ES Xeon Discussion

Frank173

Member
Feb 14, 2018
75
9
8
I am not sure what it is exactly that you are asking...

Why did you not/ don't you simply ask the seller. First off, whether the CPUs are Engineering Samples or not should be apparent from the price, charged as the difference in pricing between ES and non ES CPUs is generally huge, hence the interest in ES CPUs in this particular community/thread.

Secondly, did you ask the seller to run benchmarks and share screen shots with you or did you engage in a live Skype session of any sort? Were you satisfied when you saw the performance figures?

Keep in mind if those are ES CPUs then most always the performance will generally not equal that of retail CPUs. Most often the turbo frequencies will be lower for ES CPUs though the base frequency should match.
Stability wise, I have run a machine with dual ES CPUs for over a year non-stop now and the machine seemed perfectly solid. I run multiple virtual machines on that built and did not run into any issues.

What is it that you are concerned about? Stability? Then to be perfectly honest with you the answer is, it depends on your particular CPUs. Most that I came across turned out to be stable, but there are always a few lemons among the pack. If you are concerned about whether you received what you bargained for then you should take that up with the seller. Just my 2 cents.




Thanks.

It looks to me like they are likely engineering samples. What do you think of the stepping and QDF? I don’t have much experience interpreting that. The seller did not experience any issues when using the workstation, however I am a bit sceptical if it will stand up to heavy use over time.

Stepping: L0

QDF: QEYJ (ES2)

What do you think?
 

KNO

New Member
Feb 20, 2018
3
0
1
39
The reason for asking is that I purchased a second hand computer that was supposed to have 2 x 2690V3 CPU’s. It turns out that they are 2690V3’s on the marking but in reality they are engineering samples that run at 2.4Ghz (as opposed to 2.6Ghz). The person that sold the machine to me says he is not that computer savvy and was not aware that the CPUs are not what they are supposed to be. He also got the computer second hand and was probably duped by another seller without even knowing.

So now I am trying to assess the damage. I paid a fair price based on what the CPUs were supposed to be. I think I overpaid for engineering samples and would probably not have bought it had I known. I was not aware of the possibility that the CPUs could be engineering samples and it took med about three days to find out (after trying other thing such as bios update of the MB, searching for info, etc.). Engeering samples are not common were I come from.

I did run some test with the software I am using for simulations and with HWiNFO found that both CPU have 12 active cores. All can run @ 2.7GHz simultaneously when required. I know the real 2690V3 can do 3.5Ghz with turboboost (but that is not on all cores simultaneously I think).

I did not experience any crashes or instability and the simulations were of course much faster than my old 8 core machine. I don’t think the damage is too bad, but I still think I overpaid. I am just trying to determine what I got.
 

Frank173

Member
Feb 14, 2018
75
9
8
Well you probably got this: 2x Intel Xeon E5-2690 v3 ES, QEYJ(ES2), exactly the same CPUs I have and asked a question about on the previous page as I am intending to sell them. They run tops 2.7GHz and mine worked perfectly the entire time. Maybe let this be a lesson to always ask the seller of used motherboards/cpus to run HWiNFO, CPU-Z and show you and agree to a full refund if the purchased hardware on your end does not match. Had you done that then you would have immediately seen that the CPUs are Engineering Samples. Time to move on, I advise, unless you hopelessly overpaid and can agree with the seller on a return and full refund or a partial refund.

I wish you good luck (there is always a bit of luck involved with ES CPUs) though once they run as intended and unless you play some OC wagers in your bios then there generally should be no reason that they die any earlier than a regular chip. I recommend you run couple more stress tests, spin up a virtual machine in your OS of choice and see how they perform. If all goes according to plan I would get over it and use as intended.

The reason for asking is that I purchased a second hand computer that was supposed to have 2 x 2690V3 CPU’s. It turns out that they are 2690V3’s on the marking but in reality they are engineering samples that run at 2.4Ghz (as opposed to 2.6Ghz). The person that sold the machine to me says he is not that computer savvy and was not aware that the CPUs are not what they are supposed to be. He also got the computer second hand and was probably duped by another seller without even knowing.

So now I am trying to assess the damage. I paid a fair price based on what the CPUs were supposed to be. I think I overpaid for engineering samples and would probably not have bought it had I known. I was not aware of the possibility that the CPUs could be engineering samples and it took med about three days to find out (after trying other thing such as bios update of the MB, searching for info, etc.). Engeering samples are not common were I come from.

I did run some test with the software I am using for simulations and with HWiNFO found that both CPU have 12 active cores. All can run @ 2.7GHz simultaneously when required. I know the real 2690V3 can do 3.5Ghz with turboboost (but that is not on all cores simultaneously I think).

