Unlocking the Duron and Athlon using the pencil (2007) | Hacker News Reader

Unlocking the Duron and Athlon using the pencil (2007)

computer-communication.blogspot.com

134 pointsthunderbong85 comments

Ahh, the mention of Celeron 300A gives me a good dose of nostalgia. Sorry for the tangent. There was a local retailer and warehouse that turned out to be ran by Russian mobsters (I didn’t know, I was a naive teen), but they had everything under the sun and the prices were (probably literally) a steal, although retail packaging was curiously rare. The guy behind the counter had a box full of loose 300As and let me pick my pair, and I went back and forth between home and the warehouse a couple of times until I had a stable pair. Seeing two of those things hit 450 stable with just a flip of a BIOS setting, for only $100something a piece felt like I had pulled the wool over the world’s eyes.

That was my first SMP system, running NT4. Ended up hosting Starsiege: Tribes, Q3, and a bunch of other stuff (often simultaneously) at college. However, I dropped out after my first year and couldn’t host at home, so the machine was largely unused, and I sold it. I regret that.

Yup! The BP6 and the VP6 from Abit. Good memories!

Remember old boards from Epox? Or the TNT2 ultra?

But yeah my favorite was getting SMP for free using celerons when they weren’t supposed to be able to do it.

Yes, the Abit BP6 was the one! Nestled inside an InWin Q500 full tower case[0]. I was a pretty big Voodoo guy initially. I already had a Voodoo3 but I was working at CompUSA and befriended the Creative Labs rep. He fixed it so I got the TNT2U and I was impressed. I flipped back and forth on the two based on game optimization. When GeForce hit the ground I never looked back.

The late 90s/early 2000s were exciting times for PCs.

0 - http://www.dansdata.com/q500.htm

3Dfx had the best looking box art by far.

I was way into computers around this time. I remember the race to 1ghz. I remember the monumental change when AMD put the memory controller on die. Such awesome memories.

I had durons, coppermines, slotted CPUs, dual Celerons hacked to work all sorts of fun. Heck I was around when you could take a 300A and set the multi to 4.5 and the fsb to 100mhz and get a 150% OC!

My first PC was a Commodore 64, then my dad’s hand me down 486. What a beautiful time to be a nerd. 10k 9.1GB loud and hot SCSI drives. Such cool stuff back then

I had a 300A@450MHz running for years. Those were great!

This reminds me of the time XBox 360s were hacked by drilling the processor at a precise location, at a precise depth [1].

At some point, a tool was made to signal if you drilled far enough.

[1] https://m.youtube.com/live/RyW0lXnoFOA

Also see the infamous Wii tweezer attack which could dump the system ROM with just a pair of tweezers.

Sometimes I wonder how these sorts of exploits are discovered. It must take some serious in-depth knowledge of the target system and a lot of trial and error.

The DS's firmware could also be written to by bridging SL1, a point on the board accessible by removing the battery cover. This was used before flashcarts were popular to write FlashMe. It was an alternate firmware that allowed for booting files from a slot-2 device such as a GBAMP.

> The only drawback to this is that the potential for customers to be ripped of by untrustworthy dealers selling overclocked CPUs to the public at the price of the actual CPU speed is increased by a large margin. Intel long ago locked the multiplier of their CPUs to keep such things from happening with their processors

Is the author being charitable or do people really believe this was to protect consumers rather than constrain them?

The constraint was real, sure. But also, making lower binned/spec'd chips act like faster ones and then selling them for the same price, that was also something that happened.

If someone paid $500, they should get the $500 chip, not a $220 made to work like the $500 one. So both things can be true.

Lying about the chip (ie saying that it's a $500 intel chip when it's actually an overclocked $200 chip), is clearly unethical. That said the realities of chip binning make this weird, as you could also argue that it's unethical and wasteful to artificially lock down some hardware's potential. It's not only chips that suffer from this, like when BMW tried to apply a monthly subscriptions for using the heated seats that came with your car.

saying that chip binning is unethical assumes the binning is truly for purely artificial reasons.

