 4 is so different and so much faster,
but you have to throw big complex tasks at it to show this

for example, note launches and performs the same on a P2 400
asn it does on a p3 or AMD 1000
but we all know the 1 ghz CPU's are faster,
just becasue a simple program like notepad, or a loop unoptimized benchmark does not show this does not mean
what we all know is true the 1 ghz CPU's are faster.

SAME WITH THE P4

running office apps, or simple tasks will not saturate the CPU\memory pipe of the P4 like it does with P3
so you have to start at the top, and say copy a 50 meg file, and render a 3d image in the background while being on the internet and encoding a MP3 file

that takes bandwidth and power, and when you do something like that, the P3 and AMD tend to slow down and choke while the P4 executes it handily..


a new set of benchmarks by all the companies are being made
because of the P4 , its bandwidth with rambus,
and the special new code it requires..

hope this is helpful info
CAMERON




CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[spud]

great cyber thats great news i knew about the review but since its not tommy boy approved i decided not to add the testing to my other post but now that you have seen it as well and support it as valid data we can start to fairly compare the new intel chip architecture to the new amd architecture. but as i have mentioned before there is no technical or specification breifs on the amd architecture's so it will be hard to show the facts but at least we can show a better light on the p4 .

SPUD

Just some advice from your friendly neighborhood blue man

 

[Guest]

P4 SUCK. PERIOD!
=)

TOO expensive, TOO slow, INCOMPATIBLE problems with GF3. TOO much BUGS,.... EVEN P3 is better.

 

[spud]

Why dont you look at nvidia site and see who there designed the g3 around http://www.nvidia.com/News.nsf and http://www.nvidia.com/Products/GeForce3.nsf/testimonials.html .

SPUD

Just some advice from you friendly neighborhood blue man

 

[Guest]

BUDDY, GF3 is support both P4(sse, sse2) and AMD athlon(3dnow). no bugs reported for ATHLON(so far) and bugs floating around for P4. MAYBE is driver related, or maybe it's hardware! could be NVIDIA N20 or P4, but since there is no bug on ATHLON working on the same driver, same hardware, wouldn't you think is the P4??? HMmmmmm reall tough to figure.

MORE power to AMD.

 

[Guest]

do have a reliable source/link showing that there is a bug or are you spreading rumors. i'll admit i even posted something the other day about an glitch between the p4 and the geforce3 [oops] but it was on bad info and so i screwed up. in regards to this issue.

<A HREF="http://guru3d.com" target="_new">http://guru3d.com</A>
<A HREF="http://pcrave.com" target="_new">http://pcrave.com</A>


Yesterday we noted a rumor that the GeForce3 is experiencing compatibility issues with Intel's Pentium 4 processor. The lads from PCRave asked Derek Perez (NVIDIA) about his matter and it seems to be not true at all:

Q.Any truth to the rumor of Pentium4 incompatibilities with Geforce3?

A. No truth to it at all. It's incorrect.
dp
Thanks Weazel !

 

[priit]

Damn, last time I checked I could do all that web-surfing, mp3-encoding, file-sharing (over 2 100Mbit NIC's) and bunch of other stuff smoothly without killing the system on my Duron700. No need for X Ghz processor and RAMBUS-memory. Oh yeah, no M$ software was used...

 

[amdchuck]

So what does all of that mean?

Do I really have to do 10 things at once for the P4 to operate effectively?

Does it suck as bad at single task operations as the reviews indicate?

 

[Guest]

YOu can have an opinion, liek they suck in your opinion,
but when you cite facts like their are expensive and slow, then you are just plain saying things that are untrue

the Pentium 4 1.3 is under $400 now, and has gone down faster than any other series of past series CPU in history
less than 1 QTR..

all new cpu's when they come out are expensive first including the AMD's, the 1.2 was $750 when it came out first..

the PEntium 4 1.3 was NEVER $750 it debuted at $550
so actually it has been more reasonable on debut

the are much faster than any AMD, see the web links I provided and check out SPEC .ORG where you can see the P4 outperforms the AMD by 30% ++


there are no reported bugs in teh P4 that are signifficant
aand all CPU's have bugs..
the P4 works with the Geforce 3, I have used them togeher
I helped test one fo them and built a machine for them
together with Ultra 320 adaptec scsi controller
and giga ethernet cards ..

trust me, we would not use them if they were not the best thing out there we do some quality high end stuff

CAMERON

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[Guest]

not and copy a 50 meg file while encoding Mpeg 3 or 4 video

look at your cpu ultizations curves and you will see pentium 4 does it faster with less effort

CAMERON

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[priit]

Those SPEC scores were achieved using Intel's latest P4-optimised compilator that isn't widely used yet, so in today's real-world applications the performance isn't nearly as good... but as you are certificated Intel reseller, you can't actually admit that some way they suck, can yo ?

