FSB: 66 vs 100 vs 133 MHz \ VOGONS

FSB: 66 vs 100 vs 133 MHz

Topic actions

Post a reply

First post, by eFatal2ty

Posted on 2014-12-02, 22:04

eFatal2ty Offline

Rank Member

Rank: Member
Posts: 145
Joined: 2014-04-09, 19:45
Location: Bratislava - Slovakia

If you have any results to compare performance between different FSB settings you can post it here.
I tested that way:

SiSoftSandra 99 (CPU Arit./CPU MM/Memory)
3DMark 99MAX and Quake2 map demo1.dm2 in 1024x768 full details on V2SLI - driver 3.02.02
3DMark 2000(default settings), Quake3Arena map Four.DM_66 in 1280x1024 full details on GF2 - driver 30.82

on My 9xBEAST machine (ASUS P3B-F, IP3_450MHz, 512M CL2 RAM, GeForce2 Ti 64MB, V2 12MB SLi, SBLive! and MS Win98SE)
- all graphics cards are on default settings, nothing changed in drivers and each test i ran only once -

and results are:

FSB_66MHz=300MHz:
Sandra: 813/399 * 721/616 * 178/165
3DM99: 2825/4615
3DM00: 2222
Q2: 43.6 fps
Q3: 30.9 fps

FSB_100MHz=450MHz:
Sandra: 1224/602 * 1082/926 * 270/250
3DM99: 3530/7037
3DM00: 3386
Q2: 44.2 fps
Q3: 44.5 fps

FSB_133MHz=600MHz:
Sandra: 1625/803 * 1437/1230 * 366/339
3DM99: 3772/9367
3DM00: 4548
Q2: 54.6 fps
Q3: 55.6 fps

*ASUS P3B-F *Intel Pentium!!! 450MHz Katmai@133fsb *Hynix 4x128MB SDR PC133 CL2 *Matrox G400MAX 32MB + Procomp Voodoo2 12MB SLi *Creative SB Live! CT4760 *3Com 3C905C-TX-M *2xSeagate 40GB 7200rpmn *EIZO T68 19"CRT * Creative FPS1000 *OS: MS Win 98SE

Reply 1 of 8, by F2bnp

Posted on 2014-12-02, 23:43

F2bnp Offline

Rank l33t

Rank: l33t
Posts: 3499
Joined: 2007-09-23, 10:19

While these are some very interesting results, I think it would be even more awesome if you could run these tests using different CPUs (a.k.a. different multipliers) at the same clock speed, but different FSB. For example, a lot of the Coppermines (if not all) on Slot 1 had 100MHz and 133MHz variants. The problem would be comparing them to a 66MHz FSB, maybe you could use a Tualatin for that purpose or even a Klamath, which usually have unlocked multipliers.

Nonetheless, good job!

Reply 2 of 8, by Matth79

Posted on 2014-12-03, 00:59

Matth79 Offline

Rank Oldbie

Rank: Oldbie
Posts: 587
Joined: 2014-05-19, 14:24

The overall clock speed has far more impact than the FSB - if that's a 450MHz (100 x 4.5) CPU, then it's underclocked, normal, overclocked.
Now if it allowed a 3x multiplier at 100 and a 2.5x at 133, that would show the difference that FSB alone makes.

Some CPUs, you can lower the multiplier on, but not raise it.

Reply 3 of 8, by eFatal2ty

Posted on 2014-12-03, 13:09

eFatal2ty Offline

Rank Member

Rank: Member
Posts: 145
Joined: 2014-04-09, 19:45
Location: Bratislava - Slovakia

I forget to turn on 32Bit in 3DMark 2000

Reply 4 of 8, by GeorgeMan

Posted on 2014-12-04, 08:04

GeorgeMan Offline

Rank Oldbie

Rank: Oldbie
Posts: 512
Joined: 2013-08-26, 16:03
Location: Athens, Greece

It would be nice as stated already to see the differences of the FSB clearly, which would be obtained only by running the CPU at the same clock and with different FSBs.

The only possible combination at exactly the same clock speed and cache would be:
Pentium III 600EB @ 600MHz (133x4.5) vs Pentium III 900 @ 600MHz (66x9) vs [Pentium III 600E @ 600 (100x6) or Pentium III 800EB @ 600 (100x6)]

There exist 3-4 other combos at roughly the same speed (eg 533EB vs 550 or 733 vs 750 etc), but the only exact one I can think of is the aforementioned. 😀

Nice try btw!

Reply 5 of 8, by eFatal2ty

Posted on 2014-12-04, 14:24

eFatal2ty Offline

Rank Member

Rank: Member
Posts: 145
Joined: 2014-04-09, 19:45
Location: Bratislava - Slovakia

yes, but i have only IP III 450, 500 and 550 and IPII 450 to Slot1 and Celeron 700 to PGA370 at home

Reply 6 of 8, by GeorgeMan

Posted on 2014-12-04, 22:52

GeorgeMan Offline

Rank Oldbie

Rank: Oldbie
Posts: 512
Joined: 2013-08-26, 16:03
Location: Athens, Greece

I know, I also have many cpus from that era, but not the correct ones! 😜

Reply 7 of 8, by Shodan486

Posted on 2014-12-16, 22:23

Shodan486 Offline

Rank Member

Rank: Member
Posts: 153
Joined: 2009-10-22, 21:19

Hey eFatal2ty!

Thought it was you 😁 - I'm still looking for those CL2 RAMs, it's gonna take a while. Apart from that I see some interesting endeavor here - I may contribute with the with that PIII 900 FCPGA mentioned above to further increase the value of this test. PM me on the details.

Regarding the topic idea - the diferrence can be felt, but I've felt it more on socket7 systems. I tend to go above 133 on these systems.

MOBO: PVI-486SP3 Rev 1.2
CPU: POD-83
RAM: 2x16MB
VIDEO: Matrox Millenium 2MB/Voodoo2 12MB/Video Blaster VT300
AUDIO: SB Vibra16 FM
SCSI: 72GB 15k RPM HDD/YAMAHA CD-RW 16x/ZIP drive + FDD drive
NIC: 3Com Etherlink III
PSU: 230W Generic
OS: Win95 OSR2.5

Reply 8 of 8, by eFatal2ty

Posted on 2015-03-02, 03:51

eFatal2ty Offline

Rank Member

Rank: Member
Posts: 145
Joined: 2014-04-09, 19:45
Location: Bratislava - Slovakia

No it wasnt me, i dont remember. Btw that GF2Ti is heavy bottlenecked by CPU isnt it?

Go to top of page Go to top of page

Back to System Specs

Main menu

Common searches