Sign in with
Sign up | Sign in
Your question

Intel should just dump the P4 and go back to P3

Last response: in CPUs
Share
June 15, 2001 4:08:28 PM

<A HREF="http://www.theinquirer.net/15060105.htm" target="_new">http://www.theinquirer.net/15060105.htm&lt;/A>

A dual Tualatin system outperforming a Dual Foster Xeon system? Sources at Intel say so. If Intel had any smarts, they would have designed a modified version of the P3 and used it as their flagship CPU. The Tualatin P3 can hold it's own with an Athlon while a frigin P4 can't. That's sad.

-MP Jesse

"Signatures Still Suck"

More about : intel dump back

June 15, 2001 4:26:30 PM

this is just an indication of the speed diff that the .13um process makes, the Northwood is gonna rock!!!

also this is probably further indication that a dual tualatin sys will blow away a dual Athlon4 setup very easily.

P3/P4, old core, new process = a$$reaming.

"AMD/VIA...you <i>still</i> are the weakest link, good bye!"
Anonymous
a b à CPUs
June 15, 2001 4:31:03 PM

Well Good AMD sucks with there garage processors. Northwood should be called A$$REAMING. Good work buddy.

! Member of the Intel Triple Threat w/ Fugger and AMDmeltdown !
Related resources
June 15, 2001 4:35:51 PM

they've got too much invested in the P4 to drop it now, even if it is an expensive, costly POS.
the only way they would go back to the PIII is if someone purchased Intel and fired management.
sounds like a plan, doesn't it?

----------------------
Independant thought is good.
It won't hurt for long.
June 15, 2001 5:08:06 PM

Okay! We all know the P4 was rushed to the market incomplete. Fine it's overpriced and depends on software optimizations. Lets see what the Northwood does before we decide to kill off the P4.

I dont know if anyone remembers but the original Pentium Pro processors sucked. People laughed at the Pentium Pro when it first came out. Give the P4 a little time and you'll be retracting that statement.

Blah, Blah Blahh, Blahh, blahh blah blahh, blah blah.
June 15, 2001 5:32:47 PM

Quote:
probably further indication that a dual tualatin sys will blow away a dual Athlon4 setup very easily.

The Tualatin is having a hard enough time keeping up with a single T-bird--even when the Tualatin is overclocked. With its crappy SMP scaling (a la GTL+ bus), it's going to lag behind the Athlon even further in dual-CPU comparisons.

Hey, sorry to be the bearer of hard facts.

Kelledin

"/join #hackerz. See the Web. DoS interesting people."
June 15, 2001 5:54:27 PM

Well, we'll never know will we. Intel has already said it's disabling SMP functionality in the Tuluatin.

-MP Jesse

"Signatures Still Suck"
June 15, 2001 5:55:09 PM

LOL- yeah. Hey, feel like taking over a company?

-MP Jesse

"Signatures Still Suck"
June 15, 2001 6:00:22 PM

That's true.. the P6 did blow- but only on Win95. It rocked on WinNT and other 32bit OS's.

-MP Jesse

"Signatures Still Suck"
June 15, 2001 6:05:22 PM

Bah, for whatever that means. Remember, Intel claimed the Celeron was "uniprocessor only," and look what happened there. :wink:

Kelledin

"/join #hackerz. See the Web. DoS interesting people."
June 15, 2001 9:37:07 PM

I completely agree with that, the real P4 should be an ass kicker, but this POS...
at least, as mpjesse said, the PPro worked damn good in server/workstation environments...

----------------------
Independant thought is good.
It won't hurt for long.
a b à CPUs
June 15, 2001 9:53:06 PM

Actually the Tolutin is beating underclocked T-birds in clock-for clock comparisons at 133MHz FSB. Just saw the article myself. Tom's comparison forced the chipset to assign the slowest AGP and memory timings, making it look worse. If you want the direct comparison at the same clock/same bus speed, look around the forums for the link, I don't feel like it.

Cast not thine pearls before the swine
June 16, 2001 4:17:14 PM

You are clueless man...get a life dork!

