Hacker News Reader

Comment by tyingq | Hacker News Reader

Called "product binning".

Though it's also common to segment products that don't have defects/differences by intentionally hobbling them, or even just using different labels when there's no actual difference.

View on HN

AMD's Phenom II let you unlock the locked fourth core, if you so dared.. allowing you to save like ~$50 on the higher chip. But it could be unstable.

Or it might not work at all (as was the case for mine when I tried it).

There definitely was a period where AMD was shipping three-core Phenom IIs with a functional fourth core (demand was higher for the three-core SKUs than their actual defect rate appeared to allow), but I don’t know how plentiful fully-functional one’s actually were. And it was never guaranteed.

There was also the original Athlon, where there was an unjoined connection on top of the die. You could fill in the connection with a pencil and overclock the chip.

Been there and done that.

"By the power of graphite!", I upgraded my cheaper CPU to a more expensive one.

Skeletor failed to show, however.

Yes, allegedly demand outstripped supply for the lower core count chips so it was a mix of disabled rejected cores and disabled working cores.

Two other older AMD CPUs could be similarly unlocked:

- AMD K6-2+ with 128K of L2 cache can be converted to a K6-3+ with 512K. The CPU has a heatspreader that can be carefully removed revealing a series of zero-ohm resistors, moving it to the right location unlocks the cache. The CPU is still bound by binning as to whether or not it can be overclocked well.

- AMD Athlon XP could be overclocked with the "pencil mod", using a pencil to draw a conductive trace between two pads on top of the CPU, no delidding required.

Isn’t a zero ohm resistor just a jumper?

In a sense that's correct, but "jumper" suggests these are deliberately components you can change whereas "Zero ohm resistor" is likely intended to be permanent.

In PCB manufacturing, a jumper wire is just a wire soldered between two points. A designer might or might not install jumper pins like those you refer to.

Regarding the use of a 0-ohm resistor instead of a jumper wire: wires are rather annoying to deal with in pick-and-place machines, while a 0-ohm jumper can be reel-loaded like any other part.

It is, but pick-and-place robots like to work with zero ohm resistors compared to small pieces of wire. Also makes it easier to change later, you just swap reels.

I still have a computer running Athlon x3 unlocked to Phenom x4 - circa 2010 this was the most cost efficient way to get a 4 core with 32GB of RAM.

You don’t even need to look for extra cores. Just CPUs being locked on different frequencies is similar enough to this concept.

And certain GeForces could be turned into Quadros (for use with CAD software): https://archive.techarp.com/showarticleefc1.html?pgno=0

This was a soft mod, but I think there was also a hardware mod involving only a pencil.

Learning about how they product bin CPUs blew my mind years ago.

iirc, the CPUs come off the same manufacturing line--but the ones they sold as 3.2ghz vs 3.4ghz were based on stability tests. It was the same chip, just that some came out the oven better than others. So they would cap it and slap a label on it, which is why overclocking is feasible but inconsistent.

This is also common a common practice with garments, bakery items, fruit, racehorses, and children.

Wait, children?

Should be an XKCD comic

Makes sense, no?

Some children are just smarter, more athletic, or better looking than their siblings.

I think they are referring to this part:

> It was the same chip, just that some came out the oven better than others.

Not the part about selling.

There are shops that buy CPUs in bulk and bin them themselves for suitability for overclocking and then sell the top performing ones at a premium.

I got my i9-9900k this way, they binned it for 5.1GHz and it had been working 24h non stop for a few years (it is still working fine, but I no longer have a need for it to run).

Silicon Lottery shut down two years back. It isn't clear to me whether manufacturer binning got tighter, overclocking just doesn’t work anymore, or auto-boosting are their lunch.

These days chips with different amounts of cores or cache may derive from the same mask. If one or more of the cores or banks of cache are defective, they're disabled.

Intel/AMD do not test every single die. Once you get stable output from the foundry you learn a max clock/wafer location distribution.

That's interesting. I always thought each one had a basic pass-fail test after packaging.

though then shouldn't the correct replacement part for the HD595 be a matched pair of drivers?

If you want a matched stereo image across the spectrum, yes. I have little experience with Sennheiser but most makers of the high end (studio grade at least) will offer both singles and matched pairs and leave it up to the buyer.

I've bought matched pairs of DJ stylus. The product descriptions waxed poetically about the reasons this was important, and if I cared, it might have even made sense. The sole reason I bought them paired was the simple fact that the pair was cheaper than buying two individuals. The exact same logic was applied when I bought a pair of monitors, and the fact that the seller offered a $50 item for free.

If you're into matched pair speakers, what about for surround systems? Do you have to have a matched 4 set or a matched 6 set?

From what I can tell, hardcore audiphiles tend to stop at stereo, so there's that. But if you're going to match speakers, I would think you'd want to at least match by type: match your fronts to each other, match your surrounds to each other, match your heights to each other, match your subs to each other. Just having all of one type be the same model is probably good enough though.