 

[Guest]

no but good question none the less

what it means is it is obviously much faster doing one thing, but comparing it to an AMD or another cpu that can also do that one thingm does not good as
since that thing is well within both CPU's ability, you see similiar results.

it is when you ask them to do something really heavy
that shows one cpu bogging down while the other reamins steady or increases in speed..

if you had 2 race cars with very fast engines but one
which you believed had 2 times the power,
and asked them to show which is faster in your driveway,
well the would perform similiarly because of the limititaions of the area or test.

race both cars down a mile lane and you can see one pull ahead with a change to demonstrate it power..

that is why they drag race on mile long + strips instead of down the street, so one car can pull a head eventually

that is more what I was trying to say..
the P4 needs alot of overhead or heavy calcs to really shine and show how much faster it is the AMD, as the AMD is fast and only begins to get overwhelmed at fairly high loads
and most reviews do not take this into account..

some of the tests the INTEL, RAMBUS, HP,
and sites like sharkey's and other as well as us,
have seen and tested this ourselves
with 3d programs , photoshop, SPEC.INT,
SYSMARK, 3d MAX,ACAD, MPEG 4,SANDRA 2000,
QUAKE 3, PREMIERE , etc etc

the P4 is so damn fast, we barely have software to really hammer it and give it a sweat, so that is why many benmarks are doing rewrites and loading up their tests heavier.

hopes this explains what I meant..

have agood weekend

CAMERON

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[priit]

Running tasks that are using 100% of processor's time, there really aren't any point looking CPU load IMHO. How does 100% on other processor is worse than 100% on P4? BTW, has RC-5 client already optimised for P4 or is it still running it on 500Mhz Athlon's speed ?

 

[Guest]

www.sharkeyextreme.com/hardware/guides/pentium4/21.shtml

this review was done correctly and shows P4's huge advantage over P3 and AMD
sometimes 2 times faster or 100%

check it out, unbiased and well done..
better site than this actually and less biased with better testing on CPU's and MB's..

some of the things TOM's has predicted and said in the past are WAY OFF, like Rambus would never sell well, it is slower, Intel will lose its lead a year ago,geforce was slower than others, etc..
sometimes we have tested stuff that he says is slower or no different that I know for afact is..
it is not black and white, it depends on high end equipment
you use and how you test..

but the tests I have cited are the most respoected in teh world like SPEC, and SYSMARK, and CADBENCH, and SANDRA,
WEBMARK, etc etc across many platforms including SGI,
SUN, ALPHA, AMD INTEL, and are designed to compare apples to apple even though you have oranges in there ))

enjoy
CAMERON


CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[Guest]

becasue AMD's 100% threshhold hits faster and soon than
P4's that is my whole point..


notepad takes the same CPU load on a P2 as it does on a p4
and is the same speed, yet you know the P4 is faster right..

what makes 100% load on a 200 mhz P2 does not make on a AMD
1000 or a 1500 P4,
you see what I mean?



CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[Guest]

which RC are you referring to?????
C-

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[priit]

AFAIK Intel targeted P4 into home computers not servers, so heavy multitasking isn't where P4 should shine (and I really doubt it really shines at all). Shaky's site has always been heavly biased, sharky's Durons are as slow as Celeron's too

 

[priit]

Under RC-5 I meant RC-5 64 cracking client ( http://www.distributed.net ) That little program uses mostly integer instructions for numeric crunching so P4's double-bumped integer units should really shine there. But, at least without special optimisation, P4 performs about 10 times slower as it theoretically should...

 

[amdchuck]

gotcha, so following your car theory, the P4 is basically a P3 but with more torque?

Kind of like the difference between a pick up truck and a tow truck, while both may match speeds with no extra load, once you put some extra weight in the mix, the tow truck obviously has it's benefits.

I think I understand but it still begs the question

Is the P4 any faster than the P3 or Athlon chips when running single applications?

 

[Guest]

NO NO,

INTEL started marketing P4 for workstations and home computers at first because they could not deliver enough chips for servers until now..
Pentium 4 will go into server and replace Xeon P3,
they are waiting for the server chipsets that is all...

I do not believe SHArley's site to be as biased as TOM's
or ANDNAD sites, as they both hate INTEL for personal reasons that have been well documented by some people

CAMERON


CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[Guest]

YES AND NO

the PEntium 4 is less like the P3 than the AMD almost..
the PEntium 4 radically deviates from P6 architecture,
where P3 does not, and it is the most ground up new CPU since the P6 from the P5

the PEntium 4 is much faster when running single apps when those apps have been compiuled for the P4's more efficient architecture and sse 2 and concurrent instruction pipes..

try windows ME with DD3d 8a on some games compiled for P4,
or try photoshop 6 or a 3d rendering program..

programs are faster when they hammer the P4 like even
P3 heavy FP intensive stuff like rendering, Mpeg etc.

and having OS and API like dd3d 8a makes P4 even faster,
and having the actual program compiled for P4 makes it faster still

so with P4 against say P3 with no optimized software whatsoever, IE WIN 98, dd3d 7 and a benchmark or APP
p3 compiled, there would be only about a 25% speed increase
or ABOUT exactly linear to the increase in MHZ from
P3 1000 to P4 1500

NOW if you add P4 OS aware drivers and code like ME,
and add dd3d 8a you get about a 30-50% gain
and
if you add the actually recompiled application for P4
you would get another 10-30%..
see what I mean... it is a multi step process you
can see gains with one, but together they really smoke

cameron

CYBERIMAGE
<A HREF="http://www.4CyberImage.com " target="_new">http://www.4CyberImage.com </A>
Ultra High Performance Computers-

 

[amdchuck]

Thanks, that is helpful.

