Why is my Pentium 4 so slow ?

[@watchmakerful]

My first thought when I see such read speeds: the IDE controller is in PIO4 mode instead of UltraDMA. The most common cause of this behavior is using a 40-wire IDE cable instead of a 80-wire one (or bad quality of that cable).

[@Jerrec]

I actually thought the same. I never needed the Application Accelerator in the past. I avoided Intel crapware.

[@_Tualatin_]

40pin work with dma but only like dma 3, it's 30mbs, so it's not so bad. Yes 80 pin is better, with more modern like p4 or even PIII, but most important is manually check and configure bios. And of course use new or tested and good hdd or ssd for best speed. And don't forget install latest chipset driver like we so in a video.

[@6581punk]

PIO was awful. First time I built a computer ever I had a PIO hard disk and the machine came alive with a UDMA drive.

[@Jerrec]

@@_Tualatin_ ATA100 and ATA133, I always used 80pin cables. However they are hard to find nowadays.

[@d34dbolt20]

Around the 15 minute mark it does show the cable as an 80pin and that ultra dma 5 mode is in use.

[@justsumguy2u]

Back when the P4 came out, many critics trashed it; in some tests, the PIII was faster. The P4 was also notorious for running hot

[@Pidalin]

It was not running hot from today point of view, problem is that you jumped from 90s CPUs with ridiculously low TDP to CPU which already needs proper cooler and most of people didn't have proper cooler back in the day. Even basic Zalman cooler was able to keep it around 55°C in full load, that's not hot at all from today point of view, when you use more modern cooler which already has heat-pipes (they started appearing around 2004), it has like 40°C even in full load, compare it to today CPUs which run on 90. 😀 BTW, no, Athlon XP was not cooler, it's a lie.

[@Lurch-Bot]

The early P4 was not great. I think it took me about 6 months before I did a FSB overclock on my 1.6. I got the FSB to 150MHz (2.4GHz CPU clock) and it became a totally different machine, nearly doubling benchmark performance. I think later, I played around and got the FSB a little higher, maybe 160MHz. Anyway, it served me well for the better part of a decade and was playing new releases for about 5 years, courtesy of a GF3 Ti500 that also overclocked well. It didn't matter they ran hot because a 350nm process could handle it. Modern CPUs can't take the same kind of abuse.

[@OldFordTaurus]

The pentium d ran hot as well. It was basically two pentium 4s put together. The core 2 duo was worlds ahead

[@gokhansarioz7150]

That is for first gen of P4 which is socket 423, never use them but hear the stories and benchs. The P4 1.8A with a 512kb of cached version is something killer, but for sure such as celeron 1.7 with a 128k versions can get trouble with tualatin series even celeron itself when its clocked at 1.4.

[@XX-121]

@@Pidalin yeah but they were still pushing pentium 4's when amd was on to athlon 64 (as well as X2's) which did run a lot cooler than P4 and out performed them

[@RuruFIN]

It's a miracle that the old Maxtor DM9+ drive is still working. Those were notorious of failing.

[@g4z-kb7ct]

Hehe yeah I was thinking exactly the same thing. I had a couple of 160gb Maxtor drives both failed around the same time. They are only good as a door stop.

[@ThePuuFa]

Not all Maxtor drives were that bad... My dad used his 40GB Maxtor as his only hard drive from 2001 till 2016 and last I checked it had 30k+ power on hours and still working like a champ

[@RJARRRPCGP]

@@ThePuuFa Diamond Max 8s were pretty much the only terrible ones from Maxtor. They are identified by the new-slim-look. Their other models have the more usual style housing. The 8s are the 6Exxxxx. 6Exxxxx are bad. The 6Yxxxxx is an excellent family of HDDs from Maxtor.

[@Stefan-]

My Dell Pentuim 4 from 2005 still has its original Maxtor drive, not that im using it lately but i actually used it for recording my band with 16 channels via firewire for album recordings up until a couple of years ago but that really was on the limit of what was possible, i have since replaced it with a used but less old Intel I5 which is a monster compared to the P4 although its about 12 years old now but it works just fine for the task.

[@Lurch-Bot]

Based on the sound it makes, it is likely to die soon. I have a Quantum Fireball from that era that still works but I won't put it in a build. Too small. I have quite a few old drives in good condition. I built a PIII with a Seagate Medalist drive last year for a Win 98 machine.

[@ronalddaenen834]

Hey, its the pentium 4 I bought last week. So it's you I bought it from. Nice video 👍

[@RetroSpector78]

Hehe indeed it is. If you ever sell it perhaps the video will increase its resale value :) enjoy it !

[@ronalddaenen834]

😂

[@ronalddaenen834]

I will

[@freddan6fly]

Pentium 4 was the reason that AMD came into the match from around 1999-2012.

[@timewave02012]

At the same time Intel was failing with long pipelines and high clock rates in 32 bit Netburst, AMD developed the 64 bit architecture that both AMD and Intel are using to this day. Prior implementations of 64 bit (including Intel's ill-fated Itanium) all had major disadvantages preventing PC, workstation, and low-end server adoption. Not only can an OS running in AMD64 "long mode" still run 32-bit programs, but the way the instruction set is designed, 64 bit programs can and frequently do use 32 bit instructions for performance. They have to use 64 bit pointers, but that performance hit is offset by the gains from the instruction pointer relative address mode AMD added.

