• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Would you like to have a P!!! 512K if the price drop!?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
So...

Well, i do not know anything about the "CPU-Z" Program you are using, but many (previously also very reliable) sources have reported to me, that Intel disabled the Prefetch functionality on Celeron Processors (Not knowing anything about the mobile versions though). I guess, we will not find the answer right now. (Can an inactive Prefetch Unit still be detected? I dunno...)

It's a fact, that the P-III S Processor does outperform most Celeron Processors, not talking about synthetic benchmarks. I know that 3DMarks and Quake III Arena are heavily dependant on GFX Power, but they scale well with faster CPU's too, as any of You might already have known. Of course direct comparisons can only be made with a similar graphics card, but hey: That's why i postet that PCMark2002 Pic too!

Of course the P-III S Processor is positioned in a completely different price range, but i guess that's because it's designated as a Server Class Processor. Please don't forget that. There may be significant differences in manufacturing quality, quality assurance and thermal design. Please note, that the P-III S does not produce any more heat than a Celeron does at the same clockspeed, despite the double-sized L2 cache on P-III S' Die. Let us directly compare the P-III S 1.26GHz to a Celeron 1.2GHz:

P-III S 1.26GHz: 29.5W
Celeron 1.2GHz: 29.9W

As you can see, P-III S runs even cooler than the Celeron, which is slower by 60MHz. Let us take a look at another comparison:

P-III S 1.4GHz: 31.2W
Cleron 1.3GHz: 33.4W

We can see an even increasing difference here! A 100MHz faster P-III S that runs cooler than the compared Celeron by 2.2W! Not a big difference, I KNOW, but that just gives me another argument, doesn't it? ;)

Source: Intel SSPec Sheets!
http://www.intel.com/support/processors/sspec/p3p.htm
http://www.intel.com/support/processors/sspec/icp.htm

Last but not least the choice of a Processor is also a matter of style, isn't it? (That "Not having something that anybody does possess"-Thing ;) ) And affordable server-class "style" is called P-III S, but that part's only my personal opinion of course...

Supplemental: I just forgot about that: MCI/MDT SDRAM is manufactured in Germany in quite low quantities. I was easily able to get a review sample, because i live in austria (quite CLOSE to germany ;) ) Here's Your URL: www.mci.de You can buy the memory directly and online on the manufacturer's website. It's true pc150 CL2 with no SPD errors. Runs at 170MHz 2-2-2 in 90% of all cases!
 
Last edited:
Hey Grand ad,

The reason the PIII-s runs cooler at default volts and hurtz than the cel-t is because it runs at 1.45v instead of 1.475~1.5 of the cel-t and PIII tualatin(non-s).

Hey you can scream till your blue in the face about these points but the main reason the PIII tualatin(s and non-s) gets better 3d scores and what not than the cel-t is becasue it is running at a much higher BUS speed. You need to get a clue bout this we have been researching these things for quite some time now. The only way the cel-t would have a higher IPC than the PIII cumine is because of the prefetch it has. Everything else about it is the same as a PIII except it has a l2 cache latency of 1 instead of 0 and the cumine does nopt have prefetch otherwise they are the same.
 
Ok, let's go further!

The reason that the P-III S runs cooler should in fact be the core voltage!

Well, but at it's defaults, it just DOES run cooler, and that's what matters.

The P-III S performance is higher, because it's FSB is higher and because of the L2 Cache Latency. Well, but at it's defaults, it just DOES run faster!

After all, facts count... Don't You think?

Here's my source on that "No Prefetch Unit" matter, it's computerbase, one of the largest professional european Hardware-Sites, german language:

Intel Processor History Article:

http://www.computerbase.de/artikel.php?id=70&seite=13#fortsetzung_pentium_iii

I'll quote and try to correctly translate:
"Doch nun noch einmal zurück zum Celeron mit Tualatin-Kern. Was unterscheidet ihn genau von seinem großen Bruder, dem Pentium III Tualatin? Zum einen wäre das der geminderte FSB von 100 anstatt 133 MHz und zum anderen eben genau diese DPL-Einheit, die beim Celeron von Intel deaktiviert wurde."

Translation:
"But now let's return to the Celeron with Tualatin-Core. How does it differ from it's bigger brother, the Pentium III Tualatin? On one hand there is the lower SB of 100 instead if 133 MHz and on the other hand there is exactly this DPL (Data Prefetch Logic) - Unit, that has been disabled in the Celeron by Intel"

If you REALLY WISH, I will contact the Computerbase Editor-in-Chief and ask him for his information sources! Maybe we can clean this up... Once and for all.