I did not experience any crashes or instability and the simulations were of course much faster than my old 8 core machine. I don’t think the damage is too bad, but I still think I overpaid. I am just trying to determine what I got.
 

gentlegy

New Member
Feb 20, 2018
7
0
1
37
Hello guys,

I'm new for buying an ES CPU. This thread helps me a lot to clarify stuff. With regard to CPU, I am particularly interested in Xeon E5 2666 v3 ES for the balance between high frequency and # cores. This CPU is not shown ark of Intel, and it is deployed in Amazon AWS. My main usage of this CPU is to support a workstation for deep learning and my daily usage as well. I checked a lot of information through web and found out there are actually two late ES version 2.9G with:
1) stepping 2, revision M0;
2) stepping 2, revision R2.
The sellers for both all claim these CPUs are QS. I also saw some CPU-Z images for the "retailed" versions have supported both claims. Now, I am confused about which one is better or which one is the "real" QS actually? I know the retailed CPU could turbo boost to 3.5G, but I don't which one would has this capacity more likely.
Can someone have experience with these CPUs give me some guidance? Or other guys who could give me more choices about the Xeon E5 v3 or v4 CPUs based on your usage and my requirement? Besides using as deep learning machine, I am also considering to have two of them to build a web server in future. Any suggestion or idea would really appreciate it.

Best,
Yang
 

Patriot

Moderator
Apr 18, 2011
1,311
695
113
Hello guys,

I'm new for buying an ES CPU. This thread helps me a lot to clarify stuff. With regard to CPU, I am particularly interested in Xeon E5 2666 v3 ES for the balance between high frequency and # cores. This CPU is not shown ark of Intel, and it is deployed in Amazon AWS. My main usage of this CPU is to support a workstation for deep learning and my daily usage as well. I checked a lot of information through web and found out there are actually two late ES version 2.9G with:
1) stepping 2, revision M0;
2) stepping 2, revision R2.
The sellers for both all claim these CPUs are QS. I also saw some CPU-Z images for the "retailed" versions have supported both claims. Now, I am confused about which one is better or which one is the "real" QS actually? I know the retailed CPU could turbo boost to 3.5G, but I don't which one would has this capacity more likely.
Can someone have experience with these CPUs give me some guidance? Or other guys who could give me more choices about the Xeon E5 v3 or v4 CPUs based on your usage and my requirement? Besides using as deep learning machine, I am also considering to have two of them to build a web server in future. Any suggestion or idea would really appreciate it.

Best,
Yang
Those are different dies... different core count.
What is the actual qspec?
 

b3nno

New Member
Jul 17, 2017
12
2
3
31
Has anyone with ES CPU's dared to update their BIOS with Spectre fix?
Does anyone have an idea if this would work fine with an ES cpu?
I've got a Supermicro X10SRL-F and an e5-2650 v3 ES QEYN, and wondering if I should update with the latest BIOS containing new microcode from Intel.
 

RolloZ170

Active Member
Apr 24, 2016
299
109
43
53
Has anyone with ES CPU's dared to update their BIOS with Spectre fix?
Does anyone have an idea if this would work fine with an ES cpu?
I've got a Supermicro X10SRL-F and an e5-2650 v3 ES QEYN, and wondering if I should update with the latest BIOS containing new microcode from Intel.
the QEYN is an early stepping(CPUID 306F1) and its Microcode is not supported by intel.
ES need same CPUID than production unit to get updated Microcode.
 

RolloZ170

Active Member
Apr 24, 2016
299
109
43
53

Nizmo

Member
Jan 24, 2018
101
17
18
34
Has anyone with ES CPU's dared to update their BIOS with Spectre fix?
Does anyone have an idea if this would work fine with an ES cpu?
I've got a Supermicro X10SRL-F and an e5-2650 v3 ES QEYN, and wondering if I should update with the latest BIOS containing new microcode from Intel.
Updated with NO Problems whatsoever.

ES 2695 v4
ES 2699 v3
ES 2699 v4
 

dp76

New Member
Aug 18, 2016
1
0
1
44
Hi all, first of all, thanks for this fantastic thread: ~5 yrs ago, when i start to build the my first QA88 ES Xeon homelab, i compile info about it from the myriad different sources.

