4

Intel has released Ivy Bridge Xeon processors with PCIe 3.0, ECC, and VT-d support. However, there aren't yet any PCIe 3.0 + USB 3.0 C21x (Panther Point) boards available, while B75 and Z77 desktop motherboards are widely available with a huge range of features. Since the northbridge logic / memory controller is now integrated into the CPU, it seems like it wouldn't matter which platform controller hub the Xeon is paired with.

So, is there something that would prevent an ECC-capable processor of the Ivy Bridge generation and ECC DDR3L memory from working together when installed on a B75, H77, Z75, or Z77 PCH motherboard? Are there extra traces required to carry the parity/checksum bits which don't physically exist on desktop-class motherboards? Or perhaps ECC support depends on an ECC-aware BIOS/EFI, which these boards wouldn't carry?

I found this related question, which deals with an earlier generation: Is it possible to boot a consumer i7 system with a Xeon processor and ECC memory?

Improve this question
2
  • apparently, at least one board supports it and here's a bit more supporting information to get someone started.
    – Journeyman Geek ♦
    Commented May 27, 2012 at 5:44
  • Question had nothing to do with USB 3.0, anywhere so there's definately a failure to parse the question. Its quite simply "Can i use ECC ram with a ECC capable processor with an onboard memory controller, with a compatible arbitrary non workstation/desktop motherboard"
    – Journeyman Geek ♦
    Commented May 27, 2012 at 6:18

3 Answers 3

Reset to default
2
+50

It would totally boot. The majority of xeons will support ECC ram (this is a big part of what the xeon line is for!). Things that might make it not boot, of course, are too many banks for the BIOS to parse, or a clock speed or voltage support mismatch.

That said, if your board is garbage like Acer's OEM ones, the BIOS will likely refuse to boot it for no good reason. However, I have seen them not support certain configurations until a BIOS update is done... but that's a bit of a tangent.

One thing you might need to do as well is put one non-ECC module in slot 0 to force detection somewhat.

You're correct that the ICH does not matter in this case. Considerably more important is the BIOS RAM detection routines. You may need an unlocked or modded BIOS. You could also try the Intel OEM BIOS tool, which exposes all kinds of things.

If the board lacks the extra traces for ECC, the consequence would be that the ECC feature would not be exposed. I don't believe this would cause the unit to not boot.

Improve this answer
1

So, I tried it, and it boots just fine, but according to Memtest86+ ECC is disabled :(

Configuration:

Xeon 1230v2
AsRock B75 Pro3
Crucial CT51272BD160B DDR3-1600 1.35V ECC UDIMM (pair)
Improve this answer
0

So, is there something that would prevent an ECC-capable processor of the Ivy Bridge generation and ECC DDR3L memory from working together when installed on a B75, H77, Z75, or Z77 PCH motherboard?

According to Wikipedia on Intel Chipsets ...

Not listed below is the 3450 chipset (see Xeon chipsets) which is compatible with Nehalem mainstream and high-end processors but does not claim core iX-compatibility. With either a Core i5 or i3 processor, the 3400-series chipsets enable the ECC functionality of unbuffered ECC memory.[37] Otherwise these chipsets do not enable unbuffered ECC functionality.

To answer your follow up "Can i use ECC ram with a ECC capable processor with an onboard memory controller, with a compatible arbitrary non workstation/desktop motherboard":

Short answer: No.

Long answer: It's Intel. Go with the specs they spell out. You want to use an Ivy Bridge Xeon, you need to use the Panther Point (C2xx)/Patsburg (C6xx) chipset. If the motherboard doesn't support ECC, don't bet on it working because the CPU has the memory controller on-board. You're betting a lot on the quality of the motherboard manufacturer and probably voiding the warranty on the CPU + RAM.

Will there be some specific configurations that gets ECC RAM to work correctly and be stable? Yea, maybe a couple. Is it really worth the risk? I highly doubt it.

Which leads me to believe you're thinking about doing this because it's either:

  1. Cheaper (because Intel is just so affordable (can you feel the sarcasm?))
  2. Available (no motherboards supporting features you want now)
  3. Both #1 and #2

If you're dead set on using a Xeon CPU, your best bet is to wait until they come out with a chipset that does support PCIe 3.0, USB 3, etc. etc. Now, it's Intel so if and when they decide to deliver a chipset with all those features you want, you're pretty much stuck. If time is a factor, why not go the Core i7 Ivy Bridge (non-Xeon) route that has all those features you seem to want?

Maybe this isn't the answer you want. Maybe my answer isn't very good, but I think my answer is the most realistic. Maybe you can give some more details on what you're trying to do. Maybe some people have figured out a way to do just what you're asking to do. I'd just hate to see someone gamble a bunch of money on a configuration that may or may not work. Just my $0.02. :)

Improve this answer
3
  • 2
    It's surprisingly affordable, actually. The Xeon E3-1230v3 is significantly cheaper than the i7-3770, almost 25%, for 5% slower, with just as much cache and hyperthreading, unlike the next cheaper "consumer" CPU (i5-3750). The Xeon is also a higher binned part, running cooler with a lower voltage. And I found an extremely reasonably priced motherboard with claimed Xeon support. $325 for MB+CPU. ECC is still an open question of course, so I may still have to get the Asus P8C WS when it becomes available. Hoping not, I'd prefer to spend the difference on a video card instead.
    – Ben Voigt
    Commented May 29, 2012 at 3:58
  • In practice, the warranty would not be voided in such cases, on either the RAM or the board. Improper installation is very vague, but intel (among others) do actually give certain large customers documentation on how their ASPs are supposed to triage it. The document I have in mind (without going into any NDAed detail) has pictures of various types of physical damage and says whether the ASP should deny the claim. Ever tried proving electrical damage? Not that the wrong feature set of RAM would cause any...
    – Falcon Momot
    Commented May 29, 2012 at 4:48
  • I don't know if the warranty would or would not be voided, but do you really want to risk it? You seem to be confident that this setup would work and while there's no real proof to either prove or disprove your case, my point is risk vs. reward. I don't think it's worth the effort/money to do it.
    – osij2is
    Commented May 30, 2012 at 13:57

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .