BUILD Time for a new machine

[GrumpyBear]

Parts ordered :)

Just need to read up on the various threads for updating the BIOS and LSI2308.

What do people do for their USB boot drives? The N54L has one internal USB port, so the stick is nicely hidden away inside. Was going to use 2 sticks mirrored, but ideally keep them inside the case.

I use two of these in the chassis USB2 ports. Small form factor so unobtrusive. I keep hoping that eventually they will get the bugs ironed out of USB3. I have one SSD unused in the case and cabling for a second but I reboot so infrequently that the possible speed increase in boot is IMHO not worth it.

Looked at the SODs but with this motherboard they obstruct adjacent ports and the same boot speed comment pertains too.

I like having my jails on the ZFS volume too.

 

[adrianwi]

Well, my first delivery of bits arrived today! Quite exciting as I haven't built a system for over 10 years.

Motherboard and case have been dispatched, so just a processor and memory to ship before I can put things together.

Will buy the extra drives once I've got everything working with the one.

 

[adrianwi]

And a processor too - what happened to all the pins?

 

[Ericloewe]

And a processor too - what happened to all the pins?

Intel processors haven't had pins since socket LGA 775. LGA stands for Land Grid Array - the pins are on the socket.

 

[DataKeeper]

Whaaaaaaaaaat? Really? Lol no more CPU pins? Yeah ok.. It's way to long since I last built a system!

 

[Ericloewe]

Whaaaaaaaaaat? Really? Lol no more CPU pins? Yeah ok.. It's way to long since I last built a system!

It's called risk alienation. The risk of a broken pin is now the responsibility of the motherboard manufacturer.

AMD still uses regular pins, though.

Be happy that we even have a god damned socket! Broadwell won't even see a major socketed release, only BGA stuff.

 

[cyberjock]

It's called risk alienation. The risk of a broken pin is now the responsibility of the motherboard manufacturer.

AMD still uses regular pins, though.

Be happy that we even have a god damned socket! Broadwell won't even see a major socketed release, only BGA stuff.

I had heard that the costs of going to LGA from the old pins on Intel's side of things were actually higher and that the whole "broken pin" argument was not the reason for the change. I heard the change was because you had less electrical noise from LGA connections than from the pin layout from pre-775 (aka 478 and older). No clue if it's true or not.

I could totally buy the argument that Intel was avoiding broken pin syndrome. But let's be honest here. If someone broke a pin it would be easy for Intel to not only identify that the pin was broken, but then they'd have solid irrefutable evidence that the CPU was abused and deny the warranty claim. So would they really save any money if they are denying warranties anyway? If they weren't denying warranties for broken pins and that was a large portion of RMAs, why would they not simply make it policy to deny claims with broken pins? Seems like there are much more cost effective ways to avoid "broken pin syndrome" than changing their CPU designs.

 

[Bidule0hm]

I think the noise argument is true. With pads instead of pins you have less inter-pin capacitance and more control on the impedance of the path (and probably some other advantages I didn't think of too) and at a few GHz all of this is very very important...

 

[Ericloewe]

I think the noise argument is true. With pads instead of pins you have less inter-pin capacitance and more control on the impedance of the path (and probably some other advantages I didn't think of too) and at a few GHz all of this is very very important...

But you just transferred the capacitance to the socket - unless LGA allows for shorter/smaller "pins" than a traditional socket, and therefore a smaller capacitance.

I can think of a mechanical advantage: Plastic deformation of the socket's contacts is compensated by the pressure exerted by the CPU package, which is held tightly in place by the retaining frame. Unless the pin receptacles had their electrical connections at the bottom (unlikely due to poor contact due to the limited surface), plastic deformations that degrade contact aren't naturally compensated for.

I had heard that the costs of going to LGA from the old pins on Intel's side of things were actually higher and that the whole "broken pin" argument was not the reason for the change. I heard the change was because you had less electrical noise from LGA connections than from the pin layout from pre-775 (aka 478 and older). No clue if it's true or not.

I could totally buy the argument that Intel was avoiding broken pin syndrome. But let's be honest here. If someone broke a pin it would be easy for Intel to not only identify that the pin was broken, but then they'd have solid irrefutable evidence that the CPU was abused and deny the warranty claim. So would they really save any money if they are denying warranties anyway? If they weren't denying warranties for broken pins and that was a large portion of RMAs, why would they not simply make it policy to deny claims with broken pins? Seems like there are much more cost effective ways to avoid "broken pin syndrome" than changing their CPU designs.

Sure, they could easily cut down on RMAs, but management might have preferred passing on the whole problem to a third party instead of risking bad publicity - much like motherboard vendors get when people end up with "new" motherboards that were shipped with bent pins only to end up with the RMA rejected.

Higher (steady state) manufacturing costs for LGA sounds weird, on the processor side, as I'd assume that etching a few exposed contacts on a PCB would be cheaper than carefully sticking thousands of aligned pins in place.

In any case, Intel clearly sees value in LGA, since it's had several opportunities to go back to pins since LGA775.

 

[Bidule0hm]

Yeah, exactly, shorter pins. And the mechanism to grip the CPU pins is just a nightmare on the electrical side let alone the mechanical side...

I think the cost come from the redesign (and that include highly specialized machinery), not from the details of the product.

 

[adrianwi]

Some RAM today :D

And Amazon kindly sent me a door for a Fractal Define R4 which isn't much use without the case they should have sent. Hopefully that will arrive on Monday with the motherboard and the build can begin...