[@volf3r505]

More like 1999 - 2006. Core 2 Duo rekt every competition, then there was Core i series. Thankfully we have Ryzen now.

[@pinklightninggacha]

Amd has always been in the game

[@Hypn0s2]

Yeah. Pentium 4s were space heaters and slow per MHz. Most of us that built systems - built AMD Athlon.

[@chelsona2574]

amd was always there :)

[@AlejandroRodolfoMendez]

Also the IDE cable issue is real, I had all sort of problems in a k6-1 until I changed the IDE hdd cable

[@laurpflorin]

Defragging the hard drive as well as defragging system files offline (hibernation file, page file, MFT file, and registry hives) can also have a significant impact on performance, especially with a system installed in an HDD

[@Stefan_Payne]

The Case is a Compucase made (7106), sold by Aopen. Those are prety nice for ATX Desktop Rigs...

[@SaraMorgan-ym6ue]

makes you wonder if the Pentium 3 system could have been sped up like the Pentium 4 system due to it also having software and driver problems.🤔🤔

[@Stefan_Payne]

Yes, Pentium 3 systems went to up to 1,5GHz or thereabout, but those Chips are pretty rare (Tualatin). And you need a compatible Chipset (or mod your board) as well. @@SaraMorgan-ym6ue

[@micheledelmoro5898]

what is that 2000 era case called with green triangular downfacing power button and all textured with downfacing triangles?

[@Stefan_Payne]

My first thought of this was "Pentium 4 Willamette/S432 with PC133 Memory". And it indeed is using PC-133 memory, which the P4 really doesn't like. And making it faster is practical with a more modern Board, Single Channel DDR-SDRAM at bare minimum, Dual Channel (i865PE) recommended.

[@EgoShredder]

That is what I went with on an ABIT AS8 motherboard (i865PE) and DDR dual channel 2x1GB 3200 rated? I forget now. Pentium 4 640 Prescott. Pretty good hybrid system with AGP 8x support. Never had the chance to try it with an SSD drive though.

[@nickwallette6201]

SDRAM slows down the P4, but nothing like _that._ You wouldn't notice in everyday use. It's a bottleneck, but it isn't crippling. Windows XP on an SSD moves like it's bathed in Crisco, going downhill on a Slip n' Slide. I remember the first boot on an SSD. The XP Pro loading screen had the chance to move the blue dots about 2/3rds of a trip from left to right before the screen blanked and went to the desktop. It does depend on what else you have (physically) installed, and what drivers you have loaded, as some hardware just takes a moment to initialize. In that case it doesn't matter how fast your storage, RAM, or CPU are.

[@EgoShredder]

@@nickwallette6201 I have an SSD in my older ASUS TUSL2-C board with Pentium 3 1400S and 512MB SDRAM. That machine is as quick and responsive as my modern systems, albeit using Windows 98SE and older period correct software etc.

[@wladislawmsk]

@@nickwallette6201 indeed. As I remember - the first question about P4 PC performance was its memory type. But SDRAM with P4 only meant P4 would not be much faster than P3, not slower because P3 had the same SDRAM anyway.

[@wishusknight3009]

Even with RD ram the P4 1.5 was only _slightly_ faster than the Coppermine 1ghz. It was pretty bad. And sometimes even slower. Only in things like Quake 3 and optimised applications like MP3 compression was that 1.5 able to outrun the P3. Putting that 1.5 on SDram in fact makes it _slower_ than a coppermine 1ghz machine. Its really sad how bad early P4's were. Northwood hyperthreading P4's were a pretty decent chip though.

[@RetroTechChris]

Davy, I am so glad you are back. Thanks for taking us along as you solved this, it was a great watch! And to think that machine might have spent a good portion of its life being so misconfigured, how unfortunate! Glad you fixed it up :)

[@adventureoflinkmk2]

Does Davy got his locker named Jones ;)

[@RetroTechChris]

@@adventureoflinkmk2 not sure, but he makes great retro content!

[@adventureoflinkmk2]

@@RetroTechChris yea... he does. :)

[@donixion4368]

I have a Prescot-based system ( P4 531 3Ghz HT ) with the same problem. I put it off to the side some time ago and never really fixed the problem with it. I shall now revisit the problem after watching this.

[@TheXev]

Chipset drivers for Windows XP for newer systems that XP didn't ship with drivers helped a lot. I remember moding in the nForce2 IDE controller driver into the Windows XP setup (this was before nLite came along) for my Athlon XP 2000+ system, and Windows XP would install so much faster using the nForce2 driver vs the generic driver (which didn't support DMA at all). This would reduce my installation time to 30mins on a quick format (down from 45mins, 40GB Seagate drive), or 45mins on a full format (down from 1hour on the same Seagate drive). I don't remember how (and was never able to replicate it), but after nLite came around, I reduced my WindowsXP installation down to 200MBs installed. Just removed stupid stuff like theme support and Internet Explorer (I clearly remember this breaking Steam in one attempt a few years later after that launched)... Good times.

[@hugosimoes5119]

I could be many things. Caps may look fine from outside but they may be losing capacitace or some voltage regulator maybe running too hot or some IC/option is holding system back. Even a bad power supply can perform a slowness role. When you said ~3MB/s transfers, my mind screamed DMA is off in Device Manager. Over 10 years ago, I had XP installed on a system like these. The HDD unfortunatelly was about to die.... some bad sectors or perhaps a range of bad sectors. I noticed that XP autoswitched from DMA on to off and the system became pretty slow like this video showed. Cables sometimes can make things harder when we push the cable and not the thin black plastic. One wire broken is what's needed to make a bad cable.

