I used to fix those using empty 0.5 and 0.7 mechanical pencils. You insert the pin in the metallic tube at the tip, which gives you excellent control on where you apply the bend force along the pin.
Weird. I was thinking for sure 160Mhz would do the trick - just pure brute force - but alas that FPU is even weaker than I thought. I wonder why no one iterated on it over the relatively long lifespan of socket 3. Nice work on those pins too, wow! Looking forward to your POD video...that was the last socket 3 I used, which I bought from a yard sale for $1 back in the late 90s. It was the first time I had seen Quake playable.
Back in the day I had the IBM Blue Lightning 100mhz, whichI thought was equivalent to 486 but I'm reading that it was really a souped up 386. I was able to play games with it and I believe a Hercules card (?) and a Soundblaster 64 Gold ISA (I still have these devices somewhere). Hard to believe any of us are talking about this stuff from almost 30 years ago but I do find it interesting.
I really like how you focus on tomb raider. Can you put together the fastest hardware available at the time when this game came out for pc, and show us what that is? Can you show us the fastest newest possible computer that can run this in DOS properly?
I made a video about Tomb Raider 11 month ago with a Pentium 120 and a Voodoo 1. The game was running quite well. The Voodoo 1 was just on the market and Intel had the Pentium 200 (non MMX). That probably was the best you could have had at the time Tomb Raider was released.
A Mechanical Pencil helps a lot if you can lift the pins a little bit first and then slide the pencil over it and bend the rest of the way using the pencil. But anyways, nice work!
If only for asthetics, i personally would have repaired that broken corner with a bit of milliput or something similar, and filed it down with a nail file. Not that it would make any difference in operation, but it would give it a nice finish.
@@bitsundbolts yes that's true. I think my OCD would have forced me to do it. Amazing job on those pins btw. You did a marvelous job. I remember 12 year old me building my first pc with whatever I could get my hands on cheaply, and having to do a pin straightening job on the CPU. I snapped 3 pins off, and I had 2 use a paperclip to make new ones because I couldn't think of another way to repair it. It worked though, right up until I was around 15 and could afford a better machine.
I wish he did get all the different 486 133 just to find out what is different if anything about the ADH ADY ADW ADZ AND BGC CPU'S that would be an awesome video to see why the different codes for the basically same cpu
I think the model numbers are mainly to distinguish operating temperature. I doubt we will see any difference in performance, but I may be wrong (e.g. different size in level 1 cache)
@@bitsundbolts I to doubt it but who knows till you check them out against one and other there could be some difference that no one knew about in terms of performance between the different models
Get a mechanical pencil and use the tip of it you can insert the CPU pins into the tip of the pencil and then use the pencil as leverage to bend the pin back into position
Ah, I didn't know much about this stuff back then. A friend of mind had a BIG TOWER 486 system and I was jealous about his computer. The fact that his was a DX2-66 and mine was a DX4-100 was irrelevant because I had a smaller case 😂
Hm, good question. I know what you mean. I think all the other 486 CPUs I've tested showed exactly the same behaviour in SpeedSys. I'll check the other footage
Awesome work getting that CPU restored! Those pins were in really bad shape. Using the broken socket 3 socket top as a stencil is an excellent idea, I will have to see if I can find one to use for that purpose. I have quite a few 486 chips that look fine to the eye but have a pin or two off a bit that makes inserting them feel a bit sketchy.
Hello Mike! Yes, sometimes it is just one or two pins that make the entire experience of installing a CPU in the socket less enjoyable. The socket cover I use was from a board where the CPU was installed wrongly. One of the pins melted the cover and destroyed the socket. Hope you will find also a cover and get those CPUs back into pristine condition!
It is unfortunate that many of those old CPUs are treated that way. But I guess if all you have planned for old electronics to be recycled, then it doesn't make a difference if the pins are straight or not. I am happy that this CPU got another chance to run a game from that same era.
on tomb raider 2, i absolutely need a voodoo to be able to play on my compaq socket 5 pc with a p 180 overdrive mmx, software mode is too slow i need to try tr1
By 1996 for gaming even the budget gamer didn't have a reason to get a 5x86 CPU, as a Pentium 75 PC could be built for 100 USD more than a 133 MHz 486 PC and get much better performance in games like Quake. For 1k USD you could build a Pentium 120 PC.
A Pentium 75 definitely would have a better floating point unit and would be beneficial to run Quake. But maybe not everyone was ready to replace their motherboard and potentially the memory. A CPU replacement would be much cheaper than upgrading to a Pentium system.
@@bitsundboltsthis why AM4 is the greatest platform to have been around in decades. Intel had us used to changing platforms because it would add or remove one or two pins and with the same core count. It is good to see Intels lunch being eaten.
Excellent work! Back in those days I ended up having to run my AM5x86 CPU at 120MHz (40MHz x3) as I could not force a 2x or 4x multiplier in the 486 mainboard I had at the time. I remember trying to play Tomb Raider on that machine and it was not very enjoyable and downright unplayable in places. However, any game that used the Doom engine or similar 2.5D FPS game engine all were a great time! To get decent performance in SVGA software mode in Tomb Raider you needed a Pentium 166MHz or even a 200MHz. Examples like that are one big reason why 3D accelerators like the original 3dfx Voodoo cards were such a huge deal. It made games like Tomb Raider that were mostly playable in VGA mode on a Pentium PC and allow you to run them in SVGA 640x480 resolution at a smooth (for the time) 30 FPS or higher and allowed for 16bit (65K) color modes instead of 8bit (256) in DOS as well!
