ES Xeon Discussion

_haru

Member
Mar 24, 2018
47
6
8
17
Perth, Australia
I noticed that with the x299 FTW K, there was a light near the CPU labeled FIVR and that was on. I looked up the manual and that light means "Indicates CPU integrated voltage regulation failure." They're probably dead chips.
That most likely means it's some sort of unmarked, nonfunctional mech sample. similar to that northwood p4-m marketing kit that was given out, except that one actually worked o_O
 

Klee

Well-Known Member
Jun 2, 2016
1,285
393
83
I looked at the SMT capacitors on the bottom of a E5-2603................so tiny. Looking around the web missing ones can be replaced but its just kind of tough because of the size. I may just give it a go tho. May have to take a shot of whisky to settle me down first. LOL
 

Klee

Well-Known Member
Jun 2, 2016
1,285
393
83
I just received my i9/Xeon W mystery cpu and it has all the capacitors/resistors/smt devices on the back. Also looking at the contacts none show any use at all.
 
Last edited:

vabch22

New Member
Aug 8, 2016
27
1
1
21
That must be one hell of a parts bin.
Haha not really. I collect parts when I find a good deal. Half the time I have part of a build, but always missing something. I just like to tinker. Not sure how I am going to be able to afford some DDR4 ECC Ram though for this build. Prices went crazy the last couple of years. Looks like they are finally coming back down though. I have some DDR3, but doesn't do me any good.

Haha think I have a couple of e5-2620,30,40 v3/v4 cpu’s in my parts bin... not nearly as impressive sure... suppose I should sell them but I didn’t really think they would be much in demand.
I'd also be interested if you offload anything. :) Save me from waiting on shipping from China.
 

RageBone

Active Member
Jul 11, 2017
298
78
28
the two blank CPUs i have, don't post on a X299 Board.
They aren't shorted so voltage is getting supplied by the VRMs.
Over a longer amount of time, the CPU gets hot, i assume that's because of the RAM Controllers getting voltage and such.
 

foureight84

Member
Jun 26, 2018
61
11
8
the two blank CPUs i have, don't post on a X299 Board.
They aren't shorted so voltage is getting supplied by the VRMs.
Over a longer amount of time, the CPU gets hot, i assume that's because of the RAM Controllers getting voltage and such.
Does it trigger any CPU error light on the MB?
 

RageBone

Active Member
Jul 11, 2017
298
78
28
negative.
postcode 00.
Since i'm not sure that my MSI is good, it does a 00 postcode without cpu too, i checked it in a different, confirmed working msi board without postcode display, and it didn't even detect that the cpu is quad-channel.

So it seems to be not compatible with X299.

A different question would be how Xeon Ws actually behave when you put them on X299! ?
Anyone willing and able to try?
 

_haru

Member
Mar 24, 2018
47
6
8
17
Perth, Australia
negative.
postcode 00.
Since i'm not sure that my MSI is good, it does a 00 postcode without cpu too, i checked it in a different, confirmed working msi board without postcode display, and it didn't even detect that the cpu is quad-channel.

So it seems to be not compatible with X299.

A different question would be how Xeon Ws actually behave when you put them on X299! ?
Anyone willing and able to try?
00 postcode = no cpu = probably no functional die in the blank chip.
 

RageBone

Active Member
Jul 11, 2017
298
78
28
00 does not mean = no cpu.

00 is the very first step in the POST process.
It staying there means that there are no other progressing steps being made afterwards.

Since i don't know how a Xeon W would behave on X299, it's hard to say if the platform detects the cpu but refuses to move on because it is a Xeon W, or if it is fully refusing to talk with the system, hence being dead or Disabled.


A short side note, Intel disables CPUs during shipping. Or at least there is a patent for that since like 2010, in combination with some crazy RFID stuff.
 
Last edited:

_haru

Member
Mar 24, 2018
47
6
8
17
Perth, Australia
00 does not mean = no cpu.

00 is the very first step in the POST process.
It staying there means that there are no other progressing steps being made afterwards.

Since i don't know how a Xeon W would behave on X299, it's hard to say if the platform detects the cpu but refuses to move on because it is a Xeon W, or if it is fully refusing to talk with the system, hence being dead or Disabled.


A short side note, Intel disables CPU during shipping. Or at least there is a patent for that since like 2010, in combination with some crazy RFID stuff.
interesting - had no idea they can disable the cpu o_O
 

foureight84

Member
Jun 26, 2018
61
11
8
interesting - had no idea they can disable the cpu o_O
The RFID on this chip is also known as PiRom and please correct me if I am wrong. It's used for inventory for data centers and available on Enterprise hardware like the Xeon. I was not able to scan it with any RFID reader. But I might not have compatible readers. Other than deliding, that's the only way to tell as the pirom is not enabled for desktop CPUs even though the chip is present.
 

_haru

Member
Mar 24, 2018
47
6
8
17
Perth, Australia
The RFID on this chip is also known as PiRom and please correct me if I am wrong. It's used for inventory for data centers and available on Enterprise hardware like the Xeon. I was not able to scan it with any RFID reader. But I might not have compatible readers. Other than deliding, that's the only way to tell as the pirom is not enabled for desktop CPUs even though the chip is present.
hmm, makes sense.

on another note, i did a thing :
upload_2018-11-7_12-47-36.png
bit out of the scope of this thread, but it's a xeon (b1 stepping 5070, probably QJYK) and it's an ES ;)
 

RageBone

Active Member
Jul 11, 2017
298
78
28
The RFID on this chip is also known as PiRom .....
Is there any other source other then der8auer?
And what does PiRom actually stand for?
How should deliding help?

since i already pointed at patents, here a few relevant links.
US10009339B2 - System, apparatus and method for securely protecting a processor in transit - Google Patents
US7825776B2 - Device configuration with RFID - Google Patents

US20180096177A1 - Systems, methods, and apparatuses for implementing late fusing of processor features using a non-volatile memory - Google Patents
US20130002398A1 - Apparatus, System, and Method for Providing Attribute Identity Control Associated with a Processor - Google Patents

Of cause that doesn't mean that it is implemented that way, though since those are blanks, that screams either Dummy, or OEM.
To the second one, i'm sure that THAT's actually implemented, and well, could also be that those simply weren't flashed to make them dummys.
The third and forth contain a bit more info on how to program them and how they are also used.
But i guess that's all off topic so that will be it from my side.
 

jpmomo

Active Member
Aug 12, 2018
112
37
28
I haven't received my mystery chips yet but should this friday. There is still the remote possibility that if these chips are xeon w and some form of es, it might be worth a try to test on a supermicro mb. We have seen similar behavior (not posting or not completely posting) with other es chips on non-supermicro mb but then the same chips were able to post on specific supermicro mb. Unfortunately I am having a hard time trying to find a c422 supermicro mb to test with. something like the Supermicro X11SRM-F might be worth a try.
In regards to delidding, there are simple tools to do this task. It removes the IHS from the PCB. You would then replace the TIM (thermal paste) with something like liquid metal for better heat transfer. The complicating issue here is that most xeon cpus use solder instead of thermal paste. If we did delid this cpu and found that it had thermal paste, it might suggest that this was an i9 chip and not a xeon w series. We could also see the details of the actual chip itself. I am not sure what the i9 vs the xeon w series chip (die) looks like but I am sure we could track that down somewhere. and the mystery continues...........
 

jpmomo

Active Member
Aug 12, 2018
112
37
28
I haven't received my mystery chips yet but should this friday. There is still the remote possibility that if these chips are xeon w and some form of es, it might be worth a try to test on a supermicro mb. We have seen similar behavior (not posting or not completely posting) with other es chips on non-supermicro mb but then the same chips were able to post on specific supermicro mb. Unfortunately I am having a hard time trying to find a c422 supermicro mb to test with. something like the Supermicro X11SRM-F might be worth a try.
In regards to delidding, there are simple tools to do this task. It removes the IHS from the PCB. You would then replace the TIM (thermal paste) with something like liquid metal for better heat transfer. The complicating issue here is that most xeon cpus use solder instead of thermal paste. If we did delid this cpu and found that it had thermal paste, it might suggest that this was an i9 chip and not a xeon w series. We could also see the details of the actual chip itself. I am not sure what the i9 vs the xeon w series chip (die) looks like but I am sure we could track that down somewhere. and the mystery continues...........
I received the cpus a bit earlier than expected. I took a good look at them using a magnifying glass and made notes of a couple of things.
1. if you shine a light on the bottom of the chip, you can see a blue haze over part of the gold contacts. specifically it looks like something spilled on it then wiped off so that the gold contacts would not be blocked. I tried to clean it with so isopropyl spray but it did not seem to remove it. For those that have these chips, can you shine a light on the bottom and tilt it slightly to see if you notice this blue film?
2. the small chip on the top side, upper right corner (when gold triangle is on the bottom left) has a marking of 642
X8

It also has another marking lower right side of J20146 01 A99448
I still can't tell whether this is a skylake x or xeon w. I did get 2 so I may wind up delidding and looking under the cover!
I may be able to get an asus (both x299 and c422) to test but I would prefer to try on a supermicro if possible. the supermicro x11sra should be compatible with a real xeon w chip.
 

nero243

Active Member
Oct 28, 2018
108
76
28
So i finally bought the problem plagued E5-2696v3 QEY8 on ebay, declared as defective not booting on a X99.

First i noticed i don't have ddr4 rdimms lying around (sadly they were ddr3 rdimms), but all the x99 mainboard manufacturers say the e5 xeons should be working with udimms. Did anyone had problems with unbuffered memory and ES xeons? I tried it with 2 different sets of DDR4 Udimms, no success.

So i started trying around with a whole bunch of different microcodes (and no microcode), i always ended up with the Q-code 55 (no memory installed). Without any actual memory installed i end up with Q-Code 53 (memory initialization failed). So yeah...

I already wanted to give up here but luckily i've read here on the forum, that the QEY8 should work on a Asus rampage V with an older bios. So i gave the rampage bios a shot and just flashed it on my Asus X99 Deluxe. I mean what could go wrong?! Now i get halfway through the initialization Q-Codes and get stuck on code b0 (Runtime Set Virtual Address MAP Begin). I hoped that i get halfarsed to a post screen with the wrong bios... But it seems to work a little bit, without RAM i end back up at code 53 and with a newer rampage Bios Version and RAM installed i get back to code 55. But i have no idea what causes the initialization issues between the older and newer bios versions. If i knew i could try to switch out the responsible Bios modules from the real X99 Deluxe bios.
 
Last edited:
  • Like
Reactions: scraper

RageBone

Active Member
Jul 11, 2017
298
78
28
@nero243 in my experience e5 xeons can boot any ram that is supported by the board.
Had my 1650v3 running gskill trident z a while back, though that wasn't an ES, but i don't think Udimms are the problem.

It is nice to hear that you at least have one that posts, the most recent V3-ES i got to play with was shorted out, Postcode 00 and no Voltage from the VRMs.
To your postcodes, i currently assume that 5X so early in the Post process, before "bX" is something around "no ram detected" and bX should be Ram (init | training). 5X after the bX codes is something i have no clue about.

The last time i tried to boot a V4 Es on an Asus X99 board, it went through into the bios on the first try but after applying defaults, it was stuck somewhere. Probably Microcode related.

Maybe you want to take a look at the Asus Serverboards, Z10Pa /Pe, those, i think, have ES support. You might be able to grab the right modules from there bios.

Depending on where you are located, i could test it for you, or we could exchange or lend hardware. In my Case, EU, Germany.

@joe paolicelli you are right, if they have the crappy paste, it's most likely a dummy i9.
Though since deliding soldered cpus is a bit different then the i9s, i'm pretty sure you'd kill your sample in case that it is a soldered xeon W.

I prefer the Supermico c422.
I will check my two blanks for the shade and markings when i get back home this evening.
 

RageBone

Active Member
Jul 11, 2017
298
78
28
@joe paolicelli took my time, reread your post, and i can tell you that the blue shade is on both of mine, and i think i have something like that on older broadwell and haswell chips. So to me it seems completely normal!

The piroms also read 642 x8 on both of them, just like you said.
The Substrate marking is smiliar: J20146 01 A28029
and J20146 01 A28067