i7-4960X / Socket 2011 issues

[ddrueding]

I've clearly lost my touch with these things.

Originally I ordered a Gigabyte GA-X79-UP4 and an Intel i7-4960X. System would power cycle at 1 second intervals forever.

Suspected that the board didn't support the new i7 out of the box, I bought an i7-3820 CPU and an ASUS P9X79 Pro motherboard for troubleshooting.

When I put the 3820 in the Gigabyte board, it stayed powered on, but no video or numlock responsiveness. So I suspect that it isn't POSTing at all.

Fair enough, since the 4960 is out I installed it in the ASUS. Same result? The ASUS even has a 2 digit readout on the board that is supposed to help with boot troubleshooting. It remains at 00 which, according to the manual, is "not used".

I'm using a brand new Corsair AX1200i PSU and have tried both Kingston HyperX and Crucial Ballistic Tactical memory along with an nVidia GTX790 and a tiny video card I use for troubleshooting stuff.

 

[Chewy509]

FYI, The GA-X79-UP4 requires BIOS revision F4 for either CPU to work... (Source: http://www.gigabyte.com/support-downloads/cpu-support-popup.aspx?pid=4288 )
The ASUS P9X79 Pro requires BIOS revision 0802 for the i7-3820 and BIOS revision 4210 for the i7-4960X. (Source: http://www.asus.com/Motherboards/P9X79_PRO/#support_CPU )

Have you tried just installing just the CPU and nothing else on the boards to ensure that you get a No-Memory or No-Graphics error during POST? (The best way to see if the board is actually working).

 

[LunarMist]

So the 4960x is killing you broads?/

 

[ddrueding]

FYI, The GA-X79-UP4 requires BIOS revision F4 for either CPU to work... (Source: http://www.gigabyte.com/support-downloads/cpu-support-popup.aspx?pid=4288 )
The ASUS P9X79 Pro requires BIOS revision 0802 for the i7-3820 and BIOS revision 4210 for the i7-4960X. (Source: http://www.asus.com/Motherboards/P9X79_PRO/#support_CPU )

Have you tried just installing just the CPU and nothing else on the boards to ensure that you get a No-Memory or No-Graphics error during POST? (The best way to see if the board is actually working).

I've tried it with both the RAM and GPU in and out. From what I've been able to tell, F4 was the earliest shipping BIOS for the Gigabyte, and the ASUS shipped last Friday so should have a fairly recent BIOS. Of course I haven't been able to confirm either.

 

[Chewy509]

Maybe you've been really unlucky getting two dead boards? Next thing would be contact Gigabyte and Asus support and see what they have to say? (Have you confirmed the PSU is fine in another machine)?

 

[ddrueding]

Not yet. I also haven't tested the 3820 in the ASUS. Have to call it quits for now though, more tomorrow.

 

[Bozo]

I am having some of the same problems. I ended up buying a Celeron G450 so I could get the boards to boot up.

 

[CougTek]

If you ever are able to boot an LGA2011 motherboard using a Celeron G450, please tell us.

Ddrueding, are you sure you haven't bent one of the LGA pins?

 

[ddrueding]

I've looked a couple times, but the odds of bending a pin on both boards?

 

[Stereodude]

I'd put your power supply in the suspect category myself.

 

[ddrueding]

Swapping the PSU with a NIB Corsair HX750 didn't do it, but I did get it to boot with the 3820 in the P9X79 Pro and have confirmed.....BIOS rev 4005 :rant:

Grr. So, what is the most brain-dead, oldest 2011 CPU that is guaranteed to work in that fracking Gigabyte board regardless of revision?

 

[Bozo]

Found this list on Gigabyte site. Hope it helps

 

[Bozo]

If you ever are able to boot an LGA2011 motherboard using a Celeron G450, please tell us.

Ddrueding, are you sure you haven't bent one of the LGA pins?

I was referring to a couple of other motherboards in a different thread. Sorry.

 

[ddrueding]

Found this list on Gigabyte site. Hope it helps

Thanks Bozo, I've seen that list. The only things on it say "F4", and it includes the 4960X and the 3820. This makes me think that the board I have is running some BIOS older than F4, and it has no listing of which CPUs were supported on F3, F2, or F1.

Looks like I should take apart my home system and pull the 3970X out of it? According to Wikipedia, that was one of the first consumer 2011 CPUs?

 

[Stereodude]

Have you gotten this all sorted out yet?

 

[ddrueding]

Nope. Got the 4960X running in the ASUS for now. Planning on flashing the BIOS on the board I use at home (just noticed it is another ASUS P9X79) and swap it with the Gigabyte board at work. Then I will have three bootable systems for my 3 CPUs. Gah.

 

[Bozo]

Have you removed power and removed the battery for a while in the problem boards. The BIOS could be a little wonky from any testing the factory might have done.

 

[ddrueding]

I'm fairly certain it is just a BIOS revision issue. Once I get the original CPU freed up (3970X) I'll see if I can get the Gigabyte board to boot.

 

[joetazz]

I'm fairly certain it is just a BIOS revision issue. Once I get the original CPU freed up (3970X) I'll see if I can get the Gigabyte board to boot.

Hi,

Sorry to dig up an old thread, I am having the same issue with the 4960x and the Gigabyte X79-UP4. Did you get anywhere in the end? I'm thinking it's something to do with BIOS version, however I don't have another sky-2011 processor to put in and update the BIOS.

Would be great to hear how you resolved this.

 

[ddrueding]

Putting in the oldest 2011 CPU I could find and updating the BIOS was the only way. Note that I said oldest and not slowest, I had a 3970X in a machine that I was able to get everything handled with.

Sorry I don't have an easier or less expensive option for you.

 

[joetazz]

Putting in the oldest 2011 CPU I could find and updating the BIOS was the only way. Note that I said oldest and not slowest, I had a 3970X in a machine that I was able to get everything handled with.

Sorry I don't have an easier or less expensive option for you.

Hi,

Thanks for the reply, at least now I know what I need to do. Have had a massive headache trying to work out why the board won't post, even sent one back as faulty. Will have to see if I can borrow a 2011 CPU from somewhere.

Thanks again for the help.

 

[joetazz]

Hi,

Thanks for the reply, glad I know why it isn't posting now. Think I will have to send it off for a bios flash

 

[Tannin]

It is a massive, massive stuff-up by Intel, Dave. Intel have introduced new generations of their existing CPUs and are shipping the parts in volume but they FAILED MISERABLY to either

(a) make the new parts compatible with the old, or
(b) co-ordinate with the motherboard manufacturers to ensure that the worldwide stock of mainboards was compatible with the new part.

Huge stuff-up. Shockingly bad performance by Intel.

 

[Tannin]

Here is my epic. Note that this is with Socket 1150 Apparently Intel have screwed up everything they make, not just an isolated single part.

Bought 2 MSI motherboards & 2 CPUs, 1 x i5, 1 x i3.

1: Built i5 system. Started installing software. Error: system will not warm boot under any circumstances, otherwise works fine.
2: Swap RAM. Fail.
3: Swap PSU. Fail.
4: Remove all peripherals and repeat (2) and (3) several times.
5: Conclude that the board is faulty, remove motherboard, fit CPU to the other MSI board, reinstall in system. FAIL! Same fault.
6: Test Board #2 with the i3 CPU. Fail.
7: Swap PSU again. Swap RAM again. Test Board #1 with the i3. Fail.

8: I can't buy parts at this time of night, so go home, having completely wasted four hours. In the morning, first thing, go to a competitor and buy a motherboard at retail hoping desperately to still have the new system ready when my customer arrives after her 80km drive. Just in case, I test it on the bench with the i3 (not the suspect i5), and with completely different RAM and PSU. The new board, a Gigabyte, won't even POST, just continuously cycles power.
9: Go to a different competitor and buy a 4th motherboard, another Gigabute but a different, more expensive model. Test. Fail.
10: Nervous breakdown. Grovel to customer, who is now waiting for her system, which is not even started.
11: Go back to Competitor #2 and buy motherboard #5. Also buy an i5 CPU and ask him to plug it in and see it running before I take it away. This one works fine.
12: Spend rest of day building the system, installing software, transferring data, and grovelling to customer.
13: Spend rest of week afraid to plug anything in in case it blows something else up. Surely it can't be two fauilty CPUs? That seems absurdly unlikley. It's never the CPU. But it was.

It turns out that the new generation Intel CPUs have broken everything. They physically fit into the same Socket 1150, but they are not compatible with it. What you have to do is (somehow!) get hold of an old-model Socket 1150 CPU and use that to flash the BIOS of every motherboard you own. My Motherboard #5 just happened to work because it just happened to be a CPU from old stock. My two new CPUs wouldn't have worked in it, and neither would any other new Intel CPU. We didn't know any of this at the time though.

It's an absolute disgrace, and if someone wants to start a class action to sue the arse off Intel for the vast amounts of lost time and lost money this has cost everyone, I'll be the first one to sign up on it. They should be forced to pay damages on this one. Gross negligence. No excuses.

 

[Howell]

Is Intel responsible or all the mainboard manufacturers who allow their distibutors and retailers to sell boards and chips that don't work together?
If the bios change was frivolous then the MB makers should push back on Intel. This should be easier now that Intel doesn't make boards anymore.
Intel is big but let's get the anger pointed in the right place.

 

[Chewy509]

IMHO it's the BIOS/UEFI developers that are at fault...

The basic problem is that they don't have a safe fallback mode if an unknown CPU is installed, but instead rely on discrete lists of known CPUs and required configurations. If the BIOS on startup doesn't recognise the CPU, then they essentially triple-fault, hence the start up problem you see.

** The thing is the fallback mode could disable all ACPI functions, all power saving modes,, etc and rely on essentially the original IBM PC AT specification and give a big fat warning to the user that they need to update the BIOS, rather than rebooting with no reason. Even a BIOS number code (either on the LEDs that some motherboards have, or as beep codes) would be sufficient to tell the system builder what is going on. And since mist modern motherboards support BIOS updates direct from an attached USB stick (they don't need an OS), the process should be fairly simple.