What if the binning is genuinely because hardware is failing the required QC? Wouldn’t it be ethical to sell a unit that isn’t at risk of breaking/corrupting data etc and reusing the same units that have already been made?

Binning is a pragmatic approach to reducing waste while providing products at different price points.

Or you can think of it in the opposite manner, they could just sell all of the $500 spec'd chips as $500 chips and miss out on an entire market of people that could use a lower spec'd chip at $200.

But that's besides the point. If I'm buying a $500 chip, I want that chip. The one that passed, qualified and is guaranteed at that speed by the manufacturer. I don't want a $200 chip that you hacked to happen to be able to run at that speed.

If you want to buy that $200 chip and hack it to run at $500 speeds go for it, there's nothing wrong with that.

BMW does a lot of this kind of shenanigans, not just with their heated seats. They also charge extra to unlock the horsepower of their engines. There are several performance levels which have the exact same engine with only a few different parameters tweaked in the firmware in order to limit the horsepower output to the level each customer paid for. Tractor manufacturers do the exact same thing with horsepower output. And it's not just John Deer (famously anti-customer and anti-right-to-repair) but most (or maybe all) of the major brands are doing similar shit.

One could argue the warranty claims are likely lower on the detuned engines

A great way to extend battery life is lie about the capacity in the UI and only charge to 80%.

I bought a counterfeit Athlon XP (one modded to look like a 3200+) and spent some time troubleshooting its problems before dealing with getting my money back (from my credit card company - the vendor has already disappeared). There are definitely some advantages for the consumer in making counterfeiting difficult.

I don't remember the time frame but it was a Slot A cpu that I bought from some russian owned company in Irvine in the early 2000's that was doing exactly this practice. Since I was planning on overclocking it myself, and had taken steps to record as much about the transaction myself (because I knew this scam was going on), when reported to AMD, they had their lawyers contact me to provide an affidavit because they we're attempting to sue the shit out as many of these sellers as possible. They also replaced the cpu with the one I intended to purchase so at least the story ended nicely but what a pain in the ass it was.

IIRC, the Athlon XP processors also had these contacts, but it required an additional step to fill a gap dug between the contacts with super glue. (Probably possible because previous processors used a ceramic package while the newer ones used organic plastic.)

In Core 2 Duo days I overclocked the FSB of my E2160 Pentium CPU by 50%. This boosted the frequency from 1.8 to 2.7.

These days I am happy I got my 14700K boost to 6200 MHz on single and dual core loads and to 5600Mhz P-cores and 4400 E-cores in multi-core loads. According to benchmarks this means about 10% performance gains. But it took me a week fiddling with BIOS settings and testing.

I was in high school, rocking an E2180 with a massive Scythe Ninja cooler. Had a 3.2Ghz "summer" OC profile and a 3.45Ghz "winter" one. Those were the days.

My cousin had an E2160 running at the same 2.7Ghz, and a Radeon HD3850 that was softmodded to HD3870. This PC is still running, and it's still overclocked!

There were similar techniques to mod cores to work on multi-processor systems! Helped a number of friends with that one!

My first multiprocessor system was one of those! Two Duron CPUs, "upgraded" to think they were AthlonMPs. Worked pretty well, but it used a lot of power. I burned through two PSUs in 18 months (admittedly, both were not exactly top-shelf).

I had this too but with athlon xp dual core a real like pro :) one time my psy burned some pins in motherboard I see the atx spec and found that is the power signal only well I replaced the psu with a proper watts and made used a clip to fill the gap of the burned side of the motherboard and this running’s well for 2y after a finnaly upgraded to core 2 duo generation

Yep, those were the days. I still have the Thunderbird CPU that I overclocked using the pencil method sitting on my desk because it was such an awesome CPU.

Tangible progress every year that made it worth upgrading your parts and tech news were actually exciting back then (I still remember upgrading from my 386 with 33MHz to a 486 DX4 with 100MHz). Now the tech media is making a big deal about 10% improvements. :/

And that small 10% improvement most likely gives you the processing power of hundreds of 486DX

I almost want to blame it on how old I was in 2000, but it really was an exciting time for computing. A lot of things that weren't quite ready for prime time were finally ready. 3D games became playable, Windows graduated to a "real" OS when XP put NT in the home, chips were still getting faster...

