Tetrium wrote:

Edit:In addition to the P3-450, there was in fact also a P2-450 made 😉
Dunno what exactly the fastest Northwood was, I try to evade the s478 Preshots as they were hotter and slower then Northwood at the same frequency

Fastest Northwood was 3.4ghz- near or on the limit of what Northwood architecture could max out to. Fastest Prescott was 3.8ghz in s775 form factor.

sliderider wrote:

You do know that there is more than one 3ghz P4 and some of them sit on a 400/533 bus, right?

Actually, no. P4 3.0ghz chips were only available in 800mhz bus speeds- Northwoods and Prescotts. The 400mhz bus P4s max out at 2.8ghz, the 533Mhz bus chips at 3.06ghz, whether NW or Prescotts. So a 3ghz P4 was always a 15x 200mhz chip.

sgt76 wrote:

Fastest Northwood was 3.4ghz- near or on the limit of what Northwood architecture could max out to. Fastest Prescott was 3.8ghz […]
Show full quote

Tetrium wrote:

Edit:In addition to the P3-450, there was in fact also a P2-450 made 😉
Dunno what exactly the fastest Northwood was, I try to evade the s478 Preshots as they were hotter and slower then Northwood at the same frequency

Fastest Northwood was 3.4ghz- near or on the limit of what Northwood architecture could max out to. Fastest Prescott was 3.8ghz in s775 form factor.

sliderider wrote:

You do know that there is more than one 3ghz P4 and some of them sit on a 400/533 bus, right?

Actually, no. P4 3.0ghz chips were only available in 800mhz bus speeds- Northwoods and Prescotts. The 400mhz bus P4s max out at 2.8ghz, the 533Mhz bus chips at 3.06ghz, whether NW or Prescotts. So a 3ghz P4 was always a 15x 200mhz chip.

Please tell me what it says here.

http://www.cpu-world.com/sspec/SL/SL6YH.html

Type CPU / Microprocessor
Family Intel Pentium 4
Part number RK80532PC080512
Processor markings 3.00GHZ/512/400
Frequency (GHz) 3.0
Bus speed (MHz) ? 400
Clock multiplier ? 30
Package type 478-pin FC-PGA2
Socket type Socket 478 (mPGA478B)

Architecture / Microarchitecture / Other
CPUID 0F27h
Core stepping C1
Processor core Northwood
Manufacturing technology (micron) 0.13
L2 cache size (KB) ? 512

and here

http://www.cpu-world.com/sspec/SL/SL74Q.html

Type CPU / Microprocessor
Family Intel Pentium 4
Part number RK80532NC080512
Frequency (GHz) 3
Bus speed (MHz) ? 400
Clock multiplier ? 30
Package type 478-pin FC-PGA2
Socket type Socket 478 (mPGA478B)

Architecture / Microarchitecture / Other
Processor core Northwood
Manufacturing technology (micron) 0.13
L2 cache size (KB) ? 512

And for all practical purposes, 3.06ghz= 3ghz

Yeah I was gonna say, there is 400fsb Northwoods clocked at 3ghz. I see them pop up on ebay now and then for around 50-80 bucks or so.
http://cgi.ebay.com/Pentium-4-P4-3-0-3-0Ghz-5 … =item27bcbbc599

ProfessorProfessorson wrote:

Yeah I was gonna say, there is 400fsb Northwoods clocked at 3ghz. I see them pop up on ebay now and them for around 50-80 bucks or so.
http://cgi.ebay.com/Pentium-4-P4-3-0-3-0Ghz-5 … =item27bcbbc599

Very expensive because you need one to max out a 400 fsb motherboard. I can remember a time when I was looking for a 3.06ghz for my 533 fsb machine and couldn't get one for less than $100.

ProfessorProfessorson wrote:

Tetrium wrote:

but there was also a Coppermine 1133 made, though it was barely any faster (never saw one in the flesh btw).

I remember the Coppermine 1133 mess. When I started going on different hardware forums a lot back in early 2002 or so they were still talking about that flop of a cpu and its stability problems and crap. They had to recall those cpus. Tomshardware took a lot of flak over that since they were the main one to report on the problems with the cpu initially, and Intel started acting like a dick to them before finally admitting there was a major problem and issuing a recall.

Intel rushed the Coppermine 1133 out the door so they could say that they had the fastest CPU after AMD beat them in the 1ghz race. That chip was just an overclocked 1ghz model and that one was already reaching Coppermine's thermal limits. That's why Coppermine was replaced, because it couldn't go any faster without stability issues. Tualatin only bought Intel a little more time to develop a new architecture that could reach higher clocks.

Mixing EDO with SDRAM was only a temporary situation, while waiting for the new SDRAM (although it seemed to work just fine) With EDO RAM only I received about the same results with the K6 as with EDO mixed with SDRAM. EDO RAM speed was quite a bit faster on my P233 MMX (220 MB/sec vs 170MB/sec now with the K6), which seems kind of odd to me.

Anyway, I've now installed 2x128MB SDRAM modules 😁 There is some strange thing going on with the timings though: every now and then on a cold boot the BIOS automatically reverts to CAS3 setting. When I manually change it back to CAS2 it boots just fine and is also completely stable (Memtest86+ tested). Already tried to restore BIOS defaults, but it keeps changing it automatically 🙁 Same thing happens with another set of 2x128MB SDRAM and on FSB66 (instead of FSB75). All modules are CAS2 according to the specs.

On CAS2 setting I now get a mem speed of 190MB/sec in SpeedSys, so quite an improvement there 😀
Currently benchmarking with one V2 removed and tomorrow I will receive a PCI SATA adapter. I'm curious if I can get my old SSD working with it 😜

UPDATE: Results on 3DMark 99 MAX with one V2: 2500 3DMarks, 6400 CPU 3DMarks. Barely a difference from V2 SLI. And since I've only got one game that supports 1024x768 on 3DFX, I've decided to stay with one card if I can get the PCI SATA adapter to work properly.

mOBSCENE wrote:

Anyway, I've now installed 2x128MB SDRAM modules 😁 There is some strange thing going on with the timings though: every now and then on a cold boot the BIOS automatically reverts to CAS3 setting. When I manually change it back to CAS2 it boots just fine and is also completely stable (Memtest86+ tested). Already tried to restore BIOS defaults, but it keeps changing it automatically 🙁 Same thing happens with another set of 2x128MB SDRAM and on FSB66 (instead of FSB75). All modules are CAS2 according to the specs.

On CAS2 setting I now get a mem speed of 190MB/sec in SpeedSys, so quite an improvement there 😀
Currently benchmarking with one V2 removed and tomorrow I will receive a PCI SATA adapter. I'm curious if I can get my old SSD working with it 😜

Sounds good. Let us know what happens.

sliderider wrote:

Intel rushed the Coppermine 1133 out the door so they could say that they had the fastest CPU after AMD beat them in the 1ghz race. That chip was just an overclocked 1ghz model and that one was already reaching Coppermine's thermal limits. That's why Coppermine was replaced, because it couldn't go any faster without stability issues. Tualatin only bought Intel a little more time to develop a new architecture that could reach higher clocks.

Odd in a way though that the P6 architecture managed only about 1 Ghz while Palomino (made on the same process) reached over 1.6Ghz. Tualatin only made it to 1.4Ghz while K7 reached 2.2Ghz at the same die manufacturing process. Perhaps this is related to how little heat the P6 core emitted. If a CPU die can't loose much heat through the heatsink (like Coppermine for instance), it'll reach it's thermal limits sooner. Netburst was kinda the opposite, release a LOT of heat so it can clock higher.

