Additional 4-pin 12V plug on motherboard was made for Pentium 4 CPU's (it's even named "P4" on some PSU's) 😀

PS. I think with HT and in modern software, P4 3,0GHz w/HT will be faster than any Athlon XP.

agent_x007 wrote:

Additional 4-pin 12V plug on motherboard was made for Pentium 4 CPU's (it's even named "P4" on some PSU's) 😀

PS. I think with HT and in modern software, P4 3,0GHz w/HT will be faster than any Athlon XP.

That is good to have a modern new psu on an older motherboard that seems to really need power! (P4 at more than 110W sound quiet a lot).

386SX wrote:

That is good to have a modern new psu on an older motherboard that seems to really need power! (P4 at more than 110W sound quiet a lot).

Nah.
It's... quite OK, actually 😀
Intel even did first "double cored" CPU, out of them 😁
"2x90W" - now THAT is a lot of heat.

agent_x007 wrote:

Nah. It's... quite OK, actually :) Intel even did first "double cored" CPU, out of them :D "2x90W" - now THAT is a lot of heat. […]
Show full quote

386SX wrote:

That is good to have a modern new psu on an older motherboard that seems to really need power! (P4 at more than 110W sound quiet a lot).

Nah.
It's... quite OK, actually 😀
Intel even did first "double cored" CPU, out of them 😁
"2x90W" - now THAT is a lot of heat.

Yeah! 😁
And I thought the Barton alerady pushed my patience for fan noise and heat at its limit. 😁

The 3200+ has no chance standing today.. because of missing CPU features, here does the P4 much better. Even if the design sucks..

matze79 wrote:

The 3200+ has no chance standing today.. because of missing CPU features, here does the P4 much better. Even if the design sucks..

I understand it. But anyway trying with a lighter version of Ubuntu (Lubuntu 13.10) it's not running bad at all. Latest one with newer kernel seems a bit too much even if SSD helps a lot. But I need more ram for sure.

swaaye wrote:

I've had the displeasure of using a 2.6 GHz P4 + Radeon 9600 + 2GB RAM on Windows 7 recently and I recycled that machine. 😀 Almost anything you do on a heavy modern OS will max out the P4 for extended periods of time. Windows Updates can take ages.

BTW, with Athlons of any sort you are better off with a Nvidia chipset than VIA. You'll get better performance all around. GUI performance in XP is especially bad.

That problem with Windows 7 taking forever to check for updates went away ever since Microsoft started releasing those monthly update "packs" back in June.

Now, even my slowest full-time Win7 machines (a single core A64 @ 2.8ghz and a Dothan @ 2.72ghz) take less than a minute to find updates each month.

FGB wrote:

Sorry, I can't stand the VIA "bashing", although I always respect personal experience with certain hardware.. […]
Show full quote

Sorry, I can't stand the VIA "bashing", although I always respect personal experience with certain hardware..

VIA had very capable solutions back in the day. The KT800 was both fast and reliable at a price well below other highend offers from nVidia. I had one Athlon XP rig and it was very stable. Together with a Hercules Prophet 9800 Pro card it was such a good system for playing demanding games like Battlefield 1942.
VIA also had a good Socket 939 solution. The K8T800 Pro, found on highend boards with all the best features, like the ECS Proton KV2 Extreme or the ASUS AV8 Deluxe, was also a reliable and well designed chipset, rock stable with most setups and at least on par with the market dominant nForce3 chipset.
I think via has a kind of bad reputation, relating from much much earlier chipsets. But for the Athlon XP or the Socket 939 there is no reason to avoid VIA based boards at all.
IMO many people underestimate VIA chipsets.

I didn't say they are junk. They are fine for gaming. They make functional PCs in general. I said they have mysteriously weak GUI performance. If you run them side by side with Nvidia or Intel chipsets you will probably notice it. This includes their K8 chipsets.