Too bad you have to use that joke of an OS to fully utilize the P4

 

[priit]

Yeah, recomplie all your software so that it would run on Alpha or SUN and see how it will fly Does M$ really makes open-source software, so everyone can compile their own windowzes whenever they like and Intel shares their compilers and development software absolutely free? Like has already been proved, running typical x86 code, P4 stinks (doesn't matter is it FP or integer-intensive). That SSE2 doesn't substitute all x86 code and some places SIMD isn't useful at all.

 

[Guest]

PROOF,.... here,
http://www.reactorcritical.com/index.shtml#l658

"Voodoo5, which was physically incompatible with some of the Pentium4 boards has almost dissapered from the horizon, but it looks like we have a really nice "follower" for it... GeForce3. According to eyewitnesses, a Pentium4 system with GeForce3 is quite unstable, in some cases, even dynamic test could not be passed. Thus some of the manufacturers may delay GeForce3 cards launch and some of them has even declined to make them. We don`t know what is the reason of such behaviour of the card (drivers or a card itself), however, this fact is very annoying.

Probably, NVidia has hastened with the announce of NV20 - it was necessary to wait some time to reinsure from this kind of incidents. It is very funny - a week ago NVidia has informed about Pentium4 and SSE2 optimization for the GF3 GPU, and now it becomes clears that GeForce3 cards can`t work in Pentium4 system properly... I understand, that actually, the drivers, not cards, are optimized, but it it is great when the card simply works It seems, that there is at least one reason not to send GeForce3 samples for rewievers... "



DUDE everyone know that the P4 got bugs all over, even INTEL knew, but hey, IT a new product, it's bound to have some flaws,...... one example(so you guys won't think i'm bias) AMD 760 chipset, that new, they got to implement DDR, well,... IT's work then it had bugs, then it didn't work,... bla bla bla, NOW it work but still some tiny bug. U get the point....... so what do you DO???? FIXit. FIX the damn thing, the x86 compatible too.

AND to THIS CYBERguy, what the HELL are you talkig about FOO,..... first off I DON'T know how much the P4 1.3 debut for, so you say $550. OK $550. DO you know THAT INTEL brought out the P4 1.5 FIRST then the 1.3(to TRY to compete <PRICE wise> with AMD) so guess how much P4 1.5 debut for, i'm GUESSING <emphazise on GUESSING> around $700-1000. HAh,..... and you think your CPU is expensive.

BACK to the P4 1.3 debuing at $550 again. THAT like AMD saying OK, here you go we have a TB 1.2(100mhz) already for on the market for around $250, and i'm going to debut a TB 1Ghz(133mhz) for $200. hey LOOK a brand spanking NEW TB 1 GHZ debuing for $200 buck! omg that so cheap. OMG. what an IDIOT!

AND buddy, over at CYBERfreak. how much was you system configure with the SCSI, I just built my TB 1.2 w/ Ultra and IDE RAID 0 config, for less then $2500 that PLUS 19in MONITOR, PLUS KLIps 4.1.... and believe me those aren't cheap PARTs. and you know what, since i don't buy them in QUANTITY where it might be cheaper, I think i did alright. AND i believe my ide RAID would be right up there with the single SCSI HD. ohhh forgot, SCSI, how much did the SCSI COST. hey I'm not saying they aren't good, they just too expensive for me to play games on.

DUDE go back to HIGH-school where you can actually learn some stuff.

WTF, WHY am i writing this anyway, I don't care. I already did my research and I already got my Computer. What do i care. OK, here you go to make all you INTEL fan HAPPY, INTEL p4/RAMBUS system is so awesome,.. it beat the hell out of AMD in Memory bandwith. theere, anyway, I won't be posting anymore on this Thread, I will be checking out what other people have to say though. OH well, chow,.... GO INTEL. ya right. YAH go straigh down drain where it belongs.

UPDATE.....................

Oppppsss MY BAD turn out the incompatiblility w/ GF3 it's all fix out, CHECK this P4 1.5 is the highest 3dmark score on GF3. see I'm not posting unknown facts. BUT that at 1.5Ghz, and tested bird was runing at 1.2ghz. DON't wana say no more. <P ID="edit"><FONT SIZE=-1><EM>Edited by ehartono on 03/12/01 07:04 AM.</EM></FONT></P>

 

[Guest]

If you want a balanced review of P4 go to Ace's Hardware and read the technical evaluations (parts 1 & 2).

by the way ... I don't understand your statement about unoptimized loops not showing the increased speed of faster processors ... loops such as the following simple example DO run faster on faster processors!!

double ct, max;
time_t start;

main()
{
for (start = time(NULL),ct = 0.0, max = 1000000000.0;
ct < max; ct++);
printf("Time taken = %ld\n",time(NULL) - start);
exit(0);
}
Of course this does not excercise ALL the power of the processor and/or chipset but it is a basic indication.