Choralone422 The MMX 120 was released just after the Voodoo driver for Tomb Raider, needed that faster MMX 120 Pentium for it. Was on pentium 60, needing better build.
I admire your precision and patience to rescue a CPU in this condition. How much time did it take you to fix all pins? Thanks for your videos. Always a wonderful throwback to the past.
I fixed this CPU back in December :) but I remember spending a whole afternoon on the pins, soldering, and refinement. I must have spent three to four hour on this CPU.
At this point I think you should build a tiny hollow pin straightening tube/pen. Great effort, I had a box of CPU's in similar condition and it was not fun, the old socket thing is a good trick though!
That would be a good idea to build something like this. Although I am done for now, there is a box with socket 7 CPUs waiting :) Such a tool could reduce the time immensely because many pins would probably take a fraction of the time when using such a tube with perfect properties for a specific socket. The more difficult pins can still be fixed manually. I'll think of something.
@@bitsundbolts Maybe something like soldering needles? I'm using them to straighten legs of TH components when desoldering. Pack of 8 different sizes is ~2€ on AliExpress.
@@bitsundbolts I saw pictures of someone using Pentel Graphgear 1000 automatic pencil for AMD pins. That straightening tool should be made as pliers with inner diameter slightly smaller than the pin diameter, so you could also straighten the twisted pins. By simply closing the pliers, you'd squeeze and straighten the twist. Probably buying some bend or straight smooth jaw long nose pliers and then machining them to specs would be the best idea. You could easily punch the channel for pins on the surface of those smooth jaws.
@tezcanaslan2877 Try it. It's the most amazing automatic pencil I've ever used. And I've used many - from costing 1 Euro to even more expensive. But this Pentel is amazing.
Oh, nice tip about the finishing touches using the socket lid, I have seen something similar by an extreme overclocker that purchase old CPUs in bulk to search for very good ones. I really like your method.
It is just so beautiful to see your work on straightening the pins... And woah!? ADZ with write back L1? I thought only ADW (of those 3-letter variants) can do that... NB! 3x50 might give better results than 4x40, as with higher FSB the memory access and L2 cache access becomes faster, too. When using VLB videocard, it helps that too. Maybe worth to try?
You are absolutely right. I tested 150 MHz as well, however, in SpeedSys, 160 gave better results. I will test this once we get the Voodoo in there. I also want to try 200 MHz. At 3.45V, the CPU doesn't boot. Maybe I can try my luck at 4V.
@@bitsundbolts yeah, speedsys and other such small benchmarks usually do some small loop test or something similar. And when that test code can (mostly) fit in L1 cache, then the bus speed doesn't matter. But with "bigger" tests (for ex. Doom run) the bus will be heavily used and difference may start to appear. ;)
I want to change the motherboard later in a "tuning" video because the Soyo only works with FPM memory. I did see a big improvement in memory bandwidth when running at 50 and even 40 MHz bus speed. I agree that a proper test using some games will be much better than those synthetic benchmarks. More videos to make :)
@@bitsundboltsI was about to ask if you were going to try 50x4, so heck yea. Actually had a few 5x86's back in the day and we ran them all at 160 - and assumed they all would. Based on what I've read nowadays we just got stupid lucky with the ones we had apparently! That said we never got 200mhz to work, but we also didn't have any boards that would officially do 4v, and I imagine we'd have been too chicken to try it even if they did!
Hehe, what's the worst that can happen at 4V - the CPU dies. Oh well, content is made and the videos are online. I really hope that the CPU can do 200, that would be nice!
Just awesome! The bending adventure is real. I'd really liked the detailed review of this CPU incl. overclocking. Socket 3 is not my cup of tea (I started with socket 5) so I have a question: why are there less pins on the CPU and more possible contacts/holes on the CPU socket? Looking forward for the next VooDoo Repair; it looks really bad! Wonder what people do with their hardware? Use it as a door stopper!? Thank you for saving all this hardware!
To my knowledge, Intel designed socket 3 with future CPUs in mind. The only CPU I know of that uses the outer rows, is the Pentium OverDrive. I'll have a second video about this CPU soon once I get the fan. The outer row is mainly ground and power pins.
The extra pins are mostly ground and power. The pentium overdrive used those. There may possibly have been some other niché stuff that used them but I'm not sure (e.g. one of those evergreen upgrade things).
The pins look like gold coated steel, ie, you would have to get them glowing before they get in to the plastic stage - lol -. I would try a pointy hobby blade for that Diamond.
If you want to go wild: Try a 60MHz bus by soldering a line from the S2 pin on the clock chip to ground (or S0 from memory, it's the one who's not connected to any of the two jumpers). I've done that and gotten the board to boot with a 5x86 at 3x60 . The clock chip and the board is capable of doing it, it's just not connected since the VLB bus might start glitching, but if you do not use it ... My "record" on this motherboard booting and running doom is 150mhz (3x50), but yours looks in a way better state than mine. Actually ... mine died and two address lines have leakage current comming from somewhere glitching them.
Oh, interesting! I am always available for a good mod! I'll have a look at this. The only issue I have with this board is that it only supports FPM memory. However, I want to try something I found on a German forum: BIOS swap from a different board...
@@bitsundboltsIf you want to do that you might want to check the SIS496 stepping, my board has a non-EDO compatible earlier stepping, though information about it is not easy to come by which support what but higher letters in the alphabet is in general better afaik. The specsheet of the clockchip is on theretroweb site for this board. S2 (pin 17) is connected high, grounding it will give you the option for the 30/60/66.6/80mhz bus clocks the pll supports. There is a screwhole nearby which is easy to solder a wire to (+pinheader since the lower bus speeds need S2 to be high to allow you to disconnect it).