[@ChartreuseKitsune]

For anyone wonder how to check the mode, open Device Manager. Go to IDE ATA/ATAPI controllers. Then right click Primary IDE Channel and select Properties. Then in the window that pops up go to the Advanced Settings Tab. It'll show the current transfer mode there. I've had a similar issue with my x40 before that was likely related to a failing drive where it'd fall back into PIO mode occasionally and the solution was to uninstall the IDE controller device and restart and let it reinstall it.

[@aleksa1987]

I totally agree, bus gone to PIO mode probably, so driver reinstallation of that bus always should fix the problem.

[@hugosimoes5119]

@@aleksa1987 Not necessarily. Going back to Device Manager and check if DMA is set "if available" under ATA/ATAPI controllers/Primary channel. Deleting the Primary Channel and say yes to restart, should reinstall that devie and enumerate the hdds there connected.

[@RJARRRPCGP]

That means PATA I/O errors occurred. Usually a bad PATA cable connection.

[@dav1dbone]

Par for the course, fresh OS install - doesn't matter if its a Celeron or a 3.8Ghz Extreme Edition, give it a few weeks, it'll be grinding, shovelling and displacing bits and bytes like a gaggle of lethargic road sweepers.

[@polaxis842]

Of course, it is a Pentium 4 after all 🙂I'm amazed on how slow the system in a top noth Northwood could be, even with an SSD ^^

[@nickwallette6201]

@@polaxis842 The P4 wasn't that slow. Initially, the S423 ones weren't a great deal faster than the highest-performance Tualatin P-IIIs of the time, but those were also _really_ expensive, and never got cheap enough to be a value buy. The P4 came into its own when the clock speeds got higher. Once you got to 1.8GHz, they were worth jumping to from a Pentium III. The other problem was memory. Since Intel had to go through that messy divorce with RAMBUS, we had to wait far too long for proper DDR support. When you're stuck between expensive and clunky RDRAM and SDRAM, you're not doing the P4 CPU and Northbridge chipset any favors. Those two factors gave the P4 a bad rep that they never completely recovered from, but a Northwood with DDR performs quite well. Once you got _too_ far into the later P4s, we bumped into the practical limit on clock speed, and it was clear the whole architecture had no future. But, in its sweet spot, the P4 was still a fine performer. I had a colleague at work, at the time, who was a die-hard Athlon XP fan. When the Northwoods showed up on store shelves, he switched teams -- dubiously, but curious how far he could push them with overclocking. He got a 1.8GHz chip up to 3GHz with no trouble at all, and it ran circles around the AMD rig he was using. He flipped to Intel overnight. Not to talk smack about AMD. I think Intel would've languished without them. I just don't think it's fair to crap on the P4. I will absolutely admit, though, that the Core and Core 2 line made it clear that the P4 was a design mistake. Still, my Windows XP rig to this day is a Northwood 2.0GHz on an Abit IC7-G with 2GB DDR, and a pair of Seagate Barracuda SATA drives in RAID-0. It's _plenty_ snappy. I just wish you could still find a working Radeon 9700 Pro to compete the quintessential Blue Team build of the day. Alas, the GPUs are fundamentally flawed in the same way the Xbox 360 and PS3 Gen 1 CPUs were. RIP.

[@alexanderangelov230]

@@polaxis842 I have an Asrock P4i65g and ran it with a Prescott 3.4 and a Northwood 3.0 with Win 7 on an SSD. It previously ran XP on a HDD. It ran pretty well, though I didn't do benchmarks. Decided to retire it and got an Asrock 775i65pe and Pentium D 950 with Win 10 1511. It too runs quite well for its age.

[@polaxis842]

@@alexanderangelov230 That is quite some oldschool hardware in the meantime 🙂 From my experience: It did run Windows + Office + games from the same time period very well. But even for browsing, the CPU was heavily on edge :)

[@alexanderangelov230]

@@polaxis842 Of course Pentium has nothing on Core. But that was my first pc, a budget one, from 2005, and it came with a Celeron 2.4. The Pentium was a welcome upgrade.

[@Bewefau]

I'm glad to see you back love your sense of humor.

[@alanharkleroad4376]

My thoughts on Pentium 4 systems are that they are excellent for early to mid 2000s gaming.

[@g4z-kb7ct]

They are commonly overlooked as a retro PC. I have a Win98SE PC with a P4 and it absolutely flies with no bottlenecks :-)

[@ikbintom]

Bedankt voor de video, Peter! Heel interessant

[@Universal.G]

Your videos are made very well. Clear and easy to follow. Thank you

[@thisislocombia]

this makes me remember when i had a core 2 duo setup and my usb transfer speeds were like TOTAL TRASH and some dongles i used didn't work at all, and at the end i noticed that in the bios the ports were setup as usb 1.0 instead of 2.0

[@southernflatland]

In my experience, systems that have 3 RAM slots were generally designed with the capability to overclock the chipset if you only occupied 2 of the RAM slots. The board traces were too long to overclock with the 3rd slot occupied. Your mileage may vary, possibly significantly, because my experience was from a totally different motherboard. Still, there was a reason for 3 RAM slot systems back in the day, and it was usually a choice of running the chipset overclocked with only 2 slots occupied, or running the chipset at stock speed with all 3 slots occupied.