Now i'm start to build new system on QHV8 with X10SRH mobo. Luckily, the bios was 2.0 and it starts out-of-box, but some strange problems persists with memtest86 after some (4+ hrs run) with reporting, like
//
2018-04-05 06:39:28 - [MEM ERROR - ECC] Test: 7, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: yes, Syndrome: N/A, Channel/Slot: 0/0
2018-04-05 06:39:52 - [MEM ERROR - ECC] Test: 7, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: yes, Syndrome: N/A, Channel/Slot: 0/0
2018-04-05 11:38:50 - [MEM ERROR - ECC] Test: 11, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: yes, Syndrome: N/A, Channel/Slot: 0/0
2018-04-05 14:07:57 - [MEM ERROR - ECC] Test: 8, (Col,Row,Rank,Bank): (N/A,N/A,N/A,N/A), ECC Corrected: yes, Syndrome: N/A, Channel/Slot: 0/0
Now i disable the CPU PM features in BIOS (from "Energy efficient" to "Disable"), and suppose it can be useful.
//

Q1: If somebody have ideas about this type of ECC errors - w/o any reporting the row, col, etc - it can be good. I suppose it's originate in MSR(?) or in some other place in CPU - but not originate to memory.
I seen the some reports about ESXi host OS under v4 CPU configurations is unstable when enabled C6/C7 states in BIOS - i suppose it can be same/similar problem.

Q2: But my original question is around the BIOS update - it's the make sense for my case? I did't able to find earliest changelog in SM, except the official actual one.

Currently, i'm on 2.0, on half year before the v4 was announced:
---
2018-04-05 02:00:33 - SMBIOS BIOS INFO Vendor: American Megatrends Inc., Version: 2.0, Release Date: 12/17/2015
---

but two next updates which i find it - X10SRH6_920 (2.0?), X10SRH7_414 (2.0a), and the actual one X10SRH8_208 (2.0b) - all of them contain 0x14h rev of 406F0 microcode, dated as 02/07/2015. Looks it's the latest one.

Is the any successful runs here of X10SR* mobo with A0 CPU steppings under newest than 2.0 of bios release from Supermicro?
 

minimini

Member
Sep 9, 2016
62
13
8
49
Different point of view from software perspective -

Patching firmwares, bioses, and OS has consequences, imagine that every bank, air carrier, military simply patches their OSes as Microsoft is pleased - it would be disaster. Remember DEP rendered applications inoperable - totally not acceptable, and some manufacturers clearly specify compatibility with OS with exact SP level, no more no less.

If your system is e.g. db server behind proxies, firewalls, ssls, and content filters, etc then "meltdown" has nothing to do with this kind of server, I wouldn't even think of patching anything (ES or not ES CPU). This kind of machines don't even run antivirus.

If you are online gamer or trying "free" utils then all antiviruses and patches might not be enough to protect you.

Depends what you do with your workstation/server.
 

wiretap

Active Member
Jul 14, 2015
128
85
28
Michigan
So I ordered two E5-2667v3 (QEYA) processors from Ebay, and I received two E5-2667v3 (SR203). Are they actual ES processors, or OEM/retail? Possible seller mistake? My motherboard should arrive next week for testing.

 

wiretap

Active Member
Jul 14, 2015
128
85
28
Michigan
Aren't these soldered IHS though, or are ES processors not normally soldered so that the lid can easily be removed?
 

wiretap

Active Member
Jul 14, 2015
128
85
28
Michigan
Nevermind, you're correct, they are re-lidded. Got antsy and checked them in my X99 board. They are indeed the ES variant at 2.9GHz.
 
Dec 30, 2016
88
14
8
41
Has anyone with ES CPU's dared to update their BIOS with Spectre fix?
Does anyone have an idea if this would work fine with an ES cpu?
I've got a Supermicro X10SRL-F and an e5-2650 v3 ES QEYN, and wondering if I should update with the latest BIOS containing new microcode from Intel.
Recently updated my 2x X10SRL-F with ES CPUs and it caused the same issue on both but I don't think it had anything to do with the CPU.

Board would boot to "DXE - Intel RAS Initialization - 79" and stop. Clearing CMOS, removing RAM and IO cards, attempting BIOS recovery, nothing worked. The IPMI would still come up but despite the successful BIOS upgrade it still showed 2.0 BIOS.

Finally I set the BMC jumper to Disabled and the board boots again. Even on the new 3.65 the BMC being enabled stops the board from booting.

Supermicro support just shrugs their shoulders and says I can send in the board for $ and they'll try to fix it.

NOT a happy camper.
 

nthu9280

Well-Known Member
Feb 3, 2016
1,588
441
83
San Antonio, TX
If the board is under 3yr warranty, it's one way shipping cost with RMA. If they can't fix the issue, they will replace with another refurbished board. Not sure where you are located but in us the shipping cost is under $20 including insurance.

Sent from my Nexus 6 using Tapatalk