96.3 % of Statistics are made up.
June 16, 2001 7:35:47 PM

Whats a garage processor? Does that make the p4 a basement processor?

A little bit of knowledge is a dangerous thing!
Anonymous
a b à CPUs
June 16, 2001 8:30:59 PM

it means AMD is making crappy cpus! i could build a better one from scratch (with a dremel multi and som duct-tape)!

AMD sucks dude!
a b à CPUs
June 16, 2001 8:47:56 PM

AT best, it means that Intel put their reputation on the line buy releasing a crippled P4 in an effort to win the MHz race before the .13 micron process was available. What's worse is that they are releasing the 845 chipset for SDRAM when the CPU needs at least DDR and perferably QDR ram. One dumb move after another.

Cast not thine pearls before the swine
Anonymous
a b à CPUs
June 16, 2001 10:13:46 PM

I've been wondering about that SDRAM move, too...

Aklein

It's raining outside, and my lawn has grown a foot overnight!
June 16, 2001 11:09:04 PM

LOL, well I will agree with you on the P4 and I do like the tully. I distinctly remember an argument I had with Raystonn about the p4 Northwood to the exact regards you refer. It appears that upon its release the only motherboard (without RDram) that will be available for it will be the i845 brookdale with Sdram, and I have yet to see any conclusive evidence that suggest the northwood will differ any from the williamette other than a die shrink and the addition of more l2 cache. What ever happened to the crippled FPU getting addressed as well? Even when it gets DDR support this too will be crippled to run pc 1600 untill it gets Intells blessing, more crap to force Rdram done its users throats. I never doubted intells ability to make a good processor, what is in question is there business practices of late.
Now adressing the Tully specifically I still do not agree with you on the memory timings vs. increased FSB & AGP issue, but dammed if I can find any evidence which would prove or disprove either. Whatever the case, the tully is a good processor, it is just a shame it will only see 1.13 and 1.26 models and that you need another motherboard to support it. Maybe intel will change there minds and offer a few more speed steppings as well. There is something awfully odd about the fact that Tom and only Tom has gotten one of these to test and then the way he did go about testing it.

A little bit of knowledge is a dangerous thing!<P ID="edit"><FONT SIZE=-1><EM>Edited by ncogneto on 06/16/01 07:24 PM.</EM></FONT></P>
a b à CPUs
June 16, 2001 11:41:48 PM

Overclocking that far for Tom's test not only reduces memory timings to their lowest settings, but also reduces AGP to 2x. To make a more direct comparison Tom should have underclocked the T-Bird so that they were running the same bus speed/same clock.
I'll wait and see on the revised P4. My computer is already fast enough to last me for years, but I might upgrade anyway next year, depending on what AMD, Intel, SiS, and nVidia are producing. I like to sell my personal systems before thay loose their value, I resell them for what they originally cost me!

Cast not thine pearls before the swine
June 17, 2001 12:14:53 AM

Cas 3 at 133 outperforms cas2 at 100
so then wouldn't cas 3 at 166 outperform cas 2 a 133?

The AGP is another matter, we do not know for a fact that he had to run AGP 2x on this board, this is merly speculation do to the fact you do have to on a I815 Asus board. At worst he was running something like AGP 3x

66.6 x 4 = 266.4 AGP clock at stock FSB
99.9 x 2 = 199.8 AGP clcok at AGP 2x and FSB of 166(+33mhz)
66.6 x 2 = 133.2 AGP clcok at AGP 2x and FSB of 133(stock)

A little bit of knowledge is a dangerous thing!
June 17, 2001 12:17:05 AM

Your a funny guy, give me your address and I will send you a roll of duct tape.

A little bit of knowledge is a dangerous thing!
June 17, 2001 12:24:45 AM

Why does amd suck?

96.3 % of Statistics are made up.
a b à CPUs
June 17, 2001 12:33:41 AM

First of all, check the math. 133 is 80% of 166. Multiply that number by 133 and you get ~107MHz. Cas3 at 133 is about the same performance level as Cas2 at 106MHz.
But overclocking that far also reduces more than cas latency, it also sets the lowest memory interleave, etc. And AGP2x.
As far as I know, this is a basic BIOS feature that affects all boards with the i815 chipset, not just the Asus.