[@lucasrem]

Compaq, overclocked ? why you say that ??? new here ???

[@southernflatland]

@@lucasrem My username on other services is literally over_clox, so no, I'm most definitely not new to the scene. Probably newer to this channel though if that's what you meant.

[@southernflatland]

@@lucasrem Also, I think the system I was referring to was an early ASUS board from 2003, but I kinda forget now. Regardless, there were some unique design decisions made with systems that had an odd number of memory slots.

[@krissjacobsen9434]

The ATI2000 is probably an ATI Xpert 2000 Pro, which is based on the ATI Rage 128 Pro GPU.

[@ZipplyZane]

Hmm. I was expecting this to be about the CPU being slow, given the title. A disk being slow could happen on any computer. But, still, it was an interesting dive into the system, and interesting how much difference the drivers made. That seems much less the case today.

[@Anradak]

when i was a kid, PIII AND IV were considered beasts from another planet. i remember having a PII at 600mhz and dreaming with having one of those

[@vlogyatutorialy7796]

I had an Windows XP Machine, like 10 years ago... And GOSH IT WAS SLOW! It took like 10 minutes to boot, and it was extremely dusty also, since I've got it from a friend for free. (Who had it in like the "industrial site) But hey! I was a kid! I have really nostalgic memories of this PC. I was actually kinda sad, that a PC I bought like 4 years ago was.. well kinda fast, I DIDN'T GET THE AUTHENTIC EXPERIENCE! (It is an old school Windows me system)

[@imqqmi]

Simply installing the intel chipset drivers could make a huge difference. Microsoft often supplied outdated but stable drivers to get you going but aren't optimized. The Compaq came with good drivers preinstalled probably. Xp configuration could also impact performance like fixed or variable swap file size, stuff in autostart, virusscanners, when windows updates decides it's time to scan for updates etc.

[@Pidalin]

the only MB where installing chipset drivers was actually doing something was some MB with socket 754 (amd), but generally, I never had to install them on any socket 478 Intel motherboard with Win XP, it was always working fine, I am just installing network, graphics and sound drivers and everything works fine mostly, installing chipset drivers sometimes cause more problems than making it better

[@SomeFrenchGuy78]

I wonder why the BIOS option "PnP Operating System" i set to NO. I used to set yes if the system was Windows.

[@airfixer9461]

Nice review & good tips...well done!

[@xjr358]

Thank you! I'll check my transit P4 PC that I use to transfer files to the older machines

[@MegaUpstairs]

I remember that the motherboard drivers always had to be installed, otherwise the drives were slow. And yes, you have to reinstall the XP in every year, otherwise it will be slow.

[@thedopplereffect00]

Drivers probably never got installed because plug and play OS option was turned off

[@xsc1000]

@@thedopplereffect00 Plug and play option in BIOS has nothing to do with installed drivers. This option is about assigning resources like DMA and IRQ in BIOS.

[@charonunderground8596]

Disks are best checked with the Victoria HDD Info program which is in the bootable Hirens Boot CD.

[@EastAngliaUK]

nice system video shows them going and the inside also

[@The_Wandering_Nerd]

I guess sometimes it is worth it to install the chipset drivers and update the BIOS on these old machines

[@HappyBeezerStudios]

The first reason would be because it's a Pentium 4 But on a more serious note, there might just be something misconfigured. - At first I assume it's actually a Pentium 4 and not a Celeron, because those were cut down like crazy. - Then obviously the amount of background stuff. - Then IDE. Classic PIO mode issues. When the drive doesn't deliver proper data the system will try again. And after 5 trys it will reduce the DMA mode make sure it can get any data at all. Let that happen a couple of times and you're out of DMA modes and back to PIO. The classic fix is to remove the controller in device manager and do a restart. It should then reinstall the controler in it's proper UDMA mode. - The drive might just be on it's way out - And obviously fragmentation - The BIOS reports PC133 memory, which is probably the slowest memory any Pentium 4 could run on and would be a clear bottleneck.

[@SilverX95]

are you sure that DMA is on in XP? sometimes it likes to turn off yes the OS has a setting too. i think they fixed this in later updated installers like SP1/SP2/SP3 that and having the drivers installed and checking the bios settings.

[@xsc1000]

If those tests show transfer rate about 60MB/s, UDMA has to be active. PIO would be 3MB/s max like in first run.

[@gentuxable]

The order of driver installation is important. Sometimes Windows will just default to a generic driver and not pick up a better third party one. I think that’s why they tell you to install the driver before plugging in, that way the system can be prepared to load the right driver.

[@retrotech.oldschool]

what is memory speed on both and what is the front side bus on the compaq? just curious

[@davidp4456]

I always gauge the load time of WinXP by the number of swipes the progress bar makes. My P4 3.4Ghz Cedar Mill does it in 2.5 swipes, same as my Opteron 180. It’s not as old as your P4, so no real surprise. Intel’s application accelerator looks interesting. I’ll give it a go. They clearly had a problem that required a solution.

[@diegocipriani]

Quasi il mio primo pc, avevo il suo successore il cpc 6128, per la cpu puoi provare a vedere se magari aveva qualche piedino ossidato che non faceva ben contatto, quei socket dip alle volte sono traditori 😊. Complimenti per il canale e la preparazione al prossimo video!

