VOGONS


First post, by Soap

User metadata
Rank Newbie
Rank
Newbie

So after posting my old system that I brought back to life from my parents loft I completed some benchmarks with my Tualatin 1.4GHz and Celeron 1.3GHz, all that was changed in the computer was CPU and cooler, each benchmark was on default 3dmark settings and GPU running at stock.

Gigabyte GA-60XT intel 815EP
Colorful Geforce 2 GTS Pro AGP 32MB DDR
512 MB PC133 Samsung RAM (2x256MB)
10GB HDD ATA-66
Win98 SE USP 2.1A
230 WATT original PSU

The RAM ran at PC100 on the Celeron due to the 100MHz FSB, RAM timings was
CAS - 3
Tras/TRC - 7/9
Ras to Cas - 3
Ras Precharge - 3

The same RAM ran at PC133 on the Tualatin due to the 133MHz FSB, RAM timings was
CAS - 2
Tras/TRX - 5/7
Ras to Cas - 2
Ras precharge - 2

3DMARK 99 MAX-
Celeron 1.3GHz
7174 3DMarks
16718 CPU Marks

Tualatin 1.4GHz
7411 3dMarks
22450 CPU Marks

3Dmark 2000-
Celeron 1.3GHz
6621 3DMarks
479 CPU Marks

Tualatin
8111 3DMarks
681 CPU Marks

3dMark 2001 (cant seem to find a CPU score)-
Celeron 1.3GHz
3136 3DMarks

Tualatin 1.4GHz
3816 3DMarks

Anyway I just carried out the above out of interest and forgot all about the performance difference as this stuff has been in the loft for nearly twenty years, below I'll leave a link to the system to the post of the system complete with pictures.

Beige-Beast

Think I'll be sticking with the Tualatin for now and I'll try find the 9700 Pro, hopefully it works unlike my ATI R6 SG32M PCI 🙁

Attachments

Reply 1 of 16, by Ydee

User metadata
Rank Oldbie
Rank
Oldbie

Thank You for the comparison, I am thinking about keeping Celeron Tualatin 1 GHz or P3 Coppermine also at 1 GHz, which one will be better for late w98 games? Celeron will be able to overclock to at least 1330 (10x133) and L2 cache have the same, but will that be enough to beat P3?

Reply 2 of 16, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

Long story short - Tualatin Celeron = desktop Pentium 3 Tualatin, but with 100 Mhz FSB.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 4 of 16, by darry

User metadata
Rank l33t++
Rank
l33t++

From my tests here : Re: Tualatin Celeron vs Williamette Celeron

1300 Mhz Tualeron on a BX board (100MHz FSB) with Quadro FX1100 : 7002 3D marks in 3Dmark 2001

1050 Mhz Tualatin-S (underclocked from 1400MHz due to 100MHz FSB, but with 512K cache) on a BX board (100MHz FSB) with Quadro FX1100 : 7124 3D marks in 3Dmark 2001

So everything else being equal (including FSB), a full Tualatin chip (with 512K cache) still beats a Tualeron that is running at a 250MHz faster clock .

Reply 5 of 16, by Ydee

User metadata
Rank Oldbie
Rank
Oldbie

Thank You, darry - I have only Celeron 1000A with Tualatin core and P3 1000 Coppermine, so I choose one of them. I think, at stock clock (1 GHz both) will be slightly better P3 (faster FSB, maybe tighter L2 cache timing?), but after OC maybe Tualeron wins? Have Asus TUEP2-M with i815EP2 (or TUWE-M with i810E2, but without AGP) for Tualatin or some Medion (aka Asus CUV4X-M with Via 694X) for Cumine, all for uATX case, which I want use for it.

Reply 6 of 16, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

Comparisons between Tualatin-S and Tualeron are incorrect. Former has 512 Kb L2 cache.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 7 of 16, by darry

User metadata
Rank l33t++
Rank
l33t++
The Serpent Rider wrote on 2021-06-06, 11:56:

Comparisons between Tualatin-S and Tualeron are incorrect. Former has 512 Kb L2 cache.

Why would such a comparison be incorrect? Is there a law against it ? 😉 I did explicitly mention that I was comparing a Tualeron at a given clock speed with Tualatin-S (and I specifically mentioned its 512K cache) at a lower clock speed with both CPUs running with a 100MHz FSB on identical hardware (same system).

EDIT: Just to be clear, I do know that a Tualeron has 256K cache, as does a non S Tualatin . I thought that was obvious to everyone reading this.

Reply 8 of 16, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

Why would such a comparison be incorrect?

Because Tualeron, Coppermine and desktop Tualatin have identical L2 cache size, which is 256 Kb.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 9 of 16, by Ydee

User metadata
Rank Oldbie
Rank
Oldbie
Ydee wrote on 2021-06-06, 10:41:

I know, that "Tualeron" has disabled Data Prefetch Logic, which is enabled in P3 Tualatin.

I know I don't know anything 😁
It seems that according to this source https://www.overclockers.com/forums/showthrea … ll=1#post499107, the Tualerons also have a functioning DPL, so the performance of Tualeron at FSB 133 MHz should be +/- identical to that of Pentium 3 1,3 GHz...