Standard Def Steve wrote:

That problem with Windows 7 taking forever to check for updates went away ever since Microsoft started releasing those monthly update "packs" back in June.

Now, even my slowest full-time Win7 machines (a single core A64 @ 2.8ghz and a Dothan @ 2.72ghz) take less than a minute to find updates each month.

I think the problem still exists for a fresh install. There are some updates that need to be manually installed to prevent it. But that's not really my point. Installing just monthly updates can take some significant time. Ideally you can leave it to do its thing though. My parents had a P4 2.6 up until recently, with Win7, and I had to supervise the thing a few times while it struggled to install updates. Ugh.

I think it's a shame that Windows 8 won't run on older P4 or Athlon XP. It's such a lighter weight OS, because it has had to run on such weak tablets.

I don't doubt your experience, swaaye. But the link you put in before you were editing your posting shows a really abysmal permormance of the VIA based computer. Looks like a real bug or driver flaw, definately not normal.
If I had the hardware I would like to setup two systems for comparison. I remember building a XP3000+ computer based on a cheap ASRock K880Upgrade Motherboard. This board was on par with the nForce2 Ultra, maybe slightly slower but for sure not by this margin you experienced. But I don't recall which (version of) drivers I used..

swaaye wrote:

I didn't say they are junk. They are fine for gaming. They make functional PCs in general. I said they have mysteriously weak GUI performance. If you run them side by side with Nvidia or Intel chipsets you will probably notice it. This includes their K8 chipsets.

the KT880 whips the nforce 2 at everything except for AGP performance. On the KT880 CPU scores improve slightly, and it has better memory performance then the nforce 2 - altough this varies from board to board. For example the Asus A7V880 is slightly slower then the Asus A7N8X and both are slower then the Abit AN7. The Abit KW7 however is faster then both asus boards. It also has better memory and CPU scores then the AN7, but AGP performance is on par with the A7N8X. The A7V880 is not a very good implementation of the KT880 chipset. Be aware we're talking 2-5% performance difference here, so not enough to warrant looking for a KT880 (unless you're me).

The KT880 also has better win98 support in general, while the nforce 2 does a bit better in XP. Also, KT880 boards are less picky about ram and can run mixed-brand memory modules in dual channel as long as they are set at the slowest module's specs from BIOS. NF2 boards generally don't handle dual channel well, even with identical memory modules - so it's VIA 1 - nvidia 0 on this one.

There's also the Apollo PRO 133 vs 440BX / i810 / i815 - from my benches the VIA chipset (late revisions, 82C693A and 82C693T) have significantly better AGP performance then their intel counterparts. Also, the 440BX has better AGP performance then the i810/i815. i815 handles memory better and is a bit faster in memory benchmarks, but the via chipsets perfom better in synthetic CPU tests, being up to 10% faster in FPU julia and FPU VP8.

VIA gets a bad name because all of you probably had to deal with a cheap / shitty implementation of a VIA chipset at one point or another - including me. Try brand name VIA based boards from SOYO, Soltek, Leadtek (winfast), Asus, Abit and alike and you'll be surprised.

On topic:

The P4 - especially LGA 775 - makes a way better platform for a modern OS. It has SSE2 witch socket A athlons don't (required for Chrome) and is generally faster - and this coming from an AMD fan.

Looks like I'm not the only soul out here with some sympathies for VIA 😀

I reviewed the XP 3200+ recently in this video. Intel vs AMD S1E2 - nForce2 joins the fight

Note that the nForce is running faster timings and in dual channel memory. It does edge out a lead, but the Pentium 4 3.2 is faster altogether, even with slower memory timings.