The internet was still being figured out and wouldn't really be ready until ~2005 with widespread broadband adoption.

The other big thing that didn't happen until later was SSDs.

"Real" 3D games happened a little bit before then, though. Descent, Quake, Unreal...

This was top of the line 3D gaming as of 1998: https://retronn.de/ftp/videos/unreal_intro_voodoo2_sli.mp4

The pencil unlock seems to have made a return with the latest hack of running 9th gen Intels on 6th gen mobos. https://hackaday.com/2024/05/31/intels-anti-upgrade-tricks-d...

Good times. Back then I used this trick and paired it with an non-laser-locked modded Geforce 6800 AGP card for some gaming. I was lucky that my GPU handled having the disabled cores re-enabled without any issues, and the subsequent performance gains were huge with the additional unlocked speed from the CPU. I stuck with that GPU and CPU combo until the 9800GTX+ dropped.

Haha amazing, I did this on my Duron 700 MHz which I overclocked to 900 MHz, and my Athlon 1GHz (AXIA core IYKYK) to 1.4 GHz

Wow, those codes bring back memories. I remember trying to find the exact right type numbers on online forums, to find the ones that could be overclocked the most. Never mind that I was about 13 and didn’t have the money to buy one but I could dream ;)

Yeah I was only a couple of years older than you and barely had the money, hence the overclocking :D Tbh it was fun even to be on a forum talking about it and checking benchmarks etc, some guys had very cool watercooling setups that I was very envious of

I remember doing that back by the beginning of 2000s, unlocking one processor, and losing another identical one trying the same (guess I went too far ahead). I didn't work, and my father was not really happy to have to buy a new CPU because of some graphite (but he understood the curiosity!).

Maybe my memory is a bit hazy, but wasn't this already possible with the Athlon processors since ~2001?

Yeah, the sub-1GHz Athlon and Durons were 1999, so this article is a little late.

1999 was first gen, Slot A based. Second generation (Thunderbird) Socket A based Athlons (and Durons) described here were released 2000.

Hah! Was it 2000 already!

Got one in 2001 and I was so proud of my AMD Athlon Thunderbird 1400 that I even accepted that loud noise from the cooler I used.

Although in hindsight, the money I spend for that luxury would have been better invested in other stuff, improving my life a bit more then a loud and noisy airplane turbine.

Yep. I had a Slot A. You literally just plugged the whole HSF/CPU into the motherboard like an expansion card. Modern PC builders would be perplexed.

Here's a nice relaxing video that I like, of a guy soldering the pins instead, which is a lot of work but more reliable: https://youtu.be/Lkuf9hckV08

Celerons could be overclocked too; much cheaper than a Pentium III back in the day.

I remember doing this along with a co-worker, I think my Duron was $60 or $80? Saved me a ton of money vs a P3 and board.

Might have been long enough to have found the parts on Pricwatch.com?

I thought mechanical pencils used graphite not lead

No pencils use “lead”. It’s just called lead because the original material used looks similar to lead ore.

See manufacturing section here: https://en.m.wikipedia.org/wiki/Pencil

The words we use are funny.

I had a conversation with my wife yesterday explaining to her that "gas" (gasoline) wasn't a gas. (it can become one, but that's not where the name comes from)

And if you push the gas pedal, it does not mean more gasoline comes in, but more air (which then makes more gasoline comes in).

And people fear, that soon too many of the undividable atoms will be divided.

I was watching a YouTube video the other day that called the accelerator the "gas pedal" - which makes sense, except it was a video about an EV.

I wonder if that term will one day become disconnected from the original meaning, like "hang up" or the floppy save icon.

At least in german I saw EV cars described having a "strom pedal" (electricity/power pedal). More correct I guess, but also a bit odd. We will see, whether it will stick.

In Greek we call it "γκάζι", which means "natural gas", so it's already wrong. I'm fine with that persisting in EVs

Most of the other terms are global ones (hang up, save icon, etc), but gas pedal is pretty specific to the US as far as I know, so it’s much less likely to hang around like that.