Reply 10 of 16, by darry

User metadata
Rank l33t++
Rank
l33t++
The Serpent Rider wrote on 2021-06-06, 12:30:

Why would such a comparison be incorrect?

Because Tualeron, Coppermine and desktop Tualatin have identical L2 cache size, which is 256 Kb.

The statement about the aforementioned CPUs all having 256KB is correct, but I don't see in what way that makes comparing these said CPUs to ones with more or less cache improper or incorrect.

Comparing CPUs only has a point if there are differences between the CPUs (architecture, cache size, cache design, clock speed, FSB, etc) being compared, unless the aim is to identify sample variation between identical CPUs (obviously not the case here).

I understand that, depending on what one wants to get out of a CPU comparison, one may want to set as fixed certain variables or even to set as fixed all of them except one.

For example, one may want to

a) compare relative performance of socket 370 CPUs with cache=256KB , FSB=100MHz at various clock speeds

b) compare socket 370 CPUs at the same clock speed with different cache sizes but at the same FSB and clock speed

or any other number of comparisons with any number of variables

This thread did not explicitly set rigid criteria for variables such as cache size, the way I see it. Additionally, a Tualatin-S is still a member of the Tualatin family and is a valid choice for a build, so comparing its performance to its lower cached brethren at fixed FSB is, IMHO a very useful, relevant and valid comparison that could be made even more pertinent by adding to the comparison a non S Tualatin running at the same clock and FSB .

Feel free to disagree .

Reply 11 of 16, by melbar

User metadata
Rank Oldbie
Rank
Oldbie

Thanks for your values , Soap
😀

It gives me a good comparison to my Celeron and additionally the information how much bottleneck'ed my GF3-Ti200 is (compared to your combination , either PIII-S1400 or Cel1300 with GF2GTS)

My values:
Celeron 1000A (@1000Mhz) - GF3 Ti200
- 3DMARK 99 : 7111 , CPU 3DMarks: 14590
- 3DMARK 2000 : 6689

Celeron 1000A (@1400Mhz) - GF3 Ti200
- 3DMARK 99 : 8260 , CPU 3DMarks: 20618
- 3DMARK 2000 : 8647

#1 K6-2/500, #2 Athlon1200, #3 Celeron1000A, #4 A64-3700, #5 P4HT-3200, #6 P4-2800, #7 Am486DX2-66

Reply 12 of 16, by pshipkov

User metadata
Rank Oldbie
Rank
Oldbie

great discussion.

2 days ago i posted info here about direct comparison between "compatible" p3 and celeron processors - coppermine and tualatin models, running at same speed, same cache size, etc.
it gives direct answers to some of the questions raised here.

retro bits and bytes

Reply 14 of 16, by Oetker

User metadata
Rank Oldbie
Rank
Oldbie
darry wrote on 2021-06-06, 11:26:
From my tests here : Re: Tualatin Celeron vs Williamette Celeron […]
Show full quote

From my tests here : Re: Tualatin Celeron vs Williamette Celeron

1300 Mhz Tualeron on a BX board (100MHz FSB) with Quadro FX1100 : 7002 3D marks in 3Dmark 2001

1050 Mhz Tualatin-S (underclocked from 1400MHz due to 100MHz FSB, but with 512K cache) on a BX board (100MHz FSB) with Quadro FX1100 : 7124 3D marks in 3Dmark 2001

So everything else being equal (including FSB), a full Tualatin chip (with 512K cache) still beats a Tualeron that is running at a 250MHz faster clock .

I feel this is actually one of the most interesting tests, as anyone fully upgrading a 100MHz 440BX has to make this choice.

Reply 15 of 16, by Soap

User metadata
Rank Newbie
Rank
Newbie
melbar wrote on 2021-06-06, 13:52:
Thanks for your values , Soap :-) […]
Show full quote

Thanks for your values , Soap
😀

It gives me a good comparison to my Celeron and additionally the information how much bottleneck'ed my GF3-Ti200 is (compared to your combination , either PIII-S1400 or Cel1300 with GF2GTS)

My values:
Celeron 1000A (@1000Mhz) - GF3 Ti200
- 3DMARK 99 : 7111 , CPU 3DMarks: 14590
- 3DMARK 2000 : 6689

Celeron 1000A (@1400Mhz) - GF3 Ti200
- 3DMARK 99 : 8260 , CPU 3DMarks: 20618
- 3DMARK 2000 : 8647

Not a problem I hope the results was of some use, not the most scientific approach by me I literally just thought I'd compare before swapping CPU.

Reply 16 of 16, by Ydee

User metadata
Rank Oldbie
Rank
Oldbie

@Soap:
I hope you don't mind if I add my quick test P3 Cumine and Celeron Tualatin at 1 GHz (and overclocked Tualeron at 133 MHz FSB). Configuration, memory timing same for both CPUs - Asus TUEP2-M (i815EP2), 512 MB PC133 SDRAM, Prolink GeForce2 PRO 64MB 200/200 MHz, some IDE PATA drive 40 GB? (don´t remember brand). As expected, on their base frequency P3 is slightly faster due to the higher FSB, but overclocked Celeron at 133 MHz FSB takes clear victory.

Attachments

  • test.jpg
    Filename
    test.jpg
    File size
    71.16 KiB
    Views
    897 views
    File license
    Public domain