Still have one of these laying around somewhere... I can only imagine the pins are probably pretty smashed. It did run 98se well enough, with 56mb 72pin memory, a 512kb opti svga vlb card, pci usb card, opti sound card with a volume wheel. Played starcraft surprisingly well. Even handled MPEG2 avi video playback. Overclocked to 150 mhz. Not sure what happened with the rest of the system but the cpu is around somewhere
Nice work with the CPU pins but the best performance I think is with the pentium overdrive PODP5V83 (83MHz) OC to 120MHz or more if the CPU is stable of course
...and 160MHz was not enough. I'm not exactly surprised. I'd guess that P200mmx or K6 could run TR SW-renderer with well-tuned VGA-card. But most likely it needs something like P200mmx or K6 overlocked to at least 225/75MHz with overclocked VGA. So something Super Socket 7. Software-rendered 640*480 needs at least Pentium3 with 133MHz FSB (EB-variant) or early Athlon.
If your mobo will provide 50mhz FSB, the 3x multiplier for 150mhz runs faster than the 40x4 setting because it's just able to shove memory around faster. You end up roughly equal in most titles to a P90. It was my understanding the reason they didn't sell them as 150 or 160 is that the majority of 486 motherboards only supported 25 and 33mhz. The 40mhz was common, but true 50mhz boards (and the true 50mhz 486's) were rare. But if your board supports it, 50x3 for 150mhz is the peak performance mark. A LOT of the x4's get choked / throttled by the bus not being able to feed it fast enough.
your AMD X5 is a Pentium 586 clone. it's not a 486 processor. to play Tomb Raider properly you need at least a Pentium 133Mhz with MMX capabilities to accelerate 3D calculating. few months later, AMD provides their processors with the 3DNow technologie to beat the Intel MMX. on my part I was using a Cyrix 200Mhz (very good ans cheap processor) with a Voodoo 1 video card to replace my Pentium 100Mhz and be able to play Tomb Rider.
My first x86 processor after I sold my Amiga. It was utter garbage! Absolutely the most horrible processor I have ever met in my life. And that I have seen many different processors during my seventeen-year career as an IT technician! Performance lower than Pentium 75, completely dismal compatibility with any chipset and GW DOS was an absolutely unsolvable problem. Duke Nukem 3D, DOOM, Descent... it was all spilling into ASCII characters at random moments. The only thing that worked without problems was Windows 95 and mine sweeper... Replacing this processor with a Pentium 120 was then a ticket to a problem-free paradise.
For heat to affect viscosity of metal to the degree it bends easy, it would require much greater temperatures than what crystal and especially connecting strands can handle. It's bs, no point in heating the the pins.
It's actually incredible Tomb Raider runs on those CPUs alone. It's kind of asking a lot... VERY impressive. And I think this was before 3D Now or any of the other instruction sets that would have helped. Very impressive indeed. It's been so long I'd underestimated what those CPUs could do.
I remember visiting my friend's place, where he showed me Tomb Raider being played on a CPU like this, and it was unbearably slow (in higher resolution), in fact I was wondering how could he play it like that, but I didn't say anything because I didn't want to hurt his feelings. Also, meanwhile I think I still had only a 386SX at home, which was related to a sore disappointment of my childhood... My mother bought that 386 PC for me in used condition, and it turned out to be an SX model, not a DX. 🙈 Btw, to be honest I totally didn't get the point of 3D games for the first few years... they were so ugly compared to the already relatively refined 2D games that used sprites. I think one of the first games where 3D was appreciable for me personally was Little Big Adventure, where the characters are 3D; it was such a great game! I still remember the computer magazine where I saw it first - that was the period when games started to get pretty impressive, not much later e.g. Command & Conquer also came out. Subsequently I was lucky enough to find Little Big Adventure on our school's computer; I still remember how excited I was when I saw that three letters folder name (lba) in Norton Commander. If I remember correctly, it took an ungodly amount of floppy disks to copy, and one of them was bad, so I had to go back to the school and copy it again.
My first computer was a Packard Bell from Sears. It had a 400 MHz Pentium ii. I actually did play Tomb Raider on it. it had windows 95 which I upgraded to Win 98 SE. I only had that PC for 3 or 4 years. Got it in 1999.
Very good job with the pins. I've got some kind of 100MHz AMD 486 from a poker machine board. I find it pretty boring these days, it's neither a limited 486 nor a fast Pentium.
I played the Tomb Raider 1 demo on this CPU and it used to run perfectly, something could be wrong here. Is it possible that my slightly better graphics card with 2mbs of vram did a better job at 2d accel or something?
It's incredible how Intel seems to have been unmatched in FPU performance at least from the Pentium 1 until the Athlon arrived. Did Cyrix and AMD just not care or was it too hard to match Intel in this regard?
Have you considered using a mechanical pencil to help position it? I've used a 1.0 / 1.3, / 1.5 mechanical pencil a lot. I used the heat gun on the metal mechanical pencil to heat the pins and bend them more easily
I found a Am5x86 in a box of old parts at a local PC shop that was doing a closing down sale among other chips, funnily enough the date code puts it as 3 weeks younger than yours and thankfully only one pin has a very minor bend, although I need the rest of the system to go with it. Also found a 256k Pentium Pro there too, very keen to try that out.