[@whitesapphire5865]

Interestingly similar to my i3 Vs i5 laptops, where the i3 outstrips the i5 by a country mile. I know they're technically different, but it's eye opening to see the older machine absolutely trouncing newer one. Anyway, I may be one of those weird folk, but I always liked and enjoyed my old P4 computers. I always found them to be solid and reliable, but, not alarmingly better than my P3, which too was a strong performer. My AMD K7 was also a good and trouble free machine. All of these machines were used to digitise, process and remaster audio recordings, which, depending on how many processes were running simultaneously, was very demanding on PC resources.

[@georgemaragos2378]

HI, i find the version of XP plus the service packs makes a very significant impact on overall "snappyness" Stock XP can run in 512k happily SP1, SP2,SP3 for me means each service pack will like another 512k for each upgrade You dont need to do it, but idle usage increases as does ram requirement If you under ram, then it will just thrash around the hard drive with a slow swap file In that age i have 2 x toshiba Pentium M's and a IBM / Lenove P4, the laptops are on stock and XP1 for the other. the IBM with 2.5g ram ran fast with original XP, but i ended up taking it to the SP3 update and each step i let it run for a week or 2, and it got more and more sluggish,but luckily it was plenty of ram I also have a HP5102 mini netbook, it has 2gb ram and Atom processor, that i bought new with XP home, it ran OK in the day but 4 years ago i almost tossed it in the bin as it was extremely slow, xp start up was over 5 mins, but the hdd made no bad noises. A reinstall took almost 4 hours, and it sill performed slow. I installed linux min , faster but slow. while shopping i picked up a 128GB SSD for raspbery pi4, but decided to test it on the netbook, WOW , the ssd is faster but it turns out that the HDD has some strange fault, it is just slow no bad tracks or errors or noises - that went into a Pentium 2 and a dos 6.22 install and dual boot to 98 and it was slow in dos - it is a paperweight now but full of my dos backups ** oh some people will ask , even a spare hdd replacement in the netbook saw a 45-50 min reinstall time for xp and about a 2 to 3 min boot up time Nice P4's just go get a network card :)

[@ShrineOfLife]

hey mate, one remark -> the 845 intel chipset i capable of adressing isa, i think its the last isa compatible chipset and there are some exotic boards around, that come with isa - me, i am rocking an ibase mb800 mainboard with a not ht 2,8ghz northwood with windows 98 - with an awe64, i really love that pc!

[@osgeld]

that era of compaq pentium 3's were pretty well optimised, I used one as a garage computer up until programs (like chrome) stopped supporting it, did just fine for reading websites and PDF's

[@erikhicks07]

The Pentium 4 PC was effectively running in 'Safe Mode' (no direct disk access performance) due to the chipset drivers not being updated. Probably happened after the motherboard/CPU were upgraded but not the software.

[@menelkir]

Maybe it's fragmentation? I would defragment both machines just in case.

[@johnrickard8512]

That is a very snappy Pentium 3!

[@joesaiditstrue]

when Pentium 4 released, I was running a coppermine Celeron 533a, I think I upgraded to an AMD 1800xp, then went to a core 2 quad q6600, the memories :)

[@li-ao]

OS performace does vary with an important factor, hard drive speed. Sometimes old hardware may result in lag of program execution and OS loads, so you could check if those parts are doing wrong.

[@sandmanxo]

The Pentium 4 era, especially before Prescott was released was a truly forgettable era for Intel. At least the Pentium D made it more tolerable but until the Core 2 came out I was only using AMD at home, and at least got upgraded to a Pentium D for my work machine at the time.

[@criticscooby]

8:11 I would play with the cables for a while to hide them, it is so fun to make them look somewhat in order lol

[@InsaneWayne355]

Didn't see drive controller settings in Device Manager. Could be configured wrong.

[@spuds7677]

Back in the day, I had nothing but problems from the Matrox hard drives. Unless there is a reason you're keeping it IDE, I would get a SATA adapter and put a SSD in it. If you plan to keep the IDE drive, I would hunt down a WD drive. That P4 should run circles around the P3. Somewhere, I have a Shuttle XPC with a P4 Extreme in it. It was a few years newer than you CPU, but it ran really good! I may dig it out and see what I can do with it!

[@yyzttr6306]

Pentium 4 has a longer Pipeline for a process, although it has a greater overall clockspeed.

[@d.r.1402]

Not understanding why not using a Linux Live OS to see what the read speed of the drive is within it before stripping it down.

[@ComicSanserif]

Is "partition alignment" / "partition offset" the possible culprit here? I can't remember which Microsoft OS fixed this by default. Also, I can remember that sometimes having some "auto select" active on harddisk and then "slave" for optical media doesn't always behave. You can try setting "master" for the harddisk and temporarily remove the other drives to see if that makes a difference. Try swapping out the ide cable with the p3 machine as well. Does it have multiple connectors on the cable? Try another one. Does the motherboard have multiple cable headers for the ide cable? are they marked differently? (e.g. "IDE" vs "UDMA" or something like that). Is the (NTFS?) cluster size on the p4 machine way smaller than on the p3 machine?

[@jasonwoodruff5186]

I noticed the drive was set master. Should it not be on cable select? The jumper? Also, each computer was running a different Service pack. The P4 was running an older service pack. The P3 was running SP2, I would assume. Good video.