I found VIA chipset boards easier to find and cheaper. I also like them for socket 754 / 939. Apparently the nForce chipset has a PCI bug that causes issues with X-Fi cards. My first choice are always Intel chipsets, but the VIA chipset boards I used with Socket A and later worked fine 😀

FGB wrote:

I don't doubt your experience, swaaye. But the link you put in before you were editing your posting shows a really abysmal permormance of the VIA based computer. Looks like a real bug or driver flaw, definately not normal.
If I had the hardware I would like to setup two systems for comparison. I remember building a XP3000+ computer based on a cheap ASRock K880Upgrade Motherboard. This board was on par with the nForce2 Ultra, maybe slightly slower but for sure not by this margin you experienced. But I don't recall which (version of) drivers I used..

The problem is only GUI speed and nobody really tested that. That is all my benchmark tested. Just search the forum here for Tom2D. I think I only had a VIA Athlon 64 board at the time so it's all I benched. But the slowness is something I've always sensed with various VIA boards.

It doesn't affect games whatsoever.

PhilsComputerLab wrote:

I reviewed the XP 3200+ recently in this video. Intel vs AMD S1E2 - nForce2 joins the fight

Note that the nForce is running faster timings and in dual channel memory. It does edge out a lead, but the Pentium 4 3.2 is faster altogether, even with slower memory timings.

I found VIA chipset boards easier to find and cheaper. I also like them for socket 754 / 939. Apparently the nForce chipset has a PCI bug that causes issues with X-Fi cards. My first choice are always Intel chipsets, but the VIA chipset boards I used with Socket A and later worked fine 😀

Back in the day the when the 3200+ was out, intel only had the 2.8Ghz P4 if I remember correctly - a northwood 2.8 at that. I think the 3.2 was contemporary with the Athlon64

http://www.cpu-world.com/CPUs/K7/AMD-Athlon%2 … A3200DKV4E.html - 3200+ released in may 2003, with the rare 2333MHz version (fsb 333) released in december 2002 to OEMs.

www.cpu-world.com/CPUs/Pentium_4/Intel- ... 200E).html - intel Pentium 4 3.2 GHz released in February 2004 - a year later then the 3200+

I'm not sure about the 2.8 tough. There might have been a 3.06 fsb 533 available at the same time as the 3200+ or a few months after it. In any case, pitting a 3.2GHz P4 against a 3200+ is not period correct, but it is relevant to us retro enthusiasts since we're living in 2016 😀.

The fastest clocked P4 cpu for socket 478 is the 3.4 GHz Prescott I beleive: www.cpu-world.com/CPUs/Pentium_4/Intel- ... 400E).html - there's even a mobile variant running at 3.53 Ghz.

All this makes the P4 a better choice for modern OSes. Still, I wouldn't recommend socket 478 chips for this - if you want a modern OS on a netburst PC, go with an LGA 775 platform. The pentium D is a solid performer in windows 7 and windows 8 - it has SSE2,SSE3, EMT64, and a heap of L2 cache and x64 support. Being a dual core helps immensely as well.

386SX wrote:

agent_x007 wrote:

Nah. It's... quite OK, actually :) Intel even did first "double cored" CPU, out of them :D "2x90W" - now THAT is a lot of heat. […]
Show full quote

386SX wrote:

That is good to have a modern new psu on an older motherboard that seems to really need power! (P4 at more than 110W sound quiet a lot).

Nah.
It's... quite OK, actually 😀
Intel even did first "double cored" CPU, out of them 😁
"2x90W" - now THAT is a lot of heat.

Yeah! 😁
And I thought the Barton alerady pushed my patience for fan noise and heat at its limit. 😁

Barton's aren't hot. It makes a lot less heat then most P4's. Noise-wise it really depends on the cooler. I use a copper core aluminum Arctic Cooling Copper Lite on my Athlon 3200+ build and the noisiest thing in the machine is the video card. The CPU never goes over 61-62 celsius onder load, and hovers around 39-40c in desktop. My old socket 478 build (SL7E6 3.4GHz + gigabyte GA-8IPE1000) used to hover around 46-47C in desktop and went as high as 75 celsius in games - and that's with a full copper cooler. It eventually killed the motherboard (a mosfet popped). The only thing that could keep that CPU cool is my Tuniq Tower witch killed another motherboard (Asus P4P800 Deluxe) with it's weigh. The flimsy asus PCB had a role in it too. Long story short, I'm never building a high end socket 478 rig again. Unless I find an ABIT IS7 or AI7.... 😁