Cast not thine pearls before the swine
June 17, 2001 12:37:42 AM

Well, first of all err my math was of on the AGP bus ( oops my bad)

Should be like @ 166 pci bus is 41.5 so then AGP bus is

2x pci bus( 41.5) x 2 AGP 2x
2x pci bus( 41.5) x 4 AGP 4x

A little bit of knowledge is a dangerous thing!
a b à CPUs
June 17, 2001 12:38:00 AM

Oh, and BTW, overclocking the AGP bus has little affect on AGP speed, I own both BX (89MHz AGP) and i815 motherboards. The i815 gets consistantly better scores in gaming benchmarks. Why? My GTS is limmited by VIDEO MEMORY BANDWIDTH, not AGP bandwidth, I think this is a fairly well know issue!

Cast not thine pearls before the swine
June 17, 2001 12:48:53 AM

Look at it this way.

Take first a stick of SDRAM pc 66 cl2
latency is measured in wait states which can also be measured in ns. If we know compare it to a stick of pc133 in which the frequency is now doubled, that makes each "wait state" exactly 1/2 as long when measured in ns.( you will get two cycles at 133 to one cycle at 66hz) This would mean that a stick of pc66 cl2 would be equal in terms of latency to a stick of pc133 @ cl4.
Now, the increase in bus frequency between the two is 66 hz. Half that and you get 33. Halve the differnce in latency you get 1. Thus, pc100 cl2 equals pc 133 cl3 in terms of latency correct? But, you gained bandwidth when you increased the FSB to 133. Now wouldn't the same apply when you made a simular icrease from 133 to 166?

A little bit of knowledge is a dangerous thing!
June 17, 2001 1:05:59 AM

Quote:
Oh, and BTW, overclocking the AGP bus has little affect on AGP speed, I own both BX (89MHz AGP) and i815 motherboards. The i815 gets consistantly better scores in gaming benchmarks. Why? My GTS is limmited by VIDEO MEMORY BANDWIDTH, not AGP bandwidth, I think this is a fairly well know issue!

No I agree, I only mentioned it it up do to the fact you did. AGP 4x is hardly ever used and does not offer much of a performance increase over agp 2x ( yet)

A little bit of knowledge is a dangerous thing!
a b à CPUs
June 17, 2001 2:08:45 AM

It's hard to do the math directly, since CasLatency only affects LATENCY, or the amount of time it takes to "transfer the first byte of information", a.k.a. "lag" or "access time". It does not affect the steady throughput. So in order to calculate it, you need to know the exact size of each chunk of information being accessed. Since each program is different, it is impossible to make an exact calculation. What can be done it an experiment, using the same program each time. My experiments on 3dMark2000 showed that at around 105-107MHz, Cas2 memory yeilded the same scores as 133MHz at Cas3.
But other memory configuration issues also affect performance, such as interleave. A bios code (at least in Award BIOS) for the i815 chipsets drops all the memory settings to their slowest levels when the bus is overclocked beyond a certain point. This issue has been observed in Asus and a few other motherboards, which leads me to believe it is in the basic BIOS code for the chipset itself, and not manufacturer specific. There is a program designed for the CUSL2 that overrides that feature, but I don't know if it works on other boards or not. It is possible that certain manufacturers eliminate this feature when they modify their own BIOS, but I don't know of any.

Cast not thine pearls before the swine
June 17, 2001 3:24:01 AM

The mobile segment the northwood will be include in mobile.Sdram can cut the price for mobile who cost already $$$$$ DDR for midrange and pc1066 or pc800 for high-end and workstation.
June 17, 2001 3:27:05 AM

I found 2 link 1 that say that 815 chipset will support tuatulin and that prove that tuatulin will have 512 L2 cache and beat T-bird 1.2.
a b à CPUs
June 17, 2001 5:09:38 AM

Yes, I read that article. I think that Intel, if they wanted to sell a BUNCH of CPU's, should release the Tualatin at speeds up to 1.7GHz (if they can) and the new P4 at speed over 2.0GHz ASAP!