Here in Australia it’s the accelerator, or accelerator pedal.

Estonian word is “gaasipedaal”, which pretty much means “gas pedal”. Gasoline is “bensiin”, no relation to that. The word for pedal comes from accelerator regulating gas-mixture valve (throttle) in carburettor. “Gas-mixture” here is air mixed with atomised fuel.

Funny, in german it is "Gaspedal" and "Benzin". (And gas means gas/air in german, but I never made the connection. With cars, "gas" means "speed")

It's also literally "gas pedal" in Russian, although more often people just say "gas". To accelerate is to "give gas".

The fuel itself is also "benzin" though. As I understand it's because benzene was the original anti-knocking additive.

Unless it’s diesel, because there the accelerator causes more fuel to squirt in rather than opening the throttle body.

But can do you call diesel gas?

We do in Spain (“voy a echar gasolina” is both used for diesel and gasoline)…

In the US it’s context dependent.

If you’re running low on diesel you’d say “I need to get some gas” and you’d get it from a gas station

But if you’re talking about tractors saying that something has a “gas engine” specifically means it is not diesel

Likewise anything new. The idea that it allows more air in that then allows more fuel is true for carbureted based engines. But they don’t actually exist outside of lawnmowers/chainsaws anymore. Everything from the timing to fuel injection and air intake is computer controlled.

Really? My new-ish car (2019) just has the pedal connected to the throttle body with a cable. I’d assume that other changes to fuel injection would happen because it sensed more air coming in

They are confidentially incorrect, the best kind of incorrect, even cars with servo controlled throttle bodies (like mine) the pedal (largely) controls the position of the throttle plate. The ECU will then control the timing of the fuel injectors to achieve optimal combustion and cylinder pressures based on expected air intake.

The ECU will also take inputs directly from the pedal, but only because that provides instaneous information, rather then waiting for sensors in the intake manifold, and eventually the exhaust manifold, to catch up.

It always comes down to pedal controls air, air controls fuel.

Edit, I'm a dirty liar: It doesn't ALWAYS come down to pedal -> Air -> Fuel. On engines with turbochargers it's possible for the ECU to delay fuel injection reducing the amount of power in the cylinder and increasing the temperature/pressure of the exhaust. This causes more power to be generated by the turbine which then accelerates. This then results in the linked compressor accelerating increasing pressure in the intake manifold which means the engine can burn even more fuel generating more power.

All without the throttle plate changing position or the engine RPMs increasing.

This trick isn't used in any meaningful way in production cars to the best of my knowledge, it is used in race cars to keep the turbo spoiled up at its optimal point prior to the start of the race.

Also if your engine happens to be a turbojet instead of a reciprocating piston engine (all turbocharger, no cylinder), it's the only way you can control power.

Fuel injected engines doesn't necessarily have electronically controlled throttle. And even when they do, injection amount isn't derived directly from pedal position. What's important is amount of air sucked into the cylinder, which is calculated from manifold air pressure or mass air flow sensor readings.

Well, but in this case they're just homonyms, like how the word "bank" can refer to the edge of a river or a financial institution. The gas in your car is short for "gasoline"; you wouldn't call oxygen a type of gasoline, it's a gas.

I think the original financial institutions were actually on the rivers to handle the business around river-transported cargo.

Actually, "bank" isn't just homonyms; both senses derive from circa-Proto-Germanic benc/bank: a bench or other raised area (either a bank counter or the raised ground adjacent to a riverbed). (And yes, this etymology is also shared with "bench".)

Graphite is pretty conductive too.

2007!

I forgot about the pencil trick, but never had a relevant generation of AMD chip to do that to.

I did that with duron 600->900mhz, but it was so long ago that I don't remember exact date

My Duron 600 ran Windows 98 SE at 950mhz, on an Abit motherboard. What ever happened to Abit? Miss those mad geniuses.

They had financial problems and a class action lawsuit against them in early 2000s. The company was sold and then eventually shutdown or folded into the parent.

Those were the times. Unlocking a lot more performance just by tweaking. These days everything is locked down. I suppose this is more efficient.