I think this is a great result, how much was this CPU? Because if this was a top model than it was a bad investment even tho I think the performance increase is pretty noticeable to me.
to straighten bent pins I had used medical syringe needle with cut off sharp tilted front edge. Make sure the needle is big enough hole to fit the pin.
Wow what a substantial upgrade one could've made for their aging PC in the mid 90s to get a couple more years out of it before having to replace it. I wonder how it'd do with Quake. It might be just nipping at the heels of being able to run Quake playably (by the standards of the day).
The weakness will be the floating point unit to run Quake on a 486, but overall, I really like the socket 3 platform. So many CPUs and different flavours.
No, it isn't :) I have read somewhere that it was basically just a market name thing. It went like this: when Intel originally invented the thing called multiplier, they came up with multiplier of 2. So they marked it DX2. But for future, they wanted more granularity, so they planned to use halves, not only integer numbers, just like they were being used for many years to come in Pentium 1/2/3/n lineup :) So the next multiplier was planned to be 2.5, for which a market name DX3 was planned. But the processors with this multuplier were never released. However, it was already decided, so when multiplier of 3 came out, the processors were marked DX4. Thus, AMD did the next best thing they could when adding a new feature to an already abandoned (by Intel) platform, and marked the multiplier of 4 as X5 :) At least, this is what I heard on how these strange DX numbers arose.
What might be interresting is the Intel Pentium Overdrive. That socket 3 cpu can run at 100 Mhz and will probably outrun the 5x86@160Mhz. On the other hand...it's a Pentium. It would make that 486-system less authentic.
As a kid, I used to play Tomb Raider on an 486 CPU from AMD, clocked at 120 MHZ. In the low resolution mode, the game was much smoother than with this CPU. It must have been an almost perfect 30 fps, and I don't think it is just nostalgia or bad memory. Is there a reason why a slower 486 CPU could significantly outperform a faster one?
Maybe you had a different motherboard, better memory (I use FPM because the board does not support EDO), and better timings. The system is not yet tuned for better performance. But I would like to change the motherboard for one that supports EDO memory.
Great job straightening those pins! They were really in bad shape. So much effort to bring this CPU back to life.
Thank you! I am happy that I could save this CPU, even if it took a lot of time.
@@bitsundbolts Would have been way easier if u used the tip of a mechanical pencil to straighten those pins
The thumbnails for this video series are S tier.
Thank you!
Was about to say the same thing lol
I used to fix those using empty 0.5 and 0.7 mechanical pencils. You insert the pin in the metallic tube at the tip, which gives you excellent control on where you apply the bend force along the pin.
Let's go!
Phiiiiiil!
Love the content, it's nice to meet other people with the same passion in life for old technology
Algumas pessoas nascem com uma quantidade insana de paciência.... Trabalho incrível de recuperação da CPU. 😊
Thank you!
awesome BuB, the level of perfection in repairing stuff you try to achieve is mesmerizing to me. it's therapy to my brain
Glad to hear that! Always happy to restore old hardware that brings back memories - or provides therapeutic sessions for the brain :)
skillful hands straightened those pins. good job! liked and subscribed.
Thank you very much!
Weird. I was thinking for sure 160Mhz would do the trick - just pure brute force - but alas that FPU is even weaker than I thought. I wonder why no one iterated on it over the relatively long lifespan of socket 3. Nice work on those pins too, wow! Looking forward to your POD video...that was the last socket 3 I used, which I bought from a yard sale for $1 back in the late 90s. It was the first time I had seen Quake playable.
Great job buddy. Impressive job of patience and delicacy. 🍟
Thank you!
Chapeau! 👏 Great job and very satisfying.
Thank you
Back in the day I had the IBM Blue Lightning 100mhz, whichI thought was equivalent to 486 but I'm reading that it was really a souped up 386. I was able to play games with it and I believe a Hercules card (?) and a Soundblaster 64 Gold ISA (I still have these devices somewhere). Hard to believe any of us are talking about this stuff from almost 30 years ago but I do find it interesting.
Good job dude, really impressive
Thank you!
@BitsundBolts
As usual, well gone on the pins! This is likely the best non-Pentium CPU you can get for Socket3, especially with the easy OC.
reminds me of playing tomb raider 2018 on a intewgrated graphics on 7700k
I really like how you focus on tomb raider. Can you put together the fastest hardware available at the time when this game came out for pc, and show us what that is? Can you show us the fastest newest possible computer that can run this in DOS properly?
I made a video about Tomb Raider 11 month ago with a Pentium 120 and a Voodoo 1. The game was running quite well. The Voodoo 1 was just on the market and Intel had the Pentium 200 (non MMX). That probably was the best you could have had at the time Tomb Raider was released.
That's pretty interesting. 86box or the like wouldn't have too hard of a time emulating a pentium 200 and a voodoo1@@bitsundbolts
A Mechanical Pencil helps a lot if you can lift the pins a little bit first and then slide the pencil over it and bend the rest of the way using the pencil. But anyways, nice work!
Me going to like the video, realizing I already liked it. This comments serves as a double like. Excellent video.
Thank you so much!
Even if the pins don't get easier to bend from heating. It should, at the very least, make the metal less prone to break. In theory anyway
If only for asthetics, i personally would have repaired that broken corner with a bit of milliput or something similar, and filed it down with a nail file. Not that it would make any difference in operation, but it would give it a nice finish.
I can still do that. The corner is a bit of an eyesore, but it also is proof that this is the same CPU before and after fixing the pins.
@@bitsundbolts yes that's true. I think my OCD would have forced me to do it. Amazing job on those pins btw. You did a marvelous job. I remember 12 year old me building my first pc with whatever I could get my hands on cheaply, and having to do a pin straightening job on the CPU. I snapped 3 pins off, and I had 2 use a paperclip to make new ones because I couldn't think of another way to repair it. It worked though, right up until I was around 15 and could afford a better machine.
A very creative solution to use paperclip. Glad that it worked out for you!
I wish he did get all the different 486 133 just to find out what is different if anything about the ADH ADY ADW ADZ AND BGC CPU'S that would be an awesome video to see why the different codes for the basically same cpu
I'll try to get all of them. The ADY is possible to get, but I have never seen and ADH.
@@bitsundbolts cool be interesting to see if it has to do with better or worse floating point math or something that affects performance in some way
I think the model numbers are mainly to distinguish operating temperature. I doubt we will see any difference in performance, but I may be wrong (e.g. different size in level 1 cache)
@@bitsundbolts I to doubt it but who knows till you check them out against one and other there could be some difference that no one knew about in terms of performance between the different models
Get a mechanical pencil and use the tip of it you can insert the CPU pins into the tip of the pencil and then use the pencil as leverage to bend the pin back into position
Had this back in the day. I so wished for a Pentium :)).
Ah, I didn't know much about this stuff back then. A friend of mind had a BIG TOWER 486 system and I was jealous about his computer. The fact that his was a DX2-66 and mine was a DX4-100 was irrelevant because I had a smaller case 😂
impressive work
Thank you
Armchair engineer here, use an empty Bic ballpoint pen to lift the pins, will save you some time and effort.
I had one! It couldn't play games that required a Pentium, like Diablo 1. I got a message saying it needed a GenuineIntel(R) Pentium to load up.
Mi primer procesador, me lo encontre en ese estado y mi papa me sorprendio reparandolo era de una especie de pentium 2, sin MMX
You should look into vector fields. It kind of looks like all those bent pins. ;w;
Ha, you're absolutely right!
Are you sure the Cache was in Writeback-Mode? The Write-Graph was flat line, indicating Write-Through...
Hm, good question. I know what you mean. I think all the other 486 CPUs I've tested showed exactly the same behaviour in SpeedSys. I'll check the other footage
Awesome work getting that CPU restored! Those pins were in really bad shape. Using the broken socket 3 socket top as a stencil is an excellent idea, I will have to see if I can find one to use for that purpose. I have quite a few 486 chips that look fine to the eye but have a pin or two off a bit that makes inserting them feel a bit sketchy.
Hello Mike! Yes, sometimes it is just one or two pins that make the entire experience of installing a CPU in the socket less enjoyable. The socket cover I use was from a board where the CPU was installed wrongly. One of the pins melted the cover and destroyed the socket. Hope you will find also a cover and get those CPUs back into pristine condition!
What happened, were they using it as a toothbrush or something? It's amazing just how someone managed to do it to those poor poor pins.
It is unfortunate that many of those old CPUs are treated that way. But I guess if all you have planned for old electronics to be recycled, then it doesn't make a difference if the pins are straight or not. I am happy that this CPU got another chance to run a game from that same era.
Amazing!
on tomb raider 2, i absolutely need a voodoo to be able to play on my compaq socket 5 pc with a p 180 overdrive mmx, software mode is too slow i need to try tr1
would love to see how far you could push the overclock on this chip.
I'll try 200 Mhz at increased voltage later. I already tried at 3.45v, but the system didn't boot.
I am laughing so Hard because of the Thumbnail wkwkwkwkwk
I am glad you do!
So it was stepped on?
it does say Am5 so it must be new
YES! YOU NEED 3DFX VOODOO1
Had this bitch of cpu running on 160mhz but that was all i could get out of it
By 1996 for gaming even the budget gamer didn't have a reason to get a 5x86 CPU, as a Pentium 75 PC could be built for 100 USD more than a 133 MHz 486 PC and get much better performance in games like Quake. For 1k USD you could build a Pentium 120 PC.
A Pentium 75 definitely would have a better floating point unit and would be beneficial to run Quake. But maybe not everyone was ready to replace their motherboard and potentially the memory. A CPU replacement would be much cheaper than upgrading to a Pentium system.
@@bitsundboltsthis why AM4 is the greatest platform to have been around in decades. Intel had us used to changing platforms because it would add or remove one or two pins and with the same core count. It is good to see Intels lunch being eaten.
3.45v?
Please run Crysis on this.
try the pen metod not like this
more easy is
So satisfying, hearing the click/snap in to place!
And the single beep afer throwing the switch.
Excellent work! Back in those days I ended up having to run my AM5x86 CPU at 120MHz (40MHz x3) as I could not force a 2x or 4x multiplier in the 486 mainboard I had at the time. I remember trying to play Tomb Raider on that machine and it was not very enjoyable and downright unplayable in places. However, any game that used the Doom engine or similar 2.5D FPS game engine all were a great time!
To get decent performance in SVGA software mode in Tomb Raider you needed a Pentium 166MHz or even a 200MHz. Examples like that are one big reason why 3D accelerators like the original 3dfx Voodoo cards were such a huge deal. It made games like Tomb Raider that were mostly playable in VGA mode on a Pentium PC and allow you to run them in SVGA 640x480 resolution at a smooth (for the time) 30 FPS or higher and allowed for 16bit (65K) color modes instead of 8bit (256) in DOS as well!
this is the kind of work to make you go crazy making bent pins straight if you are not carful to take precautions.
Choralone422
The MMX 120 was released just after the Voodoo driver for Tomb Raider, needed that faster MMX 120 Pentium for it.