Cast not thine pearls before the swine
June 17, 2001 6:31:15 AM

Quote:
Whatever the case, the tully is a good processor, it is just a shame it will only see 1.13 and 1.26 models and that you need another motherboard to support it.

Your kidding right? Where are you getting this info? I can't believe this. Why in the world would intel do this? That's the most limited processor platform I've ever heard of. That basically knocks the Tully off my list.

Quote:
I have yet to see any conclusive evidence that suggest the northwood will differ any from the williamette other than a die shrink and the addition of more l2 cache. What ever happened to the crippled FPU getting addressed as well?

I can't believe this either. Why did I miss this too? I've been waiting to see if they'd fix the P4 and they're not even going to fix the FPU? Is there any hope of the hardware performance increasing enough to make the P4 practical? It seems as if the P4 is depending solely on software optimization.

<font color=red>We are going to have peace even if we have to fight for it. - Eisenhower</font color=red>
Anonymous
a b à CPUs
June 17, 2001 9:29:30 AM

.13 allows Intel to prolong, for a little while longer, the dead end P3. P3 can now max out at 1.4 GHz, maybe 1.5. Athlon will hit 1.8 GHz or higher while still on the .18u process. Northwood is still a P4. P4's suck.

P3 = old core, new process, it's almost dead
P4 = new core, new process, DOA
June 17, 2001 2:10:09 PM

It appears that my idea isn't the worst considering the recent Tuluatin benches posted by AMDMeltDown. The guy is an ass, but those benches seem to be legit. Hmm... i'll wait till good old Tom gets his hand on a 512K Tuluatin before deciding.

-MP Jesse

"Signatures Still Suck"
June 17, 2001 2:44:59 PM

Well good old tom should get some northwood samples too. I cant wait for my system. Northwood is good or on par with the athlon 4 i'll buy it. If its less... I'll go get a Sun Sparc III 733 system or Itanium 800 system.

Nice Intel and AMD users get a Cookie.... :smile: Yummy :smile:
June 17, 2001 3:53:36 PM

What? Your going to learn Unix?

A little bit of knowledge is a dangerous thing!
June 17, 2001 5:40:57 PM

I'm running Suse linux 7.1 right now.

Nice Intel and AMD users get a Cookie.... :smile: Yummy :smile:
June 17, 2001 7:23:09 PM

What the hell are you going to do with a SPARC?

-MP Jesse

"Signatures Still Suck"
June 17, 2001 9:40:52 PM

Or a $1500 Itanium?

Blah, Blah Blahh, Blahh, blahh blah blahh, blah blah.
June 17, 2001 10:10:53 PM

You really dont like AMD do u rcf84???
If northwood is on-par with Athlon 4 why woudl you go buy a Sparc III 733...hell if the Northwood is only on-par with the Athln 4 why would u get the Northwood over the Athlon 4 when the Athlon 4 will cost MUCH less ????

-MeTaL RoCkEr

AMD = Always Making Dough... =)
June 18, 2001 2:29:05 PM

ddr costs the same as sdram juin, go to crucial.com and learn to read.

~Matisaro~
"Friends don't let friends buy Pentiums"
~Tbird1.3@1.55~
Anonymous
a b à CPUs
June 19, 2001 1:16:20 AM

<<<ddr costs the same as sdram juin, go to crucial.com and learn to read.>>>

cl 2.5 ddr is the same as sdram however,
the cl 2 ddr that is used in all those amd circle jerk benchmarks is much more and hard to find.
June 19, 2001 2:08:59 AM

Can we know the reason why he suck.(p4)
June 19, 2001 3:06:32 AM

The current P4 sucks for many reasons. Why don't you learn how to read? You're such an ignorant bastard.

On the other hand, the 2GHz P4 will launch next quarter at $562. There is actually some interesting information in this <A HREF="http://www.theregister.co.uk/content/archive/18979.html" target="_new">article</A>. It should be interesting to see what happens.

