G400 AT 133MHZ FSB

[neo]

The point I made isn't that the overclocked AGP bus performs better, just that it can perform correctly if you decide to overclock your FSB past 100MHZ.

I never mentioned anything about underclocking the bus, merely overclocking it.

When my Hercules TNT2 comes in I will be able to provide a few benchmarks on all the different bus speeds available to me. Perhaps then I can try to prove my point. At this point, however, I do not have an AGP video card to test to provide concrete proof. I'll resurrect this thread at that point.

-neo

------------------
Asus P2B-F, 532MHZ Pentium II (4x133 at 2.4v), Global Win VEK12 hs/fan, 128MB Micron PC133, Maxtor 4GB, SB AWE 32, Creative DVD 2x, Mitsumi CDR 2x/8x, Sony Trinitron 17", Old Matrox Video Card, and a redhead with a pair of 36Cs (O/Ced to 38Cs)

[Arron]

Hey Gurm,

remember that the G400 is running its memory at >150MHz and its processor at >125MHz. A bus tranfer at ~100MHz should be no problem for it in theory (and I'm definitely talking out of my theory here, if not out of somewhere else ;o)

I'm also not convinced about your AGP figures. As far as I know the below is accurate:
AGPx1 = 66MHz
AGPx2 = 133MHz (uses both buses at 66MHz, hence the name AGP Sideband)
AGPx4 = 133MHz tranfers on both buses (hence x4 if it uses both buses)

[Gurm]

Arron,

Err... where did you get the whole AGP "both buses" thing? The sidebanding has nothing to do with a separate bus. Honestly, 2x = 66x2 = 133. 4x = 66x4. Really. Truly. They're not changing the AGP bus to run at 133mhz, to the best of my knowledge. Then they'd have to do a div by 4 to get the right PCI bus speed...

- Gurm

[rickT]

gurm: AGP does utilize its own bus directly to main memory and it is called sideband addressing.

[M Ragsdale]

The AGP sideband operates at 66mhz just like the AGP bus does. The difference between 1x and 2x is pretty simple, actually. In 1x mode, it does 1 per clock cycle, in 2x it does 2x per clock cycle (sideband transfers).
It DOES NOT operate at 133mhz.
Matt

------------------
Is this some kind of bust?
Yes, it's very impressive.

[Zyn]

p3-450 @ 600 4.5x133,
133 2/3 = 88 2/3
88&2/3 = 177 1/3

so basically yes, the max can take it..

[FastmanNTW]

As far as I am aware, both 1x and 2x AGP run at the same clock speed.

If you imagine a nice square ware (crap character graphic time!)

''''_ _
_| |_| |_

At 1x AGP, signals are only transmitted on the rising edge of the 66MHz clock signal, whereas at 2x AGP, signals are also transmitted on the falling edge of the 66MHz clock signal (effectively the same as 1x AGP with a 133MHz clock).

I believe the 4x AGP does some additionally jiggery-pokery to allow it to source 4 clear clock signals from a single 66MHz clock, but that's beyond the scope of my knowledge.

FastmanNTW

[This message has been edited by FastmanNTW (edited 07-23-99).]

[This message has been edited by FastmanNTW (edited 07-23-99).]

[Picard]

MAX can handle 133 MHz FSB !

I use a P3-450 overclocked to 600. So the FSB is 133 MHz. The MAX runs pretty cool. Without any problem at all!

I even ever use a 140 MHz FSB. The MAX still runs perfectly. The problem with this FSB is my RAM (PC100, LG 7 ns)). Seems it can't handle FSB higher than 133 MHz. Although the system can boot-up at 630 MHz (140 MHz FSB), the system crashed after several hours of operation. So I put back into 133 FSB (600 MHz).

The best way to know if it works or not is to try it ... :-)

[Wombat]

Gurm and Fastman have it right, I believe.
I don't know for sure, but last I heard, 4x was going to be achieved using rising-,` & falling-edge signals, as well as one on the high level, and one on low.
However, this sounds like as much of a feat as 2x133Mhz. But I can't see a piece of hardware REQUIRING a 133Mhz bus to work.

[Guyver]

133FSB - 89Mhz AGP - Running in 2X mode 100% perfect/stable. No problems.

Guyv



------------------
ABit BE6, PIII-450 OC'd 600Mhz. 128MB PC133HSDRAM, 2.0V, 41C, Matrox Millenium G400 MAX, Adaptec 2940U2W, Quantum Atlas 10K 160/m 18.1GB, Quantum Atlas II 4.5GB, Kenwood UCR-415 True-X 52X SCSI, Matshita DVD-ROM SR-8582, Memorex CRW-2642 CDRW, Iomega 100MB Zip ATAPI, 1.44MB 3.5" Floppy, 3Com/USR 56K Voice Faxmodem Pro, HP DeskJet 895CXi, Creative/Cambridge Soundworks DTT2500 5.1 Digital Surround System, Logitech Marble FX, Microsoft Natural Keyboard Pro (with 2 Port USB Hub)...