I know it sounds weird what I just wrote, but it makes sense to me in a way. K7 always ran much hotter then P6, but also clocked higher.

mOBSCENE wrote:

Anyway, I've now installed 2x128MB SDRAM modules 😁 There is some strange thing going on with the timings though: every now and then on a cold boot the BIOS automatically reverts to CAS3 setting. When I manually change it back to CAS2 it boots just fine and is also completely stable (Memtest86+ tested). Already tried to restore BIOS defaults, but it keeps changing it automatically 🙁 Same thing happens with another set of 2x128MB SDRAM and on FSB66 (instead of FSB75). All modules are CAS2 according to the specs.

You can get CAS3 PC-133 and run it @CAS2 PC-100 spec. I don't know if TX supports 256MB modules (or 128MB Single sided ones), but perhaps the TX chipset can't cope with so many banks running at such sharp timings?

You could try with just 1 module, just for testing for a while if the problem persists 😉

@ mOBSCENE

Have you tried testing 1 stick at a time of SDRAM at cas3 then cas2 timing and benchmarking. Maybe more issues with the chipset caching area? Nice thing about SDRAM is it didn't have to run in pairs 😀

I have a Socket 370 board that will recommend cas3 timings over cas2 when I force it in the bios. It won't change it on its own though. It was common for manufacturers back in the day to promise the hardware could do better than it would really do and could be the bios programmed to play it safe for stability.

@Tetrium

The Tualatin was killed off by Intel because they didn't want it to compete with the Pentium 4 line since they were being produced and sold at the time period. There was nothing stopping them running it up to 2GHz or more. As results would show that a Tualatin could beat a Williamette clocked around 400Mhz higher. Netburst was rushed out anyways to compete with the athlons and didn't compete well until the northwoods showed up. The Tualatin as well didn't support higher bandwidth for DDR and Rambus which didn't help its case.

The Pentium 3 1.13 coppermine was just OC'ed from factory as Stated. TomsHardware and HardOCP found it to not run properly then Intel recalled the CPU as to why you never see one in the flesh. The coppermine used aluminum interconnects which did limit clocking a bit before the dieshrink.

Celeron was a line of its own and I do own a few of them. The celeron 1100 coppermine core, 1300/1400 Tualeron. 366 Celeron Mendocino and a 500MHz.

Edit: Didn't mean to break the OPs topic with my post, sorry.

ProfessorProfessorson wrote:

Yeah I was gonna say, there is 400fsb Northwoods clocked at 3ghz. I see them pop up on ebay now and then for around 50-80 bucks or so.
http://cgi.ebay.com/Pentium-4-P4-3-0-3-0Ghz-5 … =item27bcbbc599

Hmm, I always thought the 3ghz 400fsb P4 was a mythical part- one you hear of but practically unavailable. Looks like I stand corrected. 😊

RogueTrip2012 wrote:

@Tetrium […]
Show full quote

@Tetrium

The Tualatin was killed off by Intel because they didn't want it to compete with the Pentium 4 line since they were being produced and sold at the time period. There was nothing stopping them running it up to 2GHz or more. As results would show that a Tualatin could beat a Williamette clocked around 400Mhz higher. Netburst was rushed out anyways to compete with the athlons and didn't compete well until the northwoods showed up. The Tualatin as well didn't support higher bandwidth for DDR and Rambus which didn't help its case.

The Pentium 3 1.13 coppermine was just OC'ed from factory as Stated. TomsHardware and HardOCP found it to not run properly then Intel recalled the CPU as to why you never see one in the flesh. The coppermine used aluminum interconnects which did limit clocking a bit before the dieshrink.

Celeron was a line of its own and I do own a few of them. The celeron 1100 coppermine core, 1300/1400 Tualeron. 366 Celeron Mendocino and a 500MHz.

Edit: Didn't mean to break the OPs topic with my post, sorry.

You are right on everything you wrote 😉

I just wanted to point out that for the Tualatin to reach 1.4Ghz, they needed to use the same fabrication process that the Northwoods (3.2Ghz) and Bartons (2.2Ghz) used. The Willamette was good in one thing:Going to 2Ghz using the same fabrication process as Coppermine.