Was on pentium 60, needing better build.
I admire your precision and patience to rescue a CPU in this condition.
How much time did it take you to fix all pins?
Thanks for your videos. Always a wonderful throwback to the past.
I fixed this CPU back in December :) but I remember spending a whole afternoon on the pins, soldering, and refinement. I must have spent three to four hour on this CPU.
My dude, they make plyers with grippy rubber teeth that wouldn't scratch the pins. Get yoself a pair.
At this point I think you should build a tiny hollow pin straightening tube/pen. Great effort, I had a box of CPU's in similar condition and it was not fun, the old socket thing is a good trick though!
That would be a good idea to build something like this. Although I am done for now, there is a box with socket 7 CPUs waiting :) Such a tool could reduce the time immensely because many pins would probably take a fraction of the time when using such a tube with perfect properties for a specific socket. The more difficult pins can still be fixed manually. I'll think of something.
@@bitsundbolts Maybe something like soldering needles? I'm using them to straighten legs of TH components when desoldering. Pack of 8 different sizes is ~2€ on AliExpress.
@@bitsundbolts
I saw pictures of someone using Pentel Graphgear 1000 automatic pencil for AMD pins.
That straightening tool should be made as pliers with inner diameter slightly smaller than the pin diameter, so you could also straighten the twisted pins. By simply closing the pliers, you'd squeeze and straighten the twist.
Probably buying some bend or straight smooth jaw long nose pliers and then machining them to specs would be the best idea. You could easily punch the channel for pins on the surface of those smooth jaws.
@@Stefan_Kawalec why would someone use such a expensive pencil? Wouldn’t any fixed-sleeve pen would do?
@tezcanaslan2877 Try it. It's the most amazing automatic pencil I've ever used. And I've used many - from costing 1 Euro to even more expensive. But this Pentel is amazing.
Oh, nice tip about the finishing touches using the socket lid, I have seen something similar by an extreme overclocker that purchase old CPUs in bulk to search for very good ones.
I really like your method.
Thank you!
It is just so beautiful to see your work on straightening the pins...
And woah!? ADZ with write back L1? I thought only ADW (of those 3-letter variants) can do that...
NB! 3x50 might give better results than 4x40, as with higher FSB the memory access and L2 cache access becomes faster, too. When using VLB videocard, it helps that too. Maybe worth to try?
You are absolutely right. I tested 150 MHz as well, however, in SpeedSys, 160 gave better results. I will test this once we get the Voodoo in there. I also want to try 200 MHz. At 3.45V, the CPU doesn't boot. Maybe I can try my luck at 4V.
@@bitsundbolts yeah, speedsys and other such small benchmarks usually do some small loop test or something similar. And when that test code can (mostly) fit in L1 cache, then the bus speed doesn't matter. But with "bigger" tests (for ex. Doom run) the bus will be heavily used and difference may start to appear. ;)
I want to change the motherboard later in a "tuning" video because the Soyo only works with FPM memory. I did see a big improvement in memory bandwidth when running at 50 and even 40 MHz bus speed. I agree that a proper test using some games will be much better than those synthetic benchmarks. More videos to make :)
@@bitsundboltsI was about to ask if you were going to try 50x4, so heck yea. Actually had a few 5x86's back in the day and we ran them all at 160 - and assumed they all would. Based on what I've read nowadays we just got stupid lucky with the ones we had apparently!
That said we never got 200mhz to work, but we also didn't have any boards that would officially do 4v, and I imagine we'd have been too chicken to try it even if they did!
Hehe, what's the worst that can happen at 4V - the CPU dies. Oh well, content is made and the videos are online. I really hope that the CPU can do 200, that would be nice!
Just awesome! The bending adventure is real. I'd really liked the detailed review of this CPU incl. overclocking. Socket 3 is not my cup of tea (I started with socket 5) so I have a question: why are there less pins on the CPU and more possible contacts/holes on the CPU socket? Looking forward for the next VooDoo Repair; it looks really bad! Wonder what people do with their hardware? Use it as a door stopper!? Thank you for saving all this hardware!
To my knowledge, Intel designed socket 3 with future CPUs in mind. The only CPU I know of that uses the outer rows, is the Pentium OverDrive. I'll have a second video about this CPU soon once I get the fan. The outer row is mainly ground and power pins.
@@bitsundbolts Thank you very much for your answer! Looking forward now to the VooDoo AND the Pentium Overdrive! .D
The extra pins are mostly ground and power. The pentium overdrive used those. There may possibly have been some other niché stuff that used them but I'm not sure (e.g. one of those evergreen upgrade things).
Guess with those bent pins the good old "bend them back by pushing the CPU into the socked"-Method would've done nothing. :D
Yeah, that would have been pretty hard :)
:-D You'd probably completely destroy both CPU and the socket in the process :-)
>39 views 6 minutes ago
Gotta love getting these as soon as they post.
From this day on though shalt be referred to as: The Pinmaster! 😆
Haha, great and what an honor! Thank you!
The pins look like gold coated steel, ie, you would have to get them glowing before they get in to the plastic stage - lol -.
I would try a pointy hobby blade for that Diamond.
If you want to go wild: Try a 60MHz bus by soldering a line from the S2 pin on the clock chip to ground (or S0 from memory, it's the one who's not connected to any of the two jumpers). I've done that and gotten the board to boot with a 5x86 at 3x60 . The clock chip and the board is capable of doing it, it's just not connected since the VLB bus might start glitching, but if you do not use it ...