Apparently they just couldn't get the 1.13 coppermine out the door. They're going to move ahead with the 1.13 Tualitin.

They also are apparently moving the Tualitin down under the low end Celeron market?!

So where does this leave us Juin, you rocket scientist you, with the setup of Intel's processors?? From what I understand, the Tualitin will take over the Celeron family, the P4 Northwood will handle everything above that, and the Coppermine will be gone. This will probably make it much easier for consumers by the time 2002 comes around, but until then there will be so many chipsets floating around supporting everything from the Celeron, Tualitin, Coppermine, Northwood, and the Williamette that it will suck to upgrade for the next year. Hopefully there will be adapters, but those are just rumors right now.

<font color=red>We are going to have peace even if we have to fight for it. - Eisenhower</font color=red>
Anonymous
a b à CPUs
June 19, 2001 5:19:58 AM

dhlucke, hah, 'you rocket scientist you' -- great stuff.
June 19, 2001 6:38:13 AM

Current lowest prices from first five entries on Pricewatch:

128MB DDR 266MHz PC2100 16x64 2.5V Non-Parity Unbuffered, 184-pin. IN STOCK- : $35
128MB Nanya PC2100/266MHz DDR SDRAM CAS 2: $37
RAMBUS 184Pin RDRAM 128MB PC800: $67 (heheheh...too bad Raystonn's not around anymore...I think he lost a bet!)
PC2100 DDR 256MB 266 Mhz DIMM - (CAS 2 / CL 2) - Gold Leads - Unbuffered 16x64-EEPROM/SPD: $59
256MB Nanya PC2100/266MHz DDR SDRAM CAS 2: $66
RAMBUS 184Pin RDRAM 256MB PC800 256 MB -Limit 5: $143

Wooow...a whole 6% difference between CL2 and CL2.5...not to mention almost twice the price for PC800 RIMMs. And all in the first five listings on Pricewatch!

<sarcasm>As a DDR user, I feel screwed!</sarcasm>

Kelledin

"/join #hackerz. See the Web. DoS interesting people."

P.S. hot damn, I just looked at my own post...yep, the cheapest 256MB DDR DIMMS are CAS2!

Let the "AMD fan circle jerk" begin! LoL :lol: 
June 19, 2001 4:35:39 PM

another retort here...
why would Intel use SDRAM in their notebooks? DDR and supposedly RDRAM(not sure) use lower voltage and consume less power than SDRAM, why limit the CPU by using slower, more power hungry memory in a laptop, where power consumption and heat are a serious issue?

----------------------
Independant thought is good.
It won't hurt for long.
Anonymous
a b à CPUs
June 19, 2001 5:34:22 PM

>I'll go get a Sun Sparc III 733 system or Itanium 800 system.

I just benchmarked my code on an 800MHz Itanium. It sucked wind. My results so far:

Athlon-C with SDRAM
1 GHZ 119.76 sec
1.13 GHz 107.94 sec
1.2 GHz 101.00 sec
1.33 GHz 92.16 sec
1.4 GHz 87.62 sec

PIII
700 MHz 199.30

PIII on serverworks MB
800 MHz 150.86 sec (looks like it likes bandwidth, I'd like to try it on a DDR Athlon)

Itanium
800 MHz 253.55 sec

Alpha EV67
667 MHz 207.79 sec (this system was under heavy load at the time. I'll try to get on again later.)

The Itanium run was compiled with gcc 2.96 (ie. RedHat) with just the -O3 option. Can anybody reccomend better compile options?

I'm still looking for a P4 for completeness sake. I should be able to get on some higher clocked PIIIs later in the week.


<i>Cognite Tute</i>
(Think for Yourself)
Anonymous
a b à CPUs
June 19, 2001 5:40:58 PM

good the tbird needs to be shot. The thing cant beat a p4. Yet it can barely keep up to a p3.

! Member of the Intel Triple Threat w/ Fugger and AMDmeltdown !
June 19, 2001 5:47:34 PM

ergeorge was it done on a 64-bit linux for the itanium.

Nice Intel and AMD users get a Cookie.... :smile: Yummy :smile:
!