Sliderider pointed out that OC Tualatin usually nets you up to around 1.6Ghz before massive overvolting is required, so personally I'm tempted to see 1.6Ghz as it's "natural ceiling".
Of course Intel could have further refined Tualatin to reach higher clockspeeds still (kinda like AMD did with Thunderbird when they created Palomino:Same process, but higher clockspeeds).

sgt76 wrote:

Hmm, I always thought the 3ghz 400fsb P4 was a mythical part- one you hear of but practically unavailable. Looks like I stand corrected. 😊

Yup, couple that CPU with a nice SDRAM motherboard and you got a real monster hehehe 😜

Still odd I think, because that CPU must be held back by it's low FSB considerably?

Tetrium wrote:

Yup, couple that CPU with a nice SDRAM motherboard and you got a real monster hehehe 😜

Still odd I think, because that CPU must be held back by it's low FSB considerably?

On a very modest motherboard with no voltage tweaks or pci/agp locks, using a stock alu heatsink and PC133 CAS3 SDRam I got 2.1ghz stable from a 1.8ghz s478 Willamette- very easily too I'll add despite the shitty stuff I had to work with.

With some better supporting hardware, I reckon at least 2.3ghz stable would have been doable- for the timeframe we're talking about (2001)- >2ghz was legendary (inefficient clock-for-clock Netburst pipeline or otherwise).

That 400fsb 3ghz P4 is stupidly priced but I can see where it can be useful, i.e. in s423 systems using a Powerleap PL4 - 3ghz in a year 2000 system - w00t!

PENTIUM 4 DISCUSSION MOVED TO HERE, Thanks to sgt76 😉

Now...Socket 7!! 😁

I should really fire up my own s7 rig again and give it a spin 😁

Tetrium wrote:

PENTIUM 4 DISCUSSION MOVED TO HERE, Thanks to sgt76 😉

Now...Socket 7!! 😁

I should really fire up my own s7 rig again and give it a spin 😁

Mind listing your S7 rig?

What was the best chipset for S7?

RogueTrip2012 wrote:

Tetrium wrote:

PENTIUM 4 DISCUSSION MOVED TO HERE, Thanks to sgt76 😉

Now...Socket 7!! 😁

I should really fire up my own s7 rig again and give it a spin 😁

Mind listing your S7 rig?

What was the best chipset for S7?

I posted it earlier, but can't remember the exact thread. Was "Post pics of your retro rigs" iirc. I'll post a link later but the specs are:

GA-5AX (the one that caches only 128MB RAM due to chipset bug)
K6-III/400 2.2v desktop chip (so all memory is cached)
TNT2/M64 AGP (only 2D part is used)
Voodoo2 12MB
256MB RAM @ 100Mhz cl2
10-ish GB harddrive
Windows ME
Can't remember what sound card I was using though

I used this rig to play through the entire Unreal campaign. And yes, it's a single Voodoo 2 (and actually my very first 3D-only Voodoo rig 😁 ).

Edit:Lol, I had to backtrack all the way to page one to find it 😜
Anyway, heres the link

....the best S7 or SS7 chipset? I do not think there is one 'hands down' winner at all events. Some can do 133 Mhz officially (but not super stable) like SiS and Aladdin 7. Aladdin 5 and VIA MVP3 are strong competitors, AGP might be better on MVP3 but as for overall overclocking and stability, Aladdin 5 is better. A FiC PA 2013 is devastating with 2 mb level 3 cache IF you manage to seriously overclock the system.
[/quote]

mOBSCENE wrote:

Mixing EDO with SDRAM was only a temporary situation, while waiting for the new SDRAM (although it seemed to work just fine) Wit […]
Show full quote

Mixing EDO with SDRAM was only a temporary situation, while waiting for the new SDRAM (although it seemed to work just fine) With EDO RAM only I received about the same results with the K6 as with EDO mixed with SDRAM. EDO RAM speed was quite a bit faster on my P233 MMX (220 MB/sec vs 170MB/sec now with the K6), which seems kind of odd to me.