[Unam]

I'm running a PIII 600B on my ASUS P3B-F. The PIII 600B runs at 133 Mhz FSB and my G400 seems fine, in some instances I do get visual anomalies that look like interlacing. A good example would be Heavy Gear 2, if you use the sniper option and you magnify the screen image 16X you will see what I mean. However the system does not lock up at all and it has run 3D Mark 99 Max demo for hours with no problems.

------------------
Asus P3B-F1, PIII 600B, 128Mb PC133 RAM, 18.2 KA drive, HP CDRW & Travan drive, SB64PCI, Intel EPro+ PnP NIC, G400.

[mathew]

Well I dont own any 133mhz rated ram but I do have my PC100 Generic Dimm running at 140Mhz, and it has been working at this speed for months.

I have had this memory running at 150Mhz on my BE6 but unfortunatly I cannot utilise the extra cpu speed of 675 Mhz because my G400 Max won't run 3D apps.

It gave me a ZD CPUmark of 52.7 with a L2 cache 2 setting,this againts 49.0 with the ram running at 140Mhz with the same L2 cache setting :P

Here are my system specs:-
G400 MAX
P3 450 running at 630 mhz (140 fsb) 2.0v s-code SL35D Costa Rica (oem)
Alpha P3 125c Copper Embedded heatsink
Abit BE6 Motherboard
128 mb pc 100 sdram (my lucky dimm)
17.3gb Fujitsu h\d Udma 66
21"Sony monitor
SB Live Value
Diamond Super Express 56i Modem
Pioneer 10X DVD Drive
Supermicro SC-750 full tower case

Test Results:
3DMark Result : 6,288 3DMarks
Synthetic CPU 3D Speed : 9,811 CPU 3DMarks
Rasterizer Score : 2,785 3DRasterMarks
Game 1 - Race: 67.8 FPS
Game 2 - First Person: 58.6 FPS
Fill Rate : 269.3 MTexels/s
Fill Rate With Multi-Texturing : 268.8 MTexels/s

2MB Texture Rendering Speed: 396.6 FPS

4MB Texture Rendering Speed: 317.2 FPS

8MB Texture Rendering Speed: 219.2 FPS

16MB Texture Rendering Speed: 165.4 FPS

32MB Texture Rendering Speed: 86.6 FPS

Bump Mapping Emboss, 3-pass: 123.1 FPS

Bump Mapping Emboss, 2-pass: 160.7 FPS

Bump Mapping Emboss, 1-pass: 288.9

So to cap it all off , If you are lucky with your memory-CPU-and other components then you can safely o\c your system without any probs at all.

Oh and my G400 Max is using Agp x1 at a 2\3 divider :P Laterz Matt {:O)

[DuRaNgO]

well I'm running 5x124=620 with 2/3 ratio and agp 2x, yes it works very nicely
jim

------------------
PIII-500mhz @ 620 ! with an Abit BE6 mobo
128mb pc-100
Mill G400 (YAHOO!!!) 32mb @ 167/208 with MGATweak-417mhz, (2.5, 2, 2.5), PD 5.41 & bios 1.5-22
Maxtor 14.3 gb Uata66 hdd
SB Live!
Winblows 98se & DX7
32 lbs. of fans, heatsinks and aluminum ductwork
3DMark Result 5954.69 3DMarks
Synthetic CPU 3D Speed 9325.69 CPU 3DMarks
Rasterizer Score 2791.72 3DRasterMarks
Game 1 - Race 65.07 FPS
Game 2 - First Person 54.89 FPS
Fill Rate 268.79 MTexels/s

[Unam]

Mathew, who made your Generic PC100 RAM? I tried my Micron PC100 at 133 with no luck. I had to go out and buy Mushkin PC133 to run this new processor. BTW I like your numbers, they are even better than mine. I get 5887 3DMarks and 9332 CPU 3DMarks, 2475 3D Rastermarks, 62.9 FPS 1st game, 55.3 FPS 2nd game and fill rates of 246.7 & 243.2. Even when I overclock to 630 I only get 6160, 9786, 2520, 66.1, 57.7, 250.6 and 247.2.

------------------
Asus P3B-F1, PIII 600B, 128Mb PC133 RAM, 18.2 KA drive, HP CDRW & Travan drive, SB64PCI, Intel EPro+ PnP NIC, G400.


[This message has been edited by Unam (edited 03 December 1999).]

[This message has been edited by Unam (edited 03 December 1999).]

[mathew]

Hi DuRaNgO

?Did you have to force your agp to run at X2 or did it just work out that way.I was told that in order to get your agp to run this way you had to run agp 1\1,but when I tried this it would not boot.

Do you see much difference with this setting,some say that it make no real difference!! Regards Matt {:O)