My "record" on this motherboard booting and running doom is 150mhz (3x50), but yours looks in a way better state than mine. Actually ... mine died and two address lines have leakage current comming from somewhere glitching them.
Oh, interesting! I am always available for a good mod! I'll have a look at this. The only issue I have with this board is that it only supports FPM memory.
However, I want to try something I found on a German forum: BIOS swap from a different board...
@@bitsundboltsIf you want to do that you might want to check the SIS496 stepping, my board has a non-EDO compatible earlier stepping, though information about it is not easy to come by which support what but higher letters in the alphabet is in general better afaik. The specsheet of the clockchip is on theretroweb site for this board. S2 (pin 17) is connected high, grounding it will give you the option for the 30/60/66.6/80mhz bus clocks the pll supports. There is a screwhole nearby which is easy to solder a wire to (+pinheader since the lower bus speeds need S2 to be high to allow you to disconnect it).
Thanks for all this info! I will have a look.
Still have one of these laying around somewhere... I can only imagine the pins are probably pretty smashed. It did run 98se well enough, with 56mb 72pin memory, a 512kb opti svga vlb card, pci usb card, opti sound card with a volume wheel. Played starcraft surprisingly well. Even handled MPEG2 avi video playback. Overclocked to 150 mhz. Not sure what happened with the rest of the system but the cpu is around somewhere
Some of these ADZs can get to 180MHz on stock voltage, and if they can, it's almost guaranteed it can do 200MHz with a slight voltage bump.
Nice work with the CPU pins but the best performance I think is with the pentium overdrive PODP5V83 (83MHz) OC to 120MHz or more if the CPU is stable of course
I hope you are right! So far, I am not happy with the performance of the 486 CPUs. Two more jokers are waiting: The Pentium OverDrive and the Voodoo 1
POD isn't interesting. Then better to go to socket 5 or 7 platform
...and 160MHz was not enough. I'm not exactly surprised. I'd guess that P200mmx or K6 could run TR SW-renderer with well-tuned VGA-card. But most likely it needs something like P200mmx or K6 overlocked to at least 225/75MHz with overclocked VGA. So something Super Socket 7. Software-rendered 640*480 needs at least Pentium3 with 133MHz FSB (EB-variant) or early Athlon.
If your mobo will provide 50mhz FSB, the 3x multiplier for 150mhz runs faster than the 40x4 setting because it's just able to shove memory around faster. You end up roughly equal in most titles to a P90.
It was my understanding the reason they didn't sell them as 150 or 160 is that the majority of 486 motherboards only supported 25 and 33mhz. The 40mhz was common, but true 50mhz boards (and the true 50mhz 486's) were rare.
But if your board supports it, 50x3 for 150mhz is the peak performance mark. A LOT of the x4's get choked / throttled by the bus not being able to feed it fast enough.
your AMD X5 is a Pentium 586 clone. it's not a 486 processor. to play Tomb Raider properly you need at least a Pentium 133Mhz with MMX capabilities to accelerate 3D calculating. few months later, AMD provides their processors with the 3DNow technologie to beat the Intel MMX. on my part I was using a Cyrix 200Mhz (very good ans cheap processor) with a Voodoo 1 video card to replace my Pentium 100Mhz and be able to play Tomb Rider.
My first x86 processor after I sold my Amiga.
It was utter garbage! Absolutely the most horrible processor I have ever met in my life. And that I have seen many different processors during my seventeen-year career as an IT technician!
Performance lower than Pentium 75, completely dismal compatibility with any chipset and GW DOS was an absolutely unsolvable problem. Duke Nukem 3D, DOOM, Descent... it was all spilling into ASCII characters at random moments. The only thing that worked without problems was Windows 95 and mine sweeper...
Replacing this processor with a Pentium 120 was then a ticket to a problem-free paradise.
For heat to affect viscosity of metal to the degree it bends easy, it would require much greater temperatures than what crystal and especially connecting strands can handle. It's bs, no point in heating the the pins.
I was already on the Intel 200mmx, iirc, and Matrox graphics.... played TR with the Matrox and a 3dfx... it was super smooth.
It's actually incredible Tomb Raider runs on those CPUs alone. It's kind of asking a lot... VERY impressive. And I think this was before 3D Now or any of the other instruction sets that would have helped. Very impressive indeed. It's been so long I'd underestimated what those CPUs could do.
I remember visiting my friend's place, where he showed me Tomb Raider being played on a CPU like this, and it was unbearably slow (in higher resolution), in fact I was wondering how could he play it like that, but I didn't say anything because I didn't want to hurt his feelings. Also, meanwhile I think I still had only a 386SX at home, which was related to a sore disappointment of my childhood... My mother bought that 386 PC for me in used condition, and it turned out to be an SX model, not a DX. 🙈 Btw, to be honest I totally didn't get the point of 3D games for the first few years... they were so ugly compared to the already relatively refined 2D games that used sprites. I think one of the first games where 3D was appreciable for me personally was Little Big Adventure, where the characters are 3D; it was such a great game! I still remember the computer magazine where I saw it first - that was the period when games started to get pretty impressive, not much later e.g. Command & Conquer also came out. Subsequently I was lucky enough to find Little Big Adventure on our school's computer; I still remember how excited I was when I saw that three letters folder name (lba) in Norton Commander. If I remember correctly, it took an ungodly amount of floppy disks to copy, and one of them was bad, so I had to go back to the school and copy it again.
My first computer was a Packard Bell from Sears. It had a 400 MHz Pentium ii. I actually did play Tomb Raider on it. it had windows 95 which I upgraded to Win 98 SE. I only had that PC for 3 or 4 years. Got it in 1999.