Anyway, I've now installed 2x128MB SDRAM modules 😁 There is some strange thing going on with the timings though: every now and then on a cold boot the BIOS automatically reverts to CAS3 setting. When I manually change it back to CAS2 it boots just fine and is also completely stable (Memtest86+ tested). Already tried to restore BIOS defaults, but it keeps changing it automatically 🙁 Same thing happens with another set of 2x128MB SDRAM and on FSB66 (instead of FSB75). All modules are CAS2 according to the specs.

On CAS2 setting I now get a mem speed of 190MB/sec in SpeedSys, so quite an improvement there 😀
Currently benchmarking with one V2 removed and tomorrow I will receive a PCI SATA adapter. I'm curious if I can get my old SSD working with it 😜

UPDATE: Results on 3DMark 99 MAX with one V2: 2500 3DMarks, 6400 CPU 3DMarks. Barely a difference from V2 SLI. And since I've only got one game that supports 1024x768 on 3DFX, I've decided to stay with one card if I can get the PCI SATA adapter to work properly.

Hey. maybe I missed this somewhere being posted but....Looking up some info on S7 chipsets coming across the Intel 430TX chipset shows it only could cache 64MB, anything more would work but show about a 5% overall performance drop. Probably impacts the cas2 timings because all the ram you are running. Might explain why performance dropped even more as you added more than 2x32MB EDO.

Found source here as I just started reading up about. http://www.thg.ru/mainboard/19971201/print.html

@ Jolaes76

Still searching but is there a better Rev. Aladdin V chipset. I see some listed up to a Rev.E?

I already have a Baby AT mobo with Via Apollo MVP3 chipset with 512kb cache. I only want to put a Pentium 1 233MMx and about 64-128MB SDRAM. It appears that the L2 on the later SS7 boards can dictate how much cachable area can be used.

Aladdin V & MVP3 chipsets.
L2 cache = Ram cachable:
512kb = 128MB
1024kb = 255MB

My MVP3 Mobo states support up to 384MB which makes this interesting.

Iirc the VIA's didn't have the same cacheable area bug that the early Ali chipsets had.
I don't know about SiS's cacheable area though. However, the amount of memory that can be cached also depends on how much L2 cache the board has.

RogueTrip2012,

as you do not plan to add more than 128 MB RAM, the cacheable area is obviously will not be an issue. That P1 system is going to be OK.
I was asking around on places with K6-2 knowhow and made dozens of benchmarks, synthetic and ingame, but unfortunately all of them using K6-2+ and K6-3+ processors on my GA-5AX 5.2 (F)...

and yes, revision E or better of the chipset is required only if you use a normal k6-2 /not k6-2+, k6-III/ processor.

@ Tetrium & RogueTrip2012: Thanks for your advices on the memory config. However 1 module of 256MB PC133 CAS2 or 128MB PC100 CAS2 (only double sided supported by my mobo) gives me the same issues with the timings.
I do notice a difference in benchmarks at CAS2 and CAS3, so it's worth switching it back when it's restored by BIOS 😀 About the cacheable area: this limit was removed when I installed the AMD K6-III+ with ondie L2 cache. Mobo L2 cache now functions as L3 cache.

The PCI SATA controller I ordered doesn't have a BIOS, so that won't really work 😜 Ordered another one from eBay which states it should be DOS bootable.

Also decided to give SS7 a try now 😁 Ordered a GA-5AX rev 4.1 off eBay this week. 35 euros including shipping and K6-2 CPU seemed quite reasonable comparing with other SS7 boards on eBay. I've also read positive experiences with this board, including here on Vogons - like on this particular page 😉 Ordered a Voodoo3 3000 AGP to go with it.
I guess my project will never ever be finished after all 😜