 

[DataKeeper]

You're gonna make me place an order soon with these pics! Hope you get everything soon and the build goes well.

 

[Tywin]

[...] But let's be honest here. If someone broke a pin it would be easy for Intel to not only identify that the pin was broken, but then they'd have solid irrefutable evidence that the CPU was abused and deny the warranty claim. So would they really save any money if they are denying warranties anyway? If they weren't denying warranties for broken pins and that was a large portion of RMAs, why would they not simply make it policy to deny claims with broken pins? Seems like there are much more cost effective ways to avoid "broken pin syndrome" than changing their CPU designs.

Re: "why would they not simply make it policy to deny claims with broken pins?"; pissed off customers, for one. I'd be pretty upset if I bought a shiny new processor, opened the box, and found one or more of the pins bent, then went to submit an RMA and found that bent pins weren't covered. I think the major gap here is your assumption that a bent/broken pin is irrefutably the fault of the end user; I refute that :P

 

[cyberjock]

Re: "why would they not simply make it policy to deny claims with broken pins?"; pissed off customers, for one. I'd be pretty upset if I bought a shiny new processor, opened the box, and found one or more of the pins bent, then went to submit an RMA and found that bent pins weren't covered. I think the major gap here is your assumption that a bent/broken pin is irrefutably the fault of the end user; I refute that :p

Ok, so you are pissed off because you broke your CPU and Intel won't do a warranty. Anyone that is reasonable should understand that Intel shouldn't be in the business of doing RMAs for stuff that isn't their fault. You don't try to return your car to the dealership if you drive it into another vehicle, do you?

And even if we assume that you are pissed off anyway. So what? What are you gonna do about it? Go to AMD? Bwahahahaha. Intel is basically the only game in town if you want high performance. So you don't have a lot of choice unless you are really going to shoot yourself in the face and go with AMD.

Its really Intel's market and Intel's choice. You and I can be upset, call it unfair, etc. They know they own you and me. They know you have zero choice when it comes to building FreeBSD servers. If they didn't know that, they wouldn't charge the prices they do for CPUs of comparable performance to AMD.

 

[rogerh]

In the UK, at least, a customer who convinced a judge he was telling the truth about getting a CPU with a bent pin would win his case against the retailer. That would not affect Intel though, who could sell to the retailer on whatever terms they liked.

 

[Tywin]

Ok, so you are pissed off because you broke your CPU and Intel won't do a warranty. Anyone that is reasonable should understand that Intel shouldn't be in the business of doing RMAs for stuff that isn't their fault. You don't try to return your car to the dealership if you drive it into another vehicle, do you?

Not sure if you are intentionally misreading my post or just lacking coffee. I didn't break the (hypothetical) CPU pin, it arrived broken. Care to try again?

 

[rogerh]

Not sure if you are intentionally misreading my post or just lacking coffee. I didn't break the (hypothetical) CPU pin, it arrived broken. Care to try again?

Can I act as mediator here? The point is that Intel will assert that whoever broke the pin, a) it was probably you in their statistical opinion, and b) they have no contractual relation with you and if they don't care to listen to you they don't have to.

The fact it arrived broken is immaterial to the outcome.

Edit: if you are buying direct from Intel you have a business contract which will say exactly what will happen if you receive damaged goods - and it probably won't be particularly favourable unless your firm is bigger than Intel.

 

[Tywin]

Can I act as mediator here? The point is that Intel will assert that whoever broke the pin, a) it was probably you in their statistical opinion, and b) they have no contractual relation with you and if they don't care to listen to you they don't have to.

This isn't mediation, it's wild speculation.

The fact it arrived broken is immaterial to the outcome.

a) Not what @cyberjock said (twice), and b) it is not immaterial. No north american company can ship multi-hundred dollar defective products, asserted to be functional working products, without warranty for manufacturer defect.

I'll take a moment to remind you that this is a hypothetical situation anyway, as @cyberjock was merely asking "why not do this"; well, in order to stay in business is a pretty good reason not to fuck over your customer base by selling them defective products.

 

[DrKK]

This isn't mediation, it's wild speculation.



a) Not what @cyberjock said (twice), and b) it is not immaterial. No north american company can ship multi-hundred dollar defective products, asserted to be functional working products, without warranty for manufacturer defect.

I'll take a moment to remind you that this is a hypothetical situation anyway, as @cyberjock was merely asking "why not do this"; well, in order to stay in business is a pretty good reason not to fuck over your customer base by selling them defective products.

I'm not sure why we're arguing about this. But, a retail boxed CPU with bent pins? Come on now. That really would be like getting a new car off the lot, driving it off, and then driving it back with a smashed quarter panel and then you claim that it came off the assembly line that way.

If pins are bent, I'm pretttttttttttttttttty sure that happened after the item left Intel's hands. Each binned item is mounted and tested, which would not be possible if the pins were bent, then the item is reinspected, then packaged in a retail box, then reinspected. Someone in the retail or shipping chain bent the pins, or the end customer. Almost certainly the end customer.

 

[rogerh]

The habit of some retailers of repackaging returns and pretending they are new might explain the bent pins! But the real point is that (unlike cars where there are laws affecting manufacturers) the manufacturer has no obligation to compensate a retail customer. In this case, unless someone could prove they came out of the factory gates with bent pins they would not lose any significant degree of reputation either.

The person who gets a new processor with bent pins certainly has a case against the retailer, and will win if believed by the court. But Intel is not bound to get involved.