 

[Tannin]

Yes it is Intel's fault. Unquestionably. When you introduce a new part, the onus is on you to make sure that it will work according to spec. If your new part doesn't work with the existing stock of industry-standard components, it is your responsibility to make sure that this is well-publicised and that there is no disruption to normal operation. Typically you do this by working with other manufacturers. Introducing a new part with stealth incompatibilities is not on. Notice that the problem applies to lots of different motherboards from many different manufacturers. It is ludicrous to suggest that they suddenly all screwed up at the exact same time. No, Intel screwed up, and Intel should pay for the damage.

The long-established industry standard expectation is that, given the multiplicity of CPU socket specs, you need to take care to match the sockets up - e.g., you can't use a Socket 1155 CPU with a Socket 1150 mainboard. That is a reasonable expectation, and that convention has pretty much always been honoured by both Intel and AMD since the end of 486 days. The only exceptions are brand-new, high-end, high-spec CPUs which push the envelope out. Anyone who has been in the industry for any length of time knows that bleeding-edge parts can require special precautions and can take a while to bed in: that's OK. What is NOT OK is this sudden stealth replacement of mainstream, everyday volume-build CPUs like i5s and i3s with new, non-compatible parts with no performance advantage or customer benefit. Check the specs for the mainstream processor range: there is no benefit, just the side-effect of severe disruption to normal business. I want to see a class action on this one. Intel, it's just not good enough.

 

[ddrueding]

Are they different part numbers?

 

[Chewy509]

@Tannin, the same would also apply to AMD as well... considering their sockets tend to have a longer life span in availability, but you pretty much always require a BIOS update if you wish to use a new CPU in an older motherboard. How is that any different in the current Intel world since socket 775 was introduced?

The issue is a software one on the mainboard since a BIOS update corrects the issue... It's a simple case that the BIOS developers don't allow for unknown CPU to be inserted and providing a fallback mode if one is. I would lay this blame on the whole chain (Intel, motherboard manufacturers, BIOS developers), not just at one entity.

PS. Not having a go at you, just trying to give a wider field of view to the problem.

 

[LunarMist]

Do the CPU/mainboard manufacturers care very much about home and small business builders?
I assume that the Apples, Dells, and Lendovos, etc. of the world have some ways around the issue.

 

[Chewy509]

Do the CPU/mainboard manufacturers care very much about home and small business builders?

Seeing how they typically treat home and SMB OEMs, I don't think they care at all.

 

[Tannin]

@Tannin, the same would also apply to AMD as well... considering their sockets tend to have a longer life span in availability, but you pretty much always require a BIOS update if you wish to use a new CPU in an older motherboard. How is that any different in the current Intel world since socket 775 was introduced?

Cheers Chewy. These are not, repeat not "older motherboards". We are talking about brand new still-in-box motherboards, current models, two different manufacturers; three different models, five different boards, every one of them bought on or after the same day the as the CPUs. From David's epic, we learn that this was not an isolated screwup: it was worldwide, and broke systems using the cream of the world's major motherboard brands, including Gigabyte, ASUS, MSI, and doubtless many others. Further, these are not bleeding edge CPUs that push the performance envelope: these are the humdrum everyday i3 and i5 parts that businesses across the world depend on for boring but vital routine tasks like accounting and process control and running cash registers. There is no excuse for breaking these things.

Short answer: Intel screwed the pooch. Big-time.

PS: Every AMD CPU I can remember having BIOS recognition issues (a) would still boot, albeit with CPU unknown errors, and (b) was an actual new part: a new speed grade or a lower voltage. (Maybe there were others I have forgotten. If so, they were few, and they were high-spec specialist ones.) Compare with these broken Intel parts which are exact functional equivalents to the old, working parts offering zero advantage to the customer.

Last point: if this was the motherboard makers' fault (which it isn't), it would be the most amazing of amazing coincidences to see every single major motherboard maker suddenly have exactly the same fault at the same time after years of trouble-free operation. There is only one common factor here: Intel's gross incompetence. They should be sued for it.

 

[CougTek]

It is stuff like this that had me quit the DIY segment and move to server-grade equipment some two years ago. There are still annoyances, but not of this level and there generally a much better level of support since the money involved is an order of magnitude higher, if not more.

I've suffered enough stories like that for a lifetime.

 

[Mercutio]

My standard practice is just to keep one of whatever the cheapest and oldest socket-compatible thing is on hand for BIOS flashes (a Sandy Bridge Pentium G for 1155, say). I went through this a couple years ago with brand new boards lacking support for Ivy Bridge CPUs. In fact, I just ran in to one of those yesterday with an Asus H61 board that I purchased two days prior and which still had the 603 BIOS on it (dated September 2011).

I also hate dealing with UEFI boot issues. About a third of the time I find that my system images don't work and then I have to go spend 15 unproductive minutes screwing around with boot options until I get a combination that will let a system start. And that's putting aside the name brand machines that don't allow for such modification.