If you are into it, sure. This is a good question and I can not quite answer it too my satisfaction despite repeated attempts, but here is my best go at. Surround sound gives a larger sweet spot than stereo (that area in the room where you can hear sounds from speakers move about the room as if they were there with you) which will compensate for less than optimal speaker matching and we also get more information to process from those extra channels so our brain can make more sense of what it is supposed to be. This along with surround sound generally being used with film where our attention is divided makes surround sound more tolerant of speaker selection.

Hopefully someone more knowledgeable in 3d sound will happen along.

Edit: Should also mention that we get a lot more leeway in general when it comes to film sound since we also have visual cues regarding the sound's location and movement.

>Edit: Should also mention that we get a lot more leeway in general when it comes to film sound since we also have visual cues regarding the sound's location and movement.

For sounds coming from on screen, maybe, but the point of rear surround speakers is to get sounds from behind you where you have no visual clues. So that sort negates this argument. Although, I'm not really sure where the whole visual cue argument comes into play when deciding how matched pair vs non-matched stereo imaging sounds to people. A matched pair vs non-matched is not going to suddenly not sound stereo.

Imagine your left speaker is down 6db above 220hz and right stays flat, you play a bar choard at A 110 on a guitar that is panned hard left. Everything in that chord bellow 220 is hard left but everything above is now more in the middle and mixing with the vocals and drums which were set in the middle making things muddy. In reality we get dips and humps all over the frequency response and if they are not matched it spreads those sounds across the image and this can be quite bad. Play your A note and the fundamental is panned where you want it, the second harmonic is closer to center, the third is somewhere between those and everything else is more right, the sound is no longer isolated in the field where it should be. Now play your A an octave up (second harmonic of the previous A) and fundamental is closer to center, second is more left and so on, everything moves subtly. Apply that to an entire band playing with all of them blurred like this, the stereo image falls apart and we only get a vague impression

These examples are not exactly accurate to the real world, meant more to demonstrate the problem but we can hear it easily, this is a big part of why some speakers/headphones sound muddy and have poor deffinition. With more simple sources this is not much of an issue but as complexity increases the lack of matching becomes more noticeable.

In surround the rear speakers can be quite bad before issues arise since the shape of our ears and our head means we are not great at pinpointing the location of sounds behind us, we hear hard left and hard right behind us well but we have a dead spot between where things are not quite so clear. I would suspect some of the rear is also mixed into the fronts which would help, hard panning tends to be avoided as it does not sound natural. As I said, the surround stuff is a best guess on my part, I suspect I am in the ballpark but lack some of the nuances.

I had a problem with a bass driver on a pair of PMC studio speakers. I asked PMC directly whether I should replace one or both drivers in the pair, and they said to only replace one, that there would be no issues.

So, in response to replacing drivers, 'it depends' is probably closer to the truth. It may well be that bass drivers are less problematic than say midrange when pairing.

Speakers are different than headphones. With speakers you have a sweet spot, if your head is not in that sweet spot than the stereo image is blurred in someway and that image is affected by room reflections, the stuff in the room, volume setting, etc. This is why headphones are common for mixing, when you lean over to adjust that preamp you take the stereo image with you instead of moving out of it. We are more tolerant of speakers in this regard (from the studio standpoint) since the stereo image with speakers changes with your position and is rarely accurate compared to headphones. Midrange is the most important part and it is where the bulk of the information is for most of our music and our hearing, we get a lot more wiggle room in the bass and the highs.

It does depend, we don't always need or want that perfect stereo image but it makes mixing easier to have it available.

Supposedly, our left and right ears aren't even equal at birth...each having different pathways to the brain. Then, of course, unequal damage over time. Makes me wonder how much tediously matched drivers matter.

Our brains are pretty good at compensating for the failings of our bodies. For general recreational listening matched pairs are not all that useful since the majority of music is mixed with the knowledge that 99% of people are not going to listen to it on something with an accurate stereo image or even flat frequency response.

I had a sad realization after finally buying the very nice 6XX with a small tube amp - that while it sounds incredible, it also makes my hearing loss/tinnitus more prominent. Probably since I was focusing much more on the intricacies of the sound. Ended up selling them because it just wasn’t an enjoyable experience

Apple and JBL measure your ear and adjust according to the measurements. Apple calls it personalized spatial audio.

We are already asymetrical, adding (inconsistent) asymetry on top of it is not going to improve on it. That's why matching is... something we can do to improve audio. At least it's doable.

The correct replacement part not only isn't matched....it's the same part for the 595 and 555.

> Quite a few people speculated in my Original thread that the more expensive HD595 headphones must also be using a more expensive driver. However, Head-fi member MCC posted the smoking gun; a picture of the original Sennheiser replacement driver labelled “HD 555 / HD 595”.

So, unless they're fucking over their 595 customers who need a repair: the drivers aren't binned.

Reply on news.ycombinator.com