Power-wise the faster P4 cpus use quite a bit more juice then their Athlon XP counterparts. A lot more. Oddly enough, entry level parts are more power efficient then AMD chips - an XP 1800+ uses a bit more power then a P4 at 1.8GHz - but the P4 makes more heat.

I see netburst chips as big honky V8's (and I love V8 engines) as opposed to the XP line witch is more like a turbocharged in line 6 - and there's only so much power you can get out of a 6-pot 😁. With LGA775 intel turbocharged their V8 monstrosities, (think pentium D and Pentium EE) making high performance chips witch were competitive with early athlon 64 chips.

kanecvr wrote:

An XP 1800+ uses a bit more power then a P4 at 1.8GHz - but the P4 makes more heat.

But that's not possible! All the power the CPU consumes eventually turns into heat, the one that consumes more always makes more heat. Or do you mean the whole system?

Problem is that Intel's "TDP", and AMD's "TDP", probably are not the same thing.
Also : If "Heat" is measured by "CPU Temp" sensor, it's WAAAY too unreliable to be useful.

PS. Heat generated by CPU is NOT determined by it's temps at idle/load.
Example : Low power chip (like Pentium G), can be hotter than Core i7 if cooling is inadequate.

PhilsComputerLab wrote:

I reviewed the XP 3200+ recently in this video. Intel vs AMD S1E2 - nForce2 joins the fight

Note that the nForce is running faster timings and in dual channel memory. It does edge out a lead, but the Pentium 4 3.2 is faster altogether, even with slower memory timings.

I found VIA chipset boards easier to find and cheaper. I also like them for socket 754 / 939. Apparently the nForce chipset has a PCI bug that causes issues with X-Fi cards. My first choice are always Intel chipsets, but the VIA chipset boards I used with Socket A and later worked fine 😀

I never had issues with VIA chipsets ethier, my Athlon XP rig (XP 3000+ FSB 400) is based on the VIA KT600 which worked fine for me

About your chipset choice, Intel Chipsets are an intel exclusive since the Pentium II/III era xD (excluding some i815 mobos with VIA C3 support). For AMD, only VIA, nVidia and SiS made the majority of chipsets for AMD, did someone had experience with SiS chipsets for AMD? i looked to some reviews on the SiS 748 and it seems to perform only sightly slower than the nForce 2 despite the SiS option only has single channel (dual channel doesn't really improve Athlon XP's perfomance though). Feature wise, the SiS 748 doesn't support SATA unlike nVidia and VIA counterparts

agent_x007 wrote:

Problem is that Intel's "TDP", and AMD's "TDP", probably are not the same thing. Also : If "Heat" is measured by "CPU Temp" sens […]
Show full quote

Problem is that Intel's "TDP", and AMD's "TDP", probably are not the same thing.
Also : If "Heat" is measured by "CPU Temp" sensor, it's WAAAY too unreliable to be useful.

PS. Heat generated by CPU is NOT determined by it's temps at idle/load.
Example : Low power chip (like Pentium G), can be hotter than Core i7 if cooling is inadequate.

Yes, you think in the right direction. What you are talking about is Newton's law of cooling.

Die temperature tells as much about TDP as maximum speed of a vehicle tells about it's engine power.

@up Would love to know Intel's TDP of this : LINK 😉
It's a really good heater tho (@load, hot air is going out even with Delta@4k RPM on TRUE cooler).

@kanecvr
First 3,2GHz Pentium 4 : LINK (June 2003)
First Intel CPU over 3GHz mark : LINK (Nov. 2002)