But again, i would like to remind everybody: After, Performance is, what counts. (Stability too, but all Intel's should be of good production quality)
 
* TASOS *

you may can´t see the higher performance of the P-III S.. well. thats no problem.
simple, run 10 seti workunits and see, how fast the t-celly realy is. btw: seti is only cpu-intensive, not gfx!
the same in 3d-modellers and raytracers.. and sound editing programms.

i said what i said.. and i stand to it.
the problem here is, that BSK21 want´s to know something about the P-III S.. so, I can´t understand, why many people say, he should use a t-celly.
may he buy a celly.. but if he jumps out the window because this cpu is not fast as all said, is this not my fault :D

the fact is, the P-III S is faster than a t-celly in all cpu-intensive apps (all benchies are not cpu-intensive only apps like seti and so on!). trust all P-III S users - i think, no one of them wants a t-celly, cause they know what they want -> CALCULATIONPOWER!

so, may we should compare seti-resaults. these scroes shows the real processing power of a cpu.
 
Last edited:
Hallo Grand Admiral T, Grüße nach Östereich
lipsrsealed.gif

Test for the MCI Memory ( I have also one ):
http://www.hardoverclock.net/html/reviews/mci_pc150_cl2/mci_1.htm

buy here:
http://www.mci.de/cgi-bin/kshop/sho...=0&code4=0&id=730934770370000&Random=77051999
 
Last edited:
Clock for clock/ fsb for fsb the PIII-S is going to be faster but there is no way I was going to run my PIII tualatin @190fsb/1700MHz on my ST6 just to keep up with my 151fsb/1.82MHz cel-t running much more stable. I now run my cel-t at 1.82GHz and yes it does only take a 1710MHz PIII tualatin to compete but like I said that would have been running a 191+FSB. That folks is pushing my barracuda a tad bit. Even 170fsb would be pushing it a little bit much. Plus finding ram that will do 170+ fsb in the states is very hard. I gave up my PIII tualatin 256k cache chip and put it in my brothers computer and run it at 1.25v with low speed 80mm fan and shroud. The hs is only passively cooled.

If you take the PIII 1.2GHz and run the vcore to 1.475v you will run the same watts as the cel-t. Check the data sheets the 256k verion of the 1.2GHz PIII runs the same watts as the 100fsb version of the 1.2GHz cel-t runnign 1.475v. The internals are the same only the cel-t have 256k of the cache disabled. Intel has to do something with the PIII-s versions of the tualatin that only 256k cache worked. They also run the latency to 1 on the cel-t but my PIII tualatin 1.2/133/1.475 ran a latency of 1 also so I think maybe alot of them are like this. My older PIII tualatin ran a latency of 0.

The prefetch is enabled on the cel-t. Where were you like 4 months ago when this was the big debate. It was proven then the prefetch is enabled on the cel-t. If you want to know something very specific call Intel like I have done many times regarding chips. They told me that they did not know if it was enabled or not. THey could not give a straight answer. CPU-Z says it is enabled. If you disable anything CPU-z will notice like if you disable your cache totally so I would guess that the prefetch spec is true. There is no other way to explain the IPC increase of the cel-t over the PIII cumine if indeed the prefetch was not enabled. Anyway we had this debate along time ago and you are now joining in. Well you are late.

The cel-t has prefetch. Just cause one site asys it is disabled is not the gospel truth cause I have seen many sites say that it is enabled. Big deal. It is enabled.
 
How about we do some of the legacy benchies I have here.

Here are some more benchies.

cel-t @ 1.82GHz

--------------------------------------------------------------------------------------------------------------------------------------------

Calculation Mark
12,478

http://www.simtel.net/autodownload....---------------------------------------------
CPU Index
2,689

http://www.simtel.net/autodownload.html?mirror=124&product=38569&key=00efbe9c8b638d67_0600
------------------------------------------------------------------------------------------------------
CPU Diagnose
Whetstone FPU----- 5,121 MWIPS
Dhrystone ALU-----5,538 MDIPS
Multimedia-----------2410 http://www.simtel.net/autodownload.html?mirror=124&product=56347&key=00ef2841a1631bba_f8e9
-------------------------------------------------------------------------------------------------------

FP intens
11,354

http://www.geocities.com/uart0/legacyfp.zip
---------------------------------------------------------------------------------------
CPUmark
150

http://www.pcekspert.com/download/index.html

We could also run some sandra too but you all think that is rigged:rolleyes:

HAHA!
 
Last edited:
How about some passmark? This will show the area's where the PIII-s shines but will be specific..

When you download this zipper run the .exe and then extract the cel-t1.82GHz .pt file over into the performence test folder. WHen done you can run it into your benchmarking platform. Here it is. You can share your benchies by making a zip file with them and posting them. I am not interestin the passmark main score only in certain other benchies.
 

Attachments

  • cel-t1[1].82ghz.zip
    519.3 KB · Views: 34
The PIII-S is faster than the
Cel-T MHz for Mhz. FACT

Cel-T is faster than the Coppermin-256 (PIII). FACT

But ol'man has a cel-t @ 1.82GHz
so unless you have a PIII-S @ 1.5-1.6GHz, the cel-t will be the faster CPU. IMHO
 
Wohoo?

ol' man, your SimTel Links didn't work!