Very good job with the pins. I've got some kind of 100MHz AMD 486 from a poker machine board. I find it pretty boring these days, it's neither a limited 486 nor a fast Pentium.
Hi. One of the better thumbnails I've seen on UA-cam recently. I laughed up to my armpits Lara straightening CPU pins with a stick ha ha ha ha
Haha, great! Glad you like it and that it made you laugh!
I usually use a mechanical pencil to straighten CPU pins...
Thankfully, now I have no CPUs with pins on them anymore! 😂
I played the Tomb Raider 1 demo on this CPU and it used to run perfectly, something could be wrong here. Is it possible that my slightly better graphics card with 2mbs of vram did a better job at 2d accel or something?
It's incredible how Intel seems to have been unmatched in FPU performance at least from the Pentium 1 until the Athlon arrived. Did Cyrix and AMD just not care or was it too hard to match Intel in this regard?
I think I had that AMD chip. Or something similar. It came as DX2/80 but I OCed it to DX4 100. Wild times!
Have you considered using a mechanical pencil to help position it?
I've used a 1.0 / 1.3, / 1.5 mechanical pencil a lot.
I used the heat gun on the metal mechanical pencil to heat the pins and bend them more easily
I vividly remember running Tombraider back in the day on a 486-SX at similar frame rates and resolution..
I found a Am5x86 in a box of old parts at a local PC shop that was doing a closing down sale among other chips, funnily enough the date code puts it as 3 weeks younger than yours and thankfully only one pin has a very minor bend, although I need the rest of the system to go with it.
Also found a 256k Pentium Pro there too, very keen to try that out.
Very nice! I'm still amazed how Pentium Pros make other CPUs look really small 😁
I remember getting the Cyrix processor for my 486. It was not much of an improvement over my 486 DX
The last time I fixed anything that bad was a K5 that of all places was from off Amazon, didn't lose any pins thankfully.
I hope to find a K5 some day as well. Never used one of them. Great job not to lose pins!
I think this is a great result, how much was this CPU? Because if this was a top model than it was a bad investment even tho I think the performance increase is pretty noticeable to me.
can my intel i9 13900k and RTX 4090 play this game? Its looks a very heavy game to me
Hi, you could use metal pen refill to fix the pins back. Much easier than to work with toothpicks.
9:27 lol. that's super weird. i'm watching this in fullscreen on an old laptop, and an even older display.
I got Tomb raider for Christmas that year, bought the MMX Pentium PC for it, better fit for my Voodoo card.
Use the shaft of a ball point pin. Did this all the time in the 90s and that method was by far better than using toothpicks or knife blades.
to straighten bent pins I had used medical syringe needle with cut off sharp tilted front edge. Make sure the needle is big enough hole to fit the pin.
Use mechanical pencil tube would helps strength enforced line up the pin
Wish I had a cheap LS486E board... my ADZ unfortunately only runs at 160MHz...
Have you tried to increase the voltage to 4V? Mine also doesn't boot when leaving the voltage at 3.45 volts
To align the legs, it is most convenient to use a syringe needle with the tip cut off
Wow what a substantial upgrade one could've made for their aging PC in the mid 90s to get a couple more years out of it before having to replace it. I wonder how it'd do with Quake. It might be just nipping at the heels of being able to run Quake playably (by the standards of the day).
The weakness will be the floating point unit to run Quake on a 486, but overall, I really like the socket 3 platform. So many CPUs and different flavours.
i found a mechanical pencil often fits snugly and allows you to bend at the base and then get the kinks out too... without tools
The old trick of increasing the bus to get great overall performance increase!
Impressive OC and funny that it even gets a higher P Rating with it. xD
it has a corner chipped and pins pressed in, way passed bent... like someone dropped it and stepped on it
Alas an amd speed is always'*'
They are never honest about heir speeds.
uh, nobody was playing this on this chipset... why did you try this?
I've always thought it should be called the DX8, because clearly the multiplier is the log2 of the DX number plus one.
No, it isn't :) I have read somewhere that it was basically just a market name thing. It went like this: when Intel originally invented the thing called multiplier, they came up with multiplier of 2. So they marked it DX2. But for future, they wanted more granularity, so they planned to use halves, not only integer numbers, just like they were being used for many years to come in Pentium 1/2/3/n lineup :) So the next multiplier was planned to be 2.5, for which a market name DX3 was planned. But the processors with this multuplier were never released. However, it was already decided, so when multiplier of 3 came out, the processors were marked DX4. Thus, AMD did the next best thing they could when adding a new feature to an already abandoned (by Intel) platform, and marked the multiplier of 4 as X5 :)
At least, this is what I heard on how these strange DX numbers arose.
What might be interresting is the Intel Pentium Overdrive. That socket 3 cpu can run at 100 Mhz and will probably outrun the 5x86@160Mhz.
On the other hand...it's a Pentium. It would make that 486-system less authentic.
I am waiting to get a fan for it. I will make a video some time in March.
I once bent the pins of my used CPU just for fun of it, now I regret it.
As a kid, I used to play Tomb Raider on an 486 CPU from AMD, clocked at 120 MHZ. In the low resolution mode, the game was much smoother than with this CPU. It must have been an almost perfect 30 fps, and I don't think it is just nostalgia or bad memory. Is there a reason why a slower 486 CPU could significantly outperform a faster one?
Maybe you had a different motherboard, better memory (I use FPM because the board does not support EDO), and better timings. The system is not yet tuned for better performance. But I would like to change the motherboard for one that supports EDO memory.