[@southernflatland]

That master vs cable select thing makes absolutely no difference, unless the drives have conflicting settings. Cable select was just introduced to make configuration easier for people that didn't know all that technical jazz.

Good video and its nice to see you back. Stuck in PIO mode sucks 😆 on the subject of Pentium 4. A equally clocked Pentium 3 tualatin is way faster then a Pentium 4 and PC133 SDRAM just add insult to injury. Pentium 4 IPC is way lower then the old P3 and Athlons, so the CPU needed every bit of data available to be efficient and with SDR its not. I would take a P3 or Athlon over a P4 all day long if they are clocked at 75% of the P4 and assuming the P4 has DDR och RAMBUS. The only thing that could make that P4 system any slower is a Celeron, Celeron netburst CPU with PC133, must be painfully slow, it sure was even with DDR266 for my friends budget PC back then 🤣

[@АлексейИванов-р8т1е]

It would also be reasonable to check the hard drive for bad sectors. The built-in disk tools, of course, try to correct and replace such areas, but all this still slows down the disk, with a large number of errors. For such a task, for example, Victoria programs by Sergei Kazansky, Norton Disk Doctor and the like are suitable.

[@Jah_Rastafari_ORIG]

You know what got me all pissed off watching this video? It reminded me that last summer, my visiting ex-wife borrowed one of my clean t-shirts (a red Compac shirt from some computer faire in the '90s) to loan to a friend who had gotten all wet in the lake; when I recently asked the friend for my shirt back, she said, "Oh sorry, it got lost...". Moral: Tell my wife I'm going to be traveling the next time she plans to come visit and don't watch videos that mention Compac.

[@pazitor]

Ah, the Pentium III. Still have my venerable Abit VP6 duallie, lovely machine for XP.

[@CasioGreg]

there's a big difference between 5400rpm drives and 7200rpm drives. just this upgrade will up the speed. also, i remember reading that up to winxp, depending on the amount of ram you have, it will determine what will be installed by the OS. After installation, you can add another stick of ram to speed things up.

[@maikgottkowski5670]

As far as I remember those days, there sometimes was a key-combination or F-key that sometimes reveals hidden insights/options in the BIOS. Not sure if that's the case with your mobo but it's worth a search and try.

[@simontay4851]

Yes, it is CTRL+F1

[@EirikrTinkerTries]

Is that a PCIe port I see? Consider adding a SATA card for an SSD OS drive, keeping a HDD for Linux boot (e.g., Archlinux32 or Debian). Install a Radeon HD 4670 AGP for graphics, and add a PCI HD 5450 and GT 610 for OpenCL and CUDA support using nvidia 390 drivers. Incorporate a Blu-ray burner via the SATA card and mount it in the front, keeping a DVD burner for booting. Add a PCI USB card and, if possible, mod the BIOS for USB and SATA boot, plus overclocking settings. Replace the CD-ROM with a 5.25” floppy drive and add a ZIP drive alongside. Opt for a Sony 2.88MB floppy drive for backward compatibility or an LS-120/LS-240 if compatible. Use ArchInstall for Archlinux32, setting your mirror(s) without changes during installation and include Python. Max out RAM to 4GB, and consider a Hauppauge PCI TV tuner to turn it into a Linux DVR. /boot on your HDD, all other folders on SSD. Debian may be easier. IDK if they have all the nvidia drivers though. Hmmmm.

[@10WA]

I had a Pentium III 900mhz, kingston 512mb ram, PNY Geforce4 ti 4600, Maxtor 80gb hhd I built in I think 2001 and used all the way till 2006. Even then it was starting to struggle with some games but overall for daily use it was still plenty fast. I think for a while I had it overclocked to 1 or 1.1ghz with the Asus board I had.

[@TomaszWiszkowski]

3MB/s is a good hint what's going on. It's the PIO 0. PIO stands for Programmed Input Output, so the CPU is used to send bytes of data to the device. 0 is the generation, the higher the number, the faster the data can be sent. So PIO0 means "use the CPU to read and write from disk, and do it at the slowest possible rate". Possibly to maintain compatibility and ensure nothing goes terribly wrong. Super unfortunate seeing BIOS applying UltraDMA 5 or 6 and then Windows rolling it back to PIO....

[@NozomuYume]

Was looking for this comment. PIO-0 maximum is 3.3 MB/sec, which is in line with what he's getting.

[@lanelesic]

Back in the day I had a Celeron Tualatin 1.2Ghz ocd to 1.34. With 384MB of sdram and a geforce 2 mx 400 later upgraded to FX5200 ultra. But more importantly I had a hardware ISDN modem, hardware sound card, 2 hdds on separate IDE branches(one for os and other for games) and a sharply tuned Windows xp operating system. So people that were buying Pentium 4 computers with like 128MB or 256MBof ram, that used windows ME, one of those dial up pci cards, etc. would not believe I didnt have a pentium 4.

[@Pidalin]

upgrade from MX400 to FX 5200, that was probably pretty bad deal back in the day, even MX440 is faster than any FX5200, only reason to use FX5200 is DX9 support

[@lanelesic]

Fx5200 was around the raw performance of a geforce 4 ti 4200@@Pidalin

[@Pidalin]

@@lanelesic No, it didn't have, you remember or wrong or you didn't test it properly, Ti 4200 is much more powerful card than FX5200. Ti 4200 should have like twice raw performance compared to FX5200, but maybe your FX5200 ultra had higher clocks so it was not such a big difference, but even FX5600 is still slower than Ti 4200 in raw performance. Ofcourse it's gonna be better for FX series when you want to run some later games which just need proper DX9 support, but try to compare them in older DX8 games, Ti 4200 is much faster.

[@lanelesic]

@@Pidalin ROFL look up some Quake 3 fps charts

[@Pidalin]

@sic Quake is OpenGL game, it can be different, but the vast majority of games run under DirectX and Ti 4200 has double raw performance there. And I don't play with some memories, I actually test a lot of old hardware now, so I know it for sure, even 64bit MX440 is better in raw performance than FX5200, but MX440 is just renamed GeForce 2, so it lacks some features you may need in 2003+ games, it doesn't even have full DX8 support, for example test Nature in 3D Mark 2001 doesn't work on it, but it works on Ti 4200.

[@johneygd]

There could be sooo many factors why a more powerful pc could run at average slower then a weaker pc,because of a slow hardrive,less and slower ram,an underclocked cpu,a poor cooled down cpu,lots of programs stored on a hdd drive being damaged to require error corrections,etc,,, or a combination of both things,so these are the reasons why a more powerful system doesn’t alwayssay everything,but it’s still interesting to think about it😁

[@wimh-e7l]

I see no thermal paste at 11:24. Was it there? That could also be why the P4 is slow. Could be thermal throttling.

[@cartman20000]

The slow speed of the drive could also be a symptom of a failing drive, used to have a sata drive that had a number of reallocated sectors on it, which absolutely killed the performance of the drive. The average read speed of the drive dropped down to around 2-3MB/s.

[@Redmage913]

Did you check SMART statistics on the drive?

[@patg108]

Could be the bios, the hdd, possibly underclocked on the p4 or overclocked on the p3. Ram may not be optimal (i've seen and have p4 with pc133 ram but most have ddr1 or ddr2 or rdram, in some cases ddr3 (basically a775 p4 on a board meant for a core2quad or 771 xeon)

[@shinya1215]

I'm just wondering what netbrust are good at (or what's in mind when design, any target like float, int performance?) and if using today's process to re-make it say with 14nm tech. The last pentium4 has 188m transistors, the first 14nm skylake has 1.9b transistors so you maybe can make a 10 cores Pentium4 with 20 threads in a chip with higher Ghz limits? It would be interesting to see how that thing can perform :)

[@0wl999]

One other factor could cause the slowdown: if the read head on the hard drive is starting to fail the hd controler may be writing bad blocks to the platter(s). A lot of that would slow it for sure as the head has to seek more inefficiently.😅

[@stevepriority4219]

I remember well the disappointment when I had first found out that the higher end PIII was much faster than the lower end P4 cpu's. The one on the left is 1.7ghz. It wasn't till about 2.2ghz that the Pentium 4 started to feel faster. By the time you got to 2.7 or 3ghz, it was much faster.

[@formatter]

I wonder what was the actual reason for the low speed? Did windows xp switch ATA transfer mode to PIO without appropriate drivers for the controller?🤔 Anyway thank you for the video!🙂

[@johnDingoFoxVelocity]

There was a update for windows xp to add support for the pentium 4

[@RWL2012]

The Pentium 4 first came out in 2000, whereas Windows XP didn't come out until 2001.

[@pianoman4Jesus]

Early Pentium processors were plagued.... the 60 and 66 MHz vintage. Once the Pentium 100 MHz arrived and the corresponding 430 TX chipset, the original Pentium finally got solid. Pentium III got very good when that era completed, 1200 MHz standard, and I had some IBM Servers with 1400 MHz processors. Pentium III chipsets with the 915 chip set was finally solid. The early vintage of Pentium 4 processors were slower than those fastest Pentium III's. Again, it took a few reversions / versions, then Pentium 4's got decent. Maxtor with 7200 rpm spindle speed. That was a pretty good specification. Naaaa.... 845 chipset era was not when Pentium 4 got decent. Intel 945 chipset era would have been much better. In my opinion, too early of a Pentium 4 era to deal with.

[@geoffk777]

The P4 was a major misstep for Intel. It was not much fadter than the P3, but ran hotter. For the Core series, Intel abandoned the P4 completly and reverted to the P3 design. On the plus side, you can use your P4 as a great space heater.

[@fhwolthuis]

Reminds me of the well known problem which I also had back in the day, that your user profile could get too big it you stored a lot of gigs of pictures and music in the default folders. Really annoying, if you're going to use XP and have a lot of data stored, then move the "music" , "photos" etc folders outside your user profile into a different directory.

[@thedopplereffect00]

That will do nothing

[@ZEMRALEX]

I have that Pentium 3 Compaq Deskpro, plastic is still white and in very good condition

[@Ilove80sever]

Which hardware fits every PC?

[@johncollins5552]

A lot of p4 era pcs are having problems due to lower quality capacitors fitted on the mobos, of course they can be replaced cheaply in a few minutes of diy soldering.

[@Carstein666]

Why are there two chapters titled ""The Exterior" in the video (6:54 & 7:42)? It looks like the second one should be titled "The Interior."

[@vojtechadame5860]

Yeah, the Intel Pentium 4 Willamette was pretty slow. The Netburst architecture needs high clock speeds to be good, below 2GHz simply isn't enough. The second reason is the RAM. The SDR RAM is just too slow for a P4, I'm sure that the system will be faster with any kind of DDR RAM or RDRAM. The performance in games will be still bad even with better memory, but the system will be way more responsive.

[@MurderMostFowl]

No thermal paste on the cpu??

[@antssaar863]

First thing is, what type of P4 is it? :D There's QUITE bit difference between willamette, northwood and prescott. Also soc 423 vs 478 and sdram vs rdram vs ddr. Willamette on 423 can easily suck behind P3 coppermine (oc) or tualatin. It was mostly thanks to bad socket design (soc 423) and weird motherboard issues/lack of good boards.

[@ABRetroCollections]

1.7 Willamette at 400 MHz BUS speed. D845 is DDR based, so there is no RDRAM. A Pentium III Tualatin would definitely smoke it in every day applications.

[@BrunoCabralPeixoto]

You use defrag in computer Hard Disk?

[@lGuileWilliamsl]

You didn't check the PIO/DMA settings in Device Manager! 😣 Also remember the computers back then were only as fast as the Front Side Bus(FSB) Memory speed is important. Lastly, Maxtor HDDs were somewhat underperforming back then. I used to have an 80GB 7200 RPM drive with 8MB cache that was fast, but for some reason the 60GB 7200RPM drive with 2 MB cache was slower. Smaller cache maybe?

[@jamespalmer5960]

it is possible that the same principle of single task oparions are quicker on a slower CPU build. where the p4 architecture could of been changed to operate with the chip structure. bug fixes leads to hardware changes (updated CPU architecure)

[@Hypn0s2]

I never had luck with Maxtor drives. I wonder if you can test it with an old ISO of Seagate Seatools - if you can find it? Those old Windows BIOS update utilities still scare me. Modern ones are OK.

[@Hypn0s2]

Also I really don't think this is totally a driver issue. When I would work on these PCs in the era, I would of checked hard drive health. Again, especially since it is a Maxtor. Seagate Seatools use to do a good job of drive self tests. I would usually do both short and long drive self tests with Seagate Seatools. Did you check the drive for bad sectors?

[@rogert151]

i have one of those ShuttleX mini boxes and when i used the same IDE channel for the HDD and Cdrom the transfer speeds and boot times slowed down to a crawl

[@morrisz2]

That P4 1.7gHz looks like an early Willamette P4. Those were actually slower per clock speed than the P3 due to the long pipelines. If you clock them at the same GHz the P3 would win.

[@longcat266]

Simple answer imo is its a willamette, not much faster at the high end than a pentium iii possibly even slower. Even worse if its an sdr equipped machine.

[@tyta1]

My first thought was dried up thermal paste since thermal throttling was introduced with the Pentium 4. But I guess you can rule that out now since you removed and reinstalled the CPU.

[@g4z-kb7ct]

That's a common misconception spread by noob youtubers. Do a real test. If the temp is 90C with dry thermal paste changing to new paste the temp will only drop about 5% or at best 10%. It's not as important as people think. PCs run for dozens of years with the same thermal paste and it still works the same way regardless. Plus or minus 5 percent isn't going to be visually noticeable. It's time to re-write your out-of-date rule book LOL!

[@ppp3435]

P4 with 256 cache and 400 MHz bus + SDR memory = extraordinary garbage. Thou, I'm glad that playing with drivers helped a bit.

[@johanzuidema353]

One of the issues with the p4 is a long,long pipeline. One wrong branchprediction and it had to flush the pipe, wichs costs countles cycles.

[@Lurch-Bot]

HDD sounds like it is dying. I've played around with P4s in recent years and ran Windows 10 on a P4 HT. I have a bunch of P4 hardware. The only P4 I actually use is a 2003 Vaio laptop with a desktop 2.8CPU. It is running the launch version of XP. I really don't use XP much but need it for the Brothers in Arms games. One of these days, I'll figure out how to crack the DRM so I can run them on one of my Windows 7 machines. Windows 7 is a great OS that I never used until recently. it will run all XP games (except those with Starforce or SecuROM) and most 9x games. Sadly, game services are abandoning it. Up until recently, I played a lot of my older Steam games in Windows 7.

[@BadManiac]

A. Pentium 4 is slower than Pentium 3 clock for clock, so if you have a Tualatin 1.4 Ghz in the compaq, it will be very similar to that 1.7Ghz P4. B. The P4 uses PC-133 SDRam, which it was never originally built for. C. the SDRam along with the limited BIOS leads me to believe it's a budget oriented motherboard, which might be limiting the performance too. Basically don't expect miracles, a good, well built Compaq P3 is absolutely not guaranteed to be slower than a budget oriented P4. If the P4 was on a high end motherboard with DDR things might be different.

[@xsc1000]

SDRAM was limit of the first 845 chipset due to Intels relation with Rambus.

[@RWL2012]

a 1.4GHz Tualatin is *way* faster than a 1.7GHz Willamette.

[@Trancelistic]

Glad I was on Duron and Athlon at this time.

[@RetroAnachronist]

The P4 was a slug from the get go. That’s why Intel eventually abandoned it and went back to the P3 design for the Core series CPUs.

[@mortrek]

Iirc, M/Core was a hybrid and also an evolution from both. They took the best parts of both and tweaked it significantly.

[@larrysrandomness3237]

Run the SFC command?