I'll try Passmark now (although i don't now a BIT about that benchmarking tool), and edit my post after completion.
 
Links should be fixed now for simtel.

Nice rio. Pretty much scales doesn't it. WIll be interesting to see what the tualatin PIII will do. I can run my FSB to 133 so we can see a better picture but for memory bandwith dependant benchies I will suffer over the PIII tualatin.

Thing I cannot understand is that I have a ata 100 mobo and ata 100 HD but it does not run at ata 100 speeds. You beat with that and vid card benchies.
 
Well, PassMark seems to be VERY synthetical too... Most of the tests are simple calculations with very little Memory load...

Unfortunately PassMark wouldn't save both results into one file, don't know why... Added both systems to save, but it didn't work. Comparsion is difficult that way, but in most of the tests the Celeron was faster. Well when doing just addition, subtraction and multiplication..... The memory performance seems to be just miscalculated... I'm a bit confused having slower memory than you, although my i815 runs at 170MHz FSB / 170MHz RAM @ 2-2-2... I would NOT trust that benchmark.

My CPUMark Score is 151, quite the same than yours, but only with about 200MHz lower Core Clock Rate! But i don't know what this benchmark really does. It just says "151"! Well, we'll better forget about that one too....

We need something, that puts a high load on both the CPU as well as the memory subsystem without using 3D accelerated graphics...

Guys, synthetic benchmarks just SUCK! I'll not try the other ones, because i guess they're all synthetic too.

How about running some Unreal Tournament at SOFTWARE Mode? That would be a bit MORE interesting than those syntho-Benches (I really hate them, cause they really tell us nothing)

Or let's do a standarized SETI Work Unit, could last a few hours, but who cares?

But no "MIPS / FLOPS" Benchmarks please. No Theory!

Edit: Don't be confused by the ZIP Filename! It's 1.62GHz, not 1.26!
 
Run the mandel and other tests above also please this is good stuff.

There really is not that much difference between the PIII-s and the cel-t.

PIII-s runs 512k cache
cel-t 256k cache

PIII-s runs cache latency of 0 usually
cel-t runs 1~2. I have seen them with a latency of two also folks but only the cel-t 1.3GHz version. All the ones under 1.2GHz should run 1:(

PIII-s has a quad quad word data cache bus
cel-t runs a double quad word cache data bus.

PIII-s runs a 133fsb default
cel-t runs a 100fsb default

These are the only differences and they are not that much. They give the PIII-s usually only about a 0~10% lead over the cel-t.
 
well, Grand Admiral T has right..

however..

P-III S 1.13GHz @ 1.26GHz
147MHz FSB, 256MB MCI/MDT PC150 CL2
Abit BE6-II /w PL-iP3/T
30% standardload, cause my sys act as file-server

my score in pc mark is 116... but.. funny thing .. in legacy FP i got 14771.76 points...
see the difference?
according to the pc-mark i can´t have a higher fp scrore (I repeat: ACCORDING TO PC-MARK I CAN´T GET SUCH A HIGH SCROE IN LEGACY FP!). .that isn´t real.. its synthetic.. and thats the problem.
also I can go higher with the FP score.. becaus with every new test i got 30 to 100points more, so I got a range from 14770 to 15000.
this fact tells me, the bench can´t be real!
 
Mandel is not synthetic it has a actual purpose.

CPU mark uses memory bench alot in it which is somewhat fair.

What is your vid card cause you smoked us in 3d graphics many worlds.

If all we are going to do is run FPS benchies yeah then the PIII-s will win with a 100+MHz slower speed but that is relative to FSB once again. We are not running 170fsb. You got to know that.

All in all the IPC(MOST IMPORTENT PART OF CPU) of the PIII-s is only %10 faster than the cel-t if that(Depending on APP.). That is not much and to use that to justify the $100~$200 price hike I cannnot do.
 
Overclocker said:
well, Grand Admiral T has right..

however..

P-III S 1.13GHz @ 1.26GHz
147MHz FSB, 256MB MCI/MDT PC150 CL2
Abit BE6-II /w PL-iP3/T
30% standardload, cause my sys act as file-server

my score in pc mark is 116... but.. funny thing .. in legacy FP i got 14771.76 points...
see the difference?
according to the pc-mark i can´t have a higher fp scrore (I repeat: ACCORDING TO PC-MARK I CAN´T GET SUCH A HIGH SCROE IN LEGACY FP!). .that isn´t real.. its synthetic.. and thats the problem.
also I can go higher with the FP score.. becaus with every new test i got 30 to 100points more, so I got a range from 16770 to 17000.
this fact tells me, the bench can´t be real!

Dude youy are running a BX chipset with awsome memory throughput.

All the BX mobo's with a cel-t scored high in Mandel compared to the 815e!

Don;t you knwo the BX has a very high memory thoruhg put much better than the 815e.
 
Back