The last time in felt the same awe, although not by my own handiwork, was when I moved to an M1 MacBook and when I installed Fedora on my desktop. Everything was so fast and silent. Really amazing.

I feel like a lot of the tweaking features popularized by Abit are now available on any "gamer" board from one of the major manufacturers, Intel and AMD now even make chipsets targeted at that segment, and there are a lot more bins for CPUs these days, most of which come "pre-overclocked" even dynamically so with turbo-boost and the like.

The Athlons and Durons of this era had exposed dies, just a few SKUs, no built-in thermal throttling or speed boosting at all, and oversized high-cfm heatsinks and fans felt like a new thing.

More like these days everything is already clocked hard by default. There's basically no point overclocking modern unlocked chips, they are already clocked to 90% of their max speed. Compare that to the above example with 600 vs 900 Mhz (50%! faster)

Some locked chips are clocked pretty slow, but that's normal segmentation :(

What made them mad geniuses? I saw their equipment from time to time in catalogs but was too young to realize they had some kind of special reputation.

https://en.wikipedia.org/wiki/ABIT_BP6 is a great example. It was the first motherboard to allow the use of two unmodified Intel Celeron processors in dual Symmetric multiprocessing (SMP) configuration. Not only did this cost drastically less than purchasing two CPUs certified for SMP from Intel at the time, Celerons of the era were the first Intel chips with on-chip L2 cache which was clocked at the full speed of the CPU as opposed to the off-chip L2 cache's half-CPU clock on the Pentium II. Abit was also one of the first companies to include jumperless overclocking features in the BIOS.

If you look close, you can even see the blue thermal sensors Abit placed in the center of the CPU sockets (CPUs had no built-in thermal sensor at the time) which greatly eased overclocking.

In short, they made boards that gave you the full range of what was possible, not just what was on the marketing sheet for the CPU.

Same! Dad and I did this a few times, got around having to upgrade to get more performance when we could still get much more out of the current system.

I had tried that with Duron 600 but couldn’t overclock it. I don’t remember what went wrong. Maybe the motherboard didn’t support it.

The best way to overclock the lower clocked durons was simply to change the fsb in the bios settings. You didn’t really need the pencil trick as they had 100mhz fsbs in an era where 133 was more common.

I remember doing that on my Athlon XP. Had too much issue with cooling though. (Probably my fault)

Sure you didn’t just flip the 100->133mhz fsb setting in the motherboard? The low end durons didn’t really need the multiplier changed as they had a 100mhz fsb when the athlon 1000 had a 133mhz fsb and used the same motherboards. You could just flip the bios setting and get that exact stated overlock without the pencil trick.

I did the overclocking and overvolting thing (with a pencil) for a bit, but then I became obsessed with silent computers and went the other way: undervolting. Trying to keep it running at the lowest possible voltage without reducing the clock speed too much. Those were fun experiments. Could do passive air cooling when the overclockers were getting into water cooling.

Aside from undervolting to reduce cooling needs and thus fan noise, suspending disks on sturby rubber bands greatly reduced vibration noises, same with mounting fans with plastic plugs instead of metal screws. Getting bigger, slower fans also helped a lot.

PCs now tend to be pretty quiet these days, but loud fans and rattling still bug me from time to time.

I split the difference. I went with an overkill watercooling set up with quietness in mind, just so I could have nearly-passive cooling at high clock speeds.

I’m currently using 2x 360x360mm radiators stacked together, with 4x 180mm 300rpm fans. It’s virtually silent yet keeps my CPU and GPU quite cool, even when gaming.

Word for word the same here. I remember spending hours and hours in the silentpcreview forums looking for the latest and greatest in silent fan technology.

I've probably still got one I should throw away. Probably more than one...

The other related article which was here recently may have been related to this one showing up: https://news.ycombinator.com/item?id=40533761

Memories... Installing coolers on these chips always terrified me. There was also a weird aftermarket for "shims" that you'd put between the chip and cooler. They had a cutout for the die and contacts on the surface, but I was never sure what they were supposed to do.

